EPA/747/R-97/002
States
imental
:ion Agency
Office of
Pollution Prevention
and Toxics
EPA 747-R-97-002
March 1997
Laboratory Study of
Lead-Cleaning Efficacy
-------
March 1997
EPA 747-R-97-002
Laboratory Study of
Lead-Cleaning Efficacy
Technical Programs Branch
Chemical Management Division
Office of Pollution Prevention and Toxics
Office of Prevention, Pesticides, and Toxic Substances
United States Environmental Protection Agency
Washington DC 20460
-------
Disclaimer
The material in this document has been subject to EPA technical and policy review and approved
for publication as an EPA report. Mention of trade names, products, or services does not convey,
and should not be interpreted as conveying, official EPA approval, endorsement, or recommen-
dation.
-------
CONTRIBUTING ORGANIZATIONS
The study described in this report was funded by the United States Environmental Protection
Agency (EPA). The study was managed by EPA and conducted collaboratively by Westat and
Midwest Research Institute (MRI) under contract to EPA. Each organization's responsibilities
are listed below.
Westat, Inc.
Westat worked with EPA to design the study and, upon completion of the laboratory and
chemical analysis, performed the statistical analysis, wrote most of the report sections, and
compiled and edited the final report.
Midwest Research Institute (MRI)
MRI conducted the laboratory sampling and chemical analysis, wrote the Quality Assurance
Project Plan (QAPjP), wrote some sections of the report, and provided editing comments on the
draft and final reports.
United States Environmental Protection Agency
EPA was responsible for managing the study; providing technical oversight, guidance, and
direction; and overseeing the peer review and finalization of the report. Dr. Khoan T. Dinh was
the EPA Work Assignment Manager for the Westat task and Dr. Ben S. Lim was the Work
Assignment Manager for the MRI task. The EPA Project Officers for the Westat task were, at
different times, Mr. Philip E. Robinson, Mr. Samuel F. Brown, and Mr. John Varhol. The EPA
Project Officer for the MRI task was Ms. Jill Hacker.
-------
Contents
Executive Summary v
1. Introduction 1
1.1 Background 1
1.2 Objectives 1
1.3 Overview of the Report 2
1.4 Peer Review 2
2. Summary of Design, Findings, and Conclusions 5
2.1 Design 5
2.2 Findings and Conclusions 6
3. Overall Quality Assurance 7
4. Study Design 8
4.1 Experimental Design 8
4.2 Cleaner Selection 8
4.3 Substrate and Soil Selection 10
4.4 Cleaner Tests 10
4.5 Randomization of Cleaner Tests 13
4.6 Soil Sample, Rod Rinse, and Cleaning Wipe Tests 13
4.7 Quality Control Tests 14
5. Laboratory Data Collection 15
5.1 Preparation of the Soil and Substrates 15
5.2 Cleaner Tests 16
5.3 Randomization of Coupon Tests 16
5.4 Soil Sample, Rod Rinse, and Cleaning Wipe Tests 17
5.5 Quality Assurance 17
5.5.1 Tests to assess measurement quality 17
5.5.2 Quality of the results 17
5.6 Design Modifications During Implementation 18
6. Chemical Analysis 19
6.1 Analytical Procedures 19
6.2 Preparation QC Samples and Results 20
6.3 Instrument QC Measurements and Results 22
6.4 Corrective Actions 24
-------
Contents
7. Data Analysis Methodology 25
7.1 Calculation of Residual Wipe Lead 25
7.2 Data Processing and Data Issues 27
7.3 Statistical Models 27
7.4 Assessment of Outliers 30
7.5 Quality Assurance for the Statistical Analysis 31
8. Statistical Results 32
8.1 Sequence and Timing of Tests 32
8.2 Analysis of Soil and Rod-Rinse Samples 32
8.2.1 Analysis of soil sample measurements 33
8.2.2 Analysis of rod rinse measurements 37
8.2.3 Analysis of soil-sample minus rod-rinse measurements 37
8.3 Analysis of Residual Wipe Lead 41
8.4 Analysis of Lead-Cleaning Efficacy 44
8.4.1 Analysis of differences among cleaner mixtures 47
8.4.2 Cleaner characteristics 49
8.4.3 Cleaning efficacy as a function of cleaner characteristics 50
8.5 Percentage of Lead Removed Using Wipes Only 56
8.6 Quality Control Results 62
8.6.1 Variability between technicians 62
8.6.2 Measurement bias due to contamination 65
8.6.3 Measurement error components of variance 65
8.7 Discussion of the Statistical Results 71
Glossary of Terms Used in this Study 73
References 76
Appendix A: Test Sequence and Schedule 77
Appendix B: Details of the Laboratory Procedures 83
Appendix C: Listing of the Data 86
n
-------
Tables
Table 1 Number of cleaners selected for the study by surfactant type, pH, and
phosphate content 10
Table 2 ICP-AES instrument analysis QC procedures 23
Table 3 GFAA instrument analysis QC procedures 23
Table 4 Symbols used in statistical equations 26
Table 5 Critical values for the extreme studentized residual for identifying outliers (5
% significance level) 31
Table 6 Analysis of variance results: Soil-sample lead concentration by soil type and
soil batch 36
Table 7 Analysis of variance results: Soil-sample minus rod-rinse lead amount per
volume of soil mixture, by soil type and soil batch 40
Table 8 Quantity of lead on the coupons before cleaning (|o.g Pb per ml soil mixture)
by soil type and soil batch 41
Table 9 Analysis of variance results: Residual wipe lead by soil type and substrate
within cleaner test 43
Table 10 Geometric mean residual wipe lead by substrate and soil type (with 95 %
confidence intervals) 47
Table 11 Analysis of variance results: Cleaning efficacy differences among cleaners 48
Table 12 Analysis of covariance results: Cleaning efficacy by cleaner characteristics 54
Table 13 Results from alternative models 55
Table 14 Regression results: Cleaning efficacy by surface tension 58
Table 15 Analysis of variance results: Cleaning wipe lead by soil type and substrate 60
Table 16 Analysis of variance results: Cleaner tests comparing technicians 63
Table Al Test sequence and schedule 78
Table Cl Soil sample and rod-rinse measurements 87
Table C2 Cleaner test data 88
Table C3 Fraction of lead removed using cleaning wipes 104
Table C4 Cleaner mixture characteristics 105
in
-------
Figures
Figure 1 Processing steps and measurements for a coupon test 12
Figure 2 Percentage recoveries for NIST SRM QC samples, by preparation batch 21
Figure 3 Soil-sample measurements expressed as lead concentration for each soil
batch, separately for each soil type 35
Figure 4 Quantity of lead placed on the coupons per milliliter of soil mixture, by soil
type and soil batch 39
Figure 5 Residual wipe lead measurements (percentage) and predicted geometric
means by cleaner, soil type, and substrate 45
Figure 6 Geometric mean residual wipe lead (percentage) by soil type and substrate 46
Figure 7 Distribution of 35 cleaners across values of pH, phosphate content, and
surface tension 51
Figure 8 Contours of predicted average residual wipe lead versus pH and surface
tension for a phosphate content of 0 and 5 g P/gal, averaged across cleaner
types 53
Figure 9 Geometric mean residual wipe lead (percentage) versus cleaner surface
tension 57
Figure 10 Percentage of lead removed using one or two wipes to clean coupons 59
Figure 11 Percentage of lead removed when using two baby wipes to clean coupons, by
substrate 61
Figure 12 Residual wipe lead measurements comparing technicians, by soil type and
substrate, using cleaner 8 64
Figure 13 Distribution of test wipe, blank wipe, and blank soil residual wipe
measurements 66
Figure 14 Standard deviation of instrument response for the ICP and GFAA QC samples 68
Figure 15 ICP and GFAA instrument response for test wipes and associated
measurement precision 70
Figure Al Sequence of cleaner tests and dates of cleaner tests, wipe preparation, and
analysis 82
IV
-------
Executive Summary
The United States Environmental Protection Agency (EPA) has recommended the use of
trisodium phosphate (TSP) detergent to clean lead-contaminated dust from surfaces, both after
residential lead hazard control work to achieve post-abatement clearance standards and in
general. This recommendation has often been assumed to apply to the general cleaning of lead-
contaminated dust during ongoing exposure reduction activities. Because of the negative impact
of phosphate detergents on the ecology of aquatic ecosystems, questions have arisen as to the
scientific basis of recommending TSP and about the effectiveness of other cleaners. The
objective of this laboratory study was to compare the cleaning efficacy of many commercially
available cleaners that could be used to remove lead-contaminated dust from residential surfaces.
Thirty-four cleaners were tested in this study: 32 commercially available cleaning agents, TSP,
and tap water of average hardness. Most brands were general all-purpose cleaners, hand or
machine dishwashing products, laundry detergents, and bathroom, floor, and glass cleaners,
while some brands were lead specific cleaners. The cleaning agents were selected to represent
the range of commercial cleaning products that would reasonably be available to a consumer.
Most of the cleaning agents were purchased at a full line grocery store. Although high purity
TSP was used in this study, TSP for cleaning walls and lead abatement activities is available at
home/builders supply stores. Two of the cleaners were purchased from a professional janitorial
supply house, principally to have all-purpose cleaners with "high" phosphate content.
The cleaning agents were selected in an attempt to span a wide range of (1) phosphate content,
(2) pH, and (3) active ingredients. It was found that most cleaning agents available to household
consumers are (1) low or zero phosphate content, (2) high pH (basic), and (3) contain various
active ingredients, often more that one surface active chemical. The cleaning agent formulations
are nearly always considered proprietary, and the information on the cleaning agent label varies
in content, particularly for the surface active ingredients that enhance the cleaning performance.
In some cases, the phosphate content varies according to the geographical region in which the
product is to be sold.
The tests were conducted using five types of surfaces selected to represent those commonly
found in residential settings: vinyl tile, latex paint on drywall, enamel paint on birch, lacquer
(Fabulon) on oak, and latex paint on birch. In addition to varying the types of surfaces tested,
two types of leaded soil were used. One soil type contained vegetable oil (oily soil); the other
contained no vegetable oil (dry soil). Each lead-containing soil mixture was mixed in a solvent,
wiped on a test surface in a standardized manner, and allowed to dry before the surfaces were
cleaned.
Each of the 34 cleaning agents was tested on all combinations of surfaces and soil types using the
test procedures. First, the cleaner was mixed with water according to the manufacturer's
instructions and the mixture's surface tension was measured, then the sponge and the cleaner
solution were used to clean lead from the prepared surfaces. Baby wipes were used to sample the
cleaned surfaced in order to measure the lead left behind after the cleaning process with the
cleaner solution and sponge. This test procedure was repeated three times.
-------
Lower surface tension cleaners were associated with better cleaning; however, differences among
cleaning agents were small. Because all tested cleaning agents have lower surface tension than
tap water alone, household cleaning using one of these cleaning agents is likely to remove more
leaded soil and dust than does water alone. Phosphate content was not linked to cleaning
efficacy. In particular, several of the lower phosphate cleaners had overall cleaning efficacy
similar to or better than TSP. Differences in cleaning efficacy also depended on which
laboratory technician performed a test, suggesting that the physical effort put into cleaning may
be more important than the choice of cleaner.
Based on the primary conclusions of this study, EPA recommends that either a general all-
purpose cleaner or a cleaner made specifically for lead should be used for both general cleaning
and for post-intervention cleaning. Household cleaning using one of these cleaning agents is
likely to remove more leaded soil and dust than does water alone. Finally, the study indicates
that the effort put into the cleaning may be more important than the choice of cleaner.
The extent to which these conclusions, based on laboratory investigation, apply to homes in real-
life situations is a matter of judgment. For example, the lead-containing soil material used in this
study was mixed in a solvent, wiped on the test surfaces, and allowed to dry before cleaning.
This application method might have resulted in soil lead that is more closely bound to a surface
than loose soil or settled dust in a home. At the same time, the soil applied to the test surfaces
was not ground in, as might occur as a result of foot traffic in a home. Regardless of the
potential differences between experimental conditions and real-life settings, the results of this
comparative study do not support the recommended use of TSP for the reduction of lead dust
exposure.
VI
-------
1. Introduction
1.1 Background
In the past, the United States Environmental Protection Agency (EPA) has recommended the use
of trisodium phosphate (TSP), a high-phosphate detergent, to clean lead-contaminated dust from
surfaces, both in general and after remediation of lead-based paint. The recommendation to use
TSP has often been assumed to cover the general cleaning of lead-contaminated dust during
exposure reduction activities. Lead-contaminated dust can arise from deteriorated or disturbed
lead-based paint, lead-contaminated soil and street dust, or other sources. Because of the
negative impact of phosphate on the ecology of aquatic ecosystems, questions have arisen as to
the scientific basis of its recommended use and about the effectiveness of other cleaners, in
particular, low-phosphate cleaners.
This study was conducted to determine the relative effectiveness of different cleaning agents for
cleaning lead-contaminated soil material from surfaces similar to floors and walls. The results of
the study will be used to support EPA's recommendations to the public on methods for cleaning
surfaces with lead-contaminated dust and soil.
1.2 Objectives
The objectives of this study were to assess the relative cleaning efficacy of cleaners (commer-
cially available cleaning agents, tap water of average hardness, and TSP) as a function of the
physical and chemical characteristics of the cleaners. Cleaning efficacy was measured by the
quantity of lead picked up by a baby wipe after the surface had been cleaned. This measure of
cleaning effectiveness is used by abatement contractors to assess lead dust cleanup after
remediation of lead-based paint. Risk assessors also do wipe sampling of dust to determine if
lead hazards are present in a home. The quantity of lead picked up by the wipe is referred to as
the wipe lead.
The study was designed to determine if and how the following four cleaner characteristics affect
the relative cleaning efficacy as measured by the wipe lead:
• pH—The measurement of acidity and alkalinity of a solution. Solutions with a pH
greater than 7.0 are basic (alkaline); solutions with a pH less than 7.0 are acidic; pure
water has a pH of 7.0.
• Phosphate content—The amount, in grams, of phosphorus per gallon of cleaner mixture.
Various phosphate chemicals have traditionally been added to detergent cleaners to en-
hance the cleaning effectiveness.
• Surfactant type—The classification of the surfactant, or wetting agent, in the cleaner.
There are four major classifications of surfactants, of which two were considered in this
study: anionic and non-ionic. Anionic surfactants are water soluble and have negative
ions. Non-ionic surfactants, a class of synthetic surfactants, are the most widely used for
-------
surface cleaning and have no charge. Cleaner mixtures with a blend of anionic and non-
ionic surfactants types were also included in this study.
• Surface tension (dyne/cm)—The force acting on the surface of a liquid, tending to
minimize the area of the surface. Surface tension is a measure of how well the cleaner
mixture wets the surface to be cleaned: the lower the surface tension of the cleaner mix-
ture, the more the cleaner mixture will wet the surface. Pure water (without a cleaner)
has a high surface tension of 70 dyne/cm. Cleaners, by design, lower the surface tension
of water to low values of around 30 dyne/cm.
1.3 Overview of the Report
The study results as they apply directly to the objectives are presented in Section 2, which also
includes a brief overview of the methods. Section 3 summarizes the measures of quality for the
data and analysis. The study design is described in Section 4. Sections 5 and 6 describe the
laboratory and chemical analysis procedures, respectively. Section 7 describes the data
processing and statistical analysis method used. Finally, the statistical results are presented in
Section 8.
Although the report covers some relatively technical material, Sections 1,2, and 3 present the
background for the study, the objectives, the results, and the quality of the data in less technical
terms. The Executive Summary includes a minimum of technical details. Sections 4 through 6
describe in detail how the study was designed and implemented and contain technical details on
the laboratory procedures. Finally, the technical details and results of the statistical analysis are
described in Sections 7 and 8. In addition, a glossary of terms used in the study, a reference list,
and three appendices are included. Appendix A includes the test sequence and schedule used in
the study, Appendix B provides details of the laboratory procedures, and Appendix C includes
tables with the primary data required for statistical analysis.
1.4 Peer Review
The draft report on the initial laboratory study was reviewed by three independent external
reviewers. A summary of their comments is presented below. Responses to comments are
reflected in the report.
One reviewer commented that this was a difficult report to read because results were in three or
four different sections and procedures were buried within the results. The reviewer noted that
there were numerous statistical charts, but they were not well explained. This reviewer also
questioned the intended audience. In response to these comments, as well as comments from
EPA staff, the report was reorganized to more clearly separate the results from the procedures.
Some of the details of the laboratory procedures were moved to an appendix and referenced in
the body of the report. Explanations for the figures and tables were modified and clarified as
appropriate. Since the audience for the report consists of both nontechnical and technical
readers, the report was reorganized to better address both audiences. The Executive Summary,
the Introduction (Section 1), and Summary of Design, Findings, and Conclusions (Section 2) are
-------
written for the nontechnical audience, and the remaining sections and the appendices contain
material for the technical audience.
Two reviewers commented on the laboratory procedures used in the study. The first asked about
the method used to spread the cleaning solution, and the other asked about how the surface
tension was measured. Based on the first comment, details of how the cleaning solutions were
spread was added to Appendix B. No changes were made in the report in response to the second
comment about the measurement of the surface tension. In the response to comments from the
reviewer, it was explained that the surface tension of all solutions was measured by tensiometer
according to ASTM Method Dl 331. The individual values determined for surface tension of the
synthetic hard water control cleaning solutions were 66.8,41.2, and 58.6 dynes/cm. The surface
tension of the laboratory deionized water also was measured each working day. The surface
tension values determined for deionized water varied over a range of 61 to 76 dynes/cm. The
variation in the deionized water surface tension may account for some of the lower than expected
surface tension values for the synthetic hard water, since the synthetic hard water was made by
dissolving inorganic salt in deionized water. A set of three 1-gallon jars was used to mix three
cleaning agent solutions daily. These jars were rinsed thoroughly each day after use and allowed
to dry overnight before being reused the next day for another three cleaning agent solutions.
There may have been some carryover of a surface active agent from one day to the next.
Because the order of the cleaner tests was randomized, any cross contamination when the
cleaners were mixed would not affect the statistical results.
One reviewer suggested editing some of the figures. The appropriate figures were modified to
reflect this suggestion.
One reviewer commented that the results obtained should be viewed as being valid for the actual
materials and soil types studied and should not be generalized to include materials not studied.
To address this reviewer's concern, these limitations are now addressed in Sections 2 and 8 in the
revised report.
One reviewer marked the presentation and quality of data as "Unsatisfactory" and recommended
the report be published after minor revisions; however, no written comments were submitted.
Since no written comments were provided by this reviewer to support his rating of the report, the
reviewer was telephoned, and it was found that the reviewer thought that the measurement used
in the study was not the best one for assessing lead-cleaning efficacy. In this study, which was
performed as planned and directed by EPA, cleaning efficacy was measured indirectly by
quantifying the amount of lead remaining on the wipe after the soiled coupons were cleaned with
a sponge and cleaner. This approach is similar to that used after lead abatement activities when
wipe lead levels are compared to regulatory levels following abatement. An alternative would
have been to use a direct method to measure the quantity of lead remaining on the surfaces after
cleaning with the sponge and then the wipe. Plans are now underway to collect additional data
using a direct measurement approach.
It should be noted that EPA has established a public record for peer review under Administrative
Record 175. The record is available in the TSCA Nonconfidential Information Center located in
-------
Room NE-B607, Northeast Mall, 401 M Street, SW, Washington DC. The Center is open from
12:00 noon to 4:00 pm, Monday through Friday, except for legal holidays.
-------
2. Summary of Design, Findings, and Conclusions
2.1 Design
Thirty-four cleaners1 (32 commercially available cleaning agents, TSP, and tap water of average
hardness) were tested in this study. The cleaning agents were selected to represent a range of
physical and chemical characteristics including:
• Surfactant type—the type of active ingredient in the cleaning agent, categorized as
anionic or non-ionic.
• Phosphate content—the amount of phosphorus per gallon of cleaner mixture.
• pH—a measure of acidity. Solutions with pH less than 7.0 are acidic. Solutions with pH
greater than 7.0 are basic. Pure water has a pH of 7.0.
The following additional physical characteristic of the cleaners was measured after the cleaning
agents were mixed with water:
• Surface tension, which measures how well the cleaning mixture wets a surface. A drop
of a liquid with low surface tension will wet the surface and tend to spread out over the
clean surface. In contrast, when a high surface tension liquid is used, the drop will not
wet the surface and, consequently, will not spread out and may retain a spherical (drop-
like) shape.
The laboratory staff prepared five types of surfaces, each one foot square, and placed soil
material that contained lead onto the surfaces. The types of surfaces selected for "contamination"
with the soil were intended to simulate household floors with tracked in dirt and dust and other
household surfaces with lead-containing dust. Two types of soil were used. One contained some
vegetable oil to simulate dirt contaminated with oils from cooking or human contact. The other,
a "dry" soil, contained no added oil. The quantity of lead placed on the test surfaces was similar
to quantities found in older homes, which tend to have higher lead levels in the dust.
The selected cleaning agents were mixed with water, according to the manufacturers' instruc-
tions. The laboratory technicians then cleaned the surfaces to remove the soil material and the
associated lead. The amount of lead remaining on the surfaces was then measured. When lead
paint is removed from a home, the contractor is required to clean the lead-contaminated dust. A
surface is considered clean if the quantity of lead picked up by a baby wipe is below regulatory
limits2. Risk assessors also do wipe sampling of dust to determine if lead hazards are present in a
home. The same approach was used in this laboratory study. Cleaners that remove most of the
Terms in bold font are defined here and in the glossary. They are used throughout the report.
2 The limits from Chapter 5, Risk Assessment, of the HUD Guidelines are 100 ng/ft2 for carpets and hardwood
floors, 500 ug/ft2 for interior window sills, and 800 ng/ft2 for window throughs.
-------
lead will leave less lead to be picked up by the wipe. Therefore, for the most effective cleaners,
the quantity of lead picked up by a baby wipe will be smaller than for other cleaners.
Each cleaner was evaluated for its ability to clean each of the surface types. The results for all of
those tests using the same cleaner were combined to obtain an overall measure of how well each
cleaner cleaned the lead-contaminated soil. This combined measure is referred to as cleaning
efficacy. The data were analyzed to determine how the surfactant type, phosphate content, pH,
and surface tension affected the cleaning efficacy.
2.2 Findings and Conclusions
The analyses suggests that the cleaners tested in this study removed most of the lead applied to
the tested surfaces. Based on a comparison of coupons cleaned with either two baby wipes or
with a sponge and cleaner:
• Roughly 91 percent or more of the applied lead was removed by cleaning.
• Roughly 2 percent was removed by the test wipe after cleaning.
• Therefore, roughly 7 percent or less of the applied lead remained on the test surfaces.
Although these percentages are approximate and vary among surface types, they serve to
illustrate that thorough cleaning methods similar to those in this study can remove most of the
lead on the test surface, regardless of the choice of cleaner.
The analysis of the relationship between cleaning efficacy and surfactant type, phosphate
content, pH, and surface tension provided the following results:
• Only surface tension was related to how well the cleaners removed the lead-containing
soil from the surfaces, and even then only minimally related. Cleaners with lower surface
tension appear to clean the soiled surfaces slightly better than cleaners with higher surface
tension (in particular, water of average hardness).
• There was no evidence that high-phosphate cleaners, and particularly TSP, have better
cleaning efficacy than other cleaners tested.
Finally, in a set of separate cleaner tests, half of the tests were completed by one technician and
the rest by another. The results from the two technicians were different. One technician tended
to wipe more lead from the surfaces than the other. This suggests that
• In order to clean lead-containing soil from a surface, cleaning effort, cleaning method, or
other factors may be more important than the choice of cleaner.
This study was undertaken to compare the efficacy of cleaners that could be used to remove lead-
contaminated dust from residential surfaces; however, it should be kept in mind that the results
apply only to the conditions used in this study. In particular, the conclusions might have been
different if different soil material, cleaning methods, or surfaces had been used.
6
-------
3. Overall Quality Assurance
As part of the study design, steps were taken to ensure that the laboratory procedures provided
data of high quality and that no mistakes were made in the analysis of the data and the
presentation of the results. All laboratory procedures were conducted according to protocols.
The protocols were included as part of the Quality Assurance Project Plan (QAPjP) and approved
by EPA's Office of Pollution Prevention and Toxics (OPPT). A summary of the procedures used
and the results are presented below.
The design for the study was reviewed by experts outside of EPA before the study was
conducted. These reviewers provided comments on the study design and suggestions for
improvements. The reviewers comments were incorporated into the design.
The laboratory took measurements to ensure that the laboratory instruments were properly
calibrated and that the measurements achieved the precision and accuracy required by EPA.
Procedures were used to detect problems if they occurred and to correct those problems. These
procedures were followed by the laboratory personnel, and all measurements were within the
acceptable limits set by EPA. The details of these measurements are discussed in Section 6.
Quality control samples (blank, standards, etc.) were prepared to (1) measure any background
lead contamination in the lead measurement process and (2) confirm that the laboratory
measurement process could reliably measure the quantity of lead in samples for which the lead
content was known. Small quantities of lead, averaging 0.488 ug, were detected in unused baby
wipes; however, the quantity of lead in the baby wipes was so small that it did not affect the
study results. The details of the analysis of the blank and standard (quality control) samples are
discussed in Section 6.2.
Additional samples were taken to measure lead from other sources and to measure the
differences associated with different technicians. Lead contamination from other sources was
negligible, averaging 1.05 jig, and did not affect the study results. In contrast, differences
between technicians were large enough to be unlikely to be due to chance events. These
differences are discussed in Section 8.6.1 and in the findings and conclusions (Section 2.2).
The data entry was checked carefully to ensure that the computer data files had no errors. The
statistical analysis was reviewed and replicated by a second statistician to ensure that no errors
were made in the calculations.
-------
4. Study Design
This section discusses the study design, including the selection of the cleaners to be tested, tests
to be performed using each cleaner, number of tests, and randomization procedures.
4.1 Experimental Design
Thirty-four cleaners were tested in the study, including 32 commercially available cleaning
agents, tap water of average hardness, and TSP. The lead-cleaning efficacy of each cleaner was
assessed through a series of tests using the cleaner to clean surfaces soiled with lead-containing
soil. The cleaner tests were replicated three times, each time using an independently prepared
mixture of the cleaner. The results from the cleaner tests were combined to create an overall
measure of cleaning efficacy.
The overall measure of cleaning efficacy was analyzed with respect to the cleaning agent's
surfactant type and phosphate content, and the pH and surface tension of the cleaner mixture.
The pH and surface tension for each cleaner mixture were measured when the mixtures were
prepared. The surfactant type and phosphate content were determined from the labels on the
cleaning agents or by contacting the manufacturers.
4.2 Cleaner Selection
Cleaning agents that were candidates for inclusion in the study were selected by the laboratory
staff, who went to local stores and purchased as many different cleaning agents as was
conveniently possible. A total of 40 cleaning agents were obtained.
From the 40 candidate cleaning agents, 32 were included in the final study based on the design
parameters. The cleaning agents were selected to represent a range of surfactant type, phosphate
content, and pH. The study design objective specified classifying cleaners into one of eight
strata corresponding to all combinations of low or high pH, low- or high-phosphate content, and
anionic or non-ionic types of surfactant. The design objective was to select four cleaning agents
from each stratum. Cleaning agents were divided into two approximately equal-sized groups,
designated low and high pH, based on the cleaner solution's pH (i.e., cleaners with pH less than
10 or greater than 10, respectively). Similarly, cleaning agents were divided into two
approximately equal-sized groups, designated low- and high-phosphate, based on phosphate
content (i.e., cleaners with phosphate content less than or greater than 0.25 g P/gal., respec-
tively).
Some cleaning agents could not be classified into surfactant-type strata because they contained
both anionic and non-ionic surfactant types or neither surfactant type. In some cases,
information on the surfactant type was not available. These cleaning agents were classified as
containing "other/unknown" surfactant types. At the time the cleaners were sampled, the
phosphate content of one cleaner containing the "other/unknown" surfactant type was unknown.
-------
The following procedures were used for selecting cleaning agents for the eight strata envisioned
in the study design objective:
1. From the cleaning agents with the same pH, phosphate, and surfactant type as the
stratum, randomly select four, or as many as are available (if fewer than four), for
each stratum.
2. For strata with fewer than four cleaning agents after step 1, from the unselected
cleaning agents with the same pH and phosphate type (ignoring surfactant type) as the
stratum, randomly select additional cleaning agents to make a total of four, or as
many as are available, if fewer than four.
3. For strata with fewer than four cleaning agents after step 2, from the unselected
cleaning agents, randomly select additional cleaning agents to make a total of four.
Although these procedures helped to ensure consistency of pH and phosphate content for the
cleaning agents within each stratum, the original objective of selecting four cleaning agents with
similar levels of pH, phosphate content, and surfactant type for each stratum could not be
achieved.
After the cleaning agents were selected, the laboratory staff were informed by the manufacturer
that one cleaner was no longer manufactured and that it had been replaced on the market by
another cleaning agent. Since the replacement cleaner had already been selected for the study,
the steps above were repeated to select an additional cleaning agent to replace the discontinued
cleaning agent.
A ninth "control" stratum was defined, containing tap water of average hardness and anhydrous
TSP (1 % by weight). As mentioned earlier, the term cleaner is used to refer to both the
commercial cleaning agents in the eight strata and the tap water of average hardness and TSP in
the control stratum. In total, 34 cleaners were tested: 32 cleaning agents in the 8 strata (4 per
stratum) plus 2 cleaners in the control stratum.
Categorized according to their intended use, the 32 selected cleaning agents included 4 bathroom
cleaners, 2 floor cleaners, 2 glass cleaners, 3 hand dishwashing detergents, 6 hard-surface
cleaners, 10 laundry detergents, and 5 machine dishwasher detergents.
For the cleaners used in the study, Table 1 shows the number of cleaners by surfactant type, pH
level, and phosphate content.
-------
Table 1 Number of cleaners selected for the study by surfactant type, pH, and phosphate
content
Surfactant type
Anionic
Non-ionic
Other or unknown
pH level
pH<10
pH>10
pH<10
pH>10
pH<10
pH>10
Control
All
Phosphate content (g P/gal)
Below 0.25
4
3
3
2
3
2
1
18
Below 0.25
0
2
2
4
4
3
1
16
4.3 Substrate and Soil Selection
Five surfaces were selected to simulate interior surfaces in homes. The five surfaces were vinyl
tile, enamel-coated plywood, latex-painted plywood, lacquer (Fabulon)-coated plywood, and
latex-painted dry wall. All test surfaces, called coupons, were approximately 12-in square.
Two soil materials were used. The dry soil was selected to represent lead dust recently deposited
on home interior surfaces. The oily soil was selected to represent lead dust that might have
resided on home interior surfaces for an extended period of time and subsequently became mixed
with oils from cooking and/or human contact.
4.4 Cleaner Tests
The lead-cleaning efficacy was assessed through a series of tests in which one mixture of the
cleaner was used to clean the lead-containing soil from various surface types. The laboratory
procedures for the cleaner tests included: placing lead-containing soil on 1-foot-square surfaces
called coupons, cleaning the coupons with selected cleaners, wiping the coupons with baby
wipes, and analyzing the baby wipes for lead. This procedure, when applied to one coupon, is
called a coupon test. Each coupon test used one of two soil types and one of five coupon types
(surfaces), called substrates. The two soil types—referred to as oily and dry—contained the
same quantities of lead and other materials, except that the oily soil also contained vegetable oil.
The soil was mixed with a solvent/carrier (mineral spirits) for ease of application. The five
substrates were vinyl tile, latex on drywall, enamel on birch, lacquer (Fabulon) on oak, and latex
on birch.
10
-------
Each cleaner was mixed with water according to the manufacturer's instructions and tested on all
combinations of substrate and soil type. The set of all coupon tests using the same cleaner
mixture is called a cleaner test. The pH and surface tension of each cleaner mixture were
measured. Each of the 34 cleaners was tested three times (replicates) for a total of 103 cleaner
tests. After the first replicate for all cleaners was completed, the latex-on-drywall substrate was
dropped from subsequent replicates due to budget constraints. Thus, each cleaner test in the first
replication included 10 coupon tests, five substrates using each of two soil types. In the second
and third replications, each cleaner test included eight coupons, four substrates using each of two
soil types. Across all three replications, each cleaner was used to clean 26 coupons.
The laboratory procedures for each coupon test, shown schematically in Figure 1, included the
following steps:
1. Place a measured quantity of the soil mixture onto the coupon. Use 4 ml of soil mixture
for the latex-on-drywall and latex-on-birch substrates and 2 ml for other substrates. Let S
indicate the quantity of lead placed onto the coupon.
2. Use an applicator rod to spread the soil mixture across the coupon surface. A small
quantity of lead (R) remains on the applicator rod.
3. Dry the coupon overnight in a warm oven.
4. Place 10 ml of the cleaning solution onto the coupon and let it sit for 30 seconds. Wipe
the coupon with a new damp sponge, rinse the sponge, and wipe the coupon again.
5. Dry the coupon in a warm oven.
6. Wipe the coupon with a baby wipe (called a test wipe) to remove residual lead. The
quantity of lead in the test wipe is analyzed in the laboratory and is indicated by W in
Figure 1.
By design, the following factors were held constant for all tests:
• Soil application method
• Cleaning agent concentration (midpoint of the manufacturer's recommendation)
• Cleaning method
• Cleaning effort (weight applied to the sponge)
• Water temperature (room temperature)
• Water hardness (American Society for Testing Materials (ASTM) 3050 synthetic hard
water adjusted to 150 ppm hardness as calcium carbonate, referred to as tap water of
average hardness)
11
-------
Apply Soil
Coupon
Clean
Cleaning wipe--
Soiled wipe, C'
Wipe
Leaded soil, S
Applicator rod
Soiled applicator rod, R
Cleaner
X
Sponge
Soiled sponge, C
Test wipe
Soiled wipe, W
K = Lead concentration in soil mixture, measured from soil samples
V = Soil volume, 4 ml for latex on birch/drywall, 2 ml for other substrates
S = Lead placed on the coupon = K • V
R = Lead on applicator rod, measured from the rod rinse samples
K' = Effective soil lead concentration after use of applicator rod
L = Lead on coupon after soil application and before cleaning = S - R = K' • V
C = Lead removed by cleaning with cleaner and sponge
W = Lead removed on test wipe
C' = Lead removed by cleaning with baby wipes alone
Z = Lead remaining on the coupon at the end of the test = L - C' - W
Y = Residual wipe lead = W / L
Figure 1 Processing steps and measurements for a coupon test
12
-------
The tests for each cleaner were replicated three times. The measurements from each test were
combined into one measure of cleaning efficacy for that cleaner mixture. Within each
replication, all tests using the same cleaner were performed at the same time, as a group.
Therefore, the study uses a full factorial design (cleaner by substrate by soil type by replicate)
with coupon tests (the substrate-by-soil interaction) nested within the cleaner-by-replicate
interaction. The order in which the tests were performed is shown in the test sequence in
Appendix A.
4.5 Randomization of Cleaner Tests
The cleaners were tested in the order specified in the test sequence shown in Appendix A. The
three replications of the full factorial design were separated in time. The tests for cleaners within
each stratum were balanced as much as possible among the first third, middle third, and last third
of the tests performed within each replication. This balancing was intended to minimize the
effect of learning over time or other time-related factors, if any. Within this constraint, the order
in which cleaners were tested was randomized and, within each stratum, the cleaners were
randomly assigned to the cleaner identification numbers for that stratum. Because laboratory
tests on a few cleaning agents had already begun by the time the final test sequence was
prepared, the order of the first few tests did not follow the original randomized test sequence.
Appendix A shows the sequence in which the tests were actually performed.
4.6 Soil Sample, Rod Rinse, and Cleaning Wipe Tests
Separate coupon tests were performed to measure the quantity of lead applied to the coupons
before cleaning and the quantity which could be removed using baby wipes. No cleaners were
used for these tests. The procedure for placing leaded soil on a coupon included the following
steps:
1. Draw a measured amount of soil mixture into a pipette.
2. Release the soil mixture onto the coupon.
3. Spread the soil mixture across the surface of the coupon using a rod.
4. Dry the coupon in an oven to evaporate any remaining solvent.
Using the procedures below, soil and rod rinse samples were collected to determine the amount
of lead actually deposited onto the coupon surface prior to cleaning, and cleaning wipes were
collected to measure the quantity of lead removed using only baby wipes. The procedures listed
below are a modification of the test procedures listed above.
la. Draw a measured amount of soil mixture into a pipette.
2a. Place the soil mixture from the pipette into a sample bottle rather than onto a coupon.
This sample, called a soil sample, was analyzed for lead (S in Figure 1).
13
-------
3a. Place another soil sample onto a coupon using steps 1 through 3. Rinse the applicator rod
into a sample bottle after using the rod to spread the soil mixture across the surface of the
coupon. This sample, called a rod-rinse sample, was analyzed for lead (R in Figure 1).
4a. After oven-drying the soiled coupon from step 3a, wipe the coupon using either one or
two baby wipes to remove as much soil material as possible. These are called the clean-
ing wipes. For most of these coupons, two wipes were used. The cleaning wipes were
analyzed for lead (C'in Figure 1).
The difference in the amounts of lead in the soil sample from step 2a and rod rinse sample
obtained in step 3a provides the amount of lead on the coupons before cleaning. The amount of
lead on the cleaning wipes is an approximate measure of the quantity of lead which can be easily
removed from the coupons.
Soil-sample, rod-rinse, and cleaning-wipe tests were performed using coupons with all
combinations of substrate and soil type. The design called for five replications of each
combination of substrate and soil type (50 tests). The design was followed except that for three
combinations of substrate and soil type, six replications were completed.
4.7 Quality Control Tests
The laboratory work also included the following tests to evaluate the quality of the measure-
ments:
• Sixteen coupon tests using one of the cleaners were performed to check for variation
between the two laboratory technicians. For these tests, one technician performed all
processing steps associated with each coupon. In contrast, for the coupons in the cleaner
tests, the technicians divided up the various processing steps.
• Sixteen coupon tests were used to assess potential background contamination during
testing. For these tests, the oily and dry soil mixtures were prepared without the leaded
soil component.
• Additional quality control (QC) measurements were used to assess contamination,
instrument drift, and precision in the sample digestion preparation and lead measurement
process.
14
-------
5. Laboratory Data Collection
The cleaner tests were performed according to the design described in Section 4 and the test
sequence shown in Appendix A. Implementation of the design is described below.
Laboratory procedures were designed such that factors affecting the cleaning procedures could be
controlled as much as possible. Procedures were selected and designed to incorporate nationally
recognized standard test methods and materials, such as those of the American Society for
Testing Materials (ASTM), the National Institute of Standards and Technology (NIST), and the
American Water Works Association (AWWA), to the extent possible. In addition, the laboratory
procedures were selected to be practical and to simulate what a homeowner/occupant might do or
be able to do to clean up lead dust in the home.
All laboratory procedures were conducted according to protocols. The protocols were included
as part of the Quality Assurance Project Plan (QAPjP) and approved by EPA's Office of
Pollution Prevention and Toxics (OPPT). The laboratory procedures are discussed in the
following sections.
5.1 Preparation of the Soil and Substrates
The test surfaces, called coupons, were 12x12 inch pieces of tile, painted dry wall, or painted
wood. Five types of coupons (called substrates) were used, described by the surface material and
the type of coating:
• Vinyl tile
• Latex on drywall
• Enamel on birch
• Lacquer on oak
• Latex on birch
The preparation of the five substrates is described in detail in Appendix B. The coupons were
soiled with two types of soil mixtures, one containing vegetable oil (oily soil) and one without
vegetable oil (dry soil). The recipe for the oily soil mixture included the following:
• 15 g of Standard Reference Material (SRM) 2710, lead-containing soil (with lead
concentration of 5,532 ng/g, as reported by NIST)
• 7.5 g Norit A (carbon black)
• 150 ml mineral spirits
• 6.75 g of vegetable oil
15
-------
The recipe for the dry soil was the same as above except that the dry soil contained no vegetable
oil.
Each cleaner test used the same cleaner mixture to clean coupons with all combinations of
substrate and soil type. In replication 1, each cleaner was tested using 10 coupons (5 substrates
and 2 soil types). After the first replication, the latex-on-drywall substrate was eliminated due to
budget constraints. In replications 2 and 3, 8 coupons were used for each cleaner test. There
were 34 cleaners and 3 replications of each cleaner tested, for a total of 102 cleaner tests in the
design. One cleaner test was performed using the hydrated form of TSP rather than anhydrous
TSP, as specified in the design. An additional test, using the correct form of TSP, was performed
to replace the incorrect test. Therefore, a total of 103 cleaner tests were completed. There were
892 coupon tests, 26 for each of the 34 cleaners tested, plus 8 for the extra cleaner test.
5.2 Cleaner Tests
Each cleaner was tested once in each of three replications. Each cleaner test consisted of the
following steps:
1. Mix the cleaner with water according to the midpoint of the manufacturer's recommen-
dations.
2. Measure the pH and surface tension of the cleaner mixture.
3. Use the cleaner mixture to clean test surfaces on which the leaded soil had been placed.
4. Dry the test surfaces.
5. Measure the residual lead on the cleaned surfaces by wiping the surfaces with a baby
wipe and analyzing the wipe for lead.
5.3 Randomization of Coupon Tests
The order of the cleaner tests was randomized as described in Section 4.5. On a given day,
30 coupons (5 substrates x 3 cleaners x 2 soils) as defined by the test plan were prepared and
labeled for subsequent soiling, drying, cleaning, drying, and wiping. The coupons were
appropriately batched by type of soil to be applied (i.e., 15 coupons for oily soil and 15 coupons
for dry soil). The 15 coupons in each batch were then soiled with the appropriate type of soil and
dried over night. This procedure of batching by soil type in this step of the preparation procedure
was necessary for practical purposes. The following day, the appropriate coupons were batched
by cleaners to be used on these coupons. Thus three batches were formed and the 10 coupons
within each cleaner batch were cleaned. Once cleaned, the 30 coupons were oven-dried for two
hours. Once dried, the 30 coupons were wiped, and each wipe was collected in its prelabeled
centrifuge tube. In each labor step, coupons were selected at random, by the technician, from the
coupons in a batch.
16
-------
5.4 Soil Sample, Rod Rinse, and Cleaning Wipe Tests
Soil-sample, rod-rinse, and cleaning-wipe tests were performed using coupons with all
combinations of substrate and soil type. These tests were performed to obtain an estimate of the
amount of lead applied to the coupons before cleaning. Details of the procedure are provided in
Section 4.6. No cleaners were used in these tests. The design called for five replications of each
combination of substrate and soil type (50 tests). The design was followed except that for three
combinations of substrate and soil type, six replications were completed. The order in which the
tests were performed is shown in the test sequence in Appendix A.
5.5 Quality Assurance
5.5.1 Tests to assess measurement quality
Sixteen tests were run to check for variation between the two laboratory technicians. Cleaner 8
was used on four substrates and two soil types for a total of eight tests by each of two laboratory
technicians. For these tests, each technician performed all processing steps associated with a
coupon. This procedure was in contrast to the actual cleaner tests where the technicians divided
the processing steps between them.
Sixteen procedure blanks were used to assess potential contamination during testing activities.
For these tests, the oily and dry soil mixtures were prepared without the leaded soil component.
In the first replication, cleaner 19 was used on four substrates and two soil types for a total of
eight tests. In the second replication, cleaner 8 was used on the four substrates and the two soil
types for a total of eight tests.
Replicate cleaner tests were used to estimate the precision of the cleaning efficacy measure-
ments.
5.5.2 Quality of the results
Significant differences (p=0.0008) in residual wipe lead measurements were found between the
two technicians. The two technicians had geometric mean residual wipe lead measurements of
1.0 percent and 2.6 percent, respectively. This means that one technician tended to wipe off 2.5
times as much lead as the other, either because less lead was removed during cleaning with the
sponge or more lead was removed during wiping. The geometric means for the two technicians
span the range of residual wipe lead measurements for most of the cleaners tested. Therefore,
apparent differences among cleaners may be due to differences in which technician performed
certain tasks. Without a record on which technicians performed which tasks, differences between
technicians cannot be analyzed further. These differences are assumed to contribute randomly to
the measurement error and may contribute significantly to the magnitude of the error. A more
detailed discussion and the statistical analysis results are presented in Section 8.6.1.
17
-------
Lead contamination from the wipes, cleaners, sponges, and other possible sources was assessed
by using blank wipes (wipes not used to clean a coupon) and by running tests with soil mixtures
that did not contain the leaded component. The results indicate that the background lead from
sources other than the leaded soil placed on the coupons was, on average, about one-twentieth of
the lead found in the cleaner tests. Because the quantity of lead from other sources was small
and the source was not known, no blank correction was made in the analyses. A more detailed
discussion and statistical analysis results are presented in Section 8.6.2.
The coefficient of variation, a measure of precision, for the mean residual wipe lead for a cleaner
was about ±20 percent. Most of the variation in the cleaning efficacy measurements came from
differences between replicate mixtures of each cleaner, with a smaller amount from variation
among coupon tests. Because differences among cleaners appeared to contribute no variation to
the cleaning efficacy measurements, cleaner characteristics are unlikely to explain differences in
cleaning efficacy. A more detailed discussion and statistical analysis results are presented in
Section 8.6.3.
5.6 Design Modifications During Implementation
After the cleaners were selected and during the laboratory phase of the project, several
modifications to the design were necessary. These modifications are summarized below.
• Because the laboratory tests had already begun on a few cleaning agents as the final test
sequence was prepared, the order of the first few tests did not follow the original random-
ized test sequence.
• One cleaner originally selected was discontinued by the manufacturer. Another cleaning
agent with a similar pH and phosphate content was substituted.
• One of the five substrates, dry wall painted with latex, was dropped after the first replica-
tion to reduce the testing costs.
• A few samples were lost due to foaming during chemical digestion prior to lead analysis.
Most of the lost samples were replaced by repeating the coupon test, using the same sub-
strate, soil, and cleaner as for the lost sample.
• One cleaner test was performed using the hydrated form of TSP rather than anhydrous
TSP, as specified in the design. A test using the correct form of TSP was performed to
replace the incorrect test.
18
-------
6. Chemical Analysis
6.1 Analytical Procedures
The baby wipes and the soil and rod-rinse samples were digested and analyzed for lead. The
centrifuge tubes containing the baby wipes were labeled with a bar code and assembled into
preparation batches. Each preparation batch contained approximately 20 wipes, plus associated
QC samples.
The samples were digested according to Modified EPA Method 3050A. The digestate was then
analyzed for lead by inductively coupled plasma-atomic emission spectrophotometry (ICP-AES).
Lead measurements that were below the ICP-AES instrument detection limit were reanalyzed by
graphite furnace atomic absorption (GFAA).
As part of the analysis procedures, sample preparation and instrument QC samples were prepared
and analyzed for lead. QC sample measurements were used to assess background contamination,
bias, instrument drift, and precision in the preparation and measurement process. The steps in
the lead analysis procedures within the analytical laboratory were as follows:
1. Group the baby wipes into batches, called preparation batches.
2. Add preparation QC samples (blank wipes, spiked wipes, and NIST standard reference
material).
3. For each wipe or NIST reference soil material sample within a preparation batch, place
the wipe into a beaker, add acid, and place the beaker on a hot plate to digest the baby
wipe and dissolve the lead.
4. Filter out any solid material remaining in the beaker and transfer the solution to a vial for
lead analysis.
5. Adjust the volume of solution in the vial to 100 ml. The content of the vial is called the
final extract.
6. Group the final extracts into batches to be analyzed as one instrument batch. These
instrument batches generally included samples from several preparation batches.
7. Add instrument QC samples (blank, calibration verification, and interference check
samples).
8. Analyze the samples using ICP-AES to determine the amount of lead in each final
extract.
9. For coupon tests with lead measurements below the ICP detection limit, reanalyze the
final extracts using the more sensitive (lower detection limit) GFAA. Also, reanalyze
selected blank wipes with ICP measurements below the detection limit using GFAA.
19
-------
6.2 Preparation QC Samples and Results
The preparation QC samples included blank wipes, spiked wipes (wipes with a known quantity
of lead solution added), and standard reference material samples (a wipe with lead-containing
soil).
The lead measurements from the blank wipes estimate the quantity of lead contamination in the
wipe or coming from the sample preparation and analysis steps. The levels of lead found in the
blank wipes averaged 0.488 (ag and were small relative to the quantities of lead found in the test
wipes, which averaged 31.99 ug. The levels of lead found in the blank wipes were small (often
below the detection limit) and were not large enough to affect the conclusions from the cleaner
tests. (See Section 8.6.3 for more details.)
The lead measurements in the spiked wipes were very close to the spike levels and indicated no
problems with the lead measurement process.
Standard reference material samples are tested to monitor data from one analysis batch to
another. These tests are performed in an ongoing fashion in the laboratory and are not project
specific. In the laboratory in which the present study was undertaken, NIST SRM 2704 (Buffalo
River Sediment), with a lead level of 161 ^g Pb/g material, was used. The procedure consisted
of inserting, in a blind manner, one subsample of the SRM into the sample stream for each
preparation batch. For this study, the sample custodian provided approximately 1 g of SRM
2704, on a wipe, for inclusion with each preparation batch. These samples were referred to as
the Performance Evaluation Samples (PES) in the QAPjP. For all cleaner tests in this study, the
substrates were soiled using a soil mixture containing SRM 2710 with a lead level of 5,532 ug
Pb/g material. This lead level is a 34-fold concentration increase over the SRM 2704 used for
the SRM samples in the digestion and lead analysis. Since 1 g of SRM was being used for the
PES as per laboratory protocol, using SRM 2710 would have required a large number of
dilutions to bring the concentration into the appropriate range. A compromise was therefore
reached in which SRM 2704 PES recoveries were used throughout the study, along with two
SRM 2710 samples for recovery comparison purposes. The assumption was that if the
recoveries for the SRM 2704 fell within the acceptable range, and if the recoveries for the two
SRM 2710 PES were adequate and comparable to those of the SRM 2704, then the recovery of
lead on the wipes was acceptable.
Shown in Figure 2 are the percentage recoveries for the NIST SRM 2704 and the NIST SRM
2710 reference materials in each preparation batch. According to the QAPjP, the data quality
objectives for the acceptable percentage recovery range (PRR) were to be between 75 percent
and 120 percent. Furthermore, a preparation batch with PRR between 70 percent and 75 percent
or between 120 percent and 125 percent is said to be outside the lower or upper warning limits,
respectively. A preparation batch with a PRR above 125 percent or below 70 percent is
considered out of control.
20
-------
i
control 1
i_
Q.
Q.
3 IT
"> CV
•«-•
E
warning
0
Q.
Q.
> c
J 0
B
1 ^~^"f
• B
1 • —
! •
1 /
•^
1 B
! ^M
B^^~—^
1 ~~~B
1 W***^
1 ^^0*^*
' H^^^
1 ^"""--B
1 m
\ ^
\ ^*~"*«
1 /
m
\ x
! B
. B
* \
• •
i /
j •
i BC.
^B
1 B'
! B
1 /
, ^
• \
' •^**^
1 X
1 "^
1 w^^
\ -*^^*^*"^
1 B
1 •
N
1 ^^i*''^
B^*^
1 i
| •—
' ' B
i
^
! m
1 ^
i ^^
i "s
• B—
1
j
! B- — "
\ NB^^
1 ^"^t
1
1 Bs^~ '
1 ^*"S'**1B
| \
' ^^'"
m
!
i i
1 B
t tf5 ^^ tfj C^ tf5 CD i^ ^^ tf
1 ^™ ^~ ^3 ^3 O5 O5 00 CO f^
1
*• —
^-•^
B|
^^
. ^
zf
i
warning
l_
1?
"^
) C
r-
-pB
I
K
O
U
^
U
1
•^
> IO C
CD C£
< O UJ
< O UJ •*
N O -
MOO
N o a.
N O ->
N Q ^
N O O
N Q U.
N Q Z
N Q 2
N Q -1
N uj n
N UJ <
N O O
N Q UJ
N 0 0
N O U
N O CD
N O <
M 0 3
N O >
N o a:
N O K
N O «
N U X
N O O
N O O
MOO.
N U Z
N O 2
N O -»
N 0 I
N U UJ
N 0 O
N CO 5
N U O
< O UJ CM
N O O
N U <
N O O
< O LU i-
N m O
N m z
N ffl >-
N ffi X
N m 3
N o a.
N < >
N < 1-
N < 3
N CO *
N < >
N m ->
N ffi <
N m 2
N CO 2
>
a>
8
'c?
o
2
6
_c
0)
•o
o
O
^
"re
m
c
o
•«=
2
CD
a.
£
a.
a>
a.
E
ro
W
o
f^.
CM
•5
(0
W
Z
\
s
(N
a:
w
t-
w
z
1
B
1
U
«
^^
a
"•M
2
a
a
$
"5.
S
A
Mi
U
o^
3
K
&">
\f2
H
o
U
V
bn
h<
£
V
9
21
-------
As shown in Figure 2, four batches had PRRs below the lower warning limit of 75 percent, with
two of these batches having PRRs at or below the lower control limit. When a PRR value falls
outside that range, the analyst is to investigate the situation and take corrective action. Given the
time pressure under this specific work assignment, real-time intervention was not possible since
the data review lagged the analysis of the samples. (This is generally true for any laboratory
task.) Also, as seen in Figure 2, there was no indication of a downward trend in percentage
recovery in the first part of the project.
Once the low values were identified, the situation was investigated. It was concluded that the
problem was with the baby wipes. New baby wipes of the same brand as previously used on
other studies were used on this program. These new baby wipes did not dissolve in the digestion
step as did the previous wipes. A decision was then made that the undissolved baby wipes
needed rinsing with additional amounts of water to capture all of the dissolved lead adhering to
the wipes. This change in the procedure improved the lead recovery from the baby wipes after
digestion as shown in Figure 2. No recovery adjustments were made to the data corresponding to
the preparation batches for which percent recovery was below 75 percent.
6.3 Instrument QC Measurements and Results
Instrument QC samples are used to assess instrument bias, drift, and precision in the measure-
ment process. Each of the various types of QC samples are described below, along with a
description of the measurement results.
The instrument calibration and QC procedures are summarized in Table 2 for ICP and in Table 3
for GFAA analyses. Table 2 outlines adaptations of SW-846 Method 6010A for ICP analysis of
lead. Table 3 outlines adaptations of the SW-846 Method 7420 for GFAA analysis of lead.
An initial calibration was conducted for each analytical run using a calibration blank and a series
of calibration standards. Lead was calibrated to 10 mg/L on the ICP and to 40 [Ag/L on the
GFAA instruments.
Initial calibration verification (ICV) was conducted immediately after instrument calibration by
analyzing an independent standard prepared at the midpoint of the calibration curve. The stocks
used to prepare the ICV standard were procured from a different commercial supplier than those
used for instrument calibration. The ICV samples were used to verify the calibration standards
used for instrument calibration. The results of the ICV analysis were within 90 percent to 110
percent of the true concentration for both ICP and GFAA analysis.
An initial calibration blank (ICB) was analyzed after the ICV. Continuing calibration blanks
(CCB) were analyzed before the first sample in the batch, after every CCV (see below), and at
the end of the run. The instrument responses for the ICB and CCB (jag/ml) had to be less than or
equal to 10 times the calculated within-run instrument detection limit, or all of the samples
analyzed after the last acceptable calibration blank had to be reanalyzed.
22
-------
Table 2 ICP-AES instrument analysis QC procedures
QC procedure
Initial calibration
Initial calibration
verification (ICV)
Initial calibration
blank (ICB)
High standard
verification
Continuing calibration
verification (CCV)
Continuing blank
verification
Interference check
standard (ICS)
High sample results
Frequency
Once per analysis
Immediately after high
standard verification
Immediately after initial
calibration verification
Immediately after initial
calibration
Every 10 samples and at
the end of a run
Every 1 0 samples and at
the end of a run
Beginning and end run
plus every 8 hours
For every analyte over
high standard response
Acceptance criteria
None
90 - 110 percent of actual
concentration
Less than detection limit or less than
10 times the within run detection limit
95 - 105 percent of actual
concentration
90 - 110 percent of actual
concentration
Less than detection limit or less than
10 times the within run detection limit
80- 120 percent of actual
concentration
Dilute the sample within the
calibration range
Table 3 GFAA instrument analysis QC procedures
QC procedure
GFAA standards
Initial calibration
Initial calibration
verification (ICV)
Initial calibration
blank (ICB)
Continuing calibration
verification (CCV)
Continuing blank
verification
High sample results
Frequency
Prepared daily
Once per analysis
Immediately after
instrument calibration
Immediately after the ICV
Every 1 0 samples and at
the end of a run
Every 10 samples and at
the end of a run
For every analyte over
high standard response
Acceptance criteria
Must be digested and prepared as
samples
Linear correlation coefficient >0.995
90 - 110 percent of actual
concentration
Less than detection limit or less than
1 0 times the within run detection limit
80-120 percent of actual
concentration
Less than detection limit or less than
10 times the within run detection limit
Dilute the sample within the
calibration range
23
-------
For ICP analysis, the high standards used for instrument calibration were reanalyzed after initial
calibration verification and before analyzing the first sample in the batch to verify the linear
range. If the result of the high standard analysis for lead was not between 95 percent and 105
percent of the true concentration, the instrument was recalibrated.
Continuing calibration verification (CCV) was conducted after the initial calibration and before
the analysis of the first sample in the batch, after the analysis of every 10 samples, and at the end
of the run. The CCV analysis consisted of analyzing the midpoint standard used for instrument
calibration. The results of the CCV analysis were within 90 percent to 110 percent of the true
concentration for ICP analysis and within 80 percent to 120 percent of the true concentration for
GFAA analysis. Lead measurements reported for all samples were completed within acceptable
CCV acceptance criteria.
An interference check standard (ICS) was performed periodically throughout the ICP runs. An
ICS sample was analyzed after the high standards and before the CCV, at the end of the run, and
twice per 8 hours of analysis time. The ICS samples were used to verify that interelement
interferences and background intensities were at minimum levels during the run. Analysis for all
samples was completed within acceptable ICS recoveries (80 percent to 120 percent) or a
reanalysis of all samples analyzed subsequent to the last acceptable ICS was performed.
6.4 Corrective Actions
During the study, two corrective actions were taken to ensure the complete transfer of lead from
the wipes to the final extract in the preparation laboratory.
The transfer for subsequent digestion of the liquid soil and the rinse samples from the rod used to
deposit material on the coupon surface was facilitated using acetone. Because of the reactivity of
acetone with acid, a modification to the digestion method was instituted in the preparation
laboratory. After the sample was transferred, the acetone was either allowed to passively
evaporate from the sample or the sample was heated at a very low temperature (~30°C) to ensure
evaporation of the acetone. The sample was then digested according to the method and analyzed.
Allowing the sample to evaporate passively or to dry at a low temperature ensured that the lead
was not volatilized and that the acetone was not brought in contact with the strongly oxidizing
acid (HNO3) employed for digestion. This also ensured that the digestion procedure was safe.
Another modification to the digestion method was the use of Buchner funnels to transfer the final
digest to the volumetric flask using approximately 50 ml of Milli-Q water to rinse the digest.
This action ensured a higher and more consistent recovery of lead from the SRM. Since the
reference material was used to correlate analytical data from one batch to the next, a more
uniform correlation was desired and was obtained through this modification.
24
-------
7. Data Analysis Methodology
The lead amounts measured in the analytical laboratory were used to calculate the residual wipe
lead, the quantity of lead removed by the wipes as a fraction of the lead on the coupon before
cleaning. The residual wipe lead measurements were then analyzed statistically to assess how
the pH, phosphate content, surface tension, and surfactant type affected the cleaning efficacy of
the cleaners. This section describes the procedures used to calculate the residual wipe lead,
process the data, and statistically analyze the data. Results of the statistical analysis are
presented in Section 8.
7.1 Calculation of Residual Wipe Lead
The primary measurement used to determine the efficacy of the cleaners was the residual wipe
lead measurement. Because the lead remaining on the coupon after cleaning might be called
"residual lead" and the measurement is the portion of the residual lead that can be removed by
wiping, the measurement is referred to as the residual wipe lead. The residual wipe lead is
expressed as a proportion of the quantity of lead on the coupon before cleaning. Cleaners with
lower residual wipe lead measurements are preferred over cleaners with higher measurements,
because a lower residual measurement indicates that more lead was removed from the surface by
cleaning. The calculation of the residual wipe lead is described below. Table 4 and Figure 1 are
quick references for the symbols used in the equations below and in Section 8.
The first step in computing the residual wipe lead measurement was determining the quantity of
lead on each coupon before cleaning. In a separate set of tests, the quantity of lead in the soil
mixture (5) placed on the coupon was measured using soil samples. The quantity of lead
adhering to the applicator rod (R) was also measured using rod-rinse samples. For the analysis of
the cleaner test data, the quantity of lead on the coupons before cleaning (L) was predicted from
an analysis ofS-R (amount of lead placed on the coupon minus the amount of lead remaining on
the applicator rod).
To calculate the residual wipe lead (Y) used to evaluate the cleaning efficacy of the cleaners, the
measured quantity of lead in the test wipes (W) was divided by the predicted quantity of lead
deposited on each coupon. Thus, Y=WIL. The percentage of lead removed by using only
cleaning wipes and no cleaner was calculated in the same way.
The log-transformed residual wipe lead measurements were analyzed statistically to assess the
cleaning efficacy of the cleaners. The log-transformed residual wipe lead measurements for
coupon tests were used to assess differences among substrates and soil types. The average of the
log-transformed residual wipe lead measurements across all coupon tests within the same cleaner
test was used to assess general differences among cleaners. This average provided a measure of
cleaning efficacy. The overall average across cleaner test replications of the average log-
transformed residual wipe lead within each cleaner test was used to assess the extent to which
pH, surface tension, phosphate content, and surfactant type can predict cleaning efficacy. The
analysis results are presented in Section 8 in the untransformed units using the geometric mean
and the associated 95 percent confidence interval.
25
-------
Table 4 Symbols used in statistical equations
Type of symbol
Measurement
Subscript
Symbol
S
V
K
R
L
K'
C
W
C
P
H
T
I
Y
u
E
S
u
C
R
B
•
T
N
M
Meaning
Quantity (^g) of lead in soil placed onto the coupon
Volume of soil mixture used (4 ml for latex on birch and
drywall, 2 ml for other substrates)
Lead concentration in soil mixture (S/V) in (j,g/g
Quantity of lead on applicator rod ((j.g), measured from the rod-
rinse samples
Quantity of lead on the coupon after using the applicator rod
before cleaning (predicted from S - R)
and
Lead quantity on the coupon per ml of soil mixture used (L/V)
Quantity of lead removed by cleaning with a cleaner and sponge
Quantity of lead removed by the test wipe after cleaning with a
cleaner and sponge
Quantity of lead removed by cleaning with only baby wipes
Phosphate content of the cleaner mixture
pH of the cleaner mixture
Surface tension of the cleaner mixture
An indicator variable; its subscript indicates how the levels are
defined
Residual wipe lead
Mean
Random error assumed to be normally distributed
Soil type, 1 or 2
Substrate, 1 through 5
Cleaner, 1 through 35
Replicate, 1 through 3
Soil batch, 1 through 14
Indicates averaging over the subscript
Cleaner type, 1 through 4
Number of cleaning wipes used, 1 or 2
Technician, 1 or 2
26
-------
In order to measure how much lead could be removed without using any cleaner, some of the
soiled coupons were cleaned using only wipes (called cleaning wipes) rather than a cleaner,
sponge, and wipe. In the discussion of the statistical results (Section 8.7), arguments are
presented to support the assumption that the quantity of lead on the cleaning wipes (C") provides
an approximate lower bound estimate for the quantity of lead removed by cleaning (C).
7.2 Data Processing and Data Issues
The laboratory that performed all lead testing and lead analyses delivered all data as both paper
reports and computer files to the statistical contractor.
The lead analysis results for each instrument batch were provided as five separate files or as a
report with five tables. In all, there were 23 ICP instrument batches and 3 GFAA instrument
batches. The five data files for each instrument batch were combined into one file and checked
to confirm that the sample information was consistent with the coding procedures employed by
the analytical laboratory. The data from all instrument batches were combined into a single file
for analysis using the Statistical Analysis System (SAS). A SAS program was written to
reformat some of the data and write the final data files.
A file containing information about the timing of the cleaner tests and characteristics of the
cleaning agents was constructed from the preliminary data used to selected the cleaners and files
provided by the analytical laboratory. Some changes were made to make the coding of the
replication numbers consistent between files and to facilitate the merging of files.
The data were printed and checked by hand against the written laboratory data reports.
Histograms and frequencies of the variables were checked to identify possible problems in the
data. As part of the analysis procedures, unusual data values were also identified and checked
with the analytical laboratory. In addition, the analytical laboratory, in its own review,
identified some corrections to the data reports that also were incorporated into the data files. The
SAS programs were modified to correct all errors in the data files. A separate data documenta-
tion report describes the data processing, checking, and corrections in detail.
7.3 Statistical Models
The primary statistical procedures used to analyze the data are called general linear models, of
which analysis of variance, analysis of covariance, and regression analysis are a subset. The
procedure for using general linear models is to (1) specify an assumed model for the data in
which the dependent variable is assumed to be predicted by a linear combination of one or more
independent variables, (2) fit the model by estimating model parameters that minimize the sum
of the squared errors, and (3) evaluate how well the model fits the data. The model may have
continuous or discrete independent variables. In an analysis of variance model, all independent
variables are discrete. For regression, all independent variables are continuous. Analysis of
covariance combines discrete and continuous independent variables.
27
-------
The model for each analysis depends on how the data were collected and what is to be estimated.
For example, one model was fit to the soil and rod-rinse data to estimate the quantity of soil on
the coupons before cleaning. Another model was fit to the residual wipe lead measurements
from the cleaned coupons to evaluate how the residual wipe lead depended on the type of coupon
surface and soil type. Finally, an overall measure of cleaning efficacy, the average residual wipe
lead, was calculated for each cleaner. A model was then fit to the average residual wipe lead to
determine how it depended on cleaner characteristics.
The choice of the model is based on the experimental design and theoretical considerations. The
independent variables in the model were designated as either basic or auxiliary variables. Basic
variables were expected to have an effect (whether statistically significant or not) on lead
measurements based on the study design or physical considerations. Auxiliary variables were
those that might have an effect on the dependent variable but were not controlled for in the
design.
The dependent variable in most models was the log-transformed quantity of lead on the wipe.
The natural logarithm was used in all cases. The choice of a transformation depends on how the
transformation affects the model, the distribution of the residuals, and the variance of the
residuals. The residuals (a statistical term unrelated to the term residual wipe lead) are the
differences between the observed and the predicted values of the independent variable. If
different transformations might fit the data equally well, the preferred transformation will result
in residuals having equal variance for all predicted values and an approximately normal
distribution.
Use of the log-transformed measurements is justified if the effect of the independent variables on
the residual wipe lead is thought to be multiplicative or can reasonably be described by a
multiplicative model. This is equivalent to assuming that the effect of the independent variables
on the log-transformed residual wipe lead is additive. Other models might be assumed to
describe the effect of the independent variables on the residual wipe lead, such as an additive
model.
The multiplicative model, equivalent to an additive model for the log-transformed residual wipe
lead, was used because (1) it explained more of the variance in the data; (2) the predicted values
from the multiplicative model are always positive, as are the residual wipe lead measurements;
(3) the residuals from the multiplicative model have roughly equal variance; however, for the
residuals from the additive model, the variance increased noticeably for higher predicted values;
(4) the distribution of the residuals from the multiplicative model was closer to normal than for
the additive model; and (5) a multiplicative model was judged to provide a reasonable
description of how the independent variables might affect residual wipe lead. The results are
presented in the untransformed units using the geometric mean, 95 percent confidence intervals,
and the coefficient of variation.3
3 The coefficient of variation corresponding to the variance (s2) in the log-transformed units is approximated as:
cv = sqrt(exp(s2)-!). This equation assumes that the residuals have a lognormal distribution.
28
-------
Various criteria were used to evaluate how well the model fit the data, including the following:
• Parameter estimates that were outside the expected range might indicate that the factors
affecting the data were not understood or that there was an error in the model.
• If the variance of the residuals was smaller for some subsets of the data and larger for
others (a characteristic called heteroscedasticity), weighted analysis or a different model
might be required.
• Unusual observations that were in error could have a significant effect on the parameter
estimates. Unusual observations that were correct might imply that the model was not
correct. Influential values and those with unusually large residuals (outliers) were inves-
tigated.
The following steps were used to identify the final model for the data:
1. Fit the model with the basic and auxiliary independent variables.
2. Remove outliers as necessary.
3. Remove the least significant of the insignificant auxiliary variables, if any; refit and
repeat if an auxiliary variable that is not statistically significant is still present.
4. Fit the final model.
5. Check that observations removed as outliers are also outliers using the final model.
Check residuals for heteroscedasticity, approximate normality, and check for serial corre-
lation or other possible problems.
6. Refit if necessary.
The final model will always include the basic variables. Auxiliary variables are included only if
they are statistically significant. The final model is presented in equation form as part of the
discussion using the symbols shown in Figure 1 and defined in Table 4.
Because the study design is balanced (that is, all combinations of substrate, soil type, replicate,
and cleaner were tested the same number of times), heteroscedasticity, if present, has very little
effect on the parameter estimates and statistical tests. Therefore, adjustments for heteroscedas-
ticity were considered only if variance differences were statistically significant at the 1 percent
level. No such differences were found.
The analysis results are presented in tables that include a description of the dependent variable,
number of observations, precision, mean response, F tests for each independent factor, analysis
of variance table, and least square means for levels of the significant factors. The overall mean
response shows the general level of the measurements. However, since the focus of the analysis
is on the significance of factors in the model rather than on the overall mean, a confidence
interval for the mean was not calculated.
29
-------
7.4 Assessment of Outliers
In common usage, an outlier is a measurement that is, in some sense, unusual relative to a
preconceived notion of typical observations. Even though they are unusual, outliers may
represent correct measurements for unusual samples. The importance of identifying outliers may
be to learn more about the unusual samples. On the other hand, outliers may be the result of
incorrect sample handling, calibrations, calculations, or transcription. Outlying observations due
to processing errors are usually of little interest and are often removed from the data to minimize
bias when summarizing the data. An outlier analysis can never detect all possible errors.
Outliers resulting from some types of error cannot be distinguished from correct measurements.
Outlier identification procedures are based on the assumption that most of the measurements
have no significant errors and are within the typical range. Thus, in practice, outliers are defined
as those measurements that are inconsistent with the majority of comparable measurements.
With few comparable measurements, identification of an outlier can be problematic.
In general, the model for the data defines what is a typical observation. Observations that are
inconsistent with the model will have large residuals. The extreme studentized residual (ESR) is
used to identify residuals that are associated with outlying observations. The ESR is the
maximum of the absolute values of the studentized residuals. The studentized residual is an
optional output from many regression programs. The ESR test assumes that the residuals have a
normal distribution.
The critical values for the ESR,4 shown in Table 5, depend on the number of observations. They
also depend on the model used to obtain the residuals and the criteria for defining the
significance level for the test. The values in Table 5 are appropriate when fitting a mean only.
Consideration of additional independent variables in the model would have slightly decreased the
critical values shown in the table; therefore, use of the values in Table 5 results in a conservative
test. The true probability of deciding that the most extreme observation is an outlier is less than
the nominal 5 percent. For numbers of observations not shown in Table 5, interpolation was
used.
This outlier test assumes that the data, excluding outliers, are normally distributed. Therefore, a
histogram of the studentized residuals was checked to confirm that the distribution of the
residuals, excluding suspected outliers, appeared to be reasonably normal.
4 Bamett and Lewis provide an equation from which conservative critical values, shown in Table 5, can be
computed. They refer to the ESR as N2 (p. 223).
30
-------
Table 5 Critical values for the extreme studentized residual for
identifying outliers (5 % significance level)
Number of observations
20
24
30
35
40
50
60
120
240
480
884
960
Critical value
2.708
2.802
2.908
2.978
3.036
3.128
3.200
3.445
3.651
3.852
4.009
4.030
In some cases there may be two or more outliers in a data set. If an outlier was identified, it was
removed from the analysis and the model was refit to determine if there were additional outliers.
With multiple outliers, the critical values in Table 5 may suggest that no outliers are present. If
two observations were judged to be unusual based on visual inspection of a histogram and if the
ESR for the observation closest to the median was judged to be an outlier after removing the
more extreme observation, then both observations were assumed to be outliers.
All values identified as outliers were checked by the analytical laboratory for errors or situations
that might explain an unusual measurement. Section 8 includes details about how the outliers
were handled in each analysis.
7.5 Quality Assurance for the Statistical Analysis
For the most part, the statistical analysis was completed using JMP, a graphically oriented
statistical analysis package from the SAS Institute. This program is both easy to use and
provides abundant plots to help assess the fit of models and the presence of outliers. The results
from JMP were reviewed by a second statistician to ensure the reasonableness of the analysis
procedures and results. Finally, a SAS program was prepared to replicate the final statistical
analysis. Although different approaches were considered and discussed, no problems were
identified in the final analysis or in the comparison of the JMP and SAS outputs.
31
-------
8. Statistical Results
The following sections present the details of the statistical analysis. Section 8.1 describes the
effect of the sequence and timing of the laboratory work on estimates of cleaning efficacy.
Section 8.2 presents the analysis of the soil and rod-rinse samples from which the quantity of
lead applied to the coupons was determined. For the analyses discussed in Sections 8.3 through
8.5, the dependent variable is the lead picked up by the wipe as a percentage of the lead applied
to the coupons (residual wipe lead). Section 8.3 presents residual wipe lead differences among
substrates and soil types. Section 8.4 presents the analysis of cleaner characteristics on cleaning
efficacy (related to the residual wipe lead averaged across coupon tests using the same cleaner).
Section 8.5 summarizes what is known about how much lead was removed by the wipe and
sponge and how much remained on the coupons. Finally, Section 8.6 presents an analysis of the
QC data, and Section 8.7 discusses some of the statistical results.
8.1 Sequence and Timing of Tests
The cleaner tests were replicated three times. The second replication began once the first
replication was completed for all cleaners, and the third replication followed the second
replication. Within each replication, each cleaner was mixed and tested once. All coupon tests
using a cleaner mixture were completed as a group. Figure Al in Appendix A shows the dates
on which the coupons for each cleaner within each replication were soiled and wiped. The figure
also shows when the wipes were prepared for analysis in the laboratory and when the final
extracts were analyzed for lead. All laboratory work was completed between June and
September 1995.
As shown in Figure Al, only a few cleaners were processed in a given preparation batch and
instrument batch. There were 103 cleaner tests, 53 preparation batches, and 23 instrument
batches. No preparation batch included wipes from more than 4 cleaners. No instrument batch
included samples from more than 8 cleaners. Wipes from replicate 1 were generally prepared for
analysis and analyzed before wipes from replicate 2, and so on. As a result, differences over
time (for example, staff familiarity with the test procedures) cannot be distinguished from
changes or trends in the preparation and sample analysis. Also, measurement differences due to
preparation or instrument batch differences are difficult to distinguish from differences among
cleaners. As a result, the analysis assumes that instrument and preparation batch effects are small
and are part of the random variation affecting each cleaner test. The QC data are consistent with
this assumption.
8.2 Analysis of Soil and Rod-Rinse Samples
In order to evaluate the performance of each cleaner, the quantity of lead on each coupon before
cleaning can be compared to the quantity picked up by the wipe after cleaning. To make this
comparison, the quantity of lead on each coupon before cleaning must be estimated. Therefore,
in this study, soil and rod-rinse samples were collected to measure the quantity of lead on the
coupons before the cleaning step.
32
-------
The quantity of lead deposited on the coupons was the quantity of lead in the soil sample less the
quantity of lead that remained on the applicator rod (measured by the rod-rinse sample).
Therefore, an analysis of the difference of the soil-sample and rod-rinse measurements was used
to estimate the quantity of lead on each coupon before cleaning. The soil-sample and rod-rinse
data were analyzed separately to identify those factors that were likely to be important in the
analysis of the difference between the two measurements. The results from the analysis of the
soil samples and rod-rinse samples, and the difference between the soil and rod-rinse results are
discussed in the following sections.
8.2.1 Analysis of soil sample measurements
Samples of the soil mixtures were measured to determine the quantity of lead used for each of
the cleaner tests. Four milliliters of the soil mixture were removed by pipette for use on the
latex-on-birch and latex-on-drywall coupons. Two milliliters of the soil mixture were removed
for use on the other substrates. The quantity (ug) of lead removed for use in the tests was
therefore either two or four times the concentration of lead (ug/ml) in the soil mixture. The
statistical analysis of the soil samples and rod-rinse samples analyzed the amount of lead per
milliliter of soil mixture used
The expected concentration of lead in the soil mixtures was approximated from the recipes for
dry and oily soil (see Section 5.1). Assuming that the dry ingredients displaced a volume in
milliliters equal to their weight in grams (i.e., that the dry ingredients had the density of water),
the final volume of a mixture of dry soil would have been 172.5 ml. The mixtures contained
5,532*15 = 82,980 ng of lead; therefore, the lead concentration in the dry soil mixture was about
481 ng/ml. The oily soil mixture also included 6.75 g of vegetable oil. Making similar
assumptions and calculations for the oily soil, the lead concentration in the oily soil was about
463 (ig/ml. Based on these calculations, the ratio of the lead concentrations in the dry to the oily
soil mixtures was expected to be 1 .04. It should be noted that the assumption that the vegetable
oil and soil had a specific gravity of 1.0 (i.e., density of water) was made only to calculate the
approximate ratio of the lead concentrations in the dry and oily soil mixtures, respectively. The
ratio is not very sensitive to this assumption and was only used in a qualitative way to show that
the measured values were close to the expected values. The data analysis is entirely valid and
does not rely on this assumption.
The soil mixtures were prepared in batches (called soil batches). Within each batch, one dry and
one oily soil mixture were prepared. There may have been differences in the lead concentration
among the soil batches due to variation during preparation. On average, each batch of oily and
dry soil provided enough soil mixture to complete eight cleaner tests. Also, on average, two soil
samples were taken from each soil mixture. Trends in the measurements within a soil batch were
used to investigate whether the lead concentrations in soil samples from the same batch were
constant. If the soil within a soil mixture was not well mixed, the lead concentration in the first
and last soil samples taken from a mixture might differ.
The basic model for the soil-sample concentrations included three factors: soil type, soil batch,
and the interaction of soil type and soil mixture. Auxiliary variables included the instrument
33
-------
batch, preparation batch, replication number (the order in which the soil samples were taken),
and the sequence of the cleaner tests within each soil batch. For each soil sample sent for lead
analysis, an associated rod-rinse sample was collected. It was assumed that the type of coupon
(substrate) used in the associated rod-rinse measurement was independent of the concentration of
lead in the soil sample. Nevertheless, substrate was included as an auxiliary variable to verify
this assumption.
The model was run on the 54 soil-sample measurements. The dot plot in Figure 3 shows the soil-
sample measurements for each soil batch, separately for each soil type, on a log scale. The soil
batch is labeled using the last two characters in the soil batch number. Two unusual outliers
were identified and can be easily spotted in Figure 3 (shown as filled circles). Relative to the
mean of the soil-sample measurements, the studentized residuals for the outliers have magnitudes
of at least 6.87 (compared to a critical value of about 3.2). The lead measurements for the
outliers are about one-quarter of the lead measurements for other soil samples.
The outliers correspond to soil samples for latex on birch, in replicate 1, and enamel on birch, in
replicate 3, both using dry soil. In both cases, other soil-sample measurements from the same
soil mixture had concentrations close to the expected lead concentrations, suggesting that the soil
batches were correctly prepared and that the two outlying soil-sample measurements were in
error. The laboratory reported that one of these samples had foamed during preparation, possibly
losing some of the lead in the sample. The laboratory found no explanation why the other
measurement was so unusual. Both outliers were judged to be incorrect measurements because
(1) both outliers were very unusual as measured by the studentized residual, (2) both were very
different from the expected lead concentration based on the recipe for the dry soil mixture, and
(3) there were problems with the preparation of one sample. The outliers were therefore removed
from the analysis.
With the two outliers removed, the final model provided a good fit to the lead concentration in
soil data. The final model included terms for the soil type, soil batch, and two-way interaction of
soil type and soil batch. The two-way interaction was statistically significant (p = 0.0295). The
variation among replicate soil-sample measurements was relatively low, with a coefficient of
variation of 3.1 percent.
The final model for soil-sample lead concentration, in equation form, was:
By restricting the analysis to soil batches with data from both oily and dry soil (48 soil-sample
measurements), the average differences between the oily and dry soil mixtures could be
estimated. The estimated ratio of the lead concentration in the dry to oily soil was 1.054 (with a
95 percent confidence interval from 1.033 to 1.075). This confidence interval includes the
approximate ratio of 1 .04 determined from the recipes for the soil mixtures. The average lead
concentration across all soil samples was 449 ug/ml, close to the average of 472 ug/ml based on
the soil mixture recipes. These analysis of variance results are shown in Table 6.
34
-------
£
«
.a
1
90
86
81
77
72
67
62
56
52
46
42
38
33
25
90
86
81
77
72
67
62
56
52
46
42
38
33
25
A
o Measurements A
n»««ii:jh*dk«j ^.
ricUlClcd O
• Outliers dp
olo
do
0 0
dp
ofo
dp
$
ob
o | o
*
i
O|D
O 0
(|>
ob
+
0 O
•
(j)
op
CJ5
• OO
_>,
0
Q.
if
0
-------
Table 6 Analysis of variance results: Soil-sample lead concentration by soil type and soil
batch
Dependent (response)
variable
Number of observations
Comments
Root mean square error
Mean response
Log-transformed lead concentration (fig/ml) (S/V)
48
Two outliers removed; two soil batches excluded
0.0310 (corresponding to a coefficient of variation in the
untransformed unit of 3.1 percent)
6. 1 074 (corresponding to a geometric mean in the
untransformed unit of 449 |J,g/ml)
Factor
Soil
Soil batch
Soil*soil batch
Degrees of
freedom
Sum of
squares
Mean
square
F ratio
Prob>F
1 0.0284 0.02843 29.52 <.0001
11 0.0435 0.00395 4.11 0.0019
11 0.0265 0.00241 2.50 0.0295
Whole-model analysis of variance test
Source
Model
Error
Corrected total
Degrees of
freedom
23
24
47
Sum of
squares
Mean
square
0.1100 0.00478
0.0231 0.00096
0.1331
F ratio Prob>F
4.97 0.0001
Factor
Soil
Level
Dry
Oily
Least square geometric mean (95 percent confidence interval)
464.0 (457.1 to 470.9) ug/ml
440.2 (434.4 to 446. 1) jig/ml
36
-------
8.2.2 Analysis of rod rinse measurements
To estimate the quantity of lead that remained on the applicator rod, the rod was rinsed into a
sample jar for lead analysis for selected coupons. The objective of the statistical analysis of the
rod-rinse measurements was to identify factors that affected the quantity of lead adhering to the
applicator rod. The measurements were scaled according to the quantity of lead deposited on the
coupons, determined from the analysis of the soil-sample measurements. Thus, the analysis
estimated the quantity of lead adhering to the applicator rod as a fraction of the lead deposited
onto the coupon.
The basic model included factors for soil type, substrate, and substrate-by-soil-type interaction.
It was assumed that the quantity of lead adhering to the applicator rod might depend on the soil
type and how the lead stuck to the coupons, which was assumed to depend on the substrate.
Because the rod-rinse measurements were adjusted for the quantity of lead in the soil mixture,
soil batch was not included in the basic model. The soil-batch-by-soil-type interaction was
included as an auxiliary variable to verify, if possible, that soil batch effects were not significant.
Other auxiliary variables were those used in the soil-sample model. Based on the results for the
soil-sample model, no auxiliary variables were expected to be statistically significant.
In the final model, substrate, soil batch, and soil-batch-by-soil-type interaction factors were
statistically significant. The differences among mixtures of soil were large and not anticipated
(rod-rinse measurements for selected soil mixtures varied by more than a factor of 5). The
differences among soil mixtures for the rod-rinse measurements were about 10 times greater than
for the soil-sample measurements. The variation among rod-rinse measurements was relatively
high, with a coefficient of variation of 28 percent. On average, 13.5 ug/ml (3.0 % of the lead
applied to the coupon) remained on the applicator rod.
There is no apparent reason why the rod-rinse measurements should vary so much among soil
mixtures. On the assumption that the significance of the soil-batch terms was due to random
factors, the basic model (the final model without the soil-batch terms) was fit to the data. With
the soil batch and soil-type-by-soil-batch interaction removed from the model, no terms are
statistically significant and the coefficient of variation among rod-rinse measurements is 55
percent.
8.2.3 Analysis of soil-sample minus rod-rinse measurements
The difference between the soil-sample and rod-rinse lead measurements for each coupon was
analyzed to estimate the quantity of lead on the coupon at the time of cleaning. For the analysis,
the soil-sample minus rod-rinse measurements were divided by the volume of soil mixture used.
Thus, the units of analysis were ug Pb on the coupon per milliliter of soil mixture used.
The basic model for the analysis included factors for differences among soil batches, soil type,
and soil-type-by-soil-batch interaction. These are the same factors that were significant in the
model for the soil samples. Substrate was included as an auxiliary factor. Because the rod-rinse
measurements averaged only 3 percent of the soil-sample measurements, only those factors that
37
-------
are significant predictors of the soil-sample data were expected to be significant when modeling
the soil-sample minus rod-rinse measurements. In the soil-sample analyses, measurements on
two samples were removed as outliers. The same two samples were removed as outliers for the
analysis of the soil-sample minus rod-rinse measurements.
In the final model, the soil-type-by-soil-batch interaction was statistically significant. On
average, the coupons had 434 fig of lead per milliliter of soil mixture. Therefore, on average,
substrates with latex paint had 1,736 jig of lead prior to cleaning; other substrates had 868 ug of
lead prior to cleaning. The variation among the measurements was relatively low, with a
coefficient of variation of 3.4 percent.
Figure 4 shows the soil-sample minus rod-rinse lead amounts per milliliter of soil mixture by soil
batch, separately for each soil type, using a log scale. The two outliers have been removed from
the data in Figure 4 to better illustrate the patterns in the data. The predicted values are indicated
by vertical lines. The soil batch is labeled using the last two characters in the soil batch number.
In equation form, the final model for the quantity of lead on the coupon per volume of soil was:
ln(K's.B) = ln((SSB - RSUB)/VU) = u + Is + IB + ISB + e •
By restricting the analysis to soil batches with data on both oily and dry soil (47 soil-sample
measurements5), the magnitude and significance of the differences between the oily and dry soil
mixtures were estimated. Differences between soil types were highly significant (p < 0.0001).
The estimated ratio of the lead concentration in the dry to oily soil was 1 .055 (with a 95 percent
confidence interval from 1.031 to 1.079). This confidence interval includes the approximate
ratio of 1.04 determined from the recipes for the soil mixtures; therefore, the ratio was not
significantly different from the expected ratio. These analysis of variance results are shown in
Table 7.
The model with the two-way interaction of soil type and soil batch corresponds to fitting a mean
to measurements from each soil mixture separately. No soil samples were available for
estimating the mean in two soil mixtures, oily soil in batch 3835-62 and dry soil in batch 3835-
90. For these soil batches, the lead concentration was based on the mean of the available data
and the average difference between the mixtures of dry and oily soil for soil batches with data on
both oily and dry soil. Table 8 shows the predicted quantity of lead per milliliter of soil mixture
for each of the soil mixtures.
5 One rod-rinse sample was lost during preparation.
38
-------
90
86
81
77
72
67
62
56
52
46
42
38
j. 33
£ 25
.Q
=5 90
w 86
81
77
72
67
62
56
52
46
42
38
33
25
(j>
OJJO
0 0
0 O
ocj o
0 | 0
o|o
0 0
0 | 0
(j>
o | o o
o o
dp
T
o cj o
0 0
4
0 I 00
II
o
o | o
(j)
o o
o|o
o | o
^
o
(0
350 400 450 500
Lead concentration (ng/ml) (log scale)
o Measurements Predicted
Figure 4 Quantity of lead placed on the coupons per milliliter of soil mixture, by soil type
and soil batch
39
-------
Table 7 Analysis of variance results: Soil-sample minus rod-rinse lead amount per volume
of soil mixture, by soil type and soil batch
Dependent (response)
variable
Number of observations
Comments
Root mean square error
Mean response
Log-transformed soil lead minus rod-rinse lead (ug) per ml of
soil mixture (ln((SSB - RSUB)/VU))
47
Two outliers removed, two soil batches excluded (one with only
dry soil and one with only oily soil)
0.0349 (corresponding to a coefficient of variation in the
untransformed unit of 3.5 percent)
6.074 (corresponding to a geometric mean in the untransformed
unit of 434 jag/ml)
Factor
Soil
Soil batch
Soil* soil batch
Degrees of
freedom
Sum of
squares
Mean
square
F ratio
Prob>F
1 0.02916 0.02916 23.9331 <.0001
11 0.04935 0.00449 3.6817 0.004
11 0.04239 0.00385 3.1626 0.0096
Whole-model analysis of variance test
Source
Model
Error
Corrected total
Degrees of
freedom
23
23
46
Sum of
squares
Mean
square
0.13672 0.005944
0.02802 0.001218
0.16474
F ratio Prob>F
4.8786 0.0002
Factor
Soil
Level
Dry
Oily
Least square geometric mean (95 percent confidence interval)
448.9 (441.4 to 456.6) u.g/ml
425.5 (419.1 to 432.0) ng/ml
40
-------
Table 8 Quantity of lead on the coupons before cleaning (ug Pb per ml soil mixture)
by soil type and soil batch
Soil batch
3835-25
3835-33
3835-38
3835-42
3835-46
3835-52
3835-56
3835-62
3835-67
3835-72
3835-77
3835-81
3835-86
3835-90
Dry soil
477.8
450.8
464.0
467.0
404.1
442.2
431.3
432.6
449.3
444.5
427.3
478.8
456.1
463.0 *
Oily soil
404.4
382.2
448.1
456.4
417.9
424.3
432.8
410.0*
419.3
421.6
429.0
448.2
427.6
438.9
* These values are based on the statistical model. No soil and rod
rinse samples were collected for these soil mixtures
8.3 Analysis of Residual Wipe Lead
For the analysis of the lead-cleaner tests, the dependent variable was the lead removed by the
wipe as a fraction of the quantity of lead on the coupon. The quantity of lead on the coupon was
predicted from the analysis of soil-sample minus rod-rinse data. It was calculated by multiplying
the appropriate value in Table 8 by the volume of soil material used.
In terms of the variables used in Table 4, the residual wipe lead, symbolized by Y, was:
YCRSU = WCRSU/LSUB ,
where LSUB = K'S.B • Vv.
In accordance with the design, all combinations of cleaner, substrate, and soil type were tested in
each replication, comprising a full factorial design with three replications. The basic model for
the coupon tests included factors for a full factorial interaction of soil type, substrate, and cleaner
and a full factorial interaction of replication and cleaner. Differences among replications were
41
-------
assumed to be random. Auxiliary factors included preparation batch, instrument batch, and soil
batch.
Although the design called for the use of the anhydrous form of TSP (cleaner 34) for all TSP
tests, the hydrated form (which has a significantly lower phosphorus content) was used by
mistake in replication 2. The laboratory ran a fourth replicate for TSP using the anhydrous form
to replace the information not obtained when the incorrect form of TSP was used. The test on
the hydrated form of TSP was designated as using cleaner 35 in the data files. To construct a
balanced set of data for the analysis, the replication number for the replacement test on cleaner
34 was changed from 4 to 2, and cleaner 35 was excluded from the analysis. These changes were
made for analysis purposes only and were not made in the data files. The residuals were checked
to verify that changing the replication number did not affect the statistical results.
One outlier was removed from the analysis.6 The analytical laboratory checked the data and
found no unusual circumstances that might explain the measurement. Relative to the predicted
value from the final model, the studentized residual for the outlier is -4.89 (compared to a critical
value of about 4.0). The effect of excluding this outlier from the analysis is discussed at the end
of this section.
Factors for soil type, substrate, and soil by substrate were statistically significant (p<0.0001),
indicating that there were consistent residual wipe lead differences between combinations of soil
type and substrate across all cleaners and replicates. Differences by soil type and substrate do
not depend on the cleaner being used. Based on the geometric mean, 2.2 percent of the lead
originally deposited on the coupons was removed by the wipes after the coupons were cleaned.
Residual wipe lead measurements using oily soil were 2.38 times higher than those for dry soil.
The geometric mean residual wipe lead differed significantly among substrates, with the lowest
residual wipe lead measurements from enamel on birch and highest from substrates with latex
paint. The ratio of the residual wipe lead from latex on birch to that from enamel on birch was
1.28. Thus, the differences among substrates are relatively small but statistically significant.
Residual wipe lead differences between dry and oily soil were greater when using the enamel-on-
birch substrate than for other substrates. The coefficient of variation of one residual wipe lead
measurement is 45 percent. Table 9 summarizes the statistical results.
In equation form, the final model for the residual wipe lead on each coupon was:
ln(YCRSU) = ja + Ic + IR + ICR + Is + ID + Isu + Tcs + Tcu
6 Cleaner 19, oily soil, lacquer (Fabulon) on oak, replication 3
42
-------
Table 9 Analysis of variance results: Residual wipe lead by soil type and substrate within
cleaner test
Dependent (response)
variable
Number of observations
Comments
Root mean square error
Mean response
Log-transformed residual wipe lead
883
One outlier removed. REPLICATE modified by receding
replicate 4 to replicate 2 for cleaner 34
0.427901 (corresponding to a coefficient of variation in the
untransformed unit of 45 percent)
-3.83917 (corresponding to a geometric mean in the
untransformed unit of 0.021 5, or 2. 1 5 percent)
Factor
Soil
Replicate
Substrate* soil
Substrate
Cleaner* replicate
Cleaner
Cleaner* soil
Cleaner* substrate
Cleaner* substrate* soil
Degrees of
freedom
1
2
4
4
66
33
33
132
132
Sum of
squares
131.093
37.300
45.316
7.671
64.531
30.987
7.503
27.640
19.453
Mean
square
131.0932
18.6500
11.3290
1.9178
0.9777
0.9390
0.2274
0.2094
0.1474
F ratio
715.968
101.857
61.873
10.474
5.340
5.128
1.242
1.144
0.805
Prob>F
<.0001
<.0001
<.0001
<.0001
<.0001
<.0001
0.1711
0.1583
0.9332
Whole-model analysis of variance test
Source
Model
Error
Corrected total
Degrees of
freedom
407
475
882
Sum of
squares
Mean
square
458.75 1.1272
86.97 0.1831
545.72
F ratio Prob>F
6.156 <.0001
43
-------
Two of the auxiliary variables, preparation batch and instrument batch, were statistically
significant when added to the model individually. However, neither was significant when both
were added. After a review of the data, it was determined that preparation batch and instrument
batch were confounded with each other and with cleaner. This indicates that the effect of
preparation batch or instrument batch on the results might be due to other effects. In the final
analysis, neither instrument nor preparation batch was included in the model. There were no
indications from the QC samples that some batches were unusual. Therefore, any differences
among preparation batches or instrument batches were assumed to contribute in a random
manner to the differences among cleaners and measurements.
Figure 5 shows a dot plot of the residual wipe lead measurements by cleaner, separately by soil
type and substrate. For each of these combinations, the residual wipe lead measurements for the
three replications are shown. For latex on drywall, only one measurement is shown because this
substrate was removed from testing after the first replication. The figure also shows the
predicted geometric mean residual wipe lead across all replications. The cleaners are ordered
from highest to lowest predicted value. The one outlier removed from the analysis is shown as a
dark circle (cleaner 19 using oily soil). Although differences among soil types and substrates
were highly significant, the variation of the measurements around the predicted values illustrates
the relatively large variation among replicate tests. Note that the x-axis uses a log scale.
Also note that, on average, cleaner 19 had the lowest residual wipe lead measurements once its
outlier was excluded. If the outlier had been included, cleaner 19 would still have had the lowest
predicted residual wipe lead. Tap water of average hardness (cleaner 33) has one of the highest
predicted values of residual wipe lead.
Figure 6 and Table 10 show the geometric mean residual wipe lead, with 95 percent confidence
intervals, by soil type and substrate.
8.4 Analysis of Lead-Cleaning Efficacy
The average of the log-transformed residual wipe lead measurements for all coupon tests using
the same cleaner mixture provides a measure of cleaning efficacy for the cleaner mixture.
Because better cleaning, that is higher cleaning efficacy, is associated with lower residual wipe
lead, the discussion implicitly assumes the following definition for cleaning efficacy:
ln(cleaning efficacy) = - average(ln(residual wipe lead)).
This is equivalent to:
Cleaning efficacy = 1 /(geometric mean residual wipe lead).
44
-------
E
3
C
0>
n
o>
Substrate
Latex on birch Lacquer on oak Enamel on birch Latex on drywall
5
33
21
23
32
31
25
12
22
6
15
17
14
29
18
4
26
9
13
28
a
2f
34
10
19
5
33
21
2!
32
31
25
168
12
22
6
15
29
18
4
26
28
a
2?
34
10
19
OQO
oc
OfO
3D
&
O O
(SO
CO
OOJO
CX>
O
Vinyl tile
00
0>
Q.
.1% 1% 10%
.1% 1% 10%
Residual
.1% 1% 10% .1% 1% 10% .1% 1% 10%
wipe lead (percent) (log scale)
o Measurements
•Predicted
• Outlier
Figure 5 Residual wipe lead measurements (percentage) and predicted geometric means
by cleaner, soil type, and substrate
45
-------
5%
s
Q.
"is
0%
Geometric mean
•95% Confidence interval
Dry soil
Oily soil
-C
e
2
m
0>
s
•>,
.E
>
-
8
_
s
•>,
•
§
o
^^
X
0)
-B
(0
_j ui
Substrate
V
-£
Figure 6 Geometric mean residual wipe lead (percentage) by soil type and substrate
46
-------
Table 10 Geometric mean residual wipe lead by substrate and soil type (with 95 %
confidence intervals)
Substrate
Dry soil
Oily soil
Enamel on birch
Vinyl tile
Lacquer (Fabulon) on oak
Latex on birch
Latex on drywall
0.76% (0.83% to 0.70%) 4.21% (4.58% to 3.87%)
1.35% (1.47% to 1.25%) 3.25% (3.53% to 2.99%)
1.59% (1.72% to 1.46%) 3.04% (3.31% to 2.80%)
1.69% (1.84% to 1.56%) 3.11% (3.38% to 2.86%)
1.88% (2.18% to 1.61%) 2.78% (3.23% to 2.40%)
Analysis of cleaning efficacy (as defined above) or analysis of the average log-transformed
residual wipe lead will reach the same conclusions about cleaning efficacy with the difference
being that lower values of residual wipe lead are associated with higher values of cleaning
efficacy. To be consistent with the analysis of residual wipe lead on individual coupons, the
cleaning efficacy results are presented in terms of average residual wipe lead. However, the
analysis may be referred to as an analysis of cleaning efficacy.
Since the interactions of cleaner with both soil and substrate, or with either individually, were not
statistically significant (see Table 9), conclusions comparing cleaners were not expected to
depend on whether the analysis was based on the average log-transformed residual wipe lead
across all coupons or across selected soil types or substrates. Therefore, it was decided to use the
average of the log-transformed measurements across all coupons in a cleaner test as the most
reasonable summary measure of cleaning efficacy. For the one cleaner mixture for which one
test was removed as an outlier (shown in Figure 5), the least square mean from the statistical
analysis was used.
8.4.1 Analysis of differences among cleaner mixtures
The replications of the cleaner tests provide independent estimates of cleaning efficacy for each
cleaner. The estimates of cleaning efficacy were used to determine the precision of the cleaning
efficacy for each cleaner.
With three replications of 34 cleaner tests, cleaning efficacy was analyzed for 102 cleaner
mixtures. The basic model assessed differences among cleaners, with two factors, cleaner and
replicate. Differences among replications were assumed to be due to unidentified changes in
how the tests were conducted over time. In equation form, the model fit to the cleaning efficacy
measurements was:
The analysis details are shown in Table 1 1 . Differences among cleaners were not statistically
significant (p = 0.171). The residual standard error of 0.342 corresponds to a coefficient of
47
-------
variation of the cleaning efficacy measurements of 35 percent. The geometric mean across three
replications, therefore, has a coefficient of variation of about 20 percent (0.20 = .035/V3). The
fact that differences among cleaners are not statistically significant suggests that no one cleaner
is clearly better than the others, as judged by cleaning efficacy, and that associations among
cleaning efficacy and cleaner characteristics are likely to be weak and not statistically significant.
Table 11 Analysis of variance results: Cleaning efficacy differences among cleaners
Dependent (response)
variable
Number of observations
Comments
Root mean square error
Mean response
Log-transformed average residual wipe lead by cleaner test
101
Cleaner 35 and replicate 2 of cleaner 20 excluded
0.3423 (corresponding to a coefficient of variation in the
untransformed unit of 35 percent)
-3.8478 (corresponding to a geometric mean in the
untransformed unit of 2.1 percent)
Factor
Cleaner
Replicate
Degrees of
freedom
Sum of
squares
Mean
square
F ratio
Prob>F
33 5.08932 0.15422 1.3164 0.1711
2 4.40768 2.20384 18.8112 <.0001
Whole-model analysis of variance test
Source
Model
Error
Corrected total
Degrees of
freedom
35
65
100
Sum of
squares
Mean
square
9.65535 0.27587
7.61513 0.11716
17.27048
F ratio Prob>F
2.3547 0.0014
48
-------
8.4.2 Cleaner characteristics
The cleaners were selected to represent cleaners with two different surfactant types (anionic, non-
ionic) and a range of phosphate content and pH. Information on the phosphate content and
surfactant type was obtained from the labels and from the manufacturers. Information on the pH
was obtained from preliminary tests or measurements. When each cleaner mixture was prepared,
its pH and surface tension were measured. The data analysis used the pH measurements for each
cleaner mixture rather than the preliminary pH estimates used to select the cleaners.
Of the cleaners selected for the study, most contained no phosphate (15 of the 34 cleaners used),
and many had very little phosphate (only 7 of the 34 cleaners had a phosphate level over 1 g
P/gal). Cleaner 11 had the highest level of phosphate at 14.4 g P/gal. The pH for the cleaners
ranged from very acidic (the lowest pH was 1.5) to very basic (the highest pH was 12.6). Most
of the cleaners were basic (24 of the 34 cleaners had a pH above 7) and the median pH was 10.
The surfactant type for many of the cleaning agents could not be categorized as either anionic or
non-ionic either because both surfactant types were present, neither surfactant type was present,
or the appropriate information could not be obtained. For 13 of the cleaning agents, the
surfactant type was classified as other/unknown. Of the cleaners with solely anionic or non-ionic
surfactant types, about one-half were non-ionic (9 of 19) and one-half were anionic (10 of 19).
Tap water of average hardness and TSP were classified as control cleaners rather than as anionic,
non-ionic, or other/unknown.
The pH and surface tension were measured in the laboratory for each cleaner mixture. The
phosphate content was calculated based on the weight of the cleaner used and the concentration
of phosphate as reported by the manufacturer. Variation in the phosphate content between
mixtures was due to variation in the quantities of cleaner and water used in the preparation of
each mixture. Variation in the pH and surface tension was due to variation in (1) the quantities
of water and cleaner; (2) other factors, such as temperature, that might affect the pH and surface
tension; and (3) measurement variation for pH and surface tension.
For cleaner 20, the pH and surface tension measurements for replication 2 are unusual compared
to replicates 1 and 3. The pH in replication 2 was 5.7 compared to pHs of 11.85 and 10.7 in
replicates 1 and 3, respectively. Differences in pH among replications of the cleaner mixtures
were anticipated; however, the differences seen here are so large that they are unlikely to be due
to chance, as judged by the laboratory. The pH of tap water is around 7; the addition of a cleaner
is likely to raise the pH for basic cleaners and lower it for acidic cleaners. It is unlikely that the
pH would be highly basic in two replications and mildly acidic in the third. The numbers were
checked and found to agree with those recorded in the laboratory notebooks. It is possible that
the measurements recorded for replicate 2 were taken on a different cleaner mixture (several
cleaners were mixed on the same day), that the cleaner was not properly mixed, or that the
measurements reflect random unknown variation. Replication 2 of cleaner 20 was removed from
the analysis summarized in Table 11. Subsequent analyses were performed both with and
without the results of this cleaner test.
49
-------
8.4.3 Cleaning efficacy as a function of cleaner characteristics
The study design called for randomly selecting four cleaners from each of eight strata defined by
pH < 10 and pH > 10, phosphate content < 0.25 and > 0.25, and anionic and non-ionic surfactant
types. Given this design, the cleaning efficacy measurements could be analyzed as a three-way
full factorial design with four replications. An equivalent analysis would use regression with
coded levels for the pH, phosphate content, and surfactant type. To the extent that the
differences in cleaning efficacy are due to finer differences in pH or phosphate content than are
represented by the two level classification of pH or phosphate content, regression analysis using
the measured pH and phosphate content will have more power for identifying significant
relationships than would an analysis of variance.
The study design for selecting the cleaners was not completely achieved because (1) some
cleaners could not be classified as either anionic or non-ionic; (2) there was no clear cutoff
distinguishing "low" versus "high" pH and phosphate content (the chosen cutoffs divided the
cleaners into two groups with approximately equal number of cleaners); (3) there were not
enough cleaners within each stratum to select four for the study, in which case other cleaners
were randomly selected; and (4) the information that was used to select cleaners for each stratum
was updated during the study so that a few cleaners would have been classified differently. As a
result, analysis of variance based on the study design was not likely to be efficient for
determining which factors predict high cleaning efficacy.
Figure 7 shows a scatter plot matrix for phosphate content, pH, and surface tension for 35
cleaners (the 34 cleaners in the design and for hydrated TSP used by mistake in one cleaner test).
For each cleaner, the pH, phosphate content, and surface tension are averages of the replicates.
Note that there are only five cleaners with a phosphate content over 2.0 g P/gal, one of which is
cleaner 35. Thus, for phosphate levels over 2.0 g P/gal, there is very little data from which to
draw conclusions about cleaning efficacy. Also, all five of the cleaners with a phosphate content
above 2.0 g P/gal have pH of about 10; therefore, any conclusions drawn about how pH,
phosphate content, and surface tension predict cleaning efficacy for cleaners with phosphate
content over 2.0 g P/gal would apply only to cleaners with pH of about 10.
For the analysis of cleaning efficacy versus cleaner characteristics, the dependent variable was
the log-transformed geometric mean cleaning efficacy across all replications. The analysis
included cleaner 35, the cleaner test completed with hydrated rather than anhydrous TSP. The
dependent variable was the log transformed geometric mean across replications, corresponding to
the least square means from the model discussed in Section 8.4.1. For cleaners without cleaner
tests in all replications (cleaner 20 and 35), the least square mean was used. For most analyses,
replication 2 of cleaner 20 was eliminated from the calculations. A weighted analysis was used,
with the weights proportional to the number of replicate cleaner tests contributing to the mean.
50
-------
16
14-
12-
10-
8
6-
4-
2-
10
8-
6-
4-
20
30 40 50
Surface tension
60
Figure 7 Distribution of 35 cleaners across values of pH, phosphate content, and surface
tension
51
-------
The design for selecting cleaners corresponds to fitting all two-way interactions of cleaner type,
pH, and phosphate content, with surface tension added as a covariate. Because cleaner type has
four levels (anionic, non-ionic, other/unknown, and control) instead of the two used in the
design, some parameters could not be estimated in this model. Possible alternative models drop
the two cleaners in the control group (TSP and water) to allow estimation of the parameters or
drop the interactions between cleaner type and other variables. In exploratory models, the
estimated mean residual wipe lead was very similar for all cleaner groups except the control
group. An alternative model is a response surface model (all two-way interactions of continuous
variables, including squared terms). The response surface model has the advantage of testing for
nonlinearity in the continuous effects.
After reviewing the possible models, a basic model was chosen that included cleaner type coded
as anionic, nonionic, other/unknown, or control (water and TSP) and all two-way interactions of
pH, surface tension, and phosphate content. This model was a compromise between the one
envisioned during the design and a response surface model that includes all squared terms.
Therefore, this model can be viewed as a simple response surface model with an added factor,
cleaner type. The model was fit to all cleaners; however, data for replicate 2 cleaner 20 was
excluded in calculating the cleaning efficacy and the average pH, phosphate content, and surface
tension among replicates. Auxiliary variables included interactions of cleaner type with pH,
surface tension, and phosphate content, as well as squared terms for pH, surface tension, and
phosphate content.
The overall F statistic for the basic model is significant at the 5 percent level, although barely
(p=0.0475). Only the term for surface tension is statistically significant (p = 0.0369). Although
only one term in the model is significant, the combination of terms making up the model
provides insight into the data.
After averaging across cleaner type, the model corresponds to a surface in four-dimensional
space (with axes of pH, surface tension, phosphate content, and cleaning efficacy). Figure 8
shows two three-dimensional cuts through the four-dimensional surface. The top plot shows
contours of residual wipe lead versus surface tension and pH when phosphate content is held
constant at 0 g P/gal. The lower plot shows the same contours when phosphate content is held
constant at 5 g P/gal. The dots shown in the plot indicate combinations of pH and surface
tension for the cleaners in the study, that is, the combinations of pH and surface tension for
which data are available to fit the predicted residual wipe lead surface. Since a lower residual
wipe lead indicates better cleaning and is preferred, the plot suggests that the preferred cleaners
will have lower surface tension. Note that any conclusions about the preferred cleaner
characteristics based on the contours in Figure 8 have little confidence. Table 12 shows the
analysis results for the basic model. None of the auxiliary variables was statistically significant.
In equation form, the basic model (used in Table 12) was as follows:
Yc... = n + H + P + T
52
-------
Phosphate concentration = 0 g P/gal.
I
60 -I
55-
50-
45-
40-
35-
30-
25
OTSP
o Cleaner 35
^v$Z
• r^—-^x?
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
PH
Phosphate concentration = 5 g P/gal.
w
I
o
o
1
(0
oTSP
o Cleaner 35
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Cleaner with
phosphate
content < 2
o Cleaner with
phosphate
content > 2
• Predicted
residual wipe
lead contour
Cleaner with
phosphate
content < 2
o Cleaner with
phosphate
content > 2
- Predicted
residual wipe
lead contour
Figure 8 Contours of predicted average residual wipe lead versus pH and surface tension
for a phosphate content of 0 and 5 g P/gal, averaged across cleaner types
53
-------
Table 12 Analysis of covariance results: Cleaning efficacy by cleaner characteristics
Dependent (response)
variable
Number of observations
Comments
Root mean square error
Mean response
Log-transformed cleaning efficacy
35
Replicate 2 of cleaner 20 excluded. Regression weight equals
the number of replicate cleaner tests/3
0.1962 (corresponding to a coefficient of variation in the
untransformed unit of 20 percent)
-3.8497 (corresponding to a geometric mean in the
untransformed unit of 2.1 percent)
Factor
Cleaner type
pH
Surface tension
Phosphate
Surface tension*pH
Phosphate*pH
Phosphate* surface tension
Degrees of
freedom
3
1
1
1
1
1
1
Sum of
squares
0.13080
0.07158
0.18699
0.00811
0.06947
0.00279
0.01138
Mean
square
0.0436
0.0716
0.1870
0.0081
0.0695
0.0028
0.0114
F ratio
1.1331
1.8602
4.8598
0.2108
1.8054
0.0724
0.2957
Prob>F
0.3548
0.1848
0.0369
0.6501
0.1911
0.7901
0.5914
Whole-model analysis of variance test
Source
Model
Error
Corrected total
Degrees of
freedom
9
25
34
Sum of
squares
Mean
square
0.80040 0.08893
0.96194 0.03848
1.76233
F ratio Prob>F
2.3113 0.0475
54
-------
Because the choice of which model to use as the basic model was not obvious, alternative models
were considered to determine how changes to the model affected the fit. Results from the basic
and alternative models are summarized in Table 13. Alternative models other than those in Table
13 were also considered. The different ways of looking at the data produced similar conclusions:
Surface tension provides the best prediction of cleaning efficacy; however, surface tension is
marginally significant and the significance depends on which model is used.
Table 13 Results from alternative models
Independent variables used in the model
Basic model (cleaner type and two way interac-
tions of pH, phosphate content, surface tension)
Basic model with squared terms for pH, phosphate
content, and surface tension
Basic model with replicate 2 of cleaner 20
included in the calculations
Basic model without the term for cleaner type
Basic model with water and TSP removed from the
analysis (i.e., the control group in cleaner type)
Basic model with interactions between cleaner
type and each of pH, surface tension, and
phosphate content
Basic model with the most unusual observation
removed (cleaner 5)
Cleaner type, pH, phosphate content, and surface
tension without interactions
Cleaner type, pH, phosphate content, and surface
tension with replicate 2 of cleaner 20 included in
the calculations
Surface tension
Surface tension with replicate 2 of cleaner 20
included in the calculations
Overall
p-value
0.0475
0.0797
0.0664
0.0267
0.1752
0.3213
0.0076
0.0231
0.0740
0.0167
0.0396
Summary of results
Cleaners with lower surface
tension preferred
Overall F test not significant
Overall F test not significant
Cleaners with lower surface
tension and higher pH preferred
Overall F test not significant
Overall F test not significant
Similar to the basic model
Similar to the basic model
Overall F test not significant
Similar to the basic model
Similar to the basic model
Finally, the least significant factors were eliminated in a stepwise fashion. The elimination
started with the preferred model with separate interactions between cleaner type and pH, surface
tension, and phosphate content. All factors with p-values above 0.05 were eliminated until the
final model was obtained. The final model predicted residual wipe lead as a function of surface
tension only. The p-value for this model is 0.0167. Figure 9 shows the regression line when
55
-------
fitting mean residual wipe lead to surface tension alone. This figure shows the measurements,
the fitted line, and the variation around the line. The line was fit using weighted regression with
replication 2 of cleaner 20 removed from the analysis. Table 14 show the regression results and
parameter estimates.
An analysis was also performed to determine whether the intended use of the cleaner (such as
dishwashing or laundry) would predict either surface tension or cleaning efficacy. Although the
intended use of the cleaner was significant for predicting surface tension, it was not significant
for predicting residual wipe lead. To the extent that surface tension predicts residual wipe lead,
no factor was identified that the consumer might use to select cleaners with lower surface
tension.
8.5 Percentage of Lead Removed Using Wipes Only
When the soil and rod-rinse samples were obtained, coupons were cleaned using wipes instead of
a cleaner and sponge. The lead measurements from these wipes are referred to as cleaning wipe
measurements. These tests measured the quantity of lead that could be removed from the soiled
coupons using wipes alone. Assuming that the wipes removed at least as much lead as did the
cleaners, the lead measurements on the wipes provide a lower bound estimate of the quantity of
lead removed using a cleaner and sponge.
Figure 10 is a dot plot of the cleaning wipe measurements (percentage of lead removed by wipes)
by substrate, separately by soil type, using a log scale. One wipe was used on the first six
coupons tested, after which two wipes were used on each coupon. The measurements taken
using one or two wipes are illustrated using a square or circle, respectively, in Figure 10.
The basic model fit to the data included factors for differences among substrates, soil types,
combinations of substrates and soil types, and the number of wipes used on the coupons.
Differences among substrates and the number of wipes were statistically significant. Using one
wipe, 79 percent of the lead was removed, based on the geometric mean. Using two wipes, 91
percent of the lead on the coupon was removed. The predicted value in Figure 10 assumes that
two wipes were used. Table 15 shows the analysis results.
In equation form, the model for the cleaning wipe measurements was:
C'SUN = H + Is + Iu + !su + IN + e .
When two wipes were used to clean the coupons, only the substrate was a significant predictor of
the quantity of lead removed. The geometric mean cleaning wipe lead using two wipes (with 95
% confidence intervals) was 98.5 percent (93.8% to 103.3%) for enamel on birch, 83.0 percent
(74.5% to 92.3%) for lacquer on oak, 91.2 percent (83.0% to 100.3%) for latex on drywall, 90.3
percent (87.1% to 93.6%) for latex on birch, and 92.9 percent (83.0% to 104.0%) for vinyl tile.
For each substrate, Figure 11 shows the geometric mean percentage of lead picked up using two
wipes and the associated 95 percent confidence interval.
56
-------
4.0%
3.5%
3.0%
-. 2.5%
I
to
I
6 2.0%
o
I
.1
1.5%
1.0%
Cleaner 20
Tap water
• Cleaner 35
• TSP
• Measurements
Regression line
Lower 95% Cl
Upper 95% Cl
20 25 30 35 40 45
Surface tension (dyne/cm)
50 55
Figure 9 Geometric mean residual wipe lead (percentage) versus cleaner surface tension
57
-------
Table 14 Regression results: Cleaning efficacy by surface tension
Dependent (response)
variable
Number of observations
Comments
Root mean square error
Mean response
Log-transformed cleaning efficacy
35
Replicate 2 of cleaner 20 excluded. Regression weight equals
the number of replicate cleaner tests/3
0.21 16 (corresponding to a coefficient of variation in the
untransformed unit of 21 percent)
-3.8497 (corresponding to a geometric mean in the
untransformed unit of 2.1 percent)
Whole-model analysis of variance test
Source
Surface tension
Error
Corrected total
Degrees of
freedom
1
33
34
Sum of
squares
Mean
square
0.28474 0.28474
1.477590 0.04476
1.76233
F ratio Prob>F
6.3593 0.0167
Parameter
Parameter estimates (95 percent confidence interval)
Intercept
Surface tension
-4.2560 (-4.592 to 3.920) ln(%)
0.0128 (0.0025 to 0.0231) ln(%) per dyne/cm
58
-------
a,
u
I
Enamel
on birch
Lacquer
on oak
Latex
on birch
Latex on
drywall
Vinyl
tile
o oo
o o
o o
o o
oo
3D CD)
o
a
Enamel
on birch
Lacquer
on oak
Latex
on birch
Latex on
drywall
Vinyl
tile
83% ' 1(JO%
Cleaning wipe lead (%) (log scale)
o Measurement with two wipes Geometric mean, two wipes
D Measurement with one wipe
Figure 10 Percentage of lead removed using one or two wipes to clean coupons
59
-------
Table 15 Analysis of variance results: Cleaning wipe lead by soil type and substrate
Dependent (response)
variable
Number of observations
Root mean square error
Mean response
Log-transformed fraction of lead removed using cleaning wipes
to clean coupons
53
0.1 1018 (corresponding to a coefficient of variation in the
untransformed unit of 1 1 percent)
-0.10946 (corresponding to a geometric mean in the
untransformed unit of 89.6 percent)
Factor
Substrate
Soil
Substrate*soil
Number of wipes
Degrees of
freedom
4
1
4
1
Sum of
squares
0.161356
0.000057
0.041158
0.092791
Mean
square
0.040339
0.000057
0.010290
0.092791
F ratio
3.3233
0.0047
0.8477
7.6445
Prob>F
0.0188
0.9456
0.5031
0.0084
Whole-model analysis of variance test
Source
Model
Error
Corrected total
Degrees of
freedom
10
42
52
Sum of
squares
Mean
square
0.343236 0.034324
0.509807 0.012138
0.85304
F ratio Prob>F
2.8277 0.0089
Factor
Number of
wipes
Level
One wipe
Two wipes
Least square geometric mean (95 percent confidence interval)
78.9% (71. 5% to 87.0%)
91. 2% (88.2% to 94.3%)
60
-------
120%--
80% - -
w
9-
1
o
o>
w
o
£ 60%
•o
(0
0>
O)
2
5
Q.
40% - -
20% - -
0%
8
O
-e-
. _ -O- - -
o
o
§
-tr
@
0
8
w
• Geometric mean
o Measurement
—95% Confidence limits
•+•
+
Enamel
on birch
Latex
on
drywall
Vinyl
tile
Substrate
Latex
on birch
Lacquer
on oak
Figure 11 Percentage of lead removed when using two baby wipes to clean coupons, by
substrate
61
-------
As discussed in Section 8.3, the geometric mean residual wipe lead for the cleaner tests was
roughly 2 percent. Assuming that the quantity of lead removed using two baby wipes was
similar to the quantity removed with a cleaner and sponge:
• Roughly 91 percent of applied lead was removed by cleaning.
• Roughly 2 percent was removed by the test wipe after cleaning.
• Therefore, roughly 7 percent of the applied lead remained on the coupon.
These percentages are approximate because (1) the percentage removed by baby wipes is, at best,
an approximation of the percentage removed by cleaning with a sponge and cleaner, and (2) there
are differences among substrates and soil types. Despite the approximations, the figures serve to
illustrate that:
• Simple cleaning methods, such as using two baby wipes, can remove most of the lead on
the coupon.
• Cleaning efficacy results are based on a statistical analysis of the roughly 2 percent of
lead removed by the wipes after cleaning.
8.6 Quality Control Results
The design included measurements to assess the quality of the data, including measurement
precision and contamination. The following sections present the results of a statistical analysis to
evaluate QC data.
8.6.1 Variability between technicians
Two extra cleaner tests were performed to assess potential differences between the two
laboratory technicians. In the standard procedure for the cleaner tests, the technicians split the
laboratory tasks (applying soil mixture to the coupons, drying the coupons, cleaning the coupons,
drying the coupons, and wiping the coupons), and no records were kept as to which technician
performed which task. In one of the extra cleaner tests, one technician performed all tasks for all
eight coupons. In the other extra cleaner tests, the second technician performed all tasks for all
eight coupons. These extra tests used cleaner number 8.
The basic model used in the analysis included terms for technician, soil type, substrate, and soil-
by-substrate interaction. All tests used soil from the same soil batch. Instrument batch was
included as an auxiliary variable. In the final model, soil type, soil by substrate, and technician
were statistically significant. A summary of the model fit is shown in Table 16. The residual
wipe lead measurements and the predicted geometric means are shown in Figure 12. The final
model in equation form was:
= H + Is + Iu + Isu + !M + e .
62
-------
Table 16 Analysis of variance results: Cleaner tests comparing technicians
Dependent (response)
variable
Number of observations
Root mean square error
Mean response
Log-transformed residual wipe lead
16
0.32796 (corresponding to a coefficient of variation in
untransformed unit of 34 percent)
the
-4.12575 (corresponding to a geometric mean in the
untransformed unit of 1 .62 percent)
Factor
Substrate
Soil
Substrate* soil
Technician
Degrees of
freedom
3
1
3
1
Sum of
squares
1.28272
4.04077
1.64666
3.34108
Mean
square
0.42758
4.04077
0.54889
3.34108
F ratio
3.9753
37.5684
5.1032
31.0632
Prob>F
0.0604
0.0005
0.0350
0.0008
Whole-model analysis of variance test
Source
Model
Error
Corrected total
Degrees of
freedom
8
7
15
Sum of
squares
Mean
square
10.3113 1.2889
0.7529 0.1076
11.0642
F ratio Prob>F
11.983 <.0001
Least square means with 95 percent confidence intervals in untransformed units
Factor
Substrate* Soil
Technician
Level
Enamel on birch, dry
Enamel on birch, oily
Lacquer on oak, dry
Lacquer on oak, oily
Latex on birch, dry
Latex on birch, oily
Vinyl tile, dry
Vinyl tile, oily
1
2
Geometric mean (95 percent confidence interval)
0.66% (0.38% to 1.1 3%)
4.97% (2.87% to 8.59%)
0.60% (0.35% to 1.04%)
1.64% (0.95% to 2.83%)
1.37% (0.79% to 2.37%)
2.79% (1.61% to 4.82%)
1.69% (0.98% to 2.92%)
2.24% (1.30% to 3.88%)
2.55% (1.94% to 3.36%)
1.02% (0.78% to 1.35%)
63
-------
2-Oily
1-Oily
2-Dry
1-Dry
2-Oily
0)
Q.
£• 1-Oily
^™
o
w 2-Dry
^3
c
* 1-Dry
o
O
0
o
= 2-Oily
| 1-Oily
| 2-Dry
*~ 1-Dry
2-Oily
1-Oily
2-Dry
1-Dry
o
0
|
O
3
O
O
4
o
I
c
.c
o
5
c:
o
OJ
£
03
0
*
4
h
0
•s
o
c
O
(U
§•
JS
"I
Id
o
"flj
03
C
LU
.003 .005
.01 .02
Residual wipe lead
o Measurement
.04 .06 .08
• Predicted
1
Figure 12 Residual wipe lead measurements comparing technicians, by soil type and
substrate, using cleaner 8
64
-------
The average residual wipe lead for the cleaner test conducted by Technician 1 was 2.5 times
higher than for the test conducted by Technician 2. The protocol for cleaning the coupon
controlled the pressure on the sponge. The effort used to clean the coupons was, therefore,
somewhat controlled, but the effort used to wipe the coupons was not specified. As a result,
variation in the residual wipe lead between technicians is suspected to be due to differences in
the wiping effort.
8.6.2 Measurement bias due to contamination
The calculation of residual wipe lead implicitly assumes that all lead in the baby wipe conies
from the soil mixture. The blank wipes (QC samples included in a preparation batch before
preparation of the samples) provided information on contamination in the preparation step or in
the wipes. An additional set of tests was performed to evaluate the magnitude of other sources of
lead. The tests used blank soil; that is, a soil mixture prepared without the leaded soil
component. The coupons were cleaned with a cleaner and subsequently wiped with a baby wipe.
The wipe lead measurements from these coupon tests, referred to as blank soil residual wipes,
provided information on contamination in all aspects of the testing.
The average lead measurement found in the blank wipes was small (0.488 |ag) and often below
the detection limit, indicating little contamination associated with baby wipes or sample
preparation. The blank soil residual wipes had a higher average lead (1.56 |ng) than the blank
wipes. Both the blank wipes and the blank soil residual wipes had a lower average lead
measurements than the test wipes from the cleaner tests (31.99 ^g). The average lead in the
blank soil residual wipes was about one-twentieth of that in the cleaner test wipes. Figure 13 is a
histogram of the lead measurements for blank wipes, blank soil residual wipes, and test wipes
from the cleaner tests.
The blank soil tests used only two cleaners, cleaner 8 and cleaner 19. Inferences about possible
contamination from the other 32 cleaners cannot be made from the blank soil residual wipes.
Since the source of contamination may have been related to the cleaner and since the level of
contamination appears to be low, no blank correction was used in any of the statistical analyses.
8.6.3 Measurement error components of variance
Multiple factors contribute to measurement error in the analyses. This section discusses the
components of variance. The analysis of variance components is not rigorous. The purpose of
the analysis is to list the variance components and to provide information on the relative
magnitudes of the variance components as they affect the analysis of the cleaner tests. The
results are discussed in terms of the coefficient of variation, rather than relative variance or
variance, because the coefficient of variation is likely to be more familiar to many readers.
65
-------
200 -I
180-
w
(0
m
"o
^
£ 160-
0)
(0
S 140 •
•O
S 12°-
.1
To
| 100-
'o
w
^ 80-
ra
5
c
«
A
0 40-
1
3
z
20-
0.
•
•
•
DTest wipe from cleaner test
i] Blank wipe
• Blank soil residual wipe
•
•
•
•
'I
&
\
\ I if)
UiBliiilliliJiHiJf
1
H n ,
coocoiooro
OT-T-CM-STCO
C> CD C) O Ci O
(omocoocomoco
co''-
Wipe lead (^g) (classes defined on a log scale and labeled by their midpoint)
Figure 13 Distribution of test wipe, blank wipe, and blank soil residual wipe measurements
66
-------
The following factors were identified as contributing to the measurement error in the relationship
between average residual wipe lead and cleaner characteristics:
• Measurement error of the ICP or GFAA instrument
• Drift of the ICP or GFAA instrument through the instrument batch
• Differences among instrument batches associated with calibration differences
• Variation among preparation batches
• Variation among coupon tests using the same cleaner
• Variation among cleaner tests
• Variation among cleaners with the same pH, surface tension, phosphate content, and
surfactant type.
QC samples, which were evaluated several times within each instrument batch, provided an
estimate of the variation of lead measurements due to both instrument error and drift. Figure 14
shows the standard deviations of the calibration blank, detection limit, interference check, and
mid calibration verification samples as a function of the known lead concentration in these
samples for the ICP and GFAA instrument batches. The plots show the observed values and a
nonlinear fit that assumes that the variance of the measurements can be expressed as the sum of
two independent terms, one with constant standard deviation and one with constant coefficient of
variation. The nonlinear model fits the ICP data quite well. The model does not fit the GFAA
data very well; however, the GFAA data are less precise because they are based on measure-
ments in only three instrument batches. The instrument response is shown on a log scale.
Figure 15 shows the predicted coefficient of variation of the instrument measurements
superimposed on a histogram of the instrument response for the test wipes from the cleaner tests.
The predicted coefficients of variation are based on the nonlinear fits. The dark portion of the
histogram indicates measurements using GFAA analysis.
Over the range of the GFAA measurements, the coefficient of variation is quite small, about 3
percent. Over the range of the ICP measurements, the coefficient of variation increases from
about 3 percent for the highest measurements to nearly 100 percent for the lowest measurements.
The majority of ICP measurements have a coefficient of variation of about 10 percent. Since for
most measurements, the instrument error is smaller than that due to other sources of variation, no
attempt was made to weight the statistical analysis to reflect the greater variation in some
measurements.
67
-------
c
o
3
M
re
0)
E
o
o
B_
.Is
0)
T3
re
•o
c
s
(0
1
4)
£
3
M
re
o>
O
a
•5
c
o
45
.s
O
•o
"S
re
•o
i
0.1 -i
0.09-
0.08-
0.07-
^0.06-
^
1 0-05-
^^ n 04 •
0.03-
0.02-
0.01 •
0 .
ICP
f
m
• c.
t O
^ ' 'ro
o » o
^^ (y _ tf^*
C J= 'C
ro o a)
& Q) .' >
c ° c
o • o
II ''' 1
5 a3 • =
ro c « ' S
o - ' V
.-••*' ^
'•§ 'E
& •— • Measurement
0)
Q - - - Nonlinear fit
-10123456
True instrument response (tag/ml)
.00050 -I
.00045 •
.00040 •
.00035 •
.00030 •
^
•§,.00025.
E
~ .00020 •
.00015-
.00010-
.00005 •
onnoo .
GFAA
g
j
• / 1
O , C
'<£5 *^ Q
^ CH ^ *^y
5J •— ' W
fl^ ^ O
^ ro
^ o
..••'* ^
#
•^
JO "^
jo ro
^ -o • Measurement
O
- - - Nonlinear fit
-0.005 0 0.005 0.01 0.015 0.02 0.025
True instrument response (mg/ml)
Figure 14 Standard deviation of instrument response for the ICP and GFAA QC samples
68
-------
The blank wipe and spiked wipe samples provide an estimate of the combined effect of
instrument variation and variation among preparation batches. With only one blank wipe and
one spiked wipe per preparation batch, the variance of the measurements is roughly equal to the
sum of the variance associated with the preparation step and the variance associated with the
instrument measurement. The mean lead measurement on blank wipes is 0.49 ug, with a
standard deviation of 2.5 ug. The standard deviation for instrument variation, estimated to be
about 2.7 ug, is of similar magnitude. Thus, the variation contributed by variation in the
preparation step appears to be small for samples with little or no lead. Note that the quantity of
lead in micrograms is 100 times the instrument response shown in Figure 15.
For the spiked wipes (with spiked lead concentrations of 100 ug), the measurement mean and
standard deviation are 97.2 ug and 8.6 ug, respectively. The coefficient of variation of the
spiked wipes is 9 percent (8.6/97.2). This variation is considerably larger than the roughly 3
percent variation associated with the instrument measurement. Therefore, the variance
contributed by variation in the preparation step is about eight times larger than that from the
measurement for the components with constant coefficient of variation. The coefficient of
variation among measurements due to variation in the measurement step and the preparation step
is about 9 percent for high measurements, increasing to larger values for the lower measure-
ments.
The coefficient of variation among coupon tests is about 45 percent (see Table 9). As compared
to a coefficient of variation of 9 percent for the measurement process, most of the variation
among coupon test measurements is associated with the coupon-testing process (applying soil,
cleaning with a sponge and wiping) rather than the preparation and lead measurement process.
The geometric mean residual wipe lead for cleaner tests was based on the average of 10 coupon
tests in replicate 1 and 8 coupon tests in replicates 2 and 3. If all of the variation in the geometric
mean for a cleaner test was due only to variation among coupons tests, the coefficient of
variation among cleaner tests would be about 45 percent/V8 - 16 percent. However, the
coefficient of variation among cleaner tests is 35 percent (see Table 11). Thus, differences
among independent mixtures of the same cleaner contribute to the variation associated with
differences among coupon tests.
Finally, the error, when predicting the overall geometric mean residual wipe lead for cleaners
using pH, phosphate content, surface tension, and surfactant type, depends on the precision of the
overall means for each cleaner and the additional differences among cleaners with similar cleaner
characteristics. If all of the variation in the overall geometric mean cleaner test replications is
due only to variation among cleaner tests, the coefficient of variation around a regression
relationship predicting cleaning efficacy from cleaner characteristics would be about 35
percent/V3 = 20 percent. Since the observed coefficient of variation around the regression
relationship is 20 percent (see Table 12), cleaning efficacy differences among cleaners with
similar cleaner characteristics apparently do not contribute to the lack of precision when
identifying which cleaner characteristics are associated with high versus low residual wipe lead.
69
-------
200
100%
+ 0%
Instrument response (ng/mL)
(Class defined on a log scale and labeled by their midpoint)
Figure 15 ICP and GFAA instrument response for test wipes and associated measurement
precision
70
-------
8.7 Discussion of the Statistical Results
Although this study was undertaken to evaluate the efficacy of cleaners that could be used to
remove lead-contaminated dust from residential surfaces, it should be kept in mind that the
results apply only to the conditions used in this study. In particular, the conclusions apply only
to the soil types and substrates tested and the lead loadings used. The extent to which the
conclusions apply to other soil types and other lead loadings has yet to be determined. For
example, the soil material used in this study was mixed in a solvent slurry, wiped across the
coupon, and dried onto the coupon. Using this application method, the soil might be more
closely bound to the coupon than would loose soil or fallen dust in a home. At the same time,
this soil was not ground into the surface as might happen as a result of foot traffic within a home.
Based on the cleaning wipe tests, one wipe removed on average 79% of the lead. The second
wipe removed an additional 22% of the lead originally applied to the coupons for a total of 91%
removed using two wipes. Because the test wipes removed only 2% of the lead applied to the
coupon, the data suggest that the sponge and cleaner removed more lead than did one wipe, that
is, more than 79% of the lead.7 Although the study provides no measure of how much more, the
fact that the second cleaning wipe removed ten times as much more lead as the test wipe suggests
that a sponge and cleaner may remove much more lead than does one wipe. Since the test wipes
removed 2% of the lead, the average percentage of lead removed by a sponge and cleaner
solution was between 79% and 98%, and probably toward the upper end of that range.
Given that the cleaning agents are expected to improve cleaning efficacy over that for water
alone, one might guess that a sponge and cleaning agent would remove more lead than do two
wipes (which do not contain the cleaning agent). Assuming that a sponge and water or a sponge
and cleaning agent remove at least as much lead as do two wipes, then most (on average 91% or
more) of the lead may be removed using any of the cleaners, including water. The cleaning
method chosen for this study was judged to reasonably reflect what a homeowner might do. One
could therefore conclude that for lead-containing soil material similar to that used in this study,
the use of a sponge, water, and a cleaning agent is likely to remove most of the lead. Although
this study did not test different cleaning methods, if a thorough cleaning method is defined as any
method that removes as much lead as does the method used in this study, one might conclude
that thorough cleaning with water and any cleaning agent will remove most of the lead.
A comparison of the replicate cleaner tests suggests that differences among cleaners are small.
The regressions using cleaner characteristics to predict overall residual wipe lead suggest that
there are marginally significant differences between cleaners and that the differences are most
closely associated with differences in surface tension. These results provide some (although not
conclusive) support for recommending the use of lower surface tension cleaners. Of the cleaners
For the cleaner tests, the coupons were dried after cleaning and before being wiped. For the cleaning wipe tests,
the coupons were not dried between wipes. The analysis above assumes that drying the coupons does not
substantially affect the quantity of lead removed by the wipe.
71
-------
tested, water of average hardness had the highest surface tension, suggesting that use of a
cleaning agent is expected to remove more lead than using water alone.
Although some cleaners may be preferable to others, the study results do not provide a confident
basis by which to select the best cleaners. In particular, the use of several low-phosphate
cleaners resulted in overall residual wipe lead measurements similar to or lower than those
obtained using TSP. This study provides no support for recommending TSP. At the same time,
there is little support for recommending other cleaners over TSP based on the overall residual
wipe lead results.
Although the intended use of a cleaner (such as a floor cleaner or dishwashing detergent) is
somewhat related to its surface tension, the intended use of the cleaner does not significantly
predict residual wipe lead. Therefore, no recommendation can be made about how a consumer
might select a cleaner with superior ability in removing lead.
The difference in the residual wipe lead results between laboratory technicians was large,
suggesting that cleaning effort, cleaning method, or other factors may be more important than the
choice of cleaner and its associated cleaner characteristics.
The conclusions from this study may, depending on the use of the results, be limited by the
following facts:
• The lead loading on the coupons was somewhat higher than is found in most homes.
• The soil material and deposition method may not simulate the soil or dust found in
homes.
• Results may depend on cleaning effort and method; only one cleaning method was used
in this study.
• This study measured residual wipe lead; for some uses, other cleaning efficacy measures
may also be of interest.
72
-------
Glossary of Terms Used in this Study
Blank wipe: Baby wipes used to measure contamination during laboratory preparation and lead
analysis. These baby wipes were analyzed for lead but not used to wipe a coupon. The blank
wipes may also be called method blanks.
Cleaner: Water of average hardness, TSP, or a commercially available cleaning agent tested in
this study.
Cleaner Characteristic: Physical and chemical characteristics of cleaners, including pH,
phosphate content, surface tension, and surfactant type. For the selection of cleaners, the
cleaners were characterized as having pH above or below 10, phosphate content above or below
0.25 g P/gal., and of the anionic or non-ionic surfactant type. See pH, phosphate content, and
surfactant type.
Cleaner Mixture: A mixture of cleaner and water mixed according to the manufacturer's
directions. The cleaner mixture was used to clean coupons that had been soiled with a lead-
containing soil mixture.
Cleaner Test: A test to measure cleaning efficacy of a cleaner. A cleaner test is the set of all
coupon tests using the same cleaner mixture.
Cleaning Agent: A commercially available cleaner, such as dishwashing detergent or laundry
detergent.
Cleaning Efficacy: A measure of the cleaner's ability to remove lead from the surface.
Cleaning efficacy is measured by the quantity of lead removed from the surface of a substrate
using a baby wipe after having cleaned the surface with a cleaner. Higher cleaning efficacy
values indicate better performance in removing lead from the coupon surface.
Cleaning Wipe: A baby wipe used in place of a cleaner to clean a coupon. Cleaning wipes
were used to determine how much lead could be removed from a coupon using only a baby wipe,
that is, without using any cleaner.
Coupon: One-foot-square surface that simulates surfaces found in homes. The coupons were
used to perform laboratory tests with individual cleaners.
Coupon Test: A test of one cleaner on one coupon. These tests consisted of cleaning the
coupon with the selected cleaner, wiping the coupons with baby wipes, and analyzing the baby
wipes for lead.
Final Extract: The sample solution used for lead analysis after digestion, filtration, and
extraction of lead from the baby wipe sample.
73
-------
Instrument Batch: A group of final extracts that were processed by a lead-detecting instrument
(ICP or GFAA) at one time.
pH: A measure of the acidity or alkalinity of a cleaner mixture. Cleaners were categorized as
having pH greater than or less than 10 for the selection of cleaners.
Phosphate Content: The amount of phosphate in a cleaner, measured as grams of phosphorus
per gallon of cleaner mixture.
Preparation Batch: A group of baby wipes from different coupon tests that were all prepared
for lead analysis at one time in the laboratory.
Replication: One of a set of cleaner tests or soil sample and rod rinse tests. Replicate tests
involve independent replication of all testing steps. In this study, tests on each cleaner were
replicated three times.
Residual Wipe Lead: The quantity of lead picked up by the test wipe expressed as a proportion
of the quantity of lead on the coupon before cleaning. Low residual wipe lead measurements
indicate greater cleaning efficacy of the cleaner, since less lead was left on the coupon after
cleaning. The average of the log-transformed residual wipe lead measurements across all coupon
tests performed using the same cleaner mixture was used as a measure of cleaning efficacy for
that cleaner.
Rod-Rinse Measurement: A measurement of the quantity of lead on the applicator rod after it
was used to spread the soil on the coupon.
Rod-Rinse Sample: A mixture of a solvent and the soil left on the applicator rod after it was
used to spread the soil on the coupon.
Soil Batch: A pair of soil mixtures prepared at the same time, one with vegetable oil and one
without.
Soil Mixture: The contents of a particular beaker containing a mixture of lead-containing soil, a
solvent, and other ingredients.
Soil Sample Measurement: The amount of lead found in a sample of a soil mixture.
Soil Sample: A sample of a soil mixture.
Soil Type: The type of laboratory-simulated soil used in this study. Soil types are oily and dry.
Substrate: The material from which a coupon is constructed. Substrates used in this study were
vinyl tile, latex on drywall, enamel on birch, latex on oak, and lacquer (Fabulon) on birch.
Surface: The outermost layer of the coupon that was cleaned. See substrate.
74
-------
Surface Tension: A property of liquids, such as water or cleaner mixtures, arising from
unbalanced cohesive forces at the surface, as a result of which the surface has properties
resembling those of an elastic membrane. A drop of a liquid with low surface tension will wet
the surface and tend to spread out over the clean surface. In contrast, when using a high surface
tension liquid, the drop will not wet the surface and consequently will not spread out and may
retain its spherical (drop like) shape.
Surfactant Type: The type of surface-active substance utilized in the cleaner. In this study,
surfactant types were classified as anionic or non-ionic. Cleaners in which neither or both
anionic and non-ionic surfactants were present were classified as other/unknown.
Tap water of average hardness: Deionized water with added chemicals to simulate tap water of
average hardness [American Society for Testing Materials (ASTM) 3050 synthetic hard water
adjusted to 150 ppm hardness as calcium carbonate].
Test Wipe: A baby wipe used to test the amount of lead left on the coupon after it had been
cleaned with a cleaner.
Wipe Lead: The amount of lead picked up by a baby wipe after the surface of the coupon had
been cleaned.
75
-------
References
1. Barnett, V., and Lewis, T. (1994). Outliers in Statistical Data, Third edition. New York: John
Wiley & Sons.
2, U.S. Department of Housing and Urban Development (HUD). June 1995. Guidelines for the
Evaluation and Control of Lead-Based Hazards in Housing. Washington DC.
3. USEPA. July 12, 1995. Quality Assurance Project Plan for Pb-Cleaning Efficacy for Lead
Abatement in Housing, Revision No. 4. Prepared under contract by Midwest Research
Institute, Kansas City, MO.
4. USEPA. September 1996. Dataset Documentation for the Laboratory Study of Lead-
Cleaning Efficacy. Prepared under contract by Westat Inc. Rockville MD.
76
-------
Appendix A: Test Sequence and Schedule
The cleaner tests were replicated three times. The second replication for all cleaners followed the
first replication for all cleaners. In a similar manner, the third replication followed the second
replication. Within each replication, each cleaner was mixed and tested once. The sequence in
which the 34 cleaners were tested was randomized within each replication to ensure that testing
order did not have an effect on the statistical analysis results.
The cleaning wipe tests were performed to measure the quantity of lead on the coupons before
the cleaning step. These tests were performed using all combinations of substrate and soil type.
The design called for five replications of each combination of substrate and soil type (50 tests).
Three combinations were actually replicated six times.
Table Al shows the test sequence and schedule for the cleaners tested within each replication.
The three pages of table Al correspond to the first, second, and third replication of cleaner tests,
respectively. The title indicates the cleaner replication number. The cleaner number column
indicates the cleaner tested and the corresponding cleaner test sequence column indicates the
order in which the cleaner was tested. The comments column provides additional information
about the purpose of the cleaner test (for example, if the cleaner was used to test for technician
differences). Also included in the third column of Table Al are the order, the substrate, soil, and
replication number of the soil sample, rod-rinse, and cleaning wipe tests. The date that the
coupon was soiled is also indicated.
Figure Al shows the dates on which the coupons for each cleaner within each replication were
soiled and wiped. The figure also shows when the wipes were prepared for lead analysis in the
laboratory and when the final extracts were analyzed for lead. The numbers along the horizontal
axis show the cleaner number, staggered to eliminate over-printing. In some cases wipes from
the same cleaner were prepared in different preparation batches or analyzed in different
instrument batches. Samples that were below the ICP instrument detection limit were re-
analyzed using GFAA. All laboratory work was completed between June and September 1995.
Figure Al also shows the preparation date of the batch from which the soil was taken for the
cleaner test. Although this information is not shown in the figure, all coupons were soiled and
cleaned in the few days between soil batch preparation and coupon wiping.
77
-------
Table Al Test sequence and schedule
Replication 1
Cleaner test
sequence
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Cleaner
number
1
2
8
18
11
32
29
13
21
3
14
33
31
25
12
28
7
17
15
23
34
5
30
4
16
10
6
19
9
22
26
24
20
27
Substrate, soil, and replicate for
soil-sample, rod-rinse, and
cleaning wipe tests
Vinyl tile, Oily 1
Latex on birch, Dry 1
Lacquer (Fabulon) on oak, Dry 1
Latex on drywall, Oily 1
Lacquer (Fabulon) on Oak, Dry 2
Vinyl tile, Dry 1
Latex on birch, Oily 1
Latex on drywall, Dry 1
Lacquer (Fabulon) on Oak, Oily 1
Latex on birch, Oily 2
Enamel on birch, Dry 1
Vinyl tile, Oily 2
Vinyl tile, Dry 2
Enamel on birch, Oily 1
Enamel on birch, Dry 2
Enamel on birch, Oily 2
Lacquer (Fabulon) on Oak, Oily 2
Date coupon
soiled
6/5
6/5
6/7
6/7
6/7
6/7
6/8
6/8
6/9
6/9
6/9
6/11
6/11
6/11
6/11
6/12
6/12
6/12
6/13
6/13
6/13
6/13
6/14
6/14
6/14
6/15
6/15
6/15
6/16
6/16
6/16
6/19
6/19
6/19
Comment
78
-------
Table Al Test sequence and schedule (Continued)
Replication 2
Cleaner test
sequence
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Cleaner
number
29
13
18
11
21
32
14
3
10
4
16
28
7
12
15
23
17
5
30
35
8
1
2
19
6
22
26
9
20
27
24
25
33
31
Substrate, soil, and replicate for
soil-sample, rod-rinse, and
cleaning wipe tests
Latex on birch, Oily 3
Latex on drywall, Oily 2
Latex on drywall, Dry 2
Enamel on birch, Dry 3
Latex on drywall, Oily 3
Latex on birch, Dry 2
Vinyl tile, Oily 3
Vinyl tile, Dry 3
Lacquer (Fabulon) on Oak, Dry 3
Latex on birch, Dry 3
Lacquer (Fabulon) on Oak, Oily 3
Enamel on birch, Oily 3
Latex on drywall, Dry 3
Latex on birch, Dry 4
Lacquer (Fabulon) on Oak, Dry 4
Vinyl tile, Oily 4
Enamel on birch, Oily 4
Date coupon
soiled
6/20
6/20
6/20
6/21
6/21
6/21
6/21
6/22
6/22
6/22
6/22
6/23
6/23
6/23
6/23
6/25
6/25
6/25
6/25
6/26
6/26
6/26
6/26
6/27
6/27
6/27
6/27
6/28
6/28
6/28
6/29
6/29
6/29
6/29
Comment
Hydrated TSP1
79
-------
Table Al Test sequence and schedule (Continued)
Replication 3,4, and QC tests
Test
sequence
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
Cleaner
number
31
25
14
3
32
13
18
29
11
21
33
23
10
4
30
34
5
7
12
28
16
17
15
26
9
22
27
24
20
6
19
19
2
8
1
34
8
8
8
7
12
Substrate, soil, and replicate
?or soil, rod-rinse, and
cleaning wipe tests
Latex on birch, Oily 4
Latex on drywall, Dry 4
Latex on drywall, Oily 4
Vinyl tile, Dry 4
Enamel on birch, Dry 4
Lacquer on Oak, Oily 4
Enamel on birch, Dry 5
Enamel on birch, Oily 6
Latex on birch, Dry 5
Latex on birch, Oily 6
Lacquer on Oak, Dry 5
Latex on drywall, Oily 5
Vinyl tile, Dry 5
Vinyl tile, Oily 5
Latex on birch, Oily 5
Enamel on birch, Oily 5
Latex on drywall, Dry 5
Lacquer on Oak, Oily 5
Enamel on birch, Dry IX
Date coupon
soiled
7/5
7/5
7/5
7/5
7/6
7/6
7/6
7/6
111
7/7
7/7
111
7/9
7/9
7/9
7/9
7/10
7/10
7/10
7/10
7/11
7/11
7/11
7/11
7/12
7/12
7/12
7/12
7/13
7/13
7/13
7/13
7/17
7/17
7/17
7/17
7/18
7/18
7/18
7/18
7/18
Comment
QC Blank— no lead soil2
Replaced cleaner 353 — Rep 4
QC Blank— no lead soil2
Technician - 1
Technician - 2s
Rerun of lost sample
Rerun of lost sample6
80
-------
Table Al Test sequence and schedule (Continued) -
Footnotes
1 Test used hydrated rather than anhydrous TSP, renumbered as cleaner 35.
2 Soil made without SRM 2710 component.
3 Replaces test on cleaner 35, replication 2, performed with the incorrect form of TSP.
Soiling, cleaning, and wipe sampling procedures were done by Technician 1.
Soiling, cleaning, and wipe sampling procedures were done by Technician 2.
Samples were reproduced because of losses during digestion.
81
-------
8: \
61
OZ
LZ
51
Zl
t>2
22
92
LI
82
oe
01
«
ee
C2
\,Z
81
62 I-
ee
02
92
et.
E
92
te
S2
Z2
oe
91.
01
s
E2
21
92
e
2e
62
£1
LZ
92
frZ
61
91
oe
91
Zl
9
62
U
BZ
SZ
ee
e
ei.
2£
91
z
0)
a
£
I
6 «
22 0)
6t £
o
ge '£
0)
•g
o
a>
c
zz i-
a>
ra
a>
•a
c
es
«
a
O.
'^
•s
XI
_w
"w
es
•a
b.
C
"w
s*.
o
a1
in
O)
I
9?
oo
5
in
o5
in
o>
CNI
oo
m
oo
m
CN
F^
m
05
m
CD
CN
CD
m
O)
CD
m
o>
55
CN
m
8JD
82
-------
Appendix B: Details of the Laboratory Procedures
Manufacture of substrate coupons
Five simulated home interior surfaces were selected for this study. The five surfaces were vinyl
tile, enamel-coated plywood, latex-painted plywood, lacquer-coated oak plywood, and latex-
painted drywall. All surfaces were approximately 12-in square.
White vinyl tiles were purchased to represent flooring often found in homes. The vinyl tiles
were V8 x 12 x 12 in. No addit
tile surfaces prior to the study.
were V8 x 12 x 12 in. No additional cleaning or coating procedures were performed on the vinyl
Cabinet-grade birch plywood sheets were purchased and cut into pieces approximately V4 x 12 x
12 in. The grain axis direction was marked on the back of each square of plywood. The front of
each piece was first coated with an oil-base enamel primer that was allowed to dry. All pieces
were then painted with a white oil-base enamel overcoat. The enamel-painted surfaces were
selected to represent interior wood surfaces in homes, such as doors, windows, and trim pieces.
Cabinet-grade birch plywood was purchased and cut, and the grain axis was marked as described
above. The front of each piece was coated with flat white latex paint. After drying, each piece
was again painted with the same paint. The latex-painted surfaces were chosen to represent
household surfaces that are difficult to clean due to the roughness of flat latex paint.
Cabinet-grade oak plywood was purchased and cut into V4 x 12 x 12 in pieces. The grain axis
direction was marked on the back of each piece. The front of each piece was sealed by spray
painting it with Fabulon clear lacquer coating, previously diluted with lacquer thinner (10 to 15
percent). After adequate drying, the front of each piece was overcoated with unthinned Fabulon.
The Fabulon coated oak substrate was chosen to represent hardwood floors in homes.
Sheets of drywall wallboard were purchased and cut into 12-in squares, and the length axis was
marked on the back of each piece. The front of each piece was painted with flat white latex paint
and allowed to dry overnight. The painted face was then lightly sanded to remove nap that might
have been raised during the painting. A second coat of the same paint was applied and allowed
to dry at ambient conditions.
Each substrate coupon was uniquely identified as to substrate type, cleaning agent, soil type, and
test replication number prior to the application of the soil material.
Preparation of soil materials
Two synthetic soil materials, both based on a soil preparation presented in ASTM D4488, were
selected for this study. Both soil materials contained NIST SRM 2710 (5,532 ^ig Pb/g of
material) as the insoluble lead dust component of the soil materials. Both soil materials also
83
-------
contained activated carbon black powder, which is difficult to clean from surfaces, and mineral
spirits, a volatile liquid carrier.
One soil material containing only the SRM 2710, carbon black, and mineral spirits was
designated as a "dry" soil. The second soil material also contained vegetable oil and was
designated as an "oily" soil.
The dry soil was selected to represent lead dust recently deposited on home interior surfaces.
The oily soil was selected to represent lead dust that might have resided on home interior
surfaces for an extended period of time and subsequently might have become contaminated with
oils from cooking and human contact.
Application of soil materials
The ingredients of both soil materials were suspended in mineral spirits, so that a pipette could
be used to measure the soil volumetrically and transfer it to the substrate coupon. The liquid soil
was then distributed as uniformly as possible over the surface of the coupon using a wire-wound
rod as discussed below.
The coupons prepared with latex paint were soiled with 4 ml of soil material. The other coupons
were soiled with 2 ml of soil material.
The applicator rod consisted of 0.010-in diameter stainless steel wire wrapped around a 3/8 -in
diameter stainless steel rod. The rod was wrapped with wire over a 13-in length. The rod was
repeatedly moved back and forth across the face of the substrate coupon to spread the liquid soil
material as uniformly as possible over an area of approximately 10 x 10-in square. The rod was
not allowed to roll on the substrate coupon surface, to minimize the amount of soil retained on
the applicator rod. The applicator rod was always drawn parallel to the grain direction of the
substrate coupon.
The substrate coupons were dried for 16 to 17hrat55°±5°Cina forced-air oven to remove the
volatile mineral spirits (solvent) carrier. The substrate coupons were dried in a horizontal
position to prevent the flow of any liquid soil during drying.
Cleaning agent preparation
Cleaning agent mixtures were prepared fresh each day of testing. Cleaning agents were added to
synthetic tap water of average hardness, 150 ppm as calcium carbonate, per ASTM D3050. The
concentration of detergent cleaning agent to be used was calculated from the midpoint of the
manufacturer's recommended use range.
A leading detergent manufacturer was contacted to determine the average water volume used
during household cleaning operations. Laundry washing machines average 17 gal of water
during the wash cycle. Machine dishwashers average 9 gal of water during the wash cycle and
use 2 to 2.5 tablespoons of detergent cleaner. Average hand dishwashing operations use 2 to
3 gal of water.
84
-------
The pH and surface tension of each freshly prepared cleaning agent mixture were measured and
reported. The phosphate content of each cleaning agent was calculated from the phosphate
content reported by the manufacturer and was calculated as grams of phosphorus per gallon of
cleaning agent mixture (g P/gal).
Cleaning of substrate coupons
The soiled and dried substrate coupons were cleaned according to a protocol developed
specifically for this research program. A new cellulose sponge was used for each substrate
coupon. The sponge was rinsed for 1 minute in flowing cold tap water and then crushed to
remove as much water as possible.
Ten milliliters of cleaning agent were spread over the surface of a substrate coupon as uniformly
as possible by dispensing it over the soiled area from a 10-mL Mohr pipet. The solution was
allowed to stand for 30 ± 5 seconds. A rinsed, new sponge was placed on a 145-g sponge holder.
The damp sponge and holder were drawn across the substrate coupon parallel to the grain axis
three times, overlapping one-half of the previous traverse on the second and third traverse.
The substrate coupon was then turned 180 degrees, in place, and the same sponge and holder
were drawn across the substrate coupon three additional times, as described above. The same
leading edge of the sponge was used for all traverses across the substrate coupon.
The sponge was rinsed under cold running tap water as described above. The substrate coupon
was rinsed using the same motions as during the cleaning described above. Care was taken that
the only pressure applied to the sponge during both the cleaning and rinse procedures was from
the weight of the sponge holder.
Wipe testing of substrates
The cleaned substrate coupons were dried for 2 hours at 55° ± 5°C in a forced-air oven. One
baby wipe was used to wipe-sample each substrate, according to the Protocol for Wipe Sampling
of Settled Dust (Appendix A of the QAPjP).
Each baby wipe was placed in a new plastic centrifuge tube and the unique number of the
substrate coupon was recorded on the centrifuge tube.
85
-------
Appendix C: Listing of the Data
The following tables present the primary data required for an analysis of lead cleaning efficacy.
Table Cl contains the soil sample and rod-rinse measurements.
Table C2 contains the residual wipe lead measurements for all cleaner test coupons.
Table C3 contains the fraction of lead removed using only wipes.
Table C4 contains the physical characteristics of the cleaner mixtures.
86
-------
Table Cl Soil sample and rod-rinse measurements
Soil
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Substrate
Enamel on birch
Enamel on birch
Enamel on birch
Enamel on birch
Enamel on birch
Fabulon on Oak
Fabulon on Oak
Fabulon on Oak
Fabulon on Oak
Fabulon on Oak
Latex on birch
Latex on birch
Latex on birch
Latex on birch
Latex on birch
Latex on Drywall
Latex on Drywall
Latex on Drywall
Latex on Drywall
Latex on Drywall
Vinyl Tile
Vinyl Tile
Vinyl Tile
Vinyl Tile
Vinyl Tile
Enamel on birch
Enamel on birch
Enamel on birch
Enamel on birch
Enamel on birch
Enamel on birch
Enamel on birch
Fabulon on Oak
Fabulon on Oak
Fabulon on Oak
Fabulon on Oak
Fabulon on Oak
Fabulon on Oak
Latex on birch
Latex on birch
Latex on birch
Latex on birch
Latex on birch
Latex on birch
Latex on Drywall
Latex on Drywall
Latex on Drywall
Latex on Drywall
Latex on Drywall
Vinyl Tile
Vinyl Tile
Vinyl Tile
Vinyl Tile
Vinyl Tile
Replicate
1
2
3
4
5
1
2
3
4
5
1
2
3
4
5
1
2
3
4
5
1
2
3
4
5
1
1
2
3
4
5
6
1
2
3
4
5
7
1
2
3
4
5
6
1
2
3
4
5
1
2
3
4
5
Soil Batch
3835-86
3835-42
3835-46
3835-72
3835-77
3835-25
3835-25
3835-56
3835-62
3835-77
3835-25
3835-52
3835-56
3835-62
3835-77
3835-33
3835-46
3835-62
3835-67
3835-86
3835-33
3835-38
3835-52
3835-72
3835-81
3835-38
3835-42
3835-42
3835-56
3835-67
3835-81
3835-77
3835-33
3835-46
3835-56
3835-72
3835-86
3835-90
3835-33
3835-33
3835-46
3835-67
3835-81
3835-77
3835-25
3835-46
3835-52
3835-72
3835-77
3835-25
3835-38
3835-52
3835-67
3835-81
Soil Sample (ng)
945.3
948.8
241.1 *
901.8
892
1030.8
977
888.9
914.8
904.6
464.5 *
1780.8
1753.6
1742.8
1740
1894.5
1700
1813.6
1827.6
1839.6
924.3
965.6
971.9
954.4
987.2
924.3
943.4
923.4
855.1
862.1
908.7
879.6
789.9
882.5
935.2
845.8
869.9
891.8
1636
1640.4
1668
1757.6
1822.4
1746.4
1878
1723.2
1757.6
1787.2
1793.2
856.7
911
860.9
847.8
917.7
Rod Rinse (ng)
28.8
14.8
50
43.2
18.8
40.7
54.7
20.6
33.9
52.2
70.6
109.9
39.7
66.2
63.4
103.8
83.6
59.5
30.5
23.7
16.3
9.7
35.7
33.8
29.7
23.2
19.4
21.5
33.9
25.9
17.9
21.5
62.4
35.6
22.8
26.2
14.6
14.1
81.4
61
Sample lost
57.6
35.1
63.6
225.2
73.8
50.2
52.5
43.7
65.3
14.9
17.3
18
12.9
These measurements were judged to be outliers and were removed from the statistical analysis.
87
-------
Table C2 Cleaner test data
Cleaner
01
01
01
01
01
01
01
01
01
01
02
02
02
02
02
02
02
02
02
02
03
03
03
03
03
03
03
03
03
03
04
04
04
04
04
04
04
04
04
04
05
05
05
05
05
05
05
05
05
05
06
06
06
06
06
06
06
06
06
06
Rep(mod)
Soil
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Substrate
Fabulon on Oak
Enamel on birch
Latex on Dry wall
Vinyl Tile
Latex on birch
•abulon on Oak
Enamel on birch
^atex on Drywall
Vinyl Tile
^atex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
7abulon on Oak
^atex on birch
Enamel on birch
•abulon on Oak
-atex on birch
Vinyl Tile
Latex on Drywall
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Latex on Drywall
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Wipe Lead (ng)
11.3
2.645
19.8
7.8
24.4
29
22.1
32
16.6
27.7
8.2
39.4
11.9
36.2
34.8
45.9
36.8
36.6
25.9
86.8
21.2
70.6
9.7
22.8
45.8
41.4
54.2
38.6
40.1
48.1
40
8.1
4.4
37.4
32.4
12.9
36.6
147.1
94.8
42.7
40.4
48.8
32.3
49.4
48.6
90.4
130.5
84.1
30.1
105.4
7.6
43.7
4.245
15.9
47.8
24.2
22.7
38.8
23.1
101.4
Residual wipe lead
0.0118
0.0028
0.0104
0.0082
0.0151
0.0359
0.0273
0.0198
0.0205
0.0145
0.0086
0.0206
0.0125
0.0379
0.0182
0.0568
0.0455
0.0226
0.0320
0.0537
0.0235
0.0392
0.0108
0.0253
0.0254
0.0542
0.0355
0.0505
0.0525
0.0315
0.0215
0.0087
0.0047
0.0403
0.0175
0.0144
0.0204
0.1642
0.1058
0.0238
0.0435
0.0263
0.0348
0.0532
0.0262
0.1009
0.0728
0.0939'
0.0336
0.0588
0.0081
0.0234
0.0045
0.0170
0.0256
0.0265
0.0124
0.0425
0.0253
0.0555
-------
Table C2 Cleaner test data (continued)
Cleaner
07
07
07
07
07
07
07
07
07
07
08
08
08
08
08
08
08
08
08
08
09
09
09
09
09
09
09
09
09
09
10
10
10
10
10
10
10
10
10
10
11
11
11
11
11
11
11
11
11
11
12
12
12
12
12
12
12
12
12
12
Rep(mod)
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
Soil
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Substrate
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Fabulon on Oak
Vinyl Tile
Latex on Drywall
Latex on birch
Enamel on birch
Latex on birch
Fabulon on Oak
Enamel on birch
Latex on Drywall
Vinyl Tile
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Latex on birch
Latex on Drywall
Vinyl Tile
Fabulon on Oak
Enamel on birch
Latex on Drywall
Vinyl Tile
Fabulon on Oak
Latex on birch
Enamel on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Wipe Lead (ng)
14
52.7
5.3
3.9
32.6
34.1
72.1
28.2
37.4
62.4
11.4
11.1
30.2
65.3
12.38
94.2
54
36
62.4
29.7
16.4
24.2
4.5
20.4
40.6
44.2
62.6
49.2
36.9
54.9
22.1
21.2
8.7
15.9
34
35.1
48.3
17.8
23.8
34.2
35.5
50.4
51.1
54.6
17.8
92.2
35.5
51.3
54.9
42.7
20.8
78.5
21.3
29
57.7
77.9
66
49.4
100.1
88.2
Residual wipe lead
0.0155
0.0292
0.0059
0.0043
0.0181
0.0446
0.0472
0.0369
0.0489
0.0408
0.0119
0.0116
0.0158
0.0342
0.0130
0.0582
0.0668
0.0445
0.0386
0.0367
0.0176
0.0130
0.0048
0.0218
0.0217
0.0484
0.0343
0.0539
0.0404
0.0301
0.0237
0.0113
0.0093
0.0170
0.0182
0.0384
0.0265
0.0195
0.0261
0.0187
0.0186
0.0264
0.0535
0.0571
0.0186
0.0570
0.0439
0.0634
0.0339
0.0528
0.0231
0.0435
0.0236
0.0322
0.0320
0.1019
0.0432
0.0646
0.1310
0.0577
89
-------
Table C2 Cleaner test data (continued)
Cleaner
13
13
13
13
13
13
13
13
13
13
14
14
14
14
14
14
14
14
14
14
15
15
15
15
15
15
15
15
15
15
16
16
16
16
16
16
16
16
16
16
17
17
17
17
17
17
17
17
17
17
18
18
18
18
18
18
18
18
18
18
Rep(mod)
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
Soil
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Substrate
Vinyl Tile
Latex on Drywall
!namel on birch
:abulon on Oak
,atex on birch
Vinyl Tile
,atex on Drywall
inamel on birch
rabulon on Oak
Latex on birch
Vinyl Tile
.atex on Drywall
inamel on birch
rabulon on Oak
Latex on birch
Vinyl Tile
,atex on Drywall
;namel on birch
•abulon on Oak
.atex on birch
Vinyl Tile
.atex on Drywall
inamel on birch
:abulon on Oak
^atex on birch
Enamel on birch
•abulon on Oak
^atex on birch
Vinyl Tile
^atex on Drywall
Vinyl Tile
^atex on Drywall
Enamel on birch
F abulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Fabulon on Oak
Latex on birch
Enamel on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Wipe Lead (ug)
18.2
219.4
23
45.1
110.4
35.8
147.5
109.8
24.8
108.6
15.5
54.6
4.6
24.9
26.2
36.6
27.3
47
59
30.5
12
50.1
11.4
14.3
36.5
79.4
25.6
59
34.7
61.8
15.4
71.4
55.6
38
64.8
25.1
164.5
64.9
107.3
83
15.1
38.6
14.7
25.9
11.3
78.7
71.2
68.5
29.7
200.2
17.8
35.2
6.3
8.1
28.2
28.4
40.8
19.8
67.8
23.7
lesidual wipe lead
0.0190
0.1148
0.0241
0.0472
0.0578
0.0443
0.0912
0.1358
0.0307
0.0671
0.0172
0.0303
0.0051
0.0276
0.0145
0.0479
0.0179
0.0615
0.0772
0.0200
0.0129
0.0270
0.0123
0.0154
0.0197
0.0886
0.0286
0.0329
0.0387
0.0345
0.0166
0.0385
0.0599
0.0409
0.0349
0.0280
0.0918
0.0724
0.1197
0.0463
0.0167
0.0214
0.0163
0.0144
0.0125
0.1030
0.0466
0.0896
0.0389
0.1310
0.0186
0.0184
0.0066
0.0085
0.0148
0.0351
0.0252
0.0245
0.0419
0.0293
90
-------
Table C2 Cleaner test data (continued)
Cleaner
19
19
19
19
19
19
19
19
19
19
20
20
20
20
20
20
20
20
20
20
21
21
21
21
21
21
21
21
21
21
22
22
22
22
22
22
22
22
22
22
23
23
23
23
23
23
23
23
23
23
24
24
24
24
24
24
24
24
24
24
Rep(mod)
1
1
1
1
1
1
1
1
Soil
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Do-
Dry
Oily
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Substrate
Fabulon on Oak
Latex on birch
Enamel on birch
Vinyl Tile
Latex on Dry wall
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Fabulon on Oak
Enamel on birch
Latex on Drywall
Vinyl Tile
Latex on birch
Fabulon on Oak
Enamel on birch
Latex on Drywall
Vinyl Tile
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Fabulon on Oak
Latex on birch
Enamel on birch
Latex on Drywall
Vinyl Tile
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Wipe Lead (mg)
9.2
28.9
3.09
11.3
13.7
48.3
10.6
38.9
17.3
11.1
71
36.9
14.7
49.9
10.7
60.7
53.5
60.8
111.1
42.4
98.3
23.2
52.2
21.5
22.7
54.1
53.1
62.4
29.5
58.6
9.8
34.4
4.3
21.2
16.3
36.7
40.2
52.6
16
65.6
24.9
30.7
2.9
37.9
15.6
58.9
84.5
101.2
33.3
116.2
10
27.9
5.4
15
26
33.7
41.1
46.8
26.9
61.6
Residual wipe lead
0.0099
0.0155
0.0033
0.0121
0.0073
0.0265
0.0116
0.0213
0.0190
0.0122
0.0439
0.0457
0.0182
0.0309
0.0132
0.0363
0.0640
0.0728
0.0665
0.0507
0.0545
0.0257
0.0289
0.0238
0.0252
0.0708
0.0347
0.0816
0.0386
0.0383
0.0105
0.0184
0.0046
0.0227
0.0087
0.0402
0.0220
0.0576
0.0175
0.0359
0.0268
0.0165
0.0031
0.0204
0.0168
0.0657
0.0471
0.1129
0.0372
0.0648
0.0124
0.0173
0.0067
0.0186
0.0161
0.0403
0.0246
0.0560
0.0322
0.0369
91
-------
Table C2 Cleaner test data (continued)
Cleaner
25
25
25
25
25
25
25
25
25
25
26
26
26
26
26
26
26
26
26
26
27
27
27
27
27
27
27
27
27
27
28
28
28
28
28
28
28
28
28
28
29
29
29
29
29
29
29
29
29
29
30
30
30
30
30
30
30
30
30
30
Rep(mod)
Soil
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Do-
Oily
Oily
Oily
Oily
Oily
Substrate
Vinyl Tile
^atex on Dry wall
-abulon on Oak
^atex on birch
Enamel on birch
Vinyl Tile
Latex on Dry wall
Enamel on birch
•abulon on Oak
-atex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
-atex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Latex on birch
Enamel on birch
Latex on Drywall
Vinyl Tile
Fabulon on Oak
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Wipe Lead (ng)
22
75
36.6
97.2
27.1
54.6
54.4
49.8
61
68.3
12.2
43.2
5.2
30.8
22.5
42
54.6
51
21.2
60.8
33.9
6.1
27.5
9.4
12
21.9
55.5
42.5
28.4
95.5
24.6
35.8
16.9
21.6
32.9
32.6
84.9
53
25.2
68
31.9
125.9
24.3
19.5
64.3
30.5
87.5
43.1
49.8
112.9
6.6
31.8
6.4
22.5
50.3
54
57.6
25.6
26.5
70.8
Residual wipe lead
0.0244
0.0416
0.0406
0.0539
0.0301
0.0714
0.0356
0.0652
0.0798
0.0447
0.0131
0.0231
0.0056
0.0330
0.0120
0.0460
0.0299
0.0559
0.0232
0.0333
0.0210
0.0075
0.0170
0.0116
0.0148
0.0262
0.0332
0.0509
0.0340
0.0571
0.0273
0.0199
0.0187
0.0240
0.0182
0.0427
0.0555
0.0693
0.0330
0.0445
0.0334
0.0659
0.0254
0.0204
0.0336
0.0377
0.0541
0.0533
0.0616
0.0698
0.0071
0.0171
0.0069
0.0242
0.0271
0.0603
0.0321
0.0286
0.0296
0.0395
92
-------
Table C2 Cleaner test data (continued)
Cleaner
31
31
31
31
31
31
31
31
31
31
32
32
32
32
32
32
32
32
32
32
33
33
33
33
33
33
33
33
33
33
34
34
34
34
34
34
34
34
34
34
Rep(mod)
1
1
1
1
Soil
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Substrate
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Vinyl Tile
Latex on Drywall
Enamel on birch
Fabulon on Oak
Latex on birch
Wipe Lead (ug)
43.9
87
19.9
13.7
60.6
54.9
73.1
30.5
47.5
95.5
39.6
68.4
9.3
31
45.4
34.7
59.2
53.6
67.4
85.8
38.3
101.9
27.9
24.7
51.6
51.9
98.3
85.6
59.4
103.4
56.2
3.8
8.3
45
7.6
27
49.8
92.1
28.5
42.2
Residual wipe lead
0.0487
0.0482
0.0221
0.0152
0.0336
0.0718
0.0478
0.0399
0.0621
0.0625
0.0414
0.0358
0.0097
0.0324
0.0238
0.0429
0.0366
0.0663
0.0833
0.0530
0.0425
0.0565
0.0309
0.0274
0.0286
0.0679
0.0643
0.1120
0.0777
0.0676
0.0303
0.0041
0.0089
0.0242
0.0082
0.0301
0.0278
0.1028
0.0318
0.0235
93
-------
Table C2 Cleaner test data (continued)
Cleaner
01
01
01
01
01
01
01
01
02
02
02
02
02
02
02
02
03
03
03
03
03
03
03
03
04
04
04
04
04
04
04
04
05
05
05
05
05
05
05
05
06
06
06
06
06
06
06
06
07
07
07
07
07
07
07
07
Rep(mod)
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
Soil
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Substrate
Vinyl Tile
Enamel on birch
•abulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
'abulon on Oak
,atex on birch
Vinyl Tile
Enamel on birch
'abulon on Oak
,atex on birch
Vinyl Tile
•namel on birch
:abulon on Oak
Latex on birch
inamel on birch
Vinyl Tile
,atex on birch
:abulon on Oak
Vinyl Tile
Enamel on birch
,atex on birch
;abulon on Oak
.atex on birch
:abulon on Oak
Enamel on birch
Vinyl Tile
Latex on birch
Fabulon on Oak
Enamel on birch
Vinyl Tile
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Latex on birch
Enamel on birch
Fabulon on Oak
Vinyl Tile
Latex on birch
Fabulon on Oak
Enamel on birch
Vinyl Tile
Latex on birch
Fabulon on Oak
Enamel on birch
Vinyl Tile
Wipe Lead (ng)
8.3
3.9
10.5
26.1
26.2
34.9
24
53.4
6.865
3.55
9.7
13.8
21.1
45.1
22.7
47
3.63
10.3
52.1
7
20.6
29.9
87.2
49.1
13.8
22.9
4.7
3.2
52.7
24.1
15.2
11.2
68.8
19.9
11.6
92.4
57.2
24.7
59.6
103.5
11.7
5.8
10.9
42.9
82.3
45.4
32
35.6
42.2
11.1
7.6
13.2
77
33
31.6
37.2
tesidual wipe lead
0.0096
0.0045
0.0122
0.0151
0.0303
0.0403
0.0277
0.0308
0.0080
0.0041
0.0112
0.0080
0.0244
0.0521
0.0262
0.0271
0.0041
0.0116
0.0295
0.0079
0.0243
0.0352
0.0514
0.0579
0.0078
0.0259
0.0053
0.0036
0.0310
0.0284
0.0179
0.0132
0.0798
0.0231
0.0134
0.0536
0.0661
0.0285
0.0689
0.0598
0.0135
0.0067
0.0126
0.0248
0.0502
0.0554
0.0390
0.0434
0.0239
0.0126
0.0086
0.0149
0.0454
0.0389
0.0372
0.0438
94
-------
Table C2 Cleaner test data (continued)
Cleaner
08
08
08
08
08
08
08
08
09
09
09
09
09
09
09
09
10
10
10
10
10
10
10
10
11
11
11
11
11
11
11
11
12
12
12
12
12
12
12
12
13
13
13
13
13
13
13
13
14
14
14
14
14
14
14
14
Rep(mod)
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
Soil
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Substrate
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Enamel on birch
Vinyl Tile
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Latex on birch
Fabulon on Oak
Enamel on birch
Vinyl Tile
Vinyl Tile
Latex on birch
Enamel on birch
Fabulon on Oak
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Latex on birch
Fabulon on Oak
Enamel on birch
Vinyl Tile
Latex on birch
Fabulon on Oak
Vinyl Tile
Enamel on birch
Vinyl Tile
Latex on birch
Fabulon on Oak
Enamel on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Fabulon on Oak
Vinyl Tile
Enamel on birch
Latex on birch
Enamel on birch
Vinyl Tile
Fabulon on Oak
Latex on birch
Wipe Lead (ng)
10.4
7.5
11.2
16.5
49.6
24.2
19.9
63.7
10.6
6
10.2
21.2
26.3
37.8
19.8
49.9
39
11.7
7.5
13.8
31.2
52.5
14.8
19.1
18.9
8.5
31.4
67
30.6
19.3
15.1
71.4
25.9
7.3
4.2
14.6
79.5
16.4
39.5
23
4.4
17.5
8.2
2.7
7.1
30.7
8.8
15.8
15.7
9.205
6.125
31.3
26.7
24.1
11.7
92.3
Residual wipe lead
0.0121
0.0087
0.0130
0.0096
0.0573
0.0280
0.0230
0.0368
0.0123
0.0069
0.0118
0.0123
0.0321
0.0461
0.0241
0.0304
0.0220
0.0132
0.0085
0.0156
0.0368
0.0309
0.0174
0.0225
0.0234
0.0105
0.0389
0.0415
0.0366
0.0231
0.0181
0.0427
0.0146
0.0083
0.0047
0.0165
0.0468
0.0193
0.0465
0.0271
0.0054
0.0108
0.0101
0.0033
0.0085
0.0367
0.0105
0.0095
0.0194
0.0114
0.0076
0.0194
0.0319
0.0288
0.0140
0.0552
95
-------
Table C2 Cleaner test data (continued)
Cleaner
15
15
15
15
15
15
15
15
16
16
16
16
16
16
16
16
17
17
17
17
17
17
17
17
IS
18
18
18
18
18
18
18
19
19
19
19
19
19
19
19
20
20
20
20
20
20
20
20
21
21
21
21
21
21
21
21
Rep(mod)
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
Soil
Dry
Dry
Dry
Dry
Oily
Oily
Oily
OHy
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Substrate
Latex on birch
Enamel on birch
Vinyl Tile
Fabulon on Oak
Latex on birch
•abulon on Oak
Enamel on birch
Vinyl Tile
.atex on birch
:abulon on Oak
Enamel on birch
Vinyl Tile
Latex on birch
•abulon on Oak
Enamel on birch
Vinyl Tile
^atex on birch
Fabulon on Oak
Enamel on birch
Vinyl Tile
Vinyl Tile
^atex on birch
"abulon on Oak
Enamel on birch
Latex on birch
Fabulon on Oak
Enamel on birch
Vinyl Tile
Latex on birch
Fabulon on Oak
Enamel on birch
Vinyl Tile
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Wipe Lead (ng)
44.8
3.895
10
6.5
86.9
35.3
34.4
37.3
32
9.7
4.9
7
38.5
26.8
31.3
21.1
60.5
12.3
15.7
18.9
25.7
58.1
33.3
27.7
106.7
23.2
10
9.4
64
38.9
7.3
12.3
12.3
7
19.4
36.7
29
40.3
41.8
58.5
8.4
2.825
9.5
22.5
9.8
22.4
29.1
32
8.7
8.8
19.6
51.7
18.8
28.1
11.4
134.5
lesidual wipe lead
0.0253
0.0044
0.0113
0.0073
0.0512
0.0416
0.0405
0.0440
0.0181
0.0110
0.0055
0.0079
0.0227
0.0316
0.0369
0.0249
0.0351
0.0143
0.0182
0.0219
0.0297
0.0336
0.0385
0.0320
0.0660
0.0287
0.0124
0.0116
0.0383
0.0465
0.0087
0.0147
0.0142
0.0081
0.0224
0.0212
0.0354
0.0491
0.0510
0.0357
0.0097
0.0033
0.0110
0.0130
0.0120
0.0273
0.0355
0.0195
0.0108
0.0109
0.0243
0.0320
0.0225
0.0336
0.0136
0.0805
96
-------
Table C2 Cleaner test data (continued)
Cleaner
22
22
22
22
22
22
22
22
23
23
23
23
23
23
23
23
24
24
24
24
24
24
24
24
25
25
25
25
25
25
25
25
26
26
26
26
26
26
26
26
27
27
27
27
27
27
27
27
28
28
28
28
28
28
28
28
Rep(mod)
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
Soil
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Substrate
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Latex on birch
Fabulon on Oak
Enamel on birch
Vinyl Tile
Latex on birch
Fabulon on Oak
Enamel on birch
Vinyl Tile
Fabulon on Oak
Latex on birch
Enamel on birch
Vinyl Tile
Vinyl Tile
Enamel on birch
Latex on birch
Fabulon on Oak
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Fabulon on Oak
Vinyl Tile
Enamel on birch
Latex on birch
Fabulon on Oak
Vinyl Tile
Enamel on birch
Latex on birch
Vinyl Tile
Fabulon on Oak
Enamel on birch
Vinyl Tile
Enamel on birch
Latex on birch
Fabulon on Oak
Wipe Lead (|ig)
7.9
10.2
5.1
26
70.9
28.3
19.4
53.2
38.5
9.5
8.9
14
66.1
17.9
22.1
22.1
5.9
26.7
6.51
6.4
34
31.1
18.4
11.1
9.8
3.1
7.5
23.8
13.3
31.1
57.1
20.9
20.7
11
7.5
6.8
30.3
31.3
37
14.5
22.2
11.46
12.7
9.9
21.5
8.1
24.9
30.8
25.4
7.2
8
3.125
12.1
13.5
33.8
10
Residual wipe lead
0.0091
0.0118
0.0059
0.0150
0.0865
0.0345
0.0237
0.0324
0.0223
0.0110
0.0103
0.0162
0.0382
0.0207
0.0255
0.0255
0.0066
0.0149
0.0072
0.0071
0.0405
0.0371
0.0110
0.0132
0.0109
0.0035
0.0083
0.0132
0.0159
0.0371
0.0681
0.0125
0.0120
0.0127
0.0087
0.0079
0.0185
0.0382
0.0451
0.0177
0.0128
0.0132
0.0147
0.0114
0.0131
0.0099
0.0304
0.0376
0.0144
0.0081
0.0090
0.0035
0.0143
0.0159
0.0199
0.0118
97
-------
Table C2 Cleaner test data (continued)
Cleaner
29
29
29
29
29
29
29
29
30
30
30
30
30
30
30
30
31
31
31
31
31
31
31
31
32
32
32
32
32
32
32
32
33
33
33
33
33
33
33
33
34
34
34
34
34
34
34
34
35
35
35
35
35
35
35
35
Rep(mod)
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
Soil
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Do-
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Substrate
Vinyl Tile
Enamel on birch
Fabulon on Oak
^atex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
;abulon on Oak
^atex on birch
Vinyl Tile
Enamel on birch
-abulon on Oak
Latex on birch
Vinyl Tile
Latex on birch
Fabulon on Oak
Enamel on birch
Vinyl Tile
Fabulon on Oak
Enamel on birch
^atex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Fabulon on Oak
Vinyl Tile
Enamel on birch
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Wipe Lead (ng)
7.155
3.14
8.55
45.3
7.7
35.2
9.125
40
29
6.8
7.595
48.4
20.4
25
14.7
50
10.8
17.5
27.6
4.2
32
26.4
54.2
28.5
7.5
2.4
18
9.9
18.4
25
8.6
109
22.9
11.4
11.5
16.7
42.3
42.7
24.7
59.3
6.5
4.6
4
36.2
9.3
5.1
6.2
27
6.635
6
7
50.3
24.2
23.4
16.5
82.5
Residual wipe lead
0.0089
0.0039
0.0106
0.0280
0.0092
0.0421
0.0109
0.0239
0.0336
0.0079
0.0088
0.0281
0.0236
0.0289
0.0170
0.0289
0.0120
0.0097
0.0307
0.0047
0.0382
0.0315
0.0646
0.0170
0.0093
0.0030
0.0223
0.0061
0.0220
0.0299
0.0103
0.0652
0.0255
0.0127
0.0128
0.0093
0.0504
0.0509
0.0295
0.0354
0.0071
0.0050
0.0044
0.0198
0.0109
0.0060
0.0072
0.0158
0.0077
0.0070
0.0081
0.0292
0.0280
0.0270
0.0191
0.0477
98
-------
Table C2 Cleaner test data (continued)
Cleaner
01
01
01
01
01
01
01
01
02
02
02
02
02
02
02
02
03
03
03
03
03
03
03
03
04
04
04
04
04
04
04
04
05
05
05
05
05
05
05
05
06
06
06
06
06
06
06
06
07
07
07
07
07
07
07
07
Rep(mod)
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
Soil
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Do-
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Substrate
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Latex on birch
Enamel on birch
Fabulon on Oak
Vinyl Tile
Latex on birch
Fabulon on Oak
Enamel on birch
Vinyl Tile
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Fabulon on Oak
Latex on birch
Enamel on birch
Vinyl Tile
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Wipe Lead (ng)
9.3
2.7
6.9
34.2
35.6
36.2
38.9
44.6
6.105
2.835
6.7
9.1
15.3
32
12
34.9
13.9
11.2
20.8
35
33.1
29.5
26.5
63.8
39.4
3.805
18.4
19.8
59.6
77.4
29.7
28.2
23.4
6
17
33.9
55
96.9
20.1
48.6
16.3
4.56
20.8
20.5
48.8
52.4
26
64.4
7.795
23.1
7.2
6.6
47.2
60.7
35.9
24.3
Residual wipe lead
0.0102
0.0030
0.0076
0.0187
0.0416
0.0423
0.0455
0.0261
0.0067
0.0031
0.0073
0.0050
0.0179
0.0374
0.0140
0.0204
0.0155
0.0125
0.0231
0.0195
0.0395
0.0352
0.0316
0.0380
0.0231
0.0045
0.0215
0.0232
0.0347
0.0902
0.0346
0.0329
0.0274
0.0070
0.0199
0.0198
0.0641
0.1129
0.0234
0.0283
0.0179
0.0050
0.0228
0.0112
0.0571
0.0613
0.0304
0.0376
0.0091
0.0135
0.0084
0.0077
0.0550
0.0707
0.0418
0.0142
99
-------
Table C2 Cleaner test data (continued)
Cleaner
08
08
08
08
08
08
08
08
09
09
09
09
09
09
09
09
10
10
10
10
10
10
10
10
11
11
11
11
11
11
11
. 11
12
12
12
12
12
12
12
12
13
13
13
13
13
13
13
13
14
14
14
14
14
14
14
14
Rep(mod)
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
Soil
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Substrate
Vinyl Tile
Enamel on birch
:abulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
.atex on birch
Vinyl Tile
Enamel on birch
:abulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
..atex on birch
Vinyl Tile
Latex on birch
Fabulon on Oak
Enamel on birch
Vinyl Tile
Fabulon on Oak
Enamel on birch
Latex on birch
^atex on birch
Fabulon on Oak
Enamel on birch
Vinyl Tile
Latex on birch
Vinyl Tile
Fabulon on Oak
Enamel on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Latex on birch
Vinyl Tile
Fabulon on Oak
Enamel on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Wipe Lead (ng)
9
4.915
24.6
24.6
43.3
53.1
43.8
40.5
6.6
10.1
32.6
10
59.4
18.7
33.4
26.9
9
10.1
7.625
4.265
13.7
13.7
6.3
20.3
19.7
5.4
8.6
14.7
46.1
29.4
28.9
21.9
9.1
3.2
9.8
17.4
16.8
11.9
13.5
39.7
8.1
3.495
2.5
25.3
26.3
29.1
17.4
32.2
9.9
3.265
6.7
30.8
29.4
51.9
22.8
68.3
Residual wipe lead
0.0099
0.0054
0.0270
0.0135
0.0506
0.0621
0.0512
0.0237
0.0069
0.0105
0.0340
0.0052
0.0663
0.0209
0.0373
0.0150
0.0105
0.0059
0.0089
0.0050
0.0160
0.0160
0.0073
0.0118
0.0111
0.0061
0.0097
0.0165
0.0273
0.0349
0.0343
0.0260
0.0106
0.0037
0.0115
0.0102
0.0098
0.0139
0.0157
0.0463
0.0091
0.0039
0.0028
0.0142
0.0312
0.0345
0.0206
0.0191
0.0110
0.0036
0.0075
0.0171
0.0351
0.0619
0.0272
0.0407
100
-------
Table C2 Cleaner test data (continued)
Cleaner
15
15
15
15
15
15
15
15
16
16
16
16
16
16
16
16
17
17
17
17
17
17
17
17
18
18
18
18
18
18
18
18
19
19
19
19
19
19
19
19
20
20
20
20
20
20
20
20
21
21
21
21
21
21
21
21
Rep(mod)
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
Soil
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Substrate
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Latex on birch
Fabulon on Oak
Enamel on birch
Latex on birch
Fabulon on Oak
Vinyl Tile
Enamel on birch
Latex on birch
Fabulon on Oak
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Fabulon on Oak
Vinyl Tile
Enamel on birch
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Latex on birch
Fabulon on Oak
Enamel on birch
Vinyl Tile
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Wipe Lead (ng)
8.6
4.035
14.1
20.9
29.1
42.4
65.6
19.6
4.895
19.3
5.4
4.87
60.9
45
14.6
19
4.18
6.2
14.2
5.9
13.1
10.3
14.4
26.6
10
6.6
64
40.3
46.3
20.8
18.3
11.8
9.2
5.9
1.26
8.3
10.8
12.9
0.54
17.6
18.4
21.6
25.9
32.1
62.5
89.3
46.3
96.7
10
6.4
20.1
27.8
40
54.2
35.1
47
Residual wipe lead
0.0090
0.0042
0.0147
0.0109
0.0325
0.0473
0.0366
0.0219
0.0051
0.0101
0.0056
0.0051
0.0679
0.0251
0.0163
0.0212
0.0044
0.0065
0.0074
0.0062
0.0146
0.0115
0.0161
0.0148
0.0112
0.0074
0.0720
0.0227
0.0275
0.0247
0.0217
0.0140
0.0050
0.0065
0.0014
0.0091
0.0126
0.0151
0.0006 *
0.0103
0.0202
0.0237
0.0284
0.0176
0.0731
0.1044
0.0541
0.0565
0.0112
0.0072
0.0226
0.0156
0.0474
0.0643
0.0416
0.0279
1 This measurement was judged to be an outlier and was removed from the statistical analysis.
101
-------
Table C2 Cleaner test data (continued)
Cleaner
22
22
22
22
22
22
22
22
23
23
23
23
23
23
23
23
24
24
24
24
24
24
24
24
25
25
25
25
25
25
25
25
26
26
26
26
26
26
26
26
27
27
27
27
27
27
27
27
28
28
28
28
28
28
28
28
Rep(mod)
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
Soil
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
[>ry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Substrate
Enamel on birch
;abulon on Oak
Vinyl Tile
.atex on birch
Vinyl Tile
Latex on birch
Fabulon on Oak
Enamel on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
_atex on birch
Vinyl Tile
Enamel on birch
•abulon on Oak
^atex on birch
Vinyl Tile
Enamel on birch
"abulon on Oak
Latex on birch
^atex on birch
Fabulon on Oak
Enamel on birch
Vinyl Tile
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Latex on birch
Fabulon on Oak
Enamel on birch
Vinyl Tile
Enamel on birch
Vinyl Tile
Latex on birch
Fabulon on Oak
Latex on birch
Fabulon on Oak
Enamel on birch
Vinyl Tile
Latex on birch
Fabulon on Oak
Enamel on birch
Vinyl Tile
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Wipe Lead (ng)
6.6
32.5
24.1
30.5
46.2
105.9
44.4
58.5
17.4
12.6
16.3
30
50.1
77.7
29.1
41.6
9.4
3.325
11.6
17.1
57.2
33.3
28.3
33.4
5.74
4.445
14.6
16.7
20.8
30.1
36.1
49.3
27.4
20.8
7.3
12.6
81.3
36.1
41.8
17.1
13.6
11.9
6.1
8.7
29.5
12.6
13
19.7
9.6
7.1
14.5
33.2
22.5
22.1
23.6
62.1
lesidual wipe lead
0.0069
0.0339
0.0252
0.0159
0.0515
0.0591
0.0495
0.0653
0.0196
0.0142
0.0183
0.0169
0.0594
0.0922
0.0345
0.0247
0.0098
0.0035
0.0121
0.0089
0.0319
0.0371
0.0316
0.0373
0.0064
0.0049
0.0162
0.0093
0.0248
0.0359
0.0430
0.0294
0.0143
0.0217
0.0076
0.0132
0.0907
0.0403
0.0233
0.0191
0.0071
0.0124
0.0064
0.0091
0.0165
0.0141
0.0145
0.0220
0.0112
0.0083
0.0170
0.0194
0.0262
0.0258
0.0275
0.0362
102
-------
Table C2 Cleaner test data (continued)
Cleaner
29
29
29
29
29
29
29
29
30
30
30
30
30
30
30
30
31
31
31
31
31
31
31
31
32
32
32
32
32
32
32
32
33
33
33
33
33
33
33
33
34
34
34
34
34
34
34
34
Rep(mod)
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
Soil
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Substrate
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Latex on birch
Fabulon on Oak
Enamel on birch
Vinyl Tile
Latex on birch
Fabulon on Oak
Enamel on birch
Vinyl Tile
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Latex on birch
Fabulon on Oak
Enamel on birch
Fabulon on Oak
Vinyl Tile
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Vinyl Tile
Enamel on birch
Fabulon on Oak
Latex on birch
Latex on birch
Fabulon on Oak
Enamel on birch
Vinyl Tile
Latex on birch
Fabulon on Oak
Enamel on birch
Vinyl Tile
Wipe Lead (ng)
7.4
5.775
15.5
12.3
18.4
22.6
10.7
28.8
15.5
8.9
5.48
8.5
31.2
11.5
31.8
17.5
7
5.4
5.9
14
25.4
33.5
19.9
44.2
11.7
5.51
29.4
20.3
72.1
41.4
31.6
87.7
30.3
11.9
25.3
31.1
30.3
39.1
14.2
59.9
25.8
16.8
14.9
20.5
43.9
26.8
29.3
7.3
Residual wipe lead
0.0083
0.0065
0.0174
0.0069
0.0218
0.0268
0.0127
0.0171
0.0091
0.0104
0.0064
0.0099
0.0182
0.0134
0.0371
0.0204
0.0078
0.0060
0.0066
0.0078
0.0303
0.0399
0.0237
0.0264
0.0132
0.0062
0.0165
0.0228
0.0855
0.0491
0.0375
0.0520
0.0341
0.0134
0.0285
0.0175
0.0359
0.0464
0.0168
0.0355
0.0151
0.0197
0.0174
0.0240
0.0256
0.0312
0.0342
0.0085
103
-------
Table C3 Fraction of lead removed using cleaning wipes
Soil
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Oily
Substrate
Enamel on birch
Enamel on birch
Enamel on birch
Enamel on birch
Enamel on birch
Fabulon on Oak
Fabulon on Oak
Fabulon on Oak
Fabulon on Oak
Fabulon on Oak
Latex on birch
Latex on birch
Latex on birch
Latex on birch
Latex on birch
Latex on Dry wall
Latex on Dry wall
Latex on Drywall
Latex on Drywall
Latex on Drywall
Vinyl Tile
Vinyl Tile
Vinyl Tile
Vinyl Tile
Vinyl Tile
Enamel on birch
Enamel on birch
Enamel on birch
Enamel on birch
Enamel on birch
Enamel on birch
Fabulon on Oak
Fabulon on Oak
Fabulon on Oak
Fabulon on Oak
Fabulon on Oak
Fabulon on Oak
Latex on birch
Latex on birch
Latex on birch
Latex on birch
Latex on birch
Latex on birch
Latex on Drywall
Latex on Drywall
Latex on Drywall
Latex on Drywall
Latex on Drywall
Vinyl Tile
Vinyl Tile
Vinyl Tile
Vinyl Tile
Vinyl Tile
Replicate
1
2
3
4
5
1
2
3
4
5
I
2
3
4
5
1
2
3
4
5
2
3
4
5
1
1
2
3
4
5
6
1
2
3
4
5
7
1
2
3
4
5
6
1
2
3
4
5
1
2
3
4
5
Soil batch
3835-86
3835-42
3835-46
3835-72
3835-77
3835-25
3835-25
3835-56
3835-62
3835-77
3835-25
3835-52
3835-56
3835-62
3835-77
3835-33
3835-46
3835-62
3835-67
3835-86
3835-38
3835-52
3835-72
3835-81
3835-33
3835-42
3835-42
3835-56
3835-67
3835-81
3835-77
3835-33
3835-46
3835-56
3835-72
3835-86
3835-90
3835-33
3835-33
3835-46
3835-67
3835-81
3835-77
3835-25
3835-46
3835-52
3835-72
3835-77
3835-25
3835-38
3835-52
3835-67
3835-81
Wipe 1 (ng)
972.9
849.8
846.7
779.7
867.9
743.5
689
667.4
785.9
631.7
1220.2
1431.6
1493.6
1516.2
1225.6
1248.8
1660.2
1383
1542
1460.6
636.4
822.3
821.6
1115.6
840.8
825.5
795
813.7
864.3
832.6
879.7
544.1
667.4
635.2
925.2
553.5
739.6
1275.4
1344.2
1395.6
1438.7
1560.8
1288.8
1399.3
1534
1455.9
1465.2
1512.8
618.3
758.1
763
727.2
810.7
Wipe2(ng)
23.8
16.5
11.2
27.1
10.1
*
*
29.5
11.5
29.1
*
139
178.7
85.5
216.8
11.9
112.7
99.9
89.5
134.3
44.2
42.4
13.9
27
*
20.7
12.4
55.3
7.2
15.7
13
21
33.9
28.5
7.8
40.2
36.2
32.1
102.4
134.6
140.3
70.2
156.1
*
117
164.4
94.5
135
*
26.8
37.9
22.6
26.4
Fraction of lead for total
1.093
0.928
1.061
0.907
1.027
0.778
0.721
0.808
0.922
0.773
0.638
0.888
0.969
0.926
0.844
0.699
1.097
0.857
0.908
0.874
0.733
0.978
0.940
1.193
0.933
0.927
0.884
1.004
1.039
0.946
1.040
0.739
0.839
0.767
1.107
0.694
0.884
0.855
0.946
0.915
0.941
0.910
0.842
0.865
0.988
0.955
0.925
0.960
0.765
0.876
0.944
0.894
0.934
' Only one wipe was used to clean these coupons
104
-------
Table C4 Cleaner mixture characteristics
Cleaner
01
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
01
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16
17
Rep(mod)
1
1
1
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
PH
9.8
10.4
10.1
5.4
5
10.85
9.8
9.5
11.35
10.25
9.8
9.95
5.95
10.3
7.6
10.85
8.75
10.7
4.45
11.85
5.8
11.75
11.85
7.5
2.05
10.9
1.5
10.35
8.6
10.8
8.97
10
4.5
13.28
10.1
11
9.6
5.4
3.75
9.95
10.2
10.4
10.35
9.8
10.58
10.58
9.05
11.85
6.1
12
9.3
Phosphate
(gP/gal)
0.97
0
0.99
0
0
1.39
0.99
0.02
0
13.73
14.4
0.03
0
0
0
0.79
0.67
4.84
0
0.53
0
0.39
1.1
0
0
0
0
0.07
0.98
0.67
0.43
0
0
7.2
0.98
0
1
0
0
1.39
0.99
0.02
0
13.7
14.4
0.03
0
0
0
0.8
0.68
Surface tension
(dyne/cm)
29.1
25.6
27.3
26
32.1
27.6
25.3
26.4
26
28.3
49.6
26.4
25
26.4
26.8
38.3
25.3
29.1
27
52
31.9
36.8
31.2
24.5
26.3
28.6
27.6
27.4
26.4
36.9
27.9
28.9
66.8
50
25.2
26.2
29.8
27.9
30.5
27.6
27.6
27.8
25
30.4
45.2
26.9
24.7
27.5
27.1
38.9
30.5
Type
Anionic
Non-ionic
Anionic
Non-ionic
Other/unknown
Mon-ionic
Other/unknown
Anionic
Anionic
Other/unknown
Other/unknown
Other/unknown
Other/unknown
Anionic
Non-ionic
Non-ionic
Other/unknown
Non-ionic
Anionic
Non-ionic
Anionic
Other/unknown
Non-ionic
Non-ionic
Anionic
Other/unknown
Other/unknown
Anionic
Other/unknown
Other/unknown
Non-ionic
Non-ionic
Control
Control
Anionic
Non-ionic
Anionic
Non-ionic
Other/unknown
Non-ionic
Other/unknown
Anionic
Anionic
Other/unknown
Other/unknown
Other/unknown
Other/unknown
Anionic
Non-ionic
Non-ionic
Other/unknown
Intended use
Laundry
riard Surface
Hard Surface
Floor
Bathroom
Laundry
Laundry
Laundry
Hard Surface
Hand Dish
Laundry
Laundry
Laundry
Glass
Hand Dish
Mach Dish
Hard Surface
Hard Surface
Bathroom
Mach Dish
Hard Surface
Mach Dish
Mach Dish
Hand Dish
Bathroom
Floor
Bathroom
Laundry
Laundry
Mach Dish
Laundry
Glass
Control
Control
Laundry
Hard Surface
Hard Surface
Floor
Bathroom
Laundry
Laundry
Laundry
Hard Surface
Hand Dish
Laundry
Laundry
Laundry
Glass
Hand Dish
Mach Dish
Hard Surface
105
-------
Table C4 Cleaner mixture characteristics (continued)
Cleaner
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
35
01
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
07
12
34
19
08
08
08
Rep(mod)
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
4
4
4
Bl
B2
Tl
T2
PH
10.9
3.5
5.7
7
10.8
10.75
6
3
9.85
1.45
10.75
10.68
10.8
9.3
11.85
5.85
12.1
10.2
11.15
10.35
5.35
3.65
10
10.25
10
10.4
9.4
9.45
10.55
7.9
10.5
6.1
12
9.3
10.6
3
10.7
5.8
10.8
10.55
6.2
3
9.7
1.4
10.75
9.6
10.85
9.85
10.7
5.85
12.25
10.2
10.5
12.4
3
10.5
10.5
10.5
Phosphate
(gP/gal)
4.82
0
0.41
0
0.38
1.1
0
0
0
0
0.07
0.96
0.67
0.43
0
0
3.1
0.98
0
1.02
0
0
1.39
0.99
0.02
0
13.73
14.4
0.03
0
0
0
0.8
0.68
4.84
0
0.51
0
0.42
1.1
0
0
0
0
0.07
0.96
0.67
0.43
0
0
7.18
0.98
0.02
7.18
0
0.02
0.02
0.02
Surface tension
(dyne/cm)
28.7
27.1
26.2
34.2
37.6
37.26
26
28.1
29.3
29.6
28.7
27.2
36.9
28.4
29.8
41.2
44.7
30.9
30.5
29.8
25.8
30.3
28.7
27.2
27.5
28.5
29.2
41.3
30.5
26
26.9
26.2
37.1
31.7
31.6
28.7
45.4
36.8
37.6
37.6
27.8
28
30.5
36
27.8
30.1
33.8
29.4
30.8
58.6
45.96
27.9
24
45.5
28.7
27.1
27.4
27.1
Type
Non-ionic
Anionic
>Ion-ionic
Anionic
Other/unknown
4on-ionic
Non-ionic
Anionic
Other/unknown
Other/unknown
Anionic
Other/unknown
Other/unknown
•Jon-ionic
•Ion-ionic
Control
Control
Anionic
Non-ionic
Anionic
•Jon-ionic
Other/unknown
^Ion-ionic
Other/unknown
Anionic
Anionic
Other/unknown
Other/unknown
Other/unknown
Other/unknown
Anionic
Non-ionic
Non-ionic
Other/unknown
Non-ionic
Anionic
Non-ionic
Anionic
Other/unknown
Non-ionic
^on-ionic
Anionic
Other/unknown
Other/unknown
Anionic
Other/unknown
Other/unknown
Non-ionic
Non-ionic
Control
Control
Other/unknown
Other/unknown
Control
Anionic
Anionic
Anionic
Anionic
ntended use
lard Surface
Jathroom
Mach Dish
lard Surface
dach Dish
tiach Dish
land Dish
Bathroom
:loor
Jathroom
.aundry
.aundry
Mach Dish
sundry
Glass
Control
Control
.aundry
Hard Surface
Hard Surface
•loor
Bathroom
Laundry
^aundry
^aundry
lard Surface
•iand Dish
^aundry
^aundry
Laundry
Glass
Hand Dish
Mach Dish
Hard Surface
Hard Surface
Bathroom
Mach Dish
Hard Surface
Mach Dish
Mach Dish
Hand Dish
Bathroom
Floor
Bathroom
Laundry
Laundry
Mach Dish
Laundry
Glass
Control
Control
Laundry
Laundry
Control
Bathroom
Laundry
Laundry
Laundry
106
-------
50272-101
REPORT DOCUMENTATION
PAGE
1. REPORT NO.
EPA 747-R-97-002
3. Recipient's Accession No.
4. Title and Subtitle
Laboratory Study of Lead-Cleaning Efficacy
5. Report Date
March 1997
6.
7. Author(s)
John Rogers and William Hartley, Westat, Inc.
Gary Cooper, Midwest Research Institute
8. Performing Organization Report No.
9. Performing Organization Name and Address
Westat, Inc.
1650 Research Boulevard
Rockville, MD 20850
Midwest Research Institute
425 Volker Boulevard
Kansas City, MO 64110
10. Project/Task/Work Unit No.
11. Contract (C) or Grant (G) No.
(C) 68-W6-0048
(C) 68-D5-0008
(068-D3-0011
12. Sponsoring Organization Name and Address
U.S. Environmental Protection Agency
Office of Pollution Prevention and Toxics
401 M Street SW
Washington, DC 20460
13. Type of Report & Period Covered
Technical Report
14.
15. Supplementary Notes
16. Abstract (Limit: 200 words)
The United States Environmental Protection Agency (EPA) has recommended the use of trisodium phosphate (TSP) detergent
to clean lead-contaminated dust from surfaces after residential lead hazard control work to achieve post-abatement clearance
standards. This recommendation has often been assumed to apply to the general cleaning of lead-contaminated dust during
ongoing exposure reduction activities. Because of the negative impact of phosphate detergents on the ecology of aquatic
ecosystems, questions have arisen as to the scientific basis of recommending TSP and about the effectiveness of other cleaners.
EPA conducted a laboratory study to evaluate the cleaning efficacy of many commercially available cleaners that could be used
to remove lead-contaminated dust from residential surfaces.
This report presents the results of the laboratory study in which surfaces were soiled with lead-containing soil, cleaned with one
of 34 cleaners, and wiped with a baby wipe which was then analyzed for lead. The measurements were analyzed to determine
how much lead remained on the surfaces. The results suggest that low surface tension cleaners remove more of the lead than
high surface tension cleaners, although the effect of surface tension is marginally statistically significant. All cleaners removed
most of the leaded soil.
1 7. Document Analysis
a. Descriptors
Environmental Contaminants
b. Identifiers/Open-Ended Terms
Lead, Clearance testing, TSP, Phosphate content, Surface tension, Wipe sampling, Cleaning Efficacy
c. COSATI Field/Group
18. Availability Statement
Unclassified
19. Security Class (This Report)
Unclassified
20. Security Class (This Page)
Unclassified
21. No. of Pages
116
22. Price
(See ANSI-239.18)
OPTIONAL FORM 272 (4-77)
(Formerly NTIS-35)
Department of Commerce
------- |