April 2004
Environmental Technology
Verification Report
BRUKER DALTONICS INC.
RAID-M ION MOBILITY SPECTROMETER
Prepared by
Battelle
Batreiie
The Business oj Innovation
Under a contract with
xvEPA
U.S. Environmental Protection Agency
-------
April 2004
Environmental Technology Verification
Report
ETV Safe Buildings Monitoring and Detection
Technologies Verification Program
Bruker Daltonics Inc. RAID-M
Ion Mobility Spectrometer
by
Kent Hofacre
Tricia Derringer
Dale Folsom
Peter Larkowski
National Security Division
Thomas Kelly
Loraine Sinnott
Cody Hamilton
Zachary Willenberg
Energy and Environment Division
Battelle
Columbus, Ohio 43201
-------
Notice
The U.S. Environmental Protection Agency (EPA), through its Office of Research and
Development, has financially supported and collaborated in the extramural program described
here. This document has been peer reviewed by the Agency. Mention of trade names or
commercial products does not constitute endorsement or recommendation by the EPA for use.
11
-------
Foreword
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 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. EPA also addresses responsibilities in homeland security
through the National Homeland Security Research Center, by means of research programs in
Drinking Water Security, Safe Buildings, and Rapid Risk Assessment.
The Environmental Technology Verification (ETV) Program has been established by the EPA to
verify the performance characteristics of innovative environmental technology across all media
and to report this objective information to permitters, buyers, and users of the technology, thus
substantially accelerating the entrance of new environmental technologies into the marketplace.
Verification organizations oversee and report verification activities based on testing and quality
assurance protocols developed with input from major stakeholders and customer groups
associated with the technology area. ETV consists of seven environmental technology centers.
Information about each of these centers can be found on the Internet at http://www.epa.gov/etv.
The ETV approach has also been applied to verification of homeland security technologies. The
verification reported herein was conducted by Battelle as part of the Safe Buildings Monitoring
and Detection Technologies Verification Program, which is funded by EPA. Information
concerning this specific environmental technology area can be found on the Internet at
http://www.epa.gov/etv/centers/centerl 1 .html.
111
-------
Acknowledgments
The authors wish to acknowledge the support of all those who helped plan and conduct the
verification test, analyze the data, and prepare this report. We also would like to thank Tom
Taylor of the Centers for Disease Control and Prevention and Mark Durno of the
U.S. Environmental Protection Agency Region 5 for their reviews of the test/quality assurance
plan and this report.
IV
-------
Contents
Page
Notice ii
Foreword iii
Acknowledgments iv
List of Abbreviations viii
1 Background 1
2 Technology Description 2
3 Test Design and Procedures 4
3.1 Introduction 4
3.2 Test Design 5
3.2.1 Chemical Test Compounds 5
3.2.2 Test Matrix 6
3.2.3 Test Locations 6
3.2.4 Test Sequence and Schedule 7
3.2.5 Reference Methods 9
3.2.6 Interferents 11
3.2.7 Materials and Equipment 13
3.3 Test Procedure 16
3.3.1 Response Time 17
3.3.2 Recovery Time 18
3.3.3 Accuracy 18
3.3.4 Repeatability 18
3.3.5 Response Threshold 18
3.3.6 Temperature and Humidity Effects 19
3.3.7 Interference Effects 19
3.3.8 Cold-/Hot-Start Behavior 20
3.3.9 Battery Life 20
3.3.10 Operational Characteristics 21
4 Quality Assurance/Quality Control 22
4.1 Equipment Calibration 22
4.1.1 Reference Methods 22
4.1.2 Instrument Checks 24
-------
4.2 Audits 24
4.2.1 Performance Evaluation Audit 24
4.2.2 Technical Systems Audit 25
4.2.3 Data Quality Audit 25
4.3 QA/QC Reporting 25
5 Statistical Methods 26
5.1 Statistical Analyses 26
5.2 Other Analyses
27
6 Test Results 28
6.1 Response Time 28
6.2 Recovery Time 30
6.3 Accuracy 30
6.4 Repeatability 31
6.5 Response Threshold 32
6.6 Temperature and Humidity Effects on Response 32
6.7 Interference Effects 33
6.8 Cold-/Hot-Start Behavior 36
6.9 Battery Life 36
6.10 Operational Characteristics 39
6.11 Cost 41
7 Performance Summary 42
8 References 45
Appendix A. Description of Statistical Analysis Methods
Appendix B. Detailed Statistical Analysis Results
Appendix C. Tabulation of Data from Verification of the Bruker RALD-M Portable Ion Mobility
Spectrometer
VI
-------
Figures
Figure 2-1. Broker Daltonics Inc. RAID-M IMS 2
Figure 3-1. Sequence of Possible TIC Verification Tests 8
Figure 3-2. Logic Diagram for Determining TIC Test Sequence 9
Figure 3-3. Test System Schematic 15
Tables
Table 3-1. Target TIC and CW Agent Challenge Concentrations 6
Table 3-2 Summary of Evaluations Conducted on the RAID-M IMS 7
Table 3-3. Test Schedule 9
Table 3-4. Summary of Primary Reference Methods 10
Table 3-5. Summary of Test Concentrations for Interferents 12
Table 3-6. Temperature and Relative Humidity Conditions for RAID-M Testing 16
Table 4-1. Summary of Performance Evaluation Audit Results 24
Table 6-1. Summary Illustrative Data from RAID-M Verification Test 29
Table 6-2. Response Threshold Data for the TIC and CW Agent Testing 33
Table 6-3. Summary Data Illustrating Interference Effects 34
Table 6-4. Summary of Cold-/Hot-Start Test Data 37
Table 6-5. Responses Recorded from the RAID-Ms in Battery Life Testing 38
vn
-------
List of Abbreviations
AC
ANOVA
C12
CG
CK
CW
DEAE
EPA
ETV
FID
FPD
GB
GC
HD
HMRC
ID
IDLH
IMS
L
LCD
mL
mm
MSD
PE
ppb
ppm
ppmC
psig
QA
QC
QMP
hydrogen cyanide
analysis of variance
chlorine
phosgene
cyanogen chloride
chemical warfare
N,N-diethylaminoethanol
U.S. Environmental Protection Agency
Environmental Technology Verification
flame ionization detection
flame photometric detection
sarin
gas chromatography
sulfur mustard
Hazardous Materials Research Center
identification
immediately dangerous to life and health
ion mobility spectrometer(ry)
liter
liquid crystal display
microgram per cubic meter
microliter
milliliter
millimeter
mass selective detection
performance evaluation
part per billion
parts per million
parts per million of carbon
pounds per square inch gauge
quality assurance
quality control
quality management plan
Vlll
-------
RH relative humidity
RSD relative standard deviation
SA arsine
THC total hydrocarbon
TIC toxic industrial chemical
TSA technical systems audit
VOC volatile organic compound
IX
-------
Chapter 1
Background
The U.S. Environmental Protection Agency (EPA) supports the Environmental Technology
Verification (ETV) Program to facilitate the deployment of innovative environmental tech-
nologies through performance verification and dissemination of information. The goal of the
ETV Program is to further environmental protection by accelerating the acceptance and use of
improved and cost-effective technologies. ETV seeks to achieve this goal by providing high-
quality, peer-reviewed data on technology performance to those involved in the design,
distribution, financing, permitting, purchase, and use of environmental technologies.
