&EPA
United States
Environmental Protection
Agency
REPORT
Environics USA Inc.
ChemPro 100
Hand-Held Chemical Detector
Office of Research and Development
National Homeland Security
Research Center
-------
EPA600/R-06/141
June 2006
Technology Evaluation Report
Environics USA Inc. ChemPro 100
Hand-Held Chemical Detector
By
Tricia Derringer, Thomas Kelly, Dale Folsom,
Robert Krile, and Zachary Willenberg
Battelle
505 King Avenue
Columbus, OH 43201
Eric Koglin
Task Order Project Officer
National Homeland Security Research Center
Office of Research and Development
U.S. Environmental Protection Agency
944 East Harmon Ave.
Las Vegas, NV 89119
-------
Notice
The U.S. Environmental Protection Agency (EPA), through its Office of Research and
Development's National Homeland Security Research Center, funded and managed this
technology evaluation through a Blanket Purchase Agreement under General Services
Administration contract number GS23F0011L-3 with Battelle. This report has been peer and
administratively reviewed and has been approved for publication as an EPA document. Mention
of trade names or commercial products does not constitute endorsement or recommendation for
use of a specific product.
-------
Preface
The U.S. Environmental Protection Agency (EPA) is charged by Congress with protecting the
nation's air, water, and land resources. Under a mandate of national environmental laws, the
Agency strives to formulate and implement actions leading to a compatible balance between
human activities and the ability of natural systems to support and nurture life. To meet this
mandate, the EPA's Office of Research and Development (ORD) provides data and science
support that can be used to solve environmental problems and to build the scientific knowledge
base needed to manage our ecological resources wisely, to understand how pollutants affect our
health, and to prevent or reduce environmental risks.
In September 2002, EPA announced the formation of the National Homeland Security Research
Center (NHSRC). The NHSRC is part of the ORD; it manages, coordinates, and supports a
variety of research and technical assistance efforts. These efforts are designed to provide
appropriate, affordable, effective, and validated technologies and methods for addressing risks
posed by chemical, biological, and radiological terrorist attacks. Research focuses on enhancing
our ability to detect, contain, and clean up in the event of such attacks.
NHSRC's team of world-renowned scientists and engineers is dedicated to understanding the
terrorist threat, communicating the risks, and mitigating the results of attacks. Guided by the
roadmap set forth in EPA's Strategic Plan for Homeland Security, NHSRC ensures rapid
production and distribution of security-related products.
The NHSRC has created the Technology Testing and Evaluation Program (TTEP) in an effort to
provide reliable information regarding the performance of homeland security-related
technologies. TTEP provides independent, quality-assured performance information that is
useful to decision makers in purchasing or applying the tested technologies. It provides potential
users with unbiased, third-party information that can supplement vendor-provided information.
Stakeholder involvement ensures that user needs and perspectives are incorporated into the test
design so that useful performance information is produced for each of the tested technologies.
The technology categories of interest include detection and monitoring, water treatment, air
purification, decontamination, and computer modeling tools for use by those responsible for
protecting buildings, drinking water supplies, and infrastructure and for decontaminating
structures and the outdoor environment.
The evaluation reported herein was conducted by Battelle as part of the TTEP program.
Information on NHSRC and TTEP can be found at http://www.epa.gov/ordnhsrc/index.htm.
in
-------
Acknowledgments
The authors wish to acknowledge the support of all those who helped plan and conduct the
evaluation, analyze the data, and prepare this report. We also would like to thank
Donald Stedman of the University of Denver and John Zimmerman of EPA/ORD for their
reviews of this report.
IV
-------
Contents
Notice ii
Preface iii
Acknowledgments iv
Abbreviations/Acronyms vii
Executive Summary viii
1.0 Introduction 1
2.0 Technology Description 5
3.0 Quality Assurance/Quality Control 6
3.1 Equipment Calibration 6
3.1.1 Reference Methods 6
3.1.2 Instrument Checks 7
3.2 Audits 8
3.2.1 Performance Evaluation Audit 8
3.2.2 Technical Systems Audit 8
3.2.3 Data Quality Audit 9
3.3 QA/QC Reporting 9
4.0 Test Results 10
4.1 Response Time 11
4.2 Recovery Time 14
4.3 Accuracy 15
4.4 Repeatability 18
4.5 Response Threshold 19
4.6 Temperature and Humidity Effects 20
4.7 Interference Effects 20
4.8 Cold-/Hot-Start Behavior 26
4.9 Battery Life 28
4.10 Operational Characteristics 30
5.0 Performance Summary 31
6.0 References 35
v
-------
Figures
Figure 2-1. Environics USA ChemPro 100 Hand-Held Chemical Detector 5
Tables
Table 1-1. Target TIC and CW Agent Challenge Concentrations 2
Table 3-1. Performance Evaluation Audit Results 7
Table 4-1. TIC Results from ChemPro 100 Evaluation 12
Table 4-2. CW Agent Results from ChemPro 100 Evaluation 13
Table 4-3. Response Threshold Data for the TIC and CW Agent Evaluation 19
Table 4-4. Interference Effects 21
Table 4-5. Start State Effects 27
Table 4-6. Responses Recorded from the ChemPro 100 in Battery Life Evaluation 29
VI
-------
Abbreviations/Acronyms
AC
CW
CK
C12
DEAE
EPA
FID
FPD
GB
GC
HD
IDLH
IMS
L
MS
ug/m3
uL
mg/m
mL
mm
MSD
NHSRC
ORD
PE
ppm
ppmC
QA
QC
QMP
RH
SA
THC
TIC
ISA
TTEP
hydrogen cyanide
chemical warfare
cyanogen chloride
chlorine
N,N-diethylaminoethanol
U.S. Environmental Protection Agency
flame ionization detection
flame photometric detection
sarin
gas chromatography
sulfur mustard
immediately dangerous to life and health
ion mobility spectrometer(ry)
liter
microgram
microgram per cubic meter
microliter
milligram per cubic meter
milliliter
millimeter
mass selective detection
National Homeland Security Research Center
Office of Research and Development
performance evaluation
parts per million
parts per million of carbon
quality assurance
quality control
quality management plan
relative humidity
arsine
total hydrocarbon
toxic industrial chemical
technical systems audit
Technology Testing and Evaluation Program
vn
-------
Executive Summary
The U.S. Environmental Protection Agency's (EPA's) National Homeland Security
Research Center (NHSRC) Technology Testing and Evaluation Program (TTEP) is
helping to protect human health and the environment from adverse impacts as a result of
acts of terror by carrying out performance tests on homeland security technologies. Under
TTEP, Battelle recently evaluated the performance of the Environics USA Inc.
ChemPro 100 Hand-Held Chemical Detector. The objective of evaluating the ChemPro
100 Hand-Held Chemical Detector was to evaluate its ability to detect toxic industrial
chemicals (TICs) and chemical warfare (CW) agents in indoor air.
The ChemPro 100 is based on Environics' open loop ion mobility spectrometry (IMS)
technology and uses an improved Ion Mobility Cell™ that is designed to increase
selectivity and sensitivity in detecting CW agents and TICs. It identifies agent class
(Nerve, Blister, or Blood), indicates relative concentration (Low, Medium, or High), and
indicates whether the concentration is increasing or decreasing.
The following performance characteristics of the ChemPro 100 were evaluated:
# Response time
# Recovery time
# Accuracy of hazard identification
# Repeatability
# Response threshold
# Temperature and humidity effects
# Interference effects
# Cold-/hot-start behavior
# Battery life
# Operational characteristics.
This evaluation addressed detection of chemicals in the vapor phase. The TICs and the
respective challenge concentrations delivered to the ChemPro 100 during the evaluation
were hydrogen cyanide [HCN; North Atlantic Treaty Organization military designation
AC; 50 milligrams per cubic meter (mg/m3)], cyanogen chloride (C1CN; designated CK;
250 mg/m3), arsine (AsH3; designated SA; 20 mg/m3), and chlorine (Ch; no military
designation; 180 mg/m3). The CW agents and concentrations were sarin (GB;
0.060 mg/m3) and sulfur mustard (HD; 0.54 mg/m3). These TIC and CW agent challenge
concentrations were established in trial runs as producing Medium response on the
ChemPro 100. The TIC challenge concentrations were equal to or greater than the Low
alarm concentrations stated by the vendor. The GB and HD challenge concentrations
were less than the Low alarm concentrations stated by the vendor for these agents. Two
ChemPro 100 units (Units 1546 and 1811) were evaluated simultaneously with the TICs;
viii
-------
one unit of the ChemPro 100 (Unit 1811) was evaluated with the CW agents. The use of
only one unit in testing with CW agents minimized the expense to the vendor, because
that unit could not be returned to the vendor after contamination with agents. Each
ChemPro 100 unit was challenged at the start of each test day with a chemical simulant
sample provided by the vendor. No test procedures were initiated unless proper response
to this challenge was obtained. This challenge was also repeated as needed during each
test day (e.g., in the case of an unexpected response during testing) before continuing the
test procedures.
The evaluation included sampling potential indoor interferents, both with and without the
target TICs and CW agents. The interferents used were latex paint fumes, air freshener
vapors, ammonia cleaner vapors, a mixture of hydrocarbons representing motor vehicle
exhaust, and diethylaminoethanol (DEAE), a boiler water additive that can enter indoor
air via steam humidification. A range of temperatures (5 to 35 °C) and relative humidities
(<20 to 80%) was used to assess the effects of these conditions.
Summary results from testing the ChemPro 100 are presented below for each
performance parameter evaluated. Results reported for CK and SA are limited due to
inconsistent responses, and few results for Cb are reported, due to lack of response found
for that chemical. Discussion of the observed performance can be found in Chapter 4 of
this report.
Response Time: When the ChemPro 100 responded to challenges, the time required to
respond to AC and CK was usually about 30 seconds or less, and response times for SA
ranged from about 20 to 80 seconds. Response times for GB were 15 seconds or less, and
for HD were usually 25 to 40 seconds, with a few results of 80 to 225 seconds. Response
times for AC, GB, and HD were not consistently affected by the temperature and relative
humidity (RH). These results do not include instances in which the ChemPro 100 failed
to respond to TIC or CW agent challenges; those instances are addressed below under
Accuracy.
Recovery Time: The time required for the ChemPro 100 to return to a baseline reading
after an alarm was typically less than 50 seconds for AC, CK, SA, and HD, and less than
about 15 seconds for GB, but in a few instances during evaluation with AC and HD,
recovery times exceeded 600 seconds. Recovery times depended only weakly on
temperature and RH, with recovery times for AC being shorter with higher temperature
and lower RH. These results exclude those instances in which the ChemPro 100 did not
respond to a TIC or agent challenge.
Accuracy: Of the 120 challenges with AC, GB, and HD used to assess accuracy, the
ChemPro 100 responded accurately to 86, with no response to 30 challenges, and four
cases of a continued alarm even when sampling clean air. Accuracy results for the target
chemicals varied from one test condition to another, and (in TIC evaluation) from one
ChemPro 100 unit to the other. Accuracy for AC was 100% in most conditions, but
ranged from 0 to 40% under conditions of high humidity. For GB, accuracy was 80 to
ix
-------
100% at most conditions, but was 0% with high humidity. Accuracy for HD was 80 to
100% at some conditions, but 0 to 40% at others, with no clear dependence on
temperature or RH. Accuracy for CK ranged from 0 to 100%, with different temperature
and RH dependence observed from the two units. For SA accuracy ranged from 0 to
100% under different conditions (from 0 to 20% for one ChemPro 100 unit), with no
apparent dependence on temperature or RH. For chlorine, only one positive response was
seen from one unit in five trials on each of the two units, so the unit accuracies were 0
and 20%.
[Failure to respond to AC challenges was also observed during cold-/hot-start and battery
life tests, but those observations were not used in the calculation of the accuracy results
noted above.]
Repeatability: When the ChemPro 100 units responded to an AC challenge, for one
unit, repeatability was perfect under all conditions of temperature and humidity (i.e., all
maximum responses were Medium). For the other unit with AC, maximum response
changed from Low to High as temperature increased, and from Medium to Low as RH
increased. For GB, maximum responses changed from High to Medium to Low as
temperature increased from low (5 °C) to room temperature to high (35 °C). No humidity
effect was seen on GB repeatability, and HD response was perfectly repeatable under all
conditions (all maximum responses were Low).
Response Threshold: For AC, the response threshold was between 3 and 6 parts per
million (ppm) (3 and 6 mg/m3) on both ChemPro 100 units. For CK the response
threshold was between 5 and 10 ppm (12.5 and 25 mg/m3) on one unit and between
10 and 20 ppm (25 and 50 mg/m3) on the other. The SA response threshold was between
3 and 6 ppm (10 and 20 mg/m3) on both units and, for Cb, was at or above about 60 ppm
(180 mg/m3). For GB the response threshold was about 0.002 ppm (0.01 mg/m3), and for
HD it was about 0.03 ppm (0.2 mg/m3).
Temperature and Humidity Effects: These effects are described in the preceding
summaries of other performance parameters.
Interference Effects: Ammonia cleaner and air freshener vapors produced false positive
responses in nearly all trials when using either the TIC or CWA library of the ChemPro
100. Latex paint fumes produced false positives in 67 to 100% of trials in the TIC library,
and in 20 to 40% of trials in the CW agent library. DEAE produced no false positive
responses, and exhaust hydrocarbons produced only one false positive out of 20 trials.
[Erroneous positive responses of a different kind (i.e., alarms while the ChemPro 100
sampled clean air) were observed in a few cases during tests of accuracy with AC and
CK.]
When added to challenge mixtures of AC, the interferences produced minimal false
negative responses for AC with one ChemPro 100 unit. However, the response accuracy
x
-------
of the other unit was reduced to 40% by the air freshener vapors and to 0% by the
ammonia cleaner vapors. False negative effects on CK and SA response were difficult to
determine because of the variability in response for these chemicals with the two
ChemPro 100 units. False negative effects on accuracy of identification for CK were seen
with DEAE, and the accuracy for S A was reduced to 0 to 20% by engine exhaust
hydrocarbons and DEAE. False negative responses with GB occurred primarily with
ammonia cleaner and exhaust hydrocarbons. False negative responses with FID occurred
with paint fumes, ammonia cleaner, and air freshener vapors. With both GB and FID, the
false negatives were primarily in the form of inaccurate responses (e.g., a response of
CHEM HAZARD rather than NERVE for GB), rather than no response at all. In these
cases the ChemPro 100 response provides a protective warning, although the threat is
incorrectly identified.
[In one challenge each with AC, GB, and HD in clean air during the evaluation of
accuracy, and in two challenges with HD in interference testing, the ChemPro 100
produced a different type of erroneous negative response in clearing its alarm while the
TIC or agent challenge was still ongoing.]
Cold-/Hot-Start Behavior: The delay time, or time to reach a ready state after start-up,
was 161 seconds and 169 seconds for the two ChemPro 100 units, respectively, when
started up from room temperature storage. The delay times were increased to 258 seconds
and 420 seconds after storage at 5 °C. Accuracy of identification of an AC challenge was
substantially reduced in initial readings after a cold start, relative to that in fully warmed
up operation. For example, one unit showed no response to AC in four of five trials after
start-up from cold storage, in all five trials after start-up from room temperature, and in
four of five trials after start-up from hot storage. In general, response times were slightly
longer, and response readings (i.e., Low/Medium/High) somewhat lower after a cold start
than in fully warmed up operation.
Battery Life: One unit of the ChemPro 100 shut down after 9 hours and 53 minutes of
continuous operation on battery power. The other unit shut down after 11 hours and
12 minutes.
Operational Characteristics: The ChemPro 100 has a large display that is easy to read
in all light conditions provided the background light (bright blue) is used. This light is
controlled from a menu within the ChemPro 100. The display indicates the response
reading of the unit (hazard identity and level), what library is being used, the date and
time, the audible alarm volume level, and the battery power level. A lighted status
indicator is green when the unit is in ready mode, and flashing red when the unit is in
alarm mode (coincident with the audible alarm). The display (when lighted) and audible
and visual alarms can be readily understood by the operator, even when wearing personal
protective equipment. When the ChemPro 100 detects a failure within its system, the
display also indicates an error message, e.g., for air intake flow or SCCell failure. The
ChemPro 100 has a "conditioning" mode that keeps the instrument from responding
while the instrument stabilizes. However, the occurrence of this mode is only apparent
xi
-------
from data displayed on a laptop computer, and is not evident to an operator using the
ChemPro 100 as a hand-held device. When the temperature or humidity condition was
changed, the ChemPro 100 may have entered conditioning mode and thus not responded
until the conditioning mode was completed. This mode may have contributed to instances
where IMS signal was observed on the laptop, but the ChemPro 100 failed to give an
alarm when challenged.
Before this evaluation began, an Environics representative trained Battelle personnel to
operate the ChemPro 100. The evaluation proceeded according to the vendor's recom-
mendations, and the vendor responded promptly when information was needed during the
evaluation. The list price of the ChemPro 100, as used in this evaluation, is
approximately $9,500.
Conclusion: The ChemPro 100 responded correctly to AC, GB, and HD in most
challenges, but responses observed with CK, SA, and Cb were less reliable. However,
even with AC, GB, and HD, observations included the absence of response to challenges,
widely different responses from two detector units challenged simultaneously, the
occasional discontinuance of a warning alarm even though a TIC or chemical agent
challenge was still present, and the failure to clear an alarm even after the challenge gas
was replaced with clean air. IMS signals recorded on laptop computers during testing
indicated that these behaviors originated with the software that interprets the IMS signal,
rather than with the IMS response itself. This finding suggests that software improve-
ments might rectify the observed responses. Both false positive and false negative
responses occurred in the presence of common indoor interferent vapors. Usually a
protective warning (albeit inaccurately identified) was present in the instances of a false
negative response caused by interferents. Elevated humidity generally produced less
accurate responses.
xn
-------
1.0 Introduction
The U.S. Environmental Protection Agency's (EPA's) National Homeland Security
Research Center (NHSRC) is helping to protect human health and the environment from
adverse impacts resulting from intentional acts of terror. With an emphasis on decontam-
ination and consequence management, water infrastructure protection, and threat and
consequence assessment, NHRSC is working to develop tools and information that will
help detect the intentional introduction of chemical or biological contaminants in build-
ings or water systems, the containment of these contaminants, the decontamination of
buildings and/or water systems, and the disposal of material resulting from clean-ups.
NHSRC's Technology Testing and Evaluation Program (TTEP) works in partnership
with recognized testing organizations; with stakeholder groups consisting of buyers,
vendor organizations, and permitters; and with the full participation of individual
technology developers in carrying out performance tests on homeland security
technologies. The program evaluates the performance of innovative homeland security
technologies by developing evaluation plans that are responsive to the needs of
stakeholders, conducting tests, collecting and analyzing data, and preparing peer-
reviewed reports. All evaluations are conducted in accordance with rigorous quality
assurance (QA) protocols to ensure that data of known and high quality are generated and
that the results are defensible. TTEP provides high-quality information that is useful to
decision makers in purchasing or applying the evaluated technologies. It provides
potential users with unbiased, third-party information that can supplement vendor-
provided information. Stakeholder involvement ensures that user needs and perspectives
are incorporated into the evaluation design so that useful performance information is
produced for each of the evaluated technologies.
Under TTEP, Battelle recently evaluated the performance of the Environics USA Inc.
ChemPro 100 Hand-Held Chemical Detector in detecting toxic industrial chemicals
(TICs) and chemical warfare (CW) agents in indoor air. This evaluation was conducted
according to a peer-reviewed test/QA plan(1) that was developed in accordance with the
requirements of the quality management plan (QMP) for TTEP.(2) The following
performance characteristics of the ChemPro 100 were evaluated:
# Response time
# Recovery time
# Accuracy of hazard identification
# Repeatability
# Response threshold
-------
# Temperature and humidity effects
# Interference effects
# Cold-/hot-start behavior
# Battery life
# Operational characteristics.
In this evaluation, two units of the ChemPro 100 (Units 1546 and 1811) were evaluated
simultaneously throughout all procedures with the TICs. In evaluating with the CW
agents, only one unit of the ChemPro 100 (Unit 1811) was used, with the other kept in
reserve. This approach minimized the expense to the vendor of the ChemPro 100 because
the unit tested with CW agents could not be returned after testing. Results are reported for
the two units separately. Each ChemPro 100 unit was challenged at the start of each test
day with a chemical simulant sample provided by the vendor. No test procedures were
initiated unless proper response to this challenge was obtained. This challenge was also
repeated as needed during each test day (e.g., in the case of an unexpected response
during testing) before continuing the test procedures.
This evaluation addressed detection of chemicals in the vapor phase, because that
application is most relevant to use in a building contamination scenario. This evaluation
took place between February 22 and August 11, 2005, in two phases: detection of TICs
(conducted in a non-surety laboratory at Battelle) and detection of CW agents (conducted
in a certified surety laboratory at Battelle's Hazardous Materials Research Center). The
TICs used were hydrogen cyanide (HCN; North Atlantic Treaty Organization military
designation AC), cyanogen chloride (C1CN; designated CK), arsine (AsH?; designated
SA), and chlorine (Cl2; no military designation). The CW agents were sarin (GB) and
sulfur mustard (HD). Most evaluation procedures were conducted with challenge
concentrations of the TIC or CW agent that were at or near immediately dangerous to life
and health (IDLH) or similar levels, as specified in the test/QA plan.(1) Table 1-1
summarizes the primary challenge concentrations used.
Table 1-1. Target TIC and CW Agent Challenge Concentrations
Chemical Concentration Type of Level
Hydrogen cyanide (AC) 50 parts per million (ppm) IDLH(a)
[50 milligrams per cubic
meter (mg/m3)]
Cyanogen chloride (CK) 100 ppm (250 mg/m3) 5 x IDLH
Arsine (SA) 6 ppm (20 mg/m3) 2 x IDLH
Chlorine (C12) 60 ppm (180 mg/m3) 6 x IDLH
Sarin (GB) 0.011 ppm (0.060 mg/m3) 0.3 x IDLH
Sulfur mustard (HD) 0.081 ppm (0.54 mg/m3) 0.9 x AEGL-2(b)
(a) IDLH = Immediately dangerous to life and health; IDLH value for CK estimated from value for AC.
(b) AEGL = Acute exposure guideline level; AEGL-2 levels are those expected to produce a serious
hindrance to efforts to escape in the general population. The AEGL-2 value of 0.09 ppm (0.6 mg/m3) for
HD is based on a 10-minute exposure.
-------
The target TIC and CW agent concentrations shown in Table 1-1 were selected because
they produced a Medium response from the ChemPro 100 units in initial trial runs at
normal temperature and relative humidity (RH) conditions (i.e., 22 °C and 50% RH)
This selection process is defined in the test/QA plan.(1) For the four TICs, the target
concentrations selected are equal to or greater than the Low alarm concentrations stated
by the ChemPro 100 vendor.(3) As Table 1-1 shows, the selected TIC concentrations also
range from 1 to 6 times the respective IDLH concentrations for the TICs. However, for
GB and HD, the target challenge concentrations in Table 1-1 are less than the Low alarm
concentrations stated by the vendor, which are 0.1 mg/m3 and 2 mg/m3, respectively.(3)
Thus, for these two CW agents, the response of the ChemPro 100 units was more
sensitive than the nominal response indicated by the vendor. In considering the results of
this evaluation, the relatively low concentrations of GB and HD, relative to the vendor's
nominal Low alarm limits, should be kept in mind. Also, note that the vendor's
informal! on(3) states that the ChemPro 100 is suited for detecting chlorine only at absolute
humidity levels below 16 g H2O/m3 [equivalent (e.g.) to 82% RH at 22 °C, and to 40%
RHat35°C].
In all evaluations, the TIC or CW agent challenge concentrations were confirmed by
means of reference analysis of the challenge air stream. The reference method for AC and
CK was a gas chromatography method using flame ionization detection (GC/FID), with
automatic sampling from the challenge air stream using a sample loop. This direct
sampling approach was supplemented by collection in gas sampling bags for a few final
samples. For SA the reference method was gas chromatography with mass selective
detection (GC/MSD), with all sampling conducted using gas sampling bags. For Cb, a
commercial electrochemical detector (Drager MiniWarn) was used as the reference
indicator. The reference method for GB and HD was gas chromatography with flame
photometric detection (GC/FPD), using bags for sample collection.
In all testing, the sample inlet of each ChemPro 100 unit was not directly plumbed to the
challenge delivery system, but sampled from a "bell" fitting through which a challenge
gas flow was supplied in excess of that required by the ChemPro 100. This was done to
avoid over- or under-pressurization of the units. The delivered challenge gas flows were
always at least twice the inlet flow of the ChemPro 100.
As described in the test/QA plan,(1) response time, recovery time, accuracy, and
repeatability were evaluated by alternately challenging the ChemPro 100s with clean air
and known vapor concentrations of target TICs and CW agents. Response thresholds
were evaluated by challenges with concentrations typically well below the target values
shown in Table 1-1. Evaluations conducted over the range of 5 to 35 °C and 20 to 80%
RH were used to establish the effects of temperature and humidity on detection
capabilities. The test apparatus allowed RH to be changed rapidly; a few minutes of
continuous operation was allowed to thoroughly flush all flow paths after a change in the
RH (with no change in temperature). On the other hand, typically two to three hours of
stabilization time were allowed after a change in the test temperature. In all cases, testing
-------
resumed only after the temperature and RH sensors in the test apparatus showed readings
stabilized within the required ranges. Throughout the stabilization period after any
change, the ChemPro 100 units remained enclosed in the test apparatus, sampling clean
air of the target RH. The effects of potential indoor interferences were assessed by
sampling selected interferences both with and without the target TICs and CW agents
present. The interferences used were latex paint fumes, ammonia floor cleaner vapors, air
freshener vapors, a mixture of gasoline exhaust hydrocarbons, and diethylaminoethanol
(DEAE), a boiler water additive potentially released to indoor air by humidification
systems. The concentrations of the interferents were checked during the evaluation by
means of a total hydrocarbon (THC) analyzer, calibrated with known concentrations of
propane. The ChemPro 100s also were evaluated with AC after a cold start (i.e., without
the usual warm-up period) from room temperature, from cold storage conditions (5°C),
and from hot storage conditions (40°C) to evaluate the delay time before readings could
be obtained and the response speed and accuracy once readings were obtained. Battery
life was determined as the time until ChemPro 100 performance degraded as battery
power was exhausted in continuous operation. Operational factors such as ease of use,
data output, and cost were assessed through observations made by test personnel and
through inquiries to the vendor.
The evaluation data were subjected to multivariate and other statistical analyses, as
described in the test/QA plan,(1)to characterize the performance of the ChemPro 100. The
data from evaluations with AC, GB, and HD were subjected to the full set of statistical
analyses; however, the data from evaluations with the other TICs were not consistent
enough to support the full set of analyses, so primarily interference effects and accuracy
of identification were evaluated for those TICs.
QA oversight of this evaluation was provided by Battelle and EPA. Battelle QA staff
conducted a technical systems audit (TSA) and a data quality audit of all the evaluation
data. A performance evaluation (PE) audit of the reference methods for AC, SA, and Cb
was also conducted.
-------
2.0 Technology Description
This report provides results for the evaluation of the ChemPro 100 hand-held chemical
detector. Following is a description of the ChemPro 100, based on information provided
by the vendor. (Contact: Rob Howard, Executive Vice President and General Manager,
Environics USA Inc., 4401 Eastport Parkway, Port Orange, Florida 32127, rob.howard@
environicsusa.com, 386-304-5252) The information provided below was not verified in
this evaluation.
The ChemPro 100 is based on Environics' open loop ion mobility spectrometry (IMS)
technology. The ChemPro 100 uses an improved Ion Mobility Cell™ that is designed to
increase selectivity and sensitivity in detecting CW agents and TICs. It identifies agent
class (Nerve, Blister, or Blood), indicates relative agent concentration (Low, Medium, or
High), and indicates whether the concentration is increasing or decreasing.
The ChemPro 100 weighs less than 700 grams (1.5 pounds) and can be powered by a
rechargeable battery pack or AA batteries. The operator interface is designed to be
operated using one hand. The user display provides the operator with a battery life
indicator, concentration bar display, agent class, agent identification, relative time-based
dose, audible alarm volume level, date,
and time. The ChemPro 100 stores agent
alarm information for retrieval at a later
time to provide a historical log of events.
The ChemPro 100, which is 102 milli-
meters (mm) by 229 mm by 51 mm
(4 inches by 9 inches by 2 inches), is
designed to be used as a personal
detector, a monitor for surveying after an
event, or a fixed installation detector. The
ChemPro 100 is designed to operate in
temperatures between -30 EC and 55 EC
(between -22 EF and 131 EF). The
ChemPro 100 is designed to operate
continuously without the need of
expendable desiccant cartridges. The
ChemPro 100 has no expendables and is
designed for low life-cycle and operating
Figure 2-1. Environics USA ChemPro
100 Hand-Held Chemical Detector
-------
costs.
-------
3.0 Quality Assurance/Quality Control
QA/quality control (QC) procedures were performed in accordance with the program
QMP(2) and the test/QA plan(1) for this evaluation.
3.1 Equipment Calibration
3.1.1 Reference Methods
As noted in Chapter 1, reference methods were used to confirm the challenge
concentrations of TICs and CW agents used in this evaluation. Calibration procedures for
the reference and other analyses are discussed in the following paragraphs.
The GC/FID reference method for AC and CK was calibrated by preparing gas mixtures
in 1-liter (L) gas sampling bags. Calibration standards for AC were prepared by diluting
1 to 4 milliliters (mL) of a certified commercial gas standard (10,000 ppm AC in
nitrogen, Scott Specialty Gases) in 800 mL of high purity air in a bag. The resulting
standards had concentrations of 12.5, 24.9, 37.4, and 49.8 ppm AC. Three samples from
each calibration bag were injected by syringe into the GC/FID. The peak areas were
recorded, and the average peak area from each set of triplicate analyses was used in a
linear regression of the calibration data. Blank samples were analyzed in the same way
and showed < 1 count peak areas. The regression of peak area versus AC standard
concentration had the form Peak Area = 1.340 (± 0.138) • AC (ppm), with a coefficient of
determination (r2) of 0.9716. Calibration standards for CK were prepared in the same
way, by diluting 1 to 4 mL of a 12,500-ppm CK compressed gas standard that had been
prepared by Battelle starting from neat CK. The resulting CK concentrations were 15.6,
31.3, 46.9, and 62.2 ppm. The resulting regression had the form Peak Area = 1.333 (±
0.169) • CK (ppm), with r2 = 0.9725.
The GC/MSD reference method for SA was calibrated by injecting 1, 3, 6, or 10 mL of a
certified commercial 997-ppm arsine standard (Linde Gas) into 1 L of high purity air in a
gas sampling bag, thereby producing standards of 1.00, 2.98, 5.95, and 9.87 ppm,
respectively. The 5-mL and 10-mL glass graduated syringes used for these injections
were preconditioned by filling them with the 997-ppm standard and storing them
overnight. This treatment minimized loss of arsine in the syringe during standard
preparation. A multipoint calibration was performed on each of the five days of testing
with SA. Each multipoint calibration included one blank sample and from one to four
7
-------
replicate analyses at the calibration concentrations noted above (i.e., from 5 to 11 total
calibration points). Response was linear, and the response to the blank samples was so
small that calculated intercepts did not appreciably differ from zero; consequently the
calibration plots were recalculated and forced through zero. The average of the
calibration results was Peak Area = 2,529,366 (± 250,138) • (SA, ppm), where the error
bar is the standard deviation of the five daily calibration slopes and is equivalent to a
9.9% relative standard deviation. For the daily calibrations, r2 values ranged from 0.9986
to 0.9998.
Cb reference analyses were conducted using a commercial electrochemical sensor
(Drager MiniWarn). The vendor-supplied calibration was used for reference deter-
minations. The upper limit of the MiniWarn was 20 ppm, and good correspondence was
observed up to that limit between the MiniWarn reading and the Cb challenge concen-
tration calculated from dilution of a certified commercial Cb standard (6,015 ppm Cb;
Praxair) in the test system. This correspondence in turn was the basis for relying on the
dilution settings of the test system in preparing Cb concentrations higher than 20 ppm.
Calibration standards for the CW agents GB and HD were prepared by diluting stock
agent to micrograms (Og) per mL concentrations and then injecting a 1-microliter (OL)
volume of each standard into the GC/FPD. Calibration was based on a regression of peak
area versus amount of agent injected.
For GB and FID testing, new calibration plots were prepared at least once a week during
detector evaluation for a total of six GB calibrations and four FID calibrations. The
concentrations of the standards used were 0.0075, 0.1, 0.25, 0.5, and 0.75 ^g/mL for GB
and 0.25, 0.5, 1.0, 2.5, 5.0, and 10 ^g/mL for HD. Low range calibrations were used to
determine agent concentrations for the response threshold and high/low tests. In all cases,
agent concentrations were determined by using the most recent calibration plot. All
calibration plots for both agents were linear, with r2 values of greater than 0.99.
The THC analyzer used to document the interferent levels provided in the evaluation was
calibrated by filling a 25-L Tedlar bag with a 33-ppm propane commercial compliance
class standard (Scott Specialty Gases). Since propane is a three-carbon molecule, this
standard constitutes a THC concentration of 99 ppm of carbon (ppmC). This standard
was used for calibrating the THC analyzer throughout the evaluation. Clean air from the
analytical laboratory was used for zeroing.
3.1.2 Instrument Checks
The ChemPro 100 was operated and maintained according to the vendor's instructions
throughout the evaluation. Maintenance was performed according to predefined
diagnostics. Daily operational check procedures were performed with vendor-supplied
simulant tubes. Proper response of the ChemPro 100 to the simulant was required before
the evaluation could proceed.
-------
3.2 Audits
3.2.1 Performance Evaluation Audit
A PE audit was conducted to assess the quality of reference measurements made in the
evaluation. For AC, SA, and Cb the PE audit was performed once during the evaluation
by diluting and analyzing a standard that was independent of the standards used during
the evaluation. In each case, the primary and audit standards were diluted in exactly the
same way, and analytical results were then compared, with allowance for differences in
the nominal concentrations of the standards. The target tolerance for this PE audit was
±20%. No PE audit was done for CK due to the lack of an independent standard.
Table 3-1 shows that the results of the PE audit were well within the target tolerance for
AC and Cb, and slightly outside the target tolerance for SA. The SA data were reviewed,
but in light of the slight exceedance of the target tolerance, no additional audits were
conducted.
Independent PE audit samples do not exist for GB and HD. Instead, for the CW agents,
check standards of GB and HD were prepared by individuals other than the staff
conducting the reference analyses. The check standards were prepared in the same way as
the reference calibration standards, i.e., by dilution of military grade agent. The results
obtained for these two sets of standards were then compared. For GB, standards were
prepared at concentrations of 0.75, 0.50, 0.25, and 0.1 Og/mL. All results were within 9%
for the separate standards made by two individuals. For HD, standards were prepared at
concentrations of 5, 2.5, 1.0, and 0.5 Og/mL. All results were within 15% for the separate
standards made by two individuals.
Table 3-1. Performance Evaluation Audit Results
TIC
AC
SA
C12
Sample
Standard (Cylinder C74059)
PE Audit Std (Cylinder LL320)
Standard (Cylinder 73486)
PE Audit Std (Cyl. KE50368)
Standard (Cylinder LL23078)
PE Audit Std (Cylinder 152836)
Date of
Audit
3/3/05
3/18/05
3/22/05
Standard
Concentration
1 0,000 ppm
1 0,000 ppm
1090 ppm
997 ppm
6015 ppm
10,200 ppm
Diluted Agreement
Result (%)
45.8 ppm
51.5 ppm
5.9 ppm
6.8 ppm
17.3 ppm
27.2 ppm
11.1
20.6
7.9
3.2.2 Technical Systems Audit
The Battelle Quality Assurance Manager conducted a TSA to ensure that the evaluation
was performed in accordance with the test/QA plan(1) and the TTEP QMP.(2) As part of
the audit, the Battelle Quality Assurance Manager reviewed the reference sampling and
analysis methods used, compared actual evaluation procedures with those specified in the
-------
test/QA plan,(1) and reviewed data acquisition and handling procedures. No significant
adverse findings were noted in this audit. The records concerning the ISA are
permanently stored with the Battelle Quality Assurance Manager.
3.2.3 Data Quality Audit
At least 10% of the data acquired during the evaluation were audited. The Battelle
Quality Assurance Manager traced the data from the initial acquisition, through reduction
and statistical analysis, to final reporting, to ensure the integrity of the reported results.
All calculations performed on the data undergoing the audit were checked.
3.3 QA/QC Reporting
Each assessment and audit was documented in accordance with the test/QA plan(1) and
the QMP.(2) Once the assessment report was prepared by the Battelle Quality Manager, it
was routed to the Test Coordinator and Battelle TTEP Program Manager for review and
approval. The Battelle Quality Manager then distributed the final assessment report to the
EPA Quality Manager and Battelle staff.
10
-------
4.0 Evaluation Results
The ChemPro 100 was evaluated with the TICs AC, CK, SA, and C12 and the CW agents
GB and HD. Test procedures were based on sets of five challenges with a TIC or CW
agent, alternating those challenges with intervals of sampling clean air.(1) Statistical
approaches were used to assess the performance parameters listed in Chapter 1 for the
ChemPro 100 for AC, GB, and HD, as specified in the test/QA plan.(1) Comparable
statistical analyses were not conducted for the other TICs due to inconsistent detector
responses with those chemicals. For those TICs, primarily interference effects and the
accuracy of identification of the TIC were evaluated. Two ChemPro 100 units (Units
1546 and 1811) were used during TIC evaluation, and one ChemPro 100 unit (Unit 1811)
was used during CW agent evaluation. The following sections summarize the findings of
this evaluation; results for both TICs and CW agents are included for each performance
parameter. Note that the target concentrations of GB and HD used in this evaluation were
less than the ChemPro vendor's nominal Low alarm concentrations,^ as described in
Chapter 1.
In all testing with TICs, the ChemPro 100 units were operated using software library
TIC 7.1, and in all CW agent tests, Unit 1811 was operated using software library
CWA-7.1.0.4. One challenge with each TIC or CW agent was also conducted in the
respective "opposite" library. AC and CK produced BLOOD responses in the CWA
library, as expected, since those TICs are in that library. Chlorine produced no response
in the CWA library, as expected, but arsine produced an unexpected BLISTER response
in the CWA library. When using the TIC library, HD produced no response, as expected,
but GB produced an unexpected CHEM HAZARD response.
It is important to note the nature of the inconsistent responses observed from the
ChemPro 100 units with some of the TICs. Throughout all evaluation procedures, the
primary data recorded from the ChemPro 100 units were the alarm indications, the
High/Medium/Low readings, and the display readings identifying the detected chemical
as NERVE, CHEM HAZARD, etc. These records were the primary evaluation data
because these are the responses that a user of the ChemPro 100 would observe during
normal hand-held operation of the instrument. However, Battelle also recorded the IMS
cell signal from each ChemPro 100 unit during all evaluations, using two laptop
computers in the test laboratory. It was often observed that the IMS cell signal would
behave as expected during a test procedure, but the ChemPro 100 alarms and visual
displays would not. For example, the IMS signal would increase when the unit was
challenged with a target chemical, but the unit would not alarm or identify the chemical.
In some instances, a unit would appear to reset itself during a chemical challenge and
11
-------
stop producing an alarm although the challenge continued, but then would sound an
alarm when the challenge gas was replaced with clean air. Based on observations of the
IMS cell signal, this behavior appears to be related to the software in the ChemPro 100
that interprets the IMS signal, rather than to the IMS principle itself.
Tables 4-1 and 4-2 summarize the results of the analysis of response time and other
performance parameters for the TICs and CW agents, respectively. These tables show
data from all evaluations for both ChemPro 100 units for illustration purposes, and the
TIC and CW agent results shown are drawn from data obtained at the target
concentrations (see Table 1-1).
4.1 Response Time
Results of the response time analysis are presented here, focusing on the temperature and
humidity effects for AC, GB, and HD. Note that only challenges in which the
ChemPro 100 actually gave a response are included in the analysis of response time. As
Tables 4-1 and 4-2 show, in 120 total challenges with AC, GB, and HD, a ChemPro 100
response occurred in 86 cases, with no response in 30 cases, and otherwise erroneous
responses (to AC) in four cases. The frequency of failure to respond was greater for CK,
SA, and C12 (23 of 60 challenges for CK, 41 of 60 challenges for SA, and 9 of 10
challenges for Cb).
Unit 1546 for AC - Across the three temperatures [low temperature (5 °C), room
temperature (22 °C), and high temperature (35 °C)] evaluated at medium humidity (50%
RH), the geometric mean time to first response was 18.0, 19.4, and 18.1 seconds,
respectively. Neither the high nor low temperature average times are statistically
significantly different than the room temperature condition. Across the two humidity
levels [low (<20% RH) and medium] that could be evaluated at room temperature, the
geometric mean time to first response was 19.2 and 19.4 seconds, respectively; again, not
a statistically significant difference. Therefore, neither temperature nor humidity had an
effect on time to first response for AC on Unit 1546. Unit 1546 did not respond to AC at
the room temperature, high humidity (80% RH) condition.
Unit 1811 for AC - Across the three temperatures (low, room, and high) evaluated at
medium humidity, the geometric mean time to first response was 15.1, 19.0, and
9.4 seconds, respectively. The high temperature average time was significantly shorter
than that of the room temperature condition. The low temperature average time was not
significantly different from that of the room temperature condition. Across the three
humidity levels (low, medium, and high) evaluated at room temperature, the geometric
mean time to first response was 18.6, 19.0, and 31.0 seconds, respectively. The high
humidity average time was substantially longer than that of the medium humidity
12
-------
Table 4-1. TIC Results from ChemPro 100 Evaluation
TIC Environmental Conditions
AC Control (22°C - 50% RH)
22°C - <20% RH
22°C - 80% RH
35°C - 50% RH
35°C - 80% RH
5°C - 50% RH
CK Control (22°C - 50% RH)
22°C - <20% RH
22°C - 80% RH
35°C - 50% RH
35°C - 80% RH
5°C - 50% RH
SA Control (22°C - 50% RH)
22°C - <20% RH
22°C - 80% RH
35°C - 50% RH
35°C - 80% RH
5°C - 50% RH
C12 Control (22°C - 50% RH)
ChemPro 100
Response
M
M
L
M(5)/H(5)
M
L(2)/M(8)
M
M
L
L(4)/M(5)
M
L(5)/M(3)
L
L
L
L
L(1)/H(5)
L
L
. , Response
Alarms „ (a)
n t- .L i *-.i • ix Time Ranee v '
(Indicated Chemical) (Seconds)
10/10 (TOXIC)
10/10 (TOXIC)
5/10 (TOXIC)
5/10 (NR)00
5/10 (TOXIC)
5/1 0(CHEM HAZARD)
2/5 (TOXIC)(d)
3/5 (NR)
10/10 (TOXIC)
4/10 (TOXIC)
6/10 (NR)
3/10 (TOXIC)
7/10 (NR)
5/10 (TOXIC)
5/10 (NR)
9/10 (TOXIC)
1/10 (NR)
3/5 (TOXIC)(e)
2/5 (NR)
8/10 (TOXIC)
2/10 (NR)
4/10 (TOXIC)
6/10 (NR)
5/10 (TOXIC)
5/10 (NR)
1/10 (TOXIC)
1/10 (CHEM HAZARD)
8/10 (NR)
1/10 (TOXIC)
9/10 (NR)
1/10 (TOXIC)
5/10 (CHEM HAZARD)
4/10 (NR)
1/10 (TOXIC)
9/10 (NR)
1/10 (TOXIC)
9/10 (NR)
18-20
18-20
30-32
6-21
18-20
7-20
26-27
31-32
54-92
20-24
23-25
23-29
52-77
35-70
79-83
38
21-82
53
71
Recovery
Time Range(a)
(Seconds)
28-33
24-27
28-41
24-600(c)
35-36
31-600
29-33
26-32
23-44
26-36
21-35
45-57
8-10
12-45
12-41
77
30-94
50
17
® Response and recovery time evaluated only when the ChemPro 100 showed response to the challenge.
^ NR = No response.
^ 600 seconds = Maximum time monitored for detector recovery time.
^ UNIT 1811 (Response not included in table) - Alarmed and cleared during first challenge (taken as an accurate
response), alarmed CHEM HAZARD High on clean air after first challenge, did not change alarm or clear for
remainder of evaluation.
(e) UNIT 1811 (Response not included in table) - Alarmed as CHEM HAZARD High prior to first challenge, did not
change alarm or clear for remainder of evaluation.
13
-------
Table 4-2. CW Agent Results from ChemPro 100 Evaluation
_,,, _ . , , _, _ ... Alarms Response
CW Environmental ChemPro 100 ., ,. ^ , _,. „ (a)
. _ .... _ (Indicated Time Range' '
Agent Conditions Response ,_,, . 1N ._, , .
Chemical) (Seconds)
GB Control (22°C - 50% RH) h
22°C - <20% RH
22°C - 80% RH
35°C - 50% RH L
35°C - 80% RH
5°C - 50% RH
HD Control (22°C - 50% RH)
22°C - <20% RH
22°C - 80% RH
35°C - 50% RH
35°C - 80% RH
5°C - 50% RH
1 (4) / H (
M
-
, (4)/M(:
-
H
L
L
L
L
L
-
1) 5/5 (NERVE)
5/5 (NERVE)
5/5 (NR)(b)
1) 5/5 (NERVE)
5/5 (NR)
5/5 (NERVE)
5/5 (BLISTER)
5/5 (BLISTER)
5/5 (BLISTER)
1/5 (BLISTER)
4/5 (NR)
2/5 (BLISTER)
3/5 (NR)
5/5 (NR)
8-12
9-10
-
12-15
-
11-14
35-42
31-34
26-31
225
82-142
-
Recovery
Time Range(a)
(Seconds)
10-15
11-13
-
6-13(c)
-
10-16
18-22(d)
24-30
28-45
14
65-600(e)
-
•! Response and recovery time evaluated only when the ChemPro 100 showed response to the challenge.
00 NR = No response.
^ During one agent challenge, the unit cleared while still being exposed to GB at 27 seconds.
*• During one agent challenge, the unit cleared while still being exposed to HD at 67 seconds.
® 600 seconds = Maximum time monitored for detector recovery time.
condition. The low humidity average time was not significantly different from the
medium humidity condition. Therefore, high temperature was linked to a lower time to
first response while high humidity was linked to a greater time to first response for AC on
Unit 1811.
Unit 1811 for GB - Across the three temperatures (low, room, and high) evaluated at
medium humidity, the geometric mean time to first response was 12.2, 9.9, and
13.4 seconds, respectively. The high temperature average response time was statistically
significantly greater than the room temperature condition. Across the two humidity levels
(low and medium) that could be evaluated at room temperature, the geometric mean time
to first response was 9.6 and 9.9 seconds, respectively. This did not represent a
statistically significant difference. Unit 1811 did not respond to GB at the room
temperature, high RH condition.
Unit 1811 for HD - Between the two temperatures (room and high) that could be
evaluated at medium humidity, the geometric mean times to first response were 37.5 and
225 seconds, respectively. The high temperature average time was statistically
significantly longer than the room temperature condition, but note that the high
temperature medium RH result is based on a single response out of five trials. Unit 1811
did not respond to HD at the low temperature, medium RH condition. Across the three
humidity levels (low, medium, and high) evaluated at room temperature, the geometric
mean time to first response was 32.4, 37.5, and 29.0 seconds, respectively. The low
14
-------
humidity and high humidity average times were statistically significantly lower than the
medium humidity average time to first response.
4.2 Recovery Time
Results of the recovery time analysis are presented below, focusing on temperature and
humidity effects for AC, GB, and HD, and results from all tests are presented in
Table 4-1 and Table 4-2. As with response time, recovery time was evaluated only when
the ChemPro 100 responded to a challenge mixture. Note that in one challenge each with
AC, GB, and HD, the ChemPro 100 cleared its alarm while the challenge was in
progress; these cases also were excluded from the evaluation of recovery time.
Unit 1546 for AC - Of the observations across the three temperatures (low, room, and
high) evaluated at medium humidity, the geometric mean recovery times were 38.7, 30.7,
and 33.0 seconds, respectively. (Note that one observation at low temperature is excluded
from this analysis because it did not clear within 600 seconds; its time to clear is,
therefore, unknown.) From these data, the low temperature average time to clear was
significantly longer than that of the room temperature condition. The high temperature
average time was not significantly different from the room temperature condition. Across
the two humidity levels (low and medium) that could be evaluated at room temperature,
the geometric mean recovery time was 25.2 and 30.7 seconds, respectively. The low
humidity average recovery time was significantly shorter than that of the medium
humidity condition. Therefore, low temperature was linked to longer recovery times
while low humidity was linked to faster recovery times for AC on Unit 1546. Unit 1546
did not respond to AC at the room temperature, high humidity condition.
Unit 1811 for AC - With this unit, one observation at the low temperature, medium
humidity condition did not clear within 600 seconds after the completion of the
challenge, and none of the observations at the high temperature, medium humidity
condition cleared within 600 seconds after the completion of the challenge. These
observations are excluded from the statistical analysis comparing recovery times. Of the
remaining observations across the two temperatures (low and room) that could be
evaluated at medium humidity, the geometric mean recovery times were 32.7 and
29.6 seconds, respectively, indicating minimal effect of temperature on recovery time for
AC. Across the three humidity levels (low, medium, and high) evaluated at room
temperature, the geometric mean recovery times were 25.2, 29.6, and 33.3 seconds,
respectively. The low humidity average recovery time was significantly shorter than that
of the room temperature, medium humidity condition. The high humidity mean recovery
time was not statistically significantly different from the medium humidity condition.
Therefore, low temperature was weakly linked to longer recovery times while low
humidity was linked to faster recovery times for AC on Unit 1811.
Unit 1811 for GB - Of the observations across the three temperatures (low, room, and
high) evaluated at medium humidity, the geometric mean recovery times were 13.4, 12.9,
15
-------
and 9.4 seconds, respectively. This does not represent a statistically significant difference
for either the high or low temperature compared to room temperature. Across the two
humidity levels (low and medium) that could be evaluated at room temperature, the
geometric mean recovery times of 11.8 and 12.9 seconds, respectively; were not
statistically significantly different.
Unit 1811 for HD - At the two temperatures (room and high) that could be evaluated at
medium humidity, the geometric mean time to clear was 19.7, and 14.0 seconds,
respectively. The high temperature time to clear was statistically significantly lower than
the room temperature time to clear, but note that the high temperature recovery time is
based on a single response out of five trials. Across the three humidity levels (low,
medium, and high) evaluated at room temperature, the geometric mean time to clear was
26.3, 19.7, and 35.2 seconds, respectively. The low humidity and high humidity average
times to clear were statistically significantly longer than the medium humidity average
time to clear.
4.3 Accuracy
Results of the accuracy analysis are summarized below and presented in Table 4-1 and
Table 4-2. The accuracy of a unit was defined as the proportion of trials in which the unit
registered an accurate response to the challenge. The ChemPro 100 was considered
accurate if it alarmed in the presence of the TIC or CW agent and correctly identified the
TIC or CW agent class. This analysis was conducted for all TICs and CW agents. For the
ChemPro 100, any level of response (Low, Medium, or High) and either "TOXIC" or
"CHEM HAZARD" were considered by the manufacturer to be accurate for evaluating
with TICs. Also, any level of response (Low, Medium, or High) and "NERVE" for GB
and "BLISTER" for HD were considered by the manufacturer to be accurate for
evaluations with CW agents. As noted in Section 4.1, in 30 of the 120 challenges with
AC, GB, and HD, no ChemPro 100 response occurred; those 30 cases are, by definition,
inaccurate responses. Lack of response and/or erroneous positive responses were also
seen with CK, SA, and C12.
Unit 1546 - For AC, Unit 1546 displayed 100% accuracy for the room temperature, low
humidity condition as well as all three temperature conditions at medium humidity. For
the room temperature, high humidity testing, 0% accuracy was observed; and for the high
temperature, high humidity condition, 40% accuracy was observed. For AC, there was no
observed effect of temperature at medium humidity on accuracy. There was a statistically
significant effect of humidity on accuracy, with lower accuracy at high humidity.
For CK, there was no statistically significant effect for temperature at medium humidity
where the accuracy values for low, room, and high temperature were 60%, 80%, and
100%, respectively. A statistically significant effect of humidity on accuracy for
operation at room temperature was observed. The low humidity condition displayed
accuracy of 40%, the medium humidity condition accuracy was 80%, and the high
16
-------
humidity condition accuracy was 0%. The accuracy of the high temperature, high
humidity condition was 60%.
For SA, the unit accuracy was very low for all evaluated conditions. At room temperature
and medium humidity, the accuracy was 0%. For all other temperature and humidity
conditions, accuracy was 20%. Neither temperature nor humidity exhibited statistically
significant effects.
For C\2, the only temperature and humidity condition evaluated was room temperature
and medium humidity. None of five trials had a response, representing 0% accuracy. No
further analysis was possible.
Unit 1811 - For AC, Unit 1811 displayed 100% accuracy for the room temperature, low
humidity condition; all three temperature conditions at medium humidity; and for room
temperature, high humidity. For the high temperature, high humidity condition, the unit
began alarming for the first trial (judged an accurate response), but continued to alarm
through the remaining four trials and the intervening clean air purges. Accuracy thus was
20% at that condition. For AC, there was no statistically significant effect for the three
evaluated temperatures at medium humidity. Similarly, no significant effect was observed
for the three evaluated humidity levels at room temperature.
For CK, there was a statistically significant effect for temperature at medium humidity
where the accuracy values for low, room, and high temperature were 100%, 0%, and
80%, respectively. There was also a statistically significant effect of humidity on
accuracy for operation at room temperature. The low humidity condition displayed
accuracy of 20%, the medium humidity condition accuracy was 0%, and the high
humidity condition accuracy was 100%. The accuracy at the high temperature, high
humidity condition was 0%, because the unit began alarming prior to the first challenge
and continued alarming through the remaining trials and the intervening clean air purges.
For SA, there was a statistically significant effect for temperature at medium humidity
where the accuracy values for low, room, and high temperature were 0%, 80%, and 0%,
respectively. No statistically significant effect was observed for humidity on accuracy for
operation at room temperature. The low and medium humidity conditions both displayed
accuracy of 80% while the high humidity condition accuracy was 20%. Oddly, while the
accuracy rates at both high temperature at medium humidity (0%) and high humidity at
room temperature (20%) were very low, the accuracy at the high temperature, high
humidity condition was 100%.
For Cb, the only temperature and humidity condition evaluated was room temperature
and medium humidity. Only one trial in five had a response, representing 20% accuracy.
No further analysis was possible.
For GB, the unit displayed 100% accuracy for the room temperature/low humidity and
room temperature/medium humidity conditions as well as the low temperature/medium
17
-------
humidity condition. For the room temperature/high humidity evaluation, 0% accuracy
was observed with no response from the detector in any of the five trials. For the high
temperature/medium humidity condition, 80% accuracy was observed. The one
inaccurate test at this condition initially responded accurately with a "NERVE" response
but the unit subsequently cleared during the challenge period. This test was consequently
identified to not be accurate and the data for time to clear was eliminated from further
analysis. However, the data for maximum response on challenge and time to first
response were retained for their respective analyses. For GB, there was not a statistically
significant effect of temperature on accuracy at medium humidity. There was a
statistically significant effect of humidity on accuracy for operation at room temperature
with the conclusion that accuracy at high humidity was significantly poorer than at low or
medium humidity. This was further supported by evaluation results at high
temperature/high humidity where accuracy of response was also 0%.
For FID, the unit displayed 80% accuracy for the room temperature/medium humidity
condition. The one inaccurate test at this condition initially responded with a "NERVE"
response but the unit subsequently cleared during the challenge period. This test was
identified to be inaccurate, and the data for time to clear was eliminated from further
analysis. However, the data for maximum response on challenge and time to first
response were retained for their respective analyses. The low temperature/medium
humidity condition had 0% accuracy while the high temperature/medium humidity
condition had 20% accuracy. For these two conditions, all inaccurate trials were
categorized as such because there was no response from the detector. The differences in
accuracy were large enough to conclude that there was a statistically significant effect of
temperature on accuracy at medium humidity. At room temperature, the low and high
humidity conditions showed better accuracy (100%) than the medium humidity condition
(80%) but the difference was not large enough to be statistically significant. Accuracy
was 40% at the high temperature and high humidity condition.
High/Low - For the high/low test, the ChemPro 100 was challenged with either a high
concentration of chemical followed by a low concentration, or a low concentration of
chemical followed by a high concentration. For AC, Unit 1811 responded with either a
low or medium "TOXIC" alarm at high concentration. At low concentration, the unit
either did not respond or responded with a low "TOXIC" alarm. This resulted in no
change in alarm level when the unit was challenged with a low concentration first, and a
change in the alarm level when the unit was challenged with a high concentration first.
For AC, Unit 1546 alarmed at medium "TOXIC" for all the high concentration
challenges and low "TOXIC" for all the low concentration challenges. The order of the
challenge did not affect the respective alarm level. For GB, Unit 1546 responded with a
medium or low "NERVE" alarm at high concentration and a low "NERVE" alarm or no
alarm at low concentration. There was always an accurate change in the level of the alarm
for GB. For HD, Unit B responded with a low "BLISTER" alarm both at high concen-
tration and low concentration. There was no distinction or change in alarm level between
the two concentrations.
18
-------
4.4 Repeatability
Results of the repeatability analysis are summarized below. As with response and
recovery times (Sections 4.1 and 4.2, respectively), the evaluation of repeatability
includes only those cases in which the ChemPro 100 responded to a TIC or CW agent
challenge. Repeatability addressed the consistency of the Low, Medium, and High
readings of the ChemPro 100.
Unit 1546 for AC - For each trial that had a response, the maximum observed response
level from the ordered progression (Low, Medium, and High) was identified. Across the
three temperatures (low, room, and high) evaluated at medium humidity, the maximum
level of alarm (Medium) for each trial was identical. Similarly, across the low and
medium humidity conditions that could be evaluated at room temperature, the maximum
level of alarm (Medium) for each trial was also identical. These results show no observed
effect of temperature or humidity on the maximum response for this unit.
Unit 1811 for AC - For each trial that had a response, the maximum observed response
level from the ordered progression (Low, Medium, and High) was identified. A
statistically significant change in maximum response was observed across temperatures
with this unit. Across the three temperatures (low, room, and high) evaluated at medium
humidity, the level of the maximum alarm increased with values of "Low" and
"Medium" at low temperature, "Medium" for all trials at room temperature, and "High"
for all trials at high temperature. A statistically significant change in maximum response
was also observed across humidity conditions. Across the three levels of humidity
evaluated at room temperature, the level of alarm decreased. The observed maximum
responses (Medium) for all trials were identical at low and medium humidity, but the
maximum response dropped to "Low" for the high humidity condition.
Unit 1811 for GB - For each trial that had a response, the maximum observed response
level from the ordered progression (Low, Medium, and High) was identified. Across the
three temperatures (low, room, and high) evaluated at medium humidity, the maximum
level of alarm appeared to decrease with increasing temperature. The lowest temperature
condition produced "High" alarms, the room temperature trials showed one "High" alarm
but 4 "Medium" alarms and the high temperature condition showed only 1 "Medium"
alarm and 4 "Low" alarms (one of these "Low" alarms was the trial that cleared during
the challenge). This effect was statistically significant. Between the two humidity
conditions with accurate responses that could be evaluated at room temperature, no
significant effect was seen in maximum response level. In one case at the high
temperature/medium humidity condition, the unit cleared during the challenge.
Unit 1811 for HD - For each trial that had a response, the maximum observed response
level from the ordered progression (Low, Medium, and High) was identified. In all cases
for HD, this response level was "Low" and therefore no temperature or humidity effects
19
-------
were seen in the level of response. In one case at the room temperature/medium humidity
condition, the unit cleared during the challenge.
4.5 Response Threshold
Response thresholds were determined by challenging the ChemPro 100 with successively
lower concentrations of TIC or CW agent until it no longer responded or the response
was not maintained during a challenge. Table 4-3 provides the results for the response
threshold tests for each ChemPro 100 unit. The concentrations used in each of these tests
are given in the table and, for most of the TICs, are well below the concentrations used in
the other evaluations. For the CW agents, the concentrations used are also below the
target concentrations used in the other evaluations.
Table 4-3. Response Threshold Data for the TIC and CW Agent Evaluation
TIC/CW Agent
(Concentration)
AC (50 ppm)
AC (25 ppm)
AC (12.5 ppm)
AC (6 ppm)
AC (3 ppm)
CK (20 ppm)
CK(lOppm)
CK (5 ppm)
SA (6 ppm)
SA (3 ppm)
C12 (60 ppm)
GB (0.01 mg/m3)
HD (0.2 mg/m3)
ChemPro 100 Identification Number
1546
L TOXIC (2) /M TOXIC (1)
L TOXIC (3)
L TOXIC (3)
No Response (1) / L TOXIC (2)
No Response (3)
No Response (2) / L TOXIC (3)
No Response (5)
No Response (3)
No Response (4) / L TOXIC (1)
No Response (3)
No Response (5)
NA
NA
1811
No Response (1) / L TOXIC (2)
L TOXIC (3)
L TOXIC (3)
No Response (2) / L TOXIC (1)
No Response (3)
No Response (4) / L TOXIC (1)
No Response (1) / L TOXIC (4)
No Response (3)
No Response (1) / L TOXIC (4)
No Response (3)
No Response (4) / L TOXIC (1)
No Response (1) / L NERVE (9)(a)
No Response (5) / L BLISTER (5)
** In six of the nine responses, the unit cleared within seconds of alarming and did not alarm again during the
remainder of the GB challenge.
Table 4-3 shows that for AC, the response threshold was between 3 and 6 ppm on both
ChemPro 100 units. For CK, the response threshold was between 10 and 20 ppm for Unit
1546 and between 5 and 10 ppm for Unit 1811. For SA, the response threshold was
between 3 and 6 ppm for both units, and for Cb, the response threshold was at or above
about 60 ppm as only Unit 1811 responded one time at 60 ppm.
For GB, the response threshold was around 0.01 mg/m3 for Unit 1811. The ChemPro 100
responded to nine out of 10 challenges at that concentration; however, in six of those
20
-------
cases, the unit stopped responding within seconds after first responding, while still being
challenged with GB. For HD, the response threshold was around 0.2 mg/m3 for Unit 1811
as it responded to five out of 10 challenges with HD at that concentration.
4.6 Temperature and Humidity Effects
The effects of temperature and humidity on the ChemPro 100 are summarized in
Sections 4.1 through 4.4.
4.7 Interference Effects
Five interferents (latex paint fumes, ammonia floor cleaner vapors, air freshener vapors,
gasoline engine exhaust hydrocarbons, and DEAE) were used in the evaluation. The
effect of these interferences on the ChemPro 100 response is summarized below and in
Table 4-4. This table summarizes results from both ChemPro units for illustration
purposes, but does not include false positive readings.
False Positive - A false positive response was noted if the ChemPro 100 responded and
provided an alarm in the presence of an interferent alone (i.e., in the absence of a TIC or
CW agent). A false positive was defined as any alarm under those conditions.
Other erroneous positive responses were observed during testing of accuracy
(Section 4.3) in the form of alarms that occurred during sampling of clean air. These are
noted in the footnotes of Table 4-1.
Unit 1546 (false positive) - In the TIC library, a false positive rate of 100% was
observed for paint fumes, ammonia cleaner, and air freshener. Engine exhaust and DEAE
did not produce any false positives in the three separate trials conducted for each
interferent.
In the CW agent library, a false positive rate of 100% was observed for ammonia cleaner
and air freshener. Paint produced a false positive rate of 40%. Engine exhaust produced a
false positive rate of 20%. DEAE did not produce any false positives in the three separate
trials conducted for this interferent in the CW agent library.
Unit 1811 (false positive) - In the TIC library, a false positive rate of 67% was observed
for paint fumes, 80% for ammonia cleaner, and 100% for air freshener. Engine exhaust
and DEAE did not produce any false positives in the three separate trials conducted for
each interferent.
21
-------
Table 4-4. Interference Effects
TICorCW ¥ , ,
Interierent
Agent
AC Control
Paint Fumes
Floor Cleaner
Air Freshener
Gasoline Engine Exhaust
DEAE
CK Control
Paint Fumes
Floor Cleaner
Air Freshener
Gasoline Engine Exhaust
DEAE
SA Control
Paint Fumes
Floor Cleaner
Air Freshener
Gasoline Engine Exhaust
DEAE
GB Control
Paint Fumes
Floor Cleaner
Air Freshener
Gasoline Engine Exhaust
DEAE
HD Control
Paint Fumes
Floor Cleaner
Air Freshener
Gasoline Engine Exhaust
DEAE
ChemPro 100
Response
M
H
H
M(5)/H(1)
L (4) / M (4)
L(7)/M(1)
M
L(3)/M(5)
M (4) / H (3)
M (2) / H (3)
L(2)/M(5)
L
L
M(2)/H(2)
L(3)/M(7)
M
L
-
M(4)/H(1)
H
M (2) / H (3)
M
L
L
L
L(2)/M(1)/H
(2)
L(1)/M(4)
M
L
L
Alarms
(Indicated Chemical)
10/10 (TOXIC)
9/10 (CHEM HAZARD)
1/10 (NR)(a)
5/10 (CHEM HAZARD)
5/10 (NR)
6/10 (CHEM HAZARD)
4/10 (NR)
8/10 (TOXIC)
2/10 (NR)
8/10 (TOXIC)
2/10 (NR)
4/10 (TOXIC)
6/10 (NR)
5/10 (CHEM HAZARD)
3/10 (TOXIC)
2/10 (NR)
7/10 (CHEM HAZARD)
3/10 (NR)
5/10 (CHEM HAZARD)
5/10 (NR)
5/10 (CHEM HAZARD)
2/10 (TOXIC)
3/10 (NR)
3/10 (TOXIC)
7/10 (NR)
4/10 (TOXIC)
6/10 (NR)
4/10 (CHEM HAZARD)
6/10 (NR)
10/10 (CHEM HAZARD)
10/10 (CHEM HAZARD)
1/10 (TOXIC)
9/10 (NR)
6/6 (NR)
5/5 (NERVE)
5/5 (NERVE)
5/5 (CHEM HAZARD)
3/5 (NERVE)
2/5 (CHEM HAZARD)
1/5 (CHEM HAZARD)
4/5 (NR)
5/5 (NERVE)
5/5 (BLISTER)
2/5 (NERVE)
3/5 (CHEM HAZARD)
1/5 (BLISTER)
4/5 (CHEM HAZARD)
5/5 (CHEM HAZARD)
5/5 (BLISTER)
5/5 (BLISTER)
Response Time
Range (Seconds)
18-20
17-20
15-18
19-21
21-25
20-23
26-27
19-108
17-19
18-26
25-83
25-35
52-77
31-64
19-24
21-26
67
-
8-12
9-10
15-18
12-20
26
12-16
35-42
10-71
18-32
22-25
27-35
39-51
Recovery Time
Range (Seconds)
28-33
66-93
31-45
13-20
13-23
17-29
29-33
18-57
19-44
17-26
26-62
18-27
8-10
146-340
36-74
33-95
11
-
10-15
145-508
18-19
12-23
6
6-12
18-22(b)
35-600(c)
20-44(d)
35-46
19-21
8-16
(a)NR = No response.
^ During one agent challenge, the unit cleared while still being exposed to HD at 67 seconds.
^ 600 seconds = Maximum time monitored for detector recovery time.
^d' During one agent challenge, the unit cleared while still being exposed to HD at 118 seconds.
22
-------
In the CW agent library, a false positive rate of 100% was observed for ammonia cleaner
and air freshener. Paint produced a false positive rate of 20%. DEAE did not produce any
false positives in the three separate trials conducted for this interferent. Engine exhaust
false positive testing was not conducted for this unit because the detector stopped
functioning before the tests could be run.
False Negative - A false negative response was noted if the presence of an interferent
masked the presence of a TIC or CW agent and the ChemPro 100 provided a lower
response or did not respond to the TIC or CW agent. Changes in response, response time,
and recovery time due to interferences are discussed in the following paragraphs.
Other erroneous negative responses (i.e., the failure to respond to a TIC or CW agent
challenge in clean air) are discussed under accuracy (Section 4.3), and also occurred
during testing of cold-/hot-start behavior (Section 4.8) and battery life (Section 4.9). In
addition, the few instances in which the ChemPro 100 cleared its alarm while a TIC or
CW agent challenge was in progress (Sections 4.2, 4.3, and below in this section) are also
erroneous negative responses.
Unit 1546 (false negative) - For this unit, the accuracy in detecting AC in the presence
of the interferents was similar to the accuracy without the interferent. Accuracy was
100% for the non-interferent evaluation as well as for the evaluations with paint and
ammonia cleaner. The accuracy for the evaluations with air freshener, engine exhaust,
and DEAE as interferents was 80%. Thus, no overall accuracy effect was observed when
interferent was added to the AC.
The interferents did exhibit a statistically significant effect on the maximum level of
response observed for AC. All five responses for the non-interferent evaluation reached a
Medium alarm level, as did all four of the responses for the engine exhaust evaluation
and three of the four responses for the air freshener evaluation. The paint and ammonia
cleaner tests each saw five of five alarms at the High level. DEAE had three of its four
observations at the Low alarm level.
The geometric mean time to first response to AC for the non-interferent evaluation was
19.4 seconds. Paint and ammonia cleaner results displayed statistically significant shorter
average response times. For paint, the average time was 17.8 seconds, and for ammonia
cleaner, the average time was 16.4 seconds. At 19.5 seconds and 20.8 seconds, the
response times for air freshener and DEAE, respectively, were comparable to that for no
interferent. Only engine exhaust showed a statistically significant longer response time, at
24.3 seconds.
The geometric mean recovery time for the non-interferent AC evaluation was
30.7 seconds. The air freshener, engine exhaust, and DEAE all produced shorter recovery
times with estimates of 17.2 seconds, 21.0 seconds, and 21.7 seconds, respectively. At
35.5 seconds, the mean recovery time for ammonia cleaner was not statistically
23
-------
significantly different from the non-interferent value. Only paint as an interferent
produced a longer recovery time at 84.8 seconds.
The accuracy in detecting CK in the presence of the interferents varied by interferent.
Accuracy was 80% for the non-interferent. For paint, ammonia cleaner, air freshener, and
engine exhaust, the accuracy was 100%. With DEAE as an interferent, the accuracy was
0%.
The accuracy in detecting SA in the presence of the interferents varied by interferent.
Accuracy was 0% for the non-interferent and for paint, engine exhaust, and DEAE. For
ammonia cleaner and air freshener, the accuracy was 100%.
Unit 1811 (false negative) - For this unit, there was some variability in the accuracy of
response for AC for this unit with different interferents. The evaluation without
interferent had 100% accuracy. An accuracy of 80% was observed for the evaluations
with paint, engine exhaust, and DEAE as interferents. Only 40% accuracy was observed
for the evaluation with air freshener as an interferent, and 0% accuracy was observed for
the ammonia cleaner evaluation. Thus, a statistically significant effect of interferents on
the accuracy of detection of AC was observed.
The interferents also exhibited a statistically significant effect in the maximum level of
response observed for AC. All five responses for the non-interferent evaluation reached a
Medium alarm level, as did both responses for the air freshener evaluation. By contrast,
all four responses for the paint evaluation showed a maximum response at the High level
and all four responses for both engine exhaust and DEAE showed a maximum response
at the Low level.
The geometric mean time to first response to AC for the non-interferent evaluation was
19.0 seconds. At 18.5 seconds and 20.5 seconds, the response times for paint and air
freshener, respectively, were comparable to that with no interferent. Both engine exhaust
and DEAE showed slightly longer response times than the non-interferent tests. The
engine exhaust time was 22.2 seconds, while the DEAE response time was 21.7 seconds.
With no accurate responses, the time to first response of ammonia cleaner could not be
analyzed.
The geometric mean recovery time for AC for the non-interferent evaluation was
29.6 seconds. The air freshener, engine exhaust, and DEAE all produced shorter recovery
times with estimates of 13.5 seconds, 14.9 seconds, and 19.5 seconds, respectively. With
no accurate responses, the recovery time for ammonia cleaner could not be analyzed.
Only paint as an interferent produced a longer recovery time at 75.3 seconds.
The accuracy in detecting CK in the presence of the interferents varied by interferent and
was clearly subject to the variability in ChemPro 100 response noted above. Accuracy
was 0% for the non-interferent evaluation, and also for air freshener. For ammonia
24
-------
cleaner and engine exhaust, the accuracy was 40%. With paint and DEAE as interferents,
the accuracy was 60%.
The accuracy in detecting SA in the presence of the interferents varied by interferent.
Accuracy was 80% for the non-interferent and for paint. For ammonia cleaner and air
freshener, the accuracy was 100%. With engine exhaust as an interferent the accuracy
was 20%, and accuracy was 0% for DEAE as an interferent.
The accuracy of Unit 1811 in detecting GB was not the same for all interferents
evaluated. The detector exhibited a range of different behaviors for the interferents
evaluated in the presence of GB.
• Paint - All five trials responded to the challenge with the appropriate "NERVE"
response. However, when the challenge was terminated and clean air passed into the
system in each of the trials, the unit produced a Low-level "CHEM HAZARD" alarm.
Since this spurious alarm was after completion of the challenge (where the correct
alarm was observed) and the unit subsequently cleared, these trials were ultimately
counted as accurate.
• Ammonia Cleaner - All five trials responded to the challenge, but in each case with
the inaccurate "CHEM HAZARD" response. Since no "NERVE" response was
observed, these trials were all counted as inaccurate.
• Air Freshener - Three of the five trials responded accurately to the challenge with a
"NERVE" response. The other two trials exhibited an inaccurate "CHEM HAZARD"
response and were counted as inaccurate.
• Engine Exhaust - One of the five trials responded to the challenge with the inaccurate
"CHEM HAZARD" response, while the other four trials resulted in no response at
all. Hence, accuracy was determined to be 0%.
• DEAE - All five trials responded accurately to the challenge in the presence of this
interferent.
In examining the individual interferent accuracy rates, only the ammonia cleaner and
engine exhaust accuracy rates were statistically significantly different than the non-
interferent evaluation.
After determining that the interferents did seem to affect the accuracy of identifying GB,
further analysis was performed on the maximum response level, time to first response,
and recovery time for each interferent compared to the non-interferent evaluation. In the
instances where the ChemPro 100 responded but produced an inaccurate response, the
maximum response level, time to first response, and time to clear from the inaccurate
responses are also provided.
The interferents showed a statistically significant effect on the maximum level of
response observed from Unit 1811 for GB. Four of the five responses for the non-
interferent evaluation reached a Medium alarm level and one reached a High alarm level.
The interferents appeared to affect the maximum response differently:
25
-------
• Paint as an interferent strengthened the maximum response level to a high alarm for
all trials.
• Ammonia cleaner alarmed two trials at medium and three at high but as "CHEM
HAZARD" rather than "NERVE" in each case.
• Air freshener alarmed at medium for all five trials but in two cases as "CHEM
HAZARD" rather than "NERVE."
• Engine exhaust seemed to cancel the response except in one trial where a low
"CHEM HAZARD" response was observed.
• DEAE appeared to reduce the maximum alarm level as all five trials exhibited a low
alarm level.
The geometric mean time to first response to GB for the non-interferent evaluation was
9.9 seconds. Ammonia cleaner, air freshener, engine exhaust, and DEAE displayed
statistically significant longer average response times. For ammonia cleaner, the average
time was 15.6 seconds. For air freshener, the average time was 15.2 seconds. For engine
exhaust, the average time was 26.0 seconds. For DEAE, the average time was
13.9 seconds. At 9.4 seconds, the average response time for paint was comparable to no
interferent.
The geometric mean recovery time for GB for the non-interferent evaluation was
12.9 seconds. Paint displayed statistically significant longer average recovery time at
236 seconds. At 18.8, 15.7, 6.0, and 9.5 seconds, the recovery times for ammonia cleaner,
air freshener, engine exhaust, and DEAE, respectively, were not significantly different
from the non-interferent evaluation.
The accuracy of Unit 1811 in detecting HD was not the same for all interferents, and the
detector exhibited a range of different behaviors for the interferents:
• Paint - All five trials responded to the challenge but two of them with the "NERVE"
response and three with the "CHEM HAZARD" response when the accurate response
to HD is "BLISTER." Therefore, this unit showed 0% accuracy in the presence of
paint.
• Ammonia Cleaner - All five trials responded to the challenge, in four cases with the
inaccurate "CHEM HAZARD" response and in the fifth case with the correct
"BLISTER" response, but in that case the unit subsequently cleared while the
challenge was going on. All these responses were judged inaccurate; therefore, this
unit showed 0% accuracy in the presence of ammonia cleaner.
• Air Freshener - All five trials responded to the challenge but all with the incorrect
"CHEM HAZARD" response. Therefore, this unit showed 0% accuracy in the
presence of air freshener.
• Engine Exhaust - The unit showed 100% accuracy despite the presence of this
interferent.
• DEAE - The unit showed 100% accuracy despite the presence of this interferent.
26
-------
In comparing the individual interferent accuracy rates, paint, ammonia cleaner, and air
freshener accuracy rates were all statistically significantly lower than that for the non-
interferent evaluation.
After determining that the interferents did seem to affect the accuracy of identifying the
HD agent, further analysis was performed on the maximum response level, time to first
response, and recovery time for each interferent compared to the non-interferent
evaluation. Note that these analyses did incorporate data from trials determined to be
inaccurate if such data were appropriate. For example, the time to first response analysis
uses data from trials that recorded an alarm, even if it was the incorrect alarm type.
The interferents exhibited a statistically significant effect on the maximum level of
response observed from Unit 1811 to HD. All five responses for the non-interferent
evaluation reached a Low alarm level as did all five responses for both the engine exhaust
and DEAE evaluations. The other three interferents generally showed higher alarm
levels. Four responses for the ammonia cleaner evaluation showed a maximum response
at the medium level with the incorrect alarm designation of "CHEM HAZARD." The air
freshener evaluation had five incorrect "CHEM HAZARD" responses, which all reached
a medium alarm level. Of the five responses to paint, two trials alarmed at the high level
for "NERVE," two alarmed at the low level for "CHEM HAZARD," and one alarmed at
the medium level for "CHEM HAZARD."
The geometric mean time to first response to HD for the non-interferent evaluation was
37.5 seconds. At 26.1, 23.8, 28.9, and 44.4 seconds, the response times for paint, air
freshener, engine exhaust, and DEAE, respectively, were comparable to no interferent. At
20.9 seconds, ammonia cleaner showed statistically significantly shorter average
response time than that of the non-interferent evaluation.
The geometric mean recovery time for HD for the non-interferent evaluation was
19.7 seconds. At 25.8, 40.0, 19.6, and 10.6 seconds, the recovery times for ammonia
cleaner, air freshener, engine exhaust, and DEAE, respectively, were comparable to no
interferent. At 107 seconds, paint showed statistically significantly longer average
recovery time than that of the non-interferent evaluation.
4.8 Cold-/Hot-Start Behavior
Analysis of the effects of insufficient warm-up time, under start-up conditions ranging
from cold (5 to 8°C) to hot (40°C), are summarized below. Table 4-5 illustrates the data
obtained in evaluating for cold-/hot-start effects, showing the ChemPro 100 units used,
the start condition, delay time, sequential experiment number, response reading, response
and recovery times, and alarm indication. Such evaluation was conducted only with AC
at the IDLH concentration.
27
-------
Table 4-5. Start State Effects
ChemPro
100 Unit
1546
1811
Start Condition
Control
Room
Temperature
(Cold Start)
Cold
Temperature
(Cold Start)
Hot
Temperature
(Cold Start)
Control
Room
Temperature
(Cold Start)
Cold
Temperature
(Cold Start)
Hot
Temperature
(Cold Start)
Delay Time
(Seconds)
NA
169
258
225
NA
161
420
169
Experiment
Number
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
2
3
4
5
1
2
3
4
5
1
2
3
4
5
ChemPro 100
Response
M
M
M
M
M
NR(a)
NR
NR
NR
NR
L
NR
NR
NR
NR
NR
NR
NR
M
NR
M
M
M
M
M
NR
L
L
L
L
NR
NR
NR
NR
NR
NR
L
L
L
NR
Response
Time
(Seconds)
20
19
20
19
19
-
-
-
-
-
23
-
-
-
-
-
-
-
19
-
18
20
19
20
18
-
31
28
29
31
-
-
-
-
-
-
22
20
21
-
Recovery
Time
(Seconds)
32
29
32
33
28
-
-
-
-
-
19
-
-
-
-
-
-
-
19
-
32
30
28
30
28
-
32
26
28
28
-
-
-
-
-
-
23
23
23
-
Alarm
(Indicated
Chemical)
TOXIC
TOXIC
TOXIC
TOXIC
TOXIC
-
-
-
-
-
TOXIC
-
-
-
-
-
-
-
TOXIC
-
TOXIC
TOXIC
TOXIC
TOXIC
TOXIC
-
TOXIC
TOXIC
TOXIC
TOXIC
-
-
-
-
-
-
TOXIC
TOXIC
TOXIC
-
(5)
NR = No response.
Unit 1546 - Delay time is the time it took the ChemPro 100 to achieve a ready state after
powering the unit on. For the room temperature cold start, the delay time was
169 seconds. For the cold temperature cold start, the delay time was 258 seconds. For the
hot temperature cold start, the delay time was 225 seconds.
28
-------
All five trials with AC with the ChemPro 100 Unit 1546 fully warmed up (i.e., the
control condition) produced a response. For the cold-start evaluations, only one of the
five trials for cold storage, none of the trials for room-temperature storage, and only one
of the five trials for high-temperature storage exhibited any response to the AC challenge.
Thus, each cold-start condition exhibited statistically significant degradation in accuracy
compared with the fully warmed-up control condition for this unit.
Unit 1811 - For the room temperature cold start, the delay time was 161 seconds. For the
cold temperature cold start, an initial error message, "Error - Check Air Intake," was
observed. At that point, the unit was restarted. At 200 and 330 seconds, additional error
messages, "Functional Exception D003," were observed. The unit then showed a ready
state at 420 seconds. For the hot temperature cold start, the delay time was 169 seconds.
All five trials with AC at the fully warmed-up control condition produced a response. For
the cold-start evaluations, none of the five trials for cold storage, four of the five trials for
room-temperature storage, and three of the five trials for hot storage exhibited a response
to the AC challenge. For this unit, only the cold-storage condition exhibited a statistically
significant degradation in accuracy compared with the fully warmed-up start condition.
When an alarm did occur from a cold start, its maximum level was always Low, while the
maximum level observed from the control condition was a Medium in all five trials. This
shows that the standard condition and the cold-start conditions do not have the same level
of maximum response.
When an alarm did occur from a cold start, it was likely to take longer to occur. With a
geometric mean of 21.0 seconds, the time to first response for a cold start from hot
storage was slightly longer than that from the control condition (geometric mean of
19.0 seconds). The response delay for the cold start from room temperature was even
greater (geometric mean of 29.7 seconds), and, thus, substantially longer than that from
the control condition.
The geometric mean recovery time for the cold start from room temperature (28.4 sec-
onds) was not significantly different than for the control condition (29.6 seconds).
However, the recovery time for the cold start from hot storage (23.0 seconds) was
significantly shorter than in the control condition.
4.9 Battery Life
The ChemPro 100 can be powered by a battery pack or AA batteries. The battery life
evaluation was conducted by placing a fully charged battery pack provided by the vendor
in the ChemPro 100. The ChemPro 100 was then powered on and allowed to warm up
fully according to the manufacturer's directions. The battery life evaluation was
conducted by successive challenges with AC at IDLH concentration delivered for
5 minutes every half hour, and the results are shown in Table 4-6. Unit 1811 frequently
29
-------
failed to respond to the AC challenge during the battery life evaluation, and Unit 1546
did so on two occasions, as shown by the entries of "No Response" in Table 4-6. For
Unit 1546, the battery indicator went from full to 3 bars at 2 hours and 45 minutes after
powering on. Then, the low battery alarm began at 9 hours and 50 minutes, and the unit
shut down at 9 hours and 53 minutes after powering on. For Unit 1811, the battery
indicator went from full to 3 bars at 3 hours and 15 minutes after powering on. The
battery indicator then went from 3 bars to 1 bar at 10 hours. The 1-bar indicator began
flashing at 11 hours. Then, the low battery alarm began at 11 hours and 10 minutes, and
the unit shut down at 11 hours and 12 minutes after powering on.
Table 4-6. Responses Recorded from the ChemPro 100 in Battery Life Evaluation(a)
Test
Start-up
1
2
o
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
Time
0815
0830
0900
0930
1000
1030
1100
1130
1200
1230
1308
1330
1400
1430
1500
1530
1600
1630
1700
1730
1800
1805
1808
1815
1830
1900
1915
1925
1927
ChemPro 100 Identification Number
1546
Response (Response
Time in Seconds)
M TOXIC (19)
M TOXIC (18)
M TOXIC (20)
M TOXIC (18)
M TOXIC (18)
M TOXIC (20)
M TOXIC (19)
M TOXIC (19)
M TOXIC (19)
M TOXIC (18)
M TOXIC (20)
No Response
M TOXIC (18)
M TOXIC (20)
L TOXIC (22)
No Response
M TOXIC (21)
M TOXIC (21)
M TOXIC (19)
M TOXIC (19)
Battery Indicator
Full
Full
Full
Full
Full
3 bars
3 bars
3 bars
3 bars
3 bars
3 bars
3 bars
3 bars
3 bars
3 bars
3 bars
3 bars
3 bars
3 bars
3 bars
Low Battery Alarm
Power Off
(9 hours, 53 minutes)
1811
Response (Response
Time in Seconds)
No Response
No Response
No Response
No Response
No Response
No Response
M TOXIC (28)
No Response
M TOXIC (28)
No Response
No Response
No Response
No Response
No Response
No Response
No Response
M TOXIC (29)
No Response
M TOXIC (18)
M TOXIC (19)
M TOXIC (21)
M TOXIC (19)
Battery Indicator
Full
Full
Full
Full
Full
Full
3 bars
3 bars
3 bars
bars
bars
bars
bars
bars
bars
bars
bars
bars
bars
3 bars
Ibar
1 bar
1 bar
1 bar (flashing)
Low Battery Alarm
Power Off
(11 hours, 12 minutes)
-1 All battery life tests were conducted with AC as the challenge TIC at the IDLH concentration of 50 ppm (50 mg/m3).
30
-------
4.10 Operational Characteristics
General performance observations noted during this evaluation were:
• Instrument Operation—The ChemPro 100 has a large display that is difficult to read
in low light conditions but very easy to read when the background light (bright blue)
is used. This light is controlled from a menu within the ChemPro 100. The display
indicates the state of the unit, what library is being used, the date and time, the
volume level, and the battery power level. Controls were easy to use, and, when lit,
the display was easily readable, even if the operator was wearing personal protective
equipment.
• Instrument Indicators—The ChemPro 100 has one lighted indicator to show the status
of the detector. This indicator is green when the unit is in ready mode and flashing
red when the unit is in alarm mode. When the ChemPro 100 alarms to a challenge, it
will sound an audible alarm and flash the red indicator light. The audible alarm has a
volume control. The visual and audible alarms were strong and readily noticeable.
The unit can also identify the type of chemical which caused the alarm. When the
ChemPro 100 detects a failure within its system, the display indicates the type of
failure.
• Warm-Up—The ChemPro 100 took about 2.7 to 7 minutes to reach a ready state after
being turned on, whether starting from room temperature storage, cold (5 to 8 °C)
storage, or hot (40 °C) storage conditions.
• Batteries—The ChemPro 100 can operate on a rechargeable battery pack or AA
batteries.
• Errors—Error messages occurred over the course of evaluating the ChemPro 100.
Several of these error messages were 'Functional Exception DOS'. Also, there were
errors for air intakes and SCCell failure.
• Conditioning Mode—The ChemPro 100 has a conditioning mode that is only
indicated on a laptop computer if connected to the ChemPro 100. The occurrence of
this mode is not shown on the display of the unit itself. The ChemPro 100 could enter
conditioning mode and would not sample until conditioning mode was completed.
The length of time that the instrument is offline in this mode would be unknown to
an operator using it as a hand-held instrument.
• Vendor Support—Before the evaluation, a vendor representative trained Battelle
employees to operate the ChemPro 100. Evaluating proceeded according to the
vendor's recommendations. The vendor responded promptly when information was
needed during the evaluation.
• Cost—The list price of the ChemPro 100 plus the Standard Accessory Kit is
approximately $9,500.
31
-------
5.0 Performance Summary
Summary results from evaluation of the ChemPro 100 are presented below for each
performance parameter evaluated. Full evaluation of test results was conducted for AC,
GB, and HD. Results reported for CK and SA are limited due to inconsistent responses,
and few results for Cb are reported, due to lack of response found for that chemical.
Discussion of the observed performance can be found in Chapter 4 of this report.
Response Time: When the ChemPro 100 responded to challenges, the time required to
respond to AC and CK was usually about 30 seconds or less, and response times for SA
ranged from about 20 to 80 seconds. Response times for GB were 15 seconds or less, and
for HD were usually 25 to 40 seconds, with a few results of 80 to 225 seconds. Response
times for AC, GB, and HD were not consistently affected by the temperature and RH.
These results do not include instances in which the ChemPro 100 failed to respond to TIC
or CW agent challenges; those instances are addressed below under Accuracy.
Recovery Time: The time required for the ChemPro 100 to return to a baseline reading
after an alarm was typically less than 50 seconds for AC, CK, SA, and HD, and less than
about 15 seconds for GB, but in a few instances during evaluation with AC and HD,
recovery times exceeded 600 seconds. Recovery times depended only weakly on
temperature and RH, with recovery times for AC being shorter with higher temperature
and lower RH. These results exclude those instances in which the ChemPro 100 did not
respond to a TIC or agent challenge.
Accuracy: Of the 120 challenges with AC, GB, and HD used to assess accuracy, the
ChemPro 100 responded accurately to 86, with no response to 30 challenges, and four
cases of a continued alarm even when sampling clean air. Accuracy results for the target
chemicals varied from one test condition to another, and (in TIC testing) from one
ChemPro 100 unit to the other. Accuracy for AC was 100% in most test conditions, but
ranged from 0 to 40% under conditions of high humidity. For GB, accuracy was 80 to
100% at most test conditions, but was 0% with high humidity. Accuracy for HD was 80
to 100% at some test conditions, but 0 to 40% at others, with no clear dependence on
temperature or RH. Accuracy for CK ranged from 0 to 100%, with different temperature
and RH dependence observed from the two units. For SA accuracy ranged from 0 to
100% under different test conditions (from 0 to 20% for one ChemPro 100 unit), with no
apparent dependence on temperature or RH. For chlorine, only one positive response was
seen from one unit in five trials on each of the two units, so the unit accuracies were 0
and 20%.
32
-------
[Failure to respond to AC challenges was also observed during cold-/hot-start and battery
life tests, but those observations were not used in the calculation of the accuracy results
noted above.]
Repeatability: When the ChemPro 100 units responded to an AC challenge, for one
unit, repeatability was perfect under all conditions of temperature and humidity (i.e., all
maximum responses were Medium). For the other unit with AC, maximum response
changed from Low to High as temperature increased, and from Medium to Low as RH
increased. For GB, maximum responses changed from High to Medium to Low as
temperature increased from low (5 °C) to room temperature to high (35 °C). No humidity
effect was seen on GB repeatability, and HD response was perfectly repeatable under all
conditions (all maximum responses were Low).
Response Threshold: For AC, the response threshold was between 3 and 6 ppm (3 and
6 mg/m3) on both ChemPro 100 units. For CK the response threshold was between 5 and
10 ppm (12.5 and 25 mg/m3) on one unit and between 10 and 20 ppm (25 and 50 mg/m3)
on the other. The SA response threshold was between 3 and 6 ppm (10 and 20 mg/m3) on
both units, and for Cb was at or above about 60 ppm (180 mg/m3). For GB the response
threshold was about 0.002 ppm (0.01 mg/m3), and for HD it was about 0.03 ppm
(0.2 mg/m3).
Temperature and Humidity Effects: These effects are described in the preceding
summaries of other performance parameters.
Interference Effects: Ammonia cleaner and air freshener vapors produced false positive
responses in nearly all trials when using either the TIC or CWA library of the ChemPro
100. Latex paint fumes produced false positives in 67 to 100% of trials in the TIC library,
and in 20 to 40% of trials in the CW agent library. DEAE produced no false positive
responses, and exhaust hydrocarbons produced only one false positive out of 20 trials.
[Erroneous positive responses of a different kind (i.e., alarms while the ChemPro 100
sampled clean air) were observed in a few cases during tests of accuracy with AC and
CK.]
When added to challenge mixtures of AC, the interferences produced minimal false
negative responses for AC with one ChemPro 100 unit. However, the response accuracy
of the other unit was reduced to 40% by the air freshener vapors and to 0% by the
ammonia cleaner vapors. False negative effects on CK and SA response were difficult to
determine because of the variability in response for these chemicals with the two
ChemPro 100 units. False negative effects on accuracy of identification for CK were seen
with DEAE, and the accuracy for S A was reduced to 0 to 20% by engine exhaust
hydrocarbons and DEAE. False negative responses with GB occurred primarily with
ammonia cleaner and exhaust hydrocarbons. False negative responses with HD occurred
with paint fumes, ammonia cleaner, and air freshener vapors. With both GB and HD, the
33
-------
false negatives were primarily in the form of inaccurate responses (e.g., a response of
CHEM HAZARD rather than NERVE for GB), rather than no response at all. In these
cases the ChemPro 100 response provides a protective warning, although the threat is
incorrectly identified.
[In one challenge each with AC, GB, and HD in clean air during the evaluation of
accuracy, and in two challenges with HD in interference testing, the ChemPro 100
produced a different type of erroneous negative response in clearing its alarm while the
TIC or agent challenge was still ongoing.]
Cold-/Hot-Start Behavior: The delay time, or time to reach a ready state after start-up,
was 161 seconds and 169 seconds for the two ChemPro 100 units, respectively, when
started up from room temperature storage. The delay times were increased to 258 seconds
and 420 seconds after storage at 5 °C. Accuracy of identification of an AC challenge was
substantially reduced in initial readings after a cold start, relative to that in fully warmed
up operation. For example, one unit showed no response to AC in four of five trials after
start-up from cold storage, in all five trials after start-up from room temperature, and in
four of five trials after start-up from hot storage. In general, response times were slightly
longer, and response readings (i.e., Low/Medium/High) somewhat lower after a cold start
than in fully warmed up operation.
Battery Life: One unit of the ChemPro 100 shut down after 9 hours and 53 minutes of
continuous operation on battery power. The other unit shut down after 11 hours and
12 minutes.
Operational Characteristics: The ChemPro 100 has a large display that is easy to read
in all light conditions provided the background light (bright blue) is used. This light is
controlled from a menu within the ChemPro 100. The display indicates the response
reading of the unit (hazard identity and level), what library is being used, the date and
time, the audible alarm volume level, and the battery power level. A lighted status
indicator is green when the unit is in ready mode, and flashing red when the unit is in
alarm mode (coincident with the audible alarm). The display (when lighted) and audible
and visual alarms can be readily understood by the operator, even when wearing personal
protective equipment. When the ChemPro 100 detects a failure within its system, the
display also indicates an error message, e.g., for air intake flow or SCCell failure. The
ChemPro 100 has a "conditioning" mode that keeps the instrument from responding
while the instrument stabilizes. However, the occurrence of this mode is only apparent
from data displayed on a laptop computer, and is not evident to an operator using the
ChemPro 100 as a hand-held device. When the temperature or humidity condition was
changed, the ChemPro 100 may have entered conditioning mode and thus not have
responded until the conditioning mode was completed. This mode may have contributed
to instances where IMS signal was observed on the laptop, but the ChemPro 100 failed to
give an alarm when challenged.
34
-------
Before this evaluation began, an Environics representative trained Battelle evaluation
personnel to operate the ChemPro 100. Evaluating proceeded according to the vendor's
recommendations, and the vendor responded promptly when information was needed
during the evaluation. The list price of the ChemPro 100 plus the Standard Accessory Kit
is approximately $9,500.
Conclusion: The ChemPro 100 responded correctly to AC, GB, and HD in most
challenges, but responses observed with CK, SA, and Cb were less reliable. However,
even with AC, GB, and HD, observations included the absence of response to challenges,
widely different responses from two units challenged simultaneously, the occasional
discontinuance of a warning alarm even though a TIC or chemical agent challenge was
still present, and the failure to clear an alarm even after the challenge gas was replaced
with clean air. IMS signals recorded on laptop computers during testing indicated that
these behaviors originated with the software that interprets the IMS signal, rather than
with the IMS response itself. This finding suggests that software improvements might
rectify the observed responses. Both false positive and false negative responses occurred
in the presence of common indoor interferent vapors. Usually a protective warning (albeit
inaccurately identified) was present in the instances of a false negative response caused
by interferents. Elevated humidity generally produced less accurate responses.
35
-------
6.0 References
1. Technology Testing and Evaluation Program Test/QA Plan for Evaluation of
Portable Ion Mobility Spectrometers for Detection of Chemicals and Chemical
Agents, Version 1, Battelle, Columbus, Ohio, February 2005.
2. Quality Management Plan (QMP) for the Technology Testing and Evaluation
Program (TTEP), Version 1, Battelle, Columbus, Ohio, January 2005.
3. Library Datasheet for SOCOM Update 3 Libraries, CP100V2-LIB-V1-TEK310105-
GasDetS 1 -CWA-7104_TIC-7 l_Precursor-71 -Library Datasheet-Final-RCM-2-8-
05.doc, Environics, Mikkeli, Finland, February 8, 2005.
36
------- |