ETV 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. The program evaluates the performance of innovative tech-
nologies by developing test plans that are responsive to the needs of stakeholders, conducting
field or laboratory tests (as appropriate), 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 adequate quality are generated and that the
results are defensible.
Subsequent to the terrorist attacks of September 11, 2001, this ETV approach has been applied
to verify the performance of homeland security technologies. Monitoring and detection
technologies for the protection of public buildings and other public spaces fall within the Safe
Buildings Monitoring and Detection Technologies Verification Program, which is funded by
EPA and conducted by Battelle. In this program, Battelle recently evaluated the performance of
the Bruker Daltonics Inc. RAID-M portable ion mobility spectrometer (IMS).
-------
Chapter 2
Technology Description
The objective of the ETV Safe Buildings Monitoring and Detection Technologies Verification
Program is to verify the performance characteristics of monitoring technologies for chemical
and/or biological contaminants that might be introduced into the building environment. This
verification report provides results for the verification testing of the RAID-M portable IMS made
by Bruker Daltonics Inc. Following is a description of the RAID-M, based on information
provided by the vendor. The information provided below was not subjected to verification in this
test.
The RAID-M is a chemical detector that uses the principle of IMS to detect, classify, quantify,
and continuously monitor concentrations of chemical warfare (CW) agents and toxic industrial
chemicals (TICs). The identity of substances detected is displayed both by class (e.g., "G," "H,"
or "T" for G series agents, H series agents, and TICs, respectively) and by specific agent,
simulant, or TIC (e.g., "GB," "HD," or "TDF). All classes can be displayed independently.
Relative concentrations are indicated by a bar display with eight increments. In addition to use
in the field, the RAID-M is designed to be capable of operating within collective protection
facilities.
The RAID-M can be operated while being held in one hand. It has no protruding parts and
weighs less than 2.80 kilograms (6.4 pounds), excluding battery. The RAID-M contains a small
radioactive sealed source that is completely housed and is such that RAID-M can be stored in
bulk. The RAID-M is 400 millimeters (mm) (15.7 inches) long, 115 mm (4.5 inches) wide, and
165 mm (6.5 inches) high. The RAID-M is of a one-tube design, with automatic polarity
switching (i.e., both positive and negative ions
are automatically monitored, in alternate
intervals of 2 to 3 seconds), and is fully
microprocessor-controlled. It has a remote
display and control option. The display shows
agent identity and a relative indication of hazard
level. The RAID-M incorporates a built-in
audible alarm to indicate agent detection, and
visual alarms to warn of a low battery and other
faults.
Figure 2-1. Bruker Daltonics Inc. RAID-M
IMS
The RAID-M is powered by an integral, primary
battery and can accept power input from a
-------
variety of sources including vehicles (12- to 24-volt direct current nominal) or from a 240-volt,
50-Hertz, alternating current power supply. A diagnostic input/output socket provides data
output, power input, personal computer connectivity, and built-in test information. The carrying
case is designed to protect the RAID-M from exposure to air blasts, thermal radiation, neutron
radiation, gamma radiation, and electromagnetic pulse.
Consumables do not need to be changed when the RAID-M detects a challenge, and
consumables are designed to have a maximum life of not less than 500 hours. There are no
scheduled preventive maintenance tasks. Daily checks are designed to not require dismantling
the equipment and to not typically exceed an average of 10 minutes per day.
-------
Chapter 3
Test Design and Procedures
3.1 Introduction
When first responders arrive at a potentially contaminated site, they need to immediately and
accurately identify chemicals that may be present. Chemicals and chemical agents that may pose
a threat in a building could include both TICs and CW agents.
The objective of this verification test of the RAID-M, a commercially available, portable IMS,
was to evaluate its ability to detect toxic chemicals and chemical agents in indoor air. This
verification focused on the scenario of a portable IMS used by first responders to identify
contaminants and guide emergency response activities after chemical contamination of a
building. This verification was conducted according to a peer-reviewed test/quality assurance
(test/QA) plan(1) that was developed according to the requirements of the Quality Management
Plan for the ETV program.(2) The following performance characteristics of the RAID-M were
evaluated:
Response time
Recovery time
Accuracy
Repeatability
Response threshold
Temperature and humidity effects
Interference effects
Cold-/hot-start behavior
Battery life
Operational characteristics.
Response time, recovery time, accuracy, and repeatability were evaluated by challenging the
RAID-Ms with known vapor concentrations of target TICs and CW agents. RAID-M
performance at low target analyte concentrations was evaluated to assess the response threshold.
Similar tests conducted over a range of temperatures and relative humidities (RH) were used to
establish the effects of these factors on detection capabilities. The effects of potential inter-
ferences in an emergency situation were assessed by sampling those interferences both with and
without the target TICs and CW agents present. The RAID-Ms were tested after a cold start (i.e.,
without the usual warm-up period) both from room temperature and from cold storage
conditions, and after hot storage, to evaluate the delay time before readings could be obtained
-------
and the response speed and accuracy of the RAID-Ms once readings were obtained. Readings of
a target TIC were obtained with the RAID-Ms operated on alternating current power, and
subsequently on battery power, to assess any differences. Battery life was determined as the time
until RAID-M performance degraded as battery power was exhausted in continuous operation.
Operational factors such as ease of use, data output, and cost were assessed by observations of
the test personnel and through inquiries to the vendor.
It was intended to conduct all tests simultaneously with two RAID-Ms. However, due to
occasional RAID-M problems, testing continued in some cases with only one RAID-M. A total
of three RAID-Ms were used during the four-month test period.
Testing was limited to detecting chemicals in the vapor phase because that mode of application
is most relevant to use by first responders. Testing was conducted 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 [HMRC]).
3.2 Test Design
3.2.1 Chemical Test Compounds
The TICs used in testing were
Hydrogen cyanide (HCN, North Atlantic Treaty Organization designation AC)
Cyanogen chloride (C1CN, designated CK)
Phosgene (COC12, designated CG)
Chlorine (C12)
Arsine (AsH3, designated SA).
The CW agents were sarin (GB) and sulfur mustard (HD).
It should be noted that the RAID-Ms tested were programmed to detect all of these compounds
except SA; neither of the software libraries in the RAID-Ms at the time of testing included SA.
Also, it should be noted that the chemical identification that the RAID-Ms displayed upon
detection of a chemical varied slightly among the target compounds. For C12, GB, and HD, the
RAID-M identifiers were "CL2," "GB," and "HD," i.e., the same as the chemical designations
themselves. For AC, the RAID-M display identifier was "CY," indicating a cyanide compound.
For CG, the RAID-M identifier was "C1X," indicating a chlorine-containing compound. For CK,
the RAID-M identifier was either "C1X" or "CY," indicating either a chlorine-containing or
cyanide-containing compound (CK contains both).
Table 3-1 summarizes the concentrations of each TIC and CW agent used in this verification
test. For the TICs AC, CK, CG, and C12, tests were conducted at both 0.1 and 1 times the
respective immediately dangerous to life and health (IDLH) level. The 0.1IDLH level was added
to the test procedure because full-scale readings were often obtained with the TICs at the IDLH
concentrations. For SA, the RAID-Ms did not respond at the IDLH level, so the 0.1 IDLH level
5
-------
was not needed. For the CW agents GB and HD, testing was conducted at a single concentration
level that produced less than full-scale readings on the RAID-Ms under normal temperature and
humidity conditions. The concentration used for GB was 0.014 ppm (0.080 mg/m3), which is 0.4
times the IDLH concentration of 0.035 ppm (0.2 mg/m3). No IDLH level has been set for HD, so
the concentration used was based on an alternative toxic effects guideline, as noted in the
footnote to Table 3-1.
Table 3-1. Target TIC and CW Agent Challenge Concentrations
Chemical Concentrations Type of Level
Hydrogen cyanide (AC) 50 ppm (50 mg/m3) and 5 ppm (5 mg/m3) 1 and 0.1 xIDLEP
Cyanogen chloride (CK) 20 ppm (50 mg/m3) and 2 ppm (5 mg/m3) 1 and 0.1 x IDLH
Phosgene (CG) 2 ppm (8 mg/m3) and 0.2 ppm (0.8 mg/m3) 1 and 0.1 x IDLH
Chlorine (C12) 10 ppm (30 mg/m3) and 1 ppm (3 mg/m3) 1 and 0.1 x IDLH
Arsine (SA) 3 ppm (10 mg/m3) 1 x IDLH
Sarin(GB) 0.014 ppm (0.080 mg/m3) 0.4 x IDLH
Sulfur mustard (HD) 0.063 ppm (0.42 mg/m3) 0.7 x AEGL-2b
(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.
3.2.2 Test Matrix
Table 3-2 summarizes the evaluations that were conducted in the verification test. As Table 3-2
indicates, except for cold-/hot-start behavior, battery life, and assessment of false positive inter-
ference effects (i.e., the interferent alone), all performance parameters were evaluated during
both the TIC and CW agent testing.
3.2.3 Test Locations
Two laboratories were used to conduct the verification tests. Testing with the non-chemical
surety materiel—TICs and interferents—was conducted in a new laboratory at Battelle's
Columbus, Ohio, campus, which has the needed vapor generation, collection, and analysis
equipment. This laboratory has been used previously to conduct IMS instrument and filter tests
using CG, AC, CK, and C12 under controlled environmental conditions. Testing with CW agents
was conducted at the HMRC at Battelle's West lefferson, Ohio, campus. Battelle's HMRC is an
ISO 9001-certified facility that provides a broad range of materials testing, system and
component evaluation, research and development, and analytical chemistry services requiring
the safe use and storage of highly toxic substances. Battelle operates the HMRC in compliance
with all applicable federal, state, and local laws and regulations, including Army regulations.
-------
Table 3-2. Summary of Evaluations Conducted on the RAID-M IMS
Performance
Parameter
Objective
Comparison Based On
Response Time
Recovery Time
Accuracy
Repeatability
Response
Threshold
Temperature and
RH Effects
Interference
Effects
Cold Start
Hot Start
Battery Operation
Determine rise time of
RAID-M response
Determine fall time of
RAID-M response
Characterize agreement of RAID-M
with reference results
Characterize consistency of RAID-M
readings with constant analyte
concentration
Estimate minimum concentration that
produces RAID-M response
Evaluate effect of temperature and
RH on RAID-M performance
Evaluate effect of building
contaminants that may
interfere on with
RAID-M performance
Characterize startup performance after
cold storage
Characterize startup performance after
hot storage
Characterize battery life and
performance
RAID-M readings with step rise in analyte
concentration
RAID-M readings with step decrease in
analyte concentration
Reference method results
RAID-M readings with constant input
Reference method results
Repeat above evaluations with different
temperature and RH
Sample interferents and TICs/CW agents
together
(and interferents alone(a))
Repeat tests with no warm-up(a)
Repeat tests with no warm-up(a)
Compare RAID-M results on battery vs.
alternating current power and duration of
operation on batteries(a)
(a)
Indicates this part of the test performed only during TIC testing.
3.2.4 Test Sequence and Schedule
The sequence of tests performed with the TICs in this study is outlined in Figure 3-1. Since
analyzer performance was not known a priori, the concentrations used in testing depended on
the results of the first few tests performed. The decision logic used to determine the actual TIC
concentrations and the test sequence is shown in Figure 3-2. After completion of TIC testing, a
similar, but slightly reduced, set of tests was performed with CW agents. Table 3-3 summarizes
the actual schedule of testing for the TICs and CW agents. As Table 3-3 indicates, testing with
AC was interrupted as a result of depletion of the AC source gas. Testing was completed on CK,
and then AC testing resumed. Testing with SA took only one day because this TIC is not
detected by the RAID-M.
-------
Test 1: Vapor Challenge with TIC
Alternating clean air with IDLH level concentration of TIC five times with IMS operating on alternating current
power, fully warmed up per manufacturer's instructions prior to testing, and room temperature (22 ± 3°C) and 50 ±
5% RH.
Test 2: Vapor Challenge with TIC at reduced concentration
Test 1 is repeated at 0.1 times the IDLH concentration.
Test 3: Vapor Challenge with TIC at increased concentration
Test 1 is repeated at roughly 10 times the IDLH concentration.
Test 4: Response Threshold of TIC
Test 1 is repeated at a concentration below IDLH. If a response is recorded, the concentration is cut in half until no
response is recorded. If no initial response is recorded, the concentration is increased by a factor of 2 until a response
is recorded.
Test 5: IDLH/0.1 IDLH/Clean Air Challenge
Test 1 is repeated by alternating IDLH, low concentration (either 0.1 IDLH or response threshold concentration. See
logic chart, Figure 3-2), and clean air three times, and alternating low concentration, IDLH, and clean air three times.
Test 6: Vapor Challenge with TIC at room temperature, low humidity
Test 1 is repeated at room temperature (22 ± 3°C) and less than 20% RH. The test is performed at the
concentration(s) determined via the logic in Figure 3-2.
Test 7: Vapor Challenge with TIC at room temperature, high humidity
Test 1 is repeated at room temperature (22 ± 3°C) and 80 ± 5% RH. The test is performed at the concentration(s)
determined via the logic in Figure 3-2.
Test 8: Vapor Challenge with TIC at high temperature, medium humidity
Test 1 is repeated at high temperature (35 ± 3°C) and 50 ± 5% RH. The test is performed at the concentration^)
determined via the logic in Figure 3-2.
Test 9: Vapor Challenge with TIC at high temperature, high humidity
Test 1 is repeated at high temperature (35 ± 3°C) and 80 ± 5% RH. The test is performed at the concentration^)
determined via the logic in Figure 3-2.
Test 10: Vapor Challenge with TIC at low temperature, medium humidity
Test 1 is repeated at low temperature (5 ± 3°C) and 50 ± 5% RH. The test is performed at the concentration(s)
determined via the logic in Figure 3-2.
Test 11: Interferent false positive tests
Test 1 is repeated alternating interferent only with clean air. The test is repeated for all interferents in both libraries.
Test 12: Interferent false negative tests
Test 1 is repeated alternating TIC and interferent with clean air. The test is repeated for all interferents.
Test 13: Opposite Library test
Test 1 is repeated for the library opposite of the one recommended by the manufacturer for TICs.
Test 14: Room Temperature, cold start behavior
Repeat Test 1 with the IMS at room temperature for a minimum of 12 hours and no warm-up.
Test 15: Cold-/Cold-start behavior
Repeat Test 1 after the IMS has been kept refrigerated (5-8°C) overnight for a minimum of 12 hours, with no
warm-up.
Test 16: HoWCold-start behavior
Repeat Test 1 after the IMS has been kept heated (40°C) overnight for a minimum of 12 hours, with no cool down or
warm up.
Test 17: Battery test
Repeat Test 1 with the IMS operating on battery power. The TIC at IDLH concentration is alternated with clean air
once every half hour until the unit stops responding or shuts down due to loss of power.
Figure 3-1. Sequence of Possible TIC Verification Tests
-------
Step 1: Perform Test 1. Depending on the results of this test, go to Step 2a, 2b, or 2c as
appropriate.
Step 2a: If there is no response in Test 1, then perform Test 3, then go to Step 4.
Step 2b: If the response in Test 1 is on scale, then skip to Step 3 and perform all
subsequent tests at the IDLH concentration.
Step 2c-l: If the response in Test 1 is full- or off-scale, then perform Test 2. Depending
on the results of Test 2, go to Step 2c-2a or 2c-2b as appropriate.
Step 2c-2a: If there is no response in Test 2, then perform Test 4. Perform all subsequent
tests at IDLH and response threshold concentrations.
Step 2c-2b: If there is a response in Test 2, then perform all subsequent tests at IDLH
and 0.1 times the IDLH concentrations.
Step 3: Perform Test 4 (if not already done), Tests 5-10, Tests 12-13 at the concen-
tration(s) determined above. For the first TIC, also perform Test 11 and Tests 14-17.
Step 4: Return to Step 1 and repeat for all other TICs.
Figure 3-2. Logic Diagram for Determining TIC Test Sequence
Table 3-3. Test Schedule
Chemical Test Dates (2003)
AC August 6-September 3
September 15-25
CK September 4-15
CG September 25-October 3
SA October 6
C12 October 13-21
GB November 19-December 8
HP December 9-18
3.2.5 Reference Methods
Table 3-4 summarizes the primary reference methods used to determine the challenge
concentrations of the target TICs and CW agents. Listed in the table are the target TICs and CW
agents, the sampling and analysis methods to be used for each compound, and the applicable
concentration range of each method. For CK and AC, low concentration samples were injected
directly for determination by gas chromatography (GC) with flame ionization detection (FID). A
gas chromatographic method for CG proved to have inadequate sensitivity, so an impinger-
based
-------
Table 3-4. Summary of Primary Reference Methods
Analyte
Concentration
Range (ppm)
Sampling Method
Analysis Method
AC
CK
CG
C12
SA
GB
HD
0.05 to 100
2 tolOO
1 to 100
0.1 to 100
0.05 to 100
0.01 to 100
0.01 to 100
Air sample injected directly
Air sample injected directly
Collection in impingers with nitrobenzyl
pyridine
Continuous electrochemical detector
with chlorine- specific sensor
Capillary gas chromatography with
syringe injection from bag sample
Air sample collected with solid sorbent
tube
Air sample collected with solid sorbent
tube
GC/FID
GC/FID
Visible absorption
at 475 nm
Continuous
detection
MSD
GC/FPD
GC/FPD
visible absorption measurement was implemented.(3) This method was satisfactory for
confirming the levels of CG provided to the RAID-Ms with sampling intervals of 20 minutes
and a sampling flow rate of one liter (L) per minute. C12 was determined by a continuous
electrochemical analyzer with a Cl2-specific sensor to allow rapid determination of C12 levels
delivered to the RALD-M during testing. SA was determined by a GC with a capillary column
and mass selective detection (MSD), using samples collected by syringe from the test apparatus.
The CW agents GB and HD were collected in solid sorbent cartridges, and determined by GC
with flame photometric detection (FPD), according to existing HMRC test procedures.
Summaries of these primary methods, and of supplemental methods also used, are as follows.
Hydrogen cyanide (AC) and cyanogen chloride (CK)—The reference method for AC and CK
was an Agilent 6890 GC with a capillary column and FID. This GC was positioned next to the
laboratory hood containing the test system during the TIC testing and sampled automatically
from the flow line delivering the challenge gases to the RAID-Ms.
Phosgene (CG)—To quantify CG, a sample of air was drawn at a known, constant flow rate
through a midget impinger containing 10 milliliter (mL) of an indicating solution, consisting of
a mixture of 4-(4'-nitrobenzyl)pyridine (0.25% w/w) and n-phenylaniline (0.5% w/w) in diethyl
phthalate.(3) In this solution, CG reacts to produce a product having a brilliant red color. The
amount of CG reacted was determined by measuring the absorbance of the indicating solution at
475 nm using a spectrophotometer. Due to the complexity of the impinger method for CG, a
Draeger Pac HI Single Gas Monitor also was used in some tests to provide rapid response. This
device was new, and its factory calibration was used in all monitoring.
Chlorine (C12)— C12 was monitored with a new Draeger MiniWarn Multi-Gas Monitor, which
was factory calibrated.
10
-------
Arsine (SA)—SA was determined using an Agilent 6890 GC coupled to a 5970 mass selective
detector. Air samples (100 |^L) were injected onto a GS-Q capillary column held at -30°C. SA
was separated from xenon at a constant flow of 3 mL/minute helium using the following
temperature program: hold at -30°C for 1 minute; ramp at 50°C/minute to 150°C; hold at 150°C
for 5 minutes (total time 9.6 minutes). The injector temperature was maintained at 100°C,
operated in splitless mode, and was purged after 0.5 minute at 5 mL/minute. Single ion
monitoring at m/z =129 and 76 was used to record signals for xenon and SA, respectively. SA
was quantified using its relative response compared with xenon as the internal standard. In
addition, to provide rapid response in SA determination, a new Draeger Miniwarn Multi-Gas
Monitor also was used. This monitor was factory calibrated.
Sarin (GB) and sulfur mustard (HD)—The analytical method for these CW agents involved
collecting the agents by sampling air through sorbent sampling tubes. The tubes were then
thermally desorbed, and the agents were determined using a capillary GC with FPD. Concen-
trations were determined based on a linear regression of peak area with the amount of agent, and
accounting for the volume of air sampled.
Total hydrocarbons—A continuous FID was used for the determination of the total hydrocarbon
(THC) content of interferent mixtures provided to the RAID-Ms during testing. The THC
concentrations characteristic of realistic interferent levels in buildings were determined, either
by direct measurement or by interpretation of published data. The interferent delivery systems
were then adjusted to achieve the desired THC indication in parts per million of carbon (ppmC)
for each interferent during testing.
3.2.6 Interferents
Interferents were selected for testing based upon their prevalence in a building. The interferents
selected were the volatile chemicals in latex paint, air freshener, and ammonia-based floor
cleaner, as well as gasoline engine exhaust and N,N-diethylaminoethanol (DEAE). DEAE is a
common additive to boiler systems and is released into the heating, ventilating, and air
conditioning system when boiler steam is used to humidify the air. These selected interferents
were tested for false positives by exposing the detectors to selected levels of the interferents in
clean air, to see whether the interferents generated a positive response from the instruments
when no TICs or CW agents were present. Each interferent also was introduced to the
instruments along with each TIC and CW agent, to determine false negatives, i.e., whether the
interferent prevents the RAID-M from indicating that the TIC or CW agent is present. The
following sections describe the materials and concentrations used for testing.
The interferents are mixtures of chemicals and determining the interferent concentration requires
the quantification of all the chemicals present. However, monitoring each component would be
time and cost prohibitive. For this reason, interferent concentrations were monitored using a
THC analyzer. THC analysis is appropriate because all the interferents consist of a significant
amount of carbon-containing compounds. Because quantification is based on carbon content,
the test concentrations are reported on a per carbon basis such as ppmC. The use of the
11
-------
hydrocarbon analyzer also provided real-time continuous monitoring of the interferent
concentration during testing.
Test concentrations for the interferents were based on direct measurements or published data.
Concentrations found in published data were converted to a per carbon basis as described below.
Table 3-5 is a summary of the test concentrations. The following sections contain a detailed
description of how the test concentrations were determined.
Table 3-5. Summary of Test Concentrations for Interferents
Interferent Test Concentration (ppmC)
Latex Paint Fumes 5-10
Floor Cleaner Vapors 10
Air Freshener Vapors 1
Gasoline Exhaust Hydrocarbons 2.5
DEAF 0.02
3.2.6.1 Latex Paint Fumes
Concentrations of latex paint fumes were evaluated directly in and around a freshly painted
office. Samples were obtained using a 25 L Teflon bag and analyzed for THC content. Each wall
in the office was painted, and the room dimensions are 11 feet by 11 feet with an alcove that is
4 feet by 10 feet and ceiling that is 12 feet high. Immediately after painting, the hydrocarbon
concentration was 170 ppmC. After 2.5 hours, the hydrocarbon concentration in the office fell to
38 ppmC. At this time, the hydrocarbon content was determined just outside the entrance to the
office and in the hallway 80 feet away from the office. Hydrocarbon content just outside the
office was 20 ppmC; in the hallway 80 feet away from the office, it was 3 ppmC. Based on these
measurements, the test concentration was maintained at 5 to 10 ppmC.
3.2.6.2 Floor Cleaner Vapors
The test concentration for the ammonia-based floor cleaner was inferred from studies on latex
paint fumes. Similar to paint, floor cleaner is applied to a surface and allowed to dry. Floor
cleaner vapors containing both ammonia and fragrances will disperse into the hallway. Because
of the similarity, a test concentration of 10 ppmC was used for the floor cleaner.
3.2.6.3 Air Freshener Vapors
Concentration levels of air freshener for interferent testing were based upon values reported at
recent indoor air quality conferences. Volatile organic compound (VOC) emission for a plug-in
air freshener was reported to be 30 to 80 milligrams per hour, resulting in a concentration of 300
to 500 micrograms per cubic meter (|ag/m3) for the average room. Assuming the VOC emitted
consists of hydrocarbons similar to limonene, a common fragrance component, the
concentration on a per carbon basis can be calculated. Limonene contains 10 carbons and has a
molecular weight of 136. A concentration of 5.56 |J,g/m3 of limonene is the same as 1 part per
12
-------
billion (ppb). With a room concentration of 500 [ig/m3 and limonene as a representative
molecule, the fragrance concentration on a per carbon basis is estimated to be 1 ppmC. This
THC level was maintained for all tests with the air freshener.
3.2.6.4 Gasoline Engine Exhaust
Of the constituents in gasoline engine exhaust fumes, the aromatic components were considered
most likely to interfere with the performance of the RAID-M. A recent study reported that urban
areas have benzene concentrations of 7 to 9 ppb with comparable concentrations of other
aromatics.(4) The test mixture used to simulate exhaust contains 61 compounds ranging in size
from 2 to 10 carbons, with an average concentration of 200 ppb for each component. To obtain a
concentration of 7 to 9 ppb for the aromatic compounds, the test mixture was diluted 30:1.
Assuming an average size of six carbons, the THC of the mixture was approximated to be
73 ppmC. After dilution, the THC content was 2.5 ppmC, and this target concentration was
maintained for all the experiments.
3.2.6.5 DEAE
DEAE is a common additive to boiler systems to prevent corrosion. When boiler steam is used
to humidify the air in a building, DEAE is released into the building as well. Generally, the
DEAE concentration is kept below 40 ppb, the threshold for odor detection. One study has
shown DEAE concentrations of 1 ppb in a building that uses direct steam injection for
humidification.(5) For testing purposes, the concentration was set at 20 ppbC, which correlates to
3.3 ppb DEAE given that DEAE contains six carbons. This concentration was not detectable by
THC analysis, so the interferent concentration was set by dilution of a concentrated standard.
3.2.7 Materials and Equipment
3.2.7.1 TICs and CWAgents
The commercial gas standards used as sources of the TICs for testing included standards of
10,000 parts per million (ppm) AC (Cylinders 151531 and 7035, Scott Specialty Gases); and
991 ppm CG (Cylinder RR37345), 997 ppm SA (Cylinder KE-50368), and 5,811 ppm C12
(Cylinder RA64239), all from Matheson Gas Products. The source of CK was a 10,000 ppm
compressed gas standard made at Battelle starting with neat CK gas. The neat CK gas was from
Atomergic Chemetals Corp. (Lot No. L6196). To prepare the CK standard, 210 mL of the neat
gas was diluted by pressurizing to 1,000 pounds per square inch gauge (psig) with hydrocarbon-
free air in a 5.9-L cylinder. The CW agents GB and HD were obtained as neat materials from the
U.S. Army under Bailment Agreement No. DAAD13-H-00-0002.
3.2.7.2 Vapor Delivery Equipment
The compressed gas mixtures noted in Section 3.2.7.1 were diluted as the vapor sources for CK,
AC, CG, C12, and SA. For the CW agents GB and HD, a diffusion cell containing the pure agent
was used. A temperature-controlled water bath was installed to control the temperature of the
diffusion cell to maintain a stable and controllable vapor generation rate. A two-way valve was
13
-------
included in the flow path downstream of the vapor generation source, so that the dilution and
test equipment could be totally isolated from the source. A schematic of the entire vapor
generation, dilution, and delivery system is shown in Figure 3-3.
3.2.7.3 Temperature/Humidity Control
The RAID-Ms were evaluated at the temperature and humidity conditions indicated by an "X"
in Table 3-6. Both the delivered air temperature and the RAID-M units were maintained within
the specified temperature range. For testing at 35°C, the vapor delivery system was warmed with
a heat-traced line, using an electronic temperature controller. For testing at 5°C, the dilution and
delivery system was enclosed in a cooled chamber to provide approximate temperature control.
For all tests, thermocouples were installed in both the clean air plenum and the challenge plenum
to provide real-time temperature monitoring.
A commercial Nafion® humidifier (Perma Pure, Inc.) was used to generate controlled high-
humidity air (50 to 100% RH), which was then mixed with dry dilution air and the target vapor
stream to obtain the target RH (< 20% to 80%) in the challenge air.
3.2.7.4 Interferent Sources
Interference test concentrations were obtained by diluting a concentrated feed with air. For latex
paint, floor cleaner, and air freshener, the concentrated feeds were made by purging the head
space of a large boiling flask containing about 100 mL of the bulk liquid of each interferent.
THC analysis of the head space samples found that the concentrated feeds contained 394, 886,
and 233 ppmC for latex paint, floor cleaner, and air freshener, respectively. Gasoline engine
exhaust was simulated using a mixture of 61 organic compounds ranging from 2 to 10 carbon
atoms (C2 to C10). This mixture was prepared by adding 1 microliter (|^L) of 51 neat liquid
components and 250 |^L of 10 gaseous components into a 15.7-L cylinder and diluting to a final
pressure of 1,200 psig with nitrogen. A concentrated standard of 1 ppm for DEAF was made by
adding zero nitrogen to 6 |^L of liquid neat DEAF to a final pressure of 1,200 psig.
3.2.7.5 Performance Evaluation Audit Materials
As part of the quality assurance effort in this verification, a performance evaluation (PE) audit
was performed on reference methods used to confirm the TIC and CW agent concentrations
provided to the RAID-Ms. This audit involved conducting analyses on independent standards,
obtained from different sources than those used for the calibration standards. The results from
the independent standards were then compared with those from the calibration standards, to
assess the degree of agreement. The target agreement in the PE audit was within 20% for TICs
and within 30% for the CW agents.
For the TICs, the PE audit standards were 967 ppm AC (Cylinder SD-10271, Matheson Gas
Products); and 997 ppm CG (Cylinder NA021189), and 5,830 ppm C12 (Cylinder 1C1857), both
from Scott Specialty Gases. A PE audit gas for SA (nominal 1,000 ppm, Scott Specialty Gases)
was obtained, but contained less than 1 ppm of SA, so no PE audit was done for arsine. Also, no
PE audit could be done for CK because of the unavailability of commercial standards for that
gas.
14
-------
r t
Figure 3-3. Test System Schematic
15
-------
Table 3-6. Temperature and Relative Humidity Conditions for RAID-M Testing
Temperature (°C)
RH (%)
< 20
50 ±5
80 ±5
5 ± 3 22 ± 3
X
X X
X
35 ±3
X
X
For the CW agents, PE audit samples were prepared at the HMRC by an analyst other than the
staff who conducted routine calibration of the reference method. The PE audit samples were
sorbent tubes, spiked with known quantities of GB or HD, starting from a different batch of
agent than that used to prepare the calibration samples. These tubes were analyzed by the same
approach used for analysis of reference samples from GB and HD testing, and the results were
compared with the spike amounts.
3.3 Test Procedure
The test system (Figure 3-3) consisted of a vapor generation system, a Nafion® humidifier, two
challenge plenums, a clean air plenum, an RH sensor, thermocouples, and mass flow meters. The
challenge vapor or gas was generated by the vapor generation system. The challenge vapor was
then mixed with the humid dilution air and flowed into the challenge plenum. Interference
vapors were added to the challenge mixtures as needed for testing.
The RH and target concentration of the challenge vapor were obtained by adjusting the mixing
ratio of the humid air (from the Nafion® humidifier) to the dry dilution air, and the mixing ratio
of the vapor generation stream to the humid dilution air, respectively. To avoid potential
corrosion or malfunction of the RH sensor from exposure to the challenge vapor, the RH meter
was installed upstream of the inlet of the vapor stream. The RH of the challenge vapor stream
was calculated based on the measured RH of the humid dilution air and the mixing ratio of the
vapor generation stream to the humid dilution air.
To establish the background readings of the two RAID-Ms being tested, a clean air plenum was
installed. Part of the humid dilution air was introduced directly into the clean air plenum. When
establishing the RAID-M background, the four-way valves connected to the two RAID-Ms were
switched to the clean air plenum to collect baseline data.
After the background measurement, the four-way valves connected to the two RAID-Ms were
switched to one of the challenge plenums to allow the RAID-Ms to sample the challenge
mixture. Switching between the challenge and clean air plenums was rapid, and the residence
time of gas in the test system was short to allow determination of the response and recovery
times of the RAID-Ms. The reference methods described in Section 3.2.5 were used to confirm
that the gas concentrations in the challenge plenums were within ± 20% of the target level.
16
-------
Concentrations outside that tolerance range triggered a repeat of any test procedures conducted
since the last analysis.
3.3.1 Response Time
To evaluate IMS response time, the environmental target conditions were established at 22 ± 3°C
and 50 ± 5% RH. Initially 10 L/min of clean humidified air were passed through the clean air
plenum. The RAID-M sampled the clean air for a minimum of 5 minutes or until a stable
reading was indicated, but not exceeding 10 minutes, to obtain a baseline. At the same time the
clean air was sampled by the RAID-M, the clean air plenum also was sampled with the
appropriate reference method. This sampling took place after the RAID-M reading was
stabilized.
Concurrent with the background measurements, the target challenge concentration in the high
challenge plenum was established. The high challenge concentration was generated at the target
environmental conditions. Adjustments were made to the generator operating conditions and the
dilution flow as needed to establish a challenge concentration within ±20% of the IDLH or other
target concentration, with a stability characterized by a relative standard deviation (RSD) of
10% or less in successive reference measurements. Reference samples were collected and
analyzed immediately to establish the challenge concentration and demonstrate stability prior to
testing. A challenge concentration was considered stable if it could be maintained within the
target challenge bounds on three consecutive reference sample measurements for AC and CK
over a minimum of 5 minutes of continuous operation prior to the test. A challenge concentra-
tion was considered stable for CG and SA if one sample taken prior to testing was consistent
with the method calibration curve. A challenge concentration was considered stable for C12 when
the continuously monitoring reference method reached a stable reading.
After a stable reading was obtained from the RAID-M on background air, and the challenge
mixture was stable and at the target concentration, the four-way valve at the RAID-M's inlet was
switched to sample from the high challenge plenum. The response of the RAID-M was recorded,
and the time to reach a stable response was determined. The RAID-M sampled from the
challenge plenum for a minimum of 5 minutes, up to a maximum of 10 minutes, unless the
RAID-M responded with an alarm of 8 bars (i.e., a full-scale response). In that case, the RAID-
M sampled from the challenge plenum for only 30 seconds. The high challenge vapor concen-
tration was determined by the reference method as frequently as possible during the procedure.
For AC, CK, and C12, a reference sample was taken prior to every challenge with the RAID-M.
For CG and SA, a reference sample was taken prior to and at the conclusion of every set of tests.
After the challenge sampling was concluded, the sample inlet four-way valve was switched to
again sample from the clean air plenum. The time required for the RAID-M to clear (i.e., the
time to return to its starting baseline or non-alarm reading) was recorded as the recovery time. A
minimum of 5 minutes was permitted to allow the RAID-M response to return to baseline. After
a maximum of 10 minutes, regardless of whether the RAID-M returned to baseline, subsequent
cycles of alternating challenge/clean air sampling were carried out, controlled by the four-way
valve. Five such challenge/clean air cycles were completed.
17
-------
The same sampling procedure (i.e., challenge/clean air cycles) was carried out at different
temperatures and RHs or challenge concentrations to evaluate temperature and RH effects and
response thresholds. For the TIC testing, the initial test was conducted at a concentration equal
to the target chemical's IDLH level. If the RAID-M alarmed at 8 bars, the five challenge/clean
air cycles also were conducted at a concentration of 0.1 times the original IDLH (see Table 3-1).
In the CW agent testing, only a single target concentration was used that produced responses
below 8 bars (Table 3-1).
Following the five challenge/clean air cycles, six cycles also were conducted in which the
RAID-M alternated sampling from the high and low challenge plenums. For the TICs, the high
challenge plenum provided the IDLH or target concentration, and the low challenge plenum
provided a concentration of 0.1 times that level. For the CW agents, two different levels below
about 0.7 IDLH were used. This procedure simulated using the RAID-M in locations having
different degrees of contamination. In three of the six cycles the high plenum was sampled first,
then the low plenum; in the other three the order was reversed (this change in order was not
carried out in testing with HD). Clean air was sampled before the first cycle, and again after
every high/low cycle. This test with alternating concentrations was conducted only at the normal
temperature and RH conditions (i.e., 22°C and 50% RH).
3.3.2 Recovery Time
The time for the RAID-M to return to its baseline reading or non-alarm state after removing a
challenge concentration was measured as described in Section 3.3.1.
3.3.3 Accuracy
In all of the response threshold and response time tests, the challenge concentration was
measured using a reference method or monitor. Reference samples were collected prior to,
during, and after RAID-M testing to ensure that a stable concentration was maintained for the
AC, CK, and C12 testing. Reference samples were collected prior to and after RAID-M testing to
ensure a stable concentration was maintained for the CG and SA testing and the CW agent tests.
The reference samples were the ground truth samples used to assess accuracy.
3.3.4. Repeatability
Repeatability was assessed using data obtained from the five repeated challenge/clean air cycles
or the six repeated high challenge/low challenge/clean air cycles. The repeated test results at the
same environmental and concentration conditions were used to quantify the repeatability of the
measurements and the effects of test conditions on repeatability.
3.3.5. Response Threshold
The response threshold of each RAID-M was evaluated by repeating the procedure in
Section 3.3.1 at successively lower concentrations. The response threshold was determined at the
baseline environmental condition of 22°C and 50% RH, in the absence of any interfering
chemicals. The manufacturer's reported detection limit (±50%) was used as the starting
18
-------
concentration. If the manufacturer did not provide a detection limit, a concentration below the
IDLH or target concentration was chosen. Three cycles of challenge/clean air were conducted at
each concentration. If the RAID-M showed a response at that challenge concentration, the
challenge concentration was decreased. The decrease in concentration was continued until the
response threshold had been bracketed. The minimum challenge concentration producing a
response was denoted as the response threshold. In some cases, the RAID-Ms continued to alarm
even at the lowest challenge concentration obtainable from the testing system. In such cases, that
minimum challenge concentration was denoted as an upper limit for the response threshold.
3.3.6 Temperature and Humidity Effects
The tests described in Section 3.3.1 were repeated at the IDLH or other selected target con-
centration of significant health concern, over the range of environmental conditions shown in
Table 3-3. Five repeat runs were performed at each set of test conditions for each target TIC or
CW agent. The data at different temperature and RH conditions were used to infer whether these
conditions affected the detection (i.e., accuracy, repeatability, response threshold) of the RAID-
M for the target chemical. The effect on response time and recovery time also was assessed.
3.3.7 Interference Effects
To evaluate the effects of the interferents described in Section 3.2.6, the test system shown in
Figure 3-3 was modified by adding an interferent vapor generator. The output from this source
was directed as needed to mix with the humidified air flowing to the challenge plenum. Inter-
ference effects were evaluated with the interferent alone and with the interferent and TIC or CW
agent together. Testing with the interferent alone allowed evaluation of false positive responses,
and testing with the interferent and target chemical together allowed evaluation of false
negatives. Interference effects on response time and recovery time also were observed.
Interferent testing involved only one interferent vapor at a time. False positive testing was done
by alternately sampling clean air and an interferent vapor, for a total of up to five times each, in
a procedure analogous to that described in Section 3.3.1. However, if no interferent effect was
observed after three such test cycles, the test was truncated. The same process was used for
testing for false negatives with interferents and TICs (or interferents and CW agents) together,
with the two compounds diluted together in humidified air delivered to the challenge plenum. In
the interference tests, all TICs were tested at their IDLH levels, and the two CW agents were
tested at the concentrations shown in Table 3-1. The interferent concentrations used were those
shown in Table 3-5. A response from the RAID-M with the interferent alone was recorded as a
false positive. The absence of a response, or a reduced response, to the TIC or CW agent in the
presence of the interferent was recorded as a false negative.
If the RAID-M alarmed at 8 bars during the false negative tests, the five challenge and
interferent/clean air cycles also were conducted at a concentration 0.1 times the original target
concentration for the TIC testing.
The replicate test runs conducted with the interferent plus TIC or CW agent allowed the
response time and recovery time of the RAID-M to be assessed with interferents present.
19
-------
Differences in response and recovery times, relative to those in previous tests with only the TIC
or CW agent present, were attributed to the effect of the interferent vapor.
3.3.8 Cold-/Hot-Start Behavior
The cold-/hot-start tests were conducted in a manner similar to the response time test in
Section 3.3.1. Prior to these tests, however, the RAID-M was not allowed to warm up per the
manufacturer's recommendation.
The cold-start test was conducted both with the RAID-M at room temperature and subsequently
at reduced temperature for a minimum of 12 hours prior to start-up. In the former test, the
RAID-M was stored at 22 ± 3 °C for at least 12 hours prior to testing. The cold-start effect was
assessed using an IDLH challenge concentration of AC, at the baseline conditions of 22°C and
50% RH. The time from powering up the RAID-M to its first readiness to provide readings was
considered the start-up delay or standby time. After challenge with AC, the response time was
measured, followed by the recovery time. Repeatability and accuracy in five replicate
challenge/clean air cycles also were noted.
For the reduced temperature cold-start test, the RAID-M was placed in a refrigerated enclosure
(5 to 8°C) for at least 12 hours overnight. At the start of the next day, the cold-start test with AC
was repeated, under the same baseline conditions (22°C and 50% RH) and again recording the
start-up delay or standby time and other performance parameters.
For the hot-start test, the RAID-M was placed in a heated enclosure at 40 ± 3°C for at least
12 hours overnight. At the start of the next test day, the hot-start test was conducted using AC in
the same fashion as in the cold-start tests, at the baseline test conditions (22°C and 50% RH).
For the cold-/hot-start tests, the RAID-M was connected to the clean air manifold and switched
on. The time between switching the RAID-M on and when it indicated it was ready to begin
providing readings was recorded as the delay or standby time. Then the RAID-M was connected
(by the four-way valve shown in Figure 3-3) to the challenge plenum, which was supplied with
the IDLH level of AC. The response time, stable reading, and recovery time of each RAID-M
were recorded for each of five successive periods of alternating challenge mixture and clean air.
The recorded data were used to evaluate whether response and recovery times, repeatability, and
accuracy were affected by a cold or hot start relative to normal (i.e., fully warmed up) operation.
Only one cold-/hot-start test was performed per day so that the RAID-M could equilibrate to
storage conditions prior to the test.
3.3.9 Battery Life
Battery life was evaluated by assessing the degradation of performance with extended
continuous operation. Fully charged batteries were installed, and the RAID-Ms were turned on
and allowed to warm up, and an initial response time test was performed (see Section 3.3.1). An
IDLH concentration of AC was used in this evaluation. The indicated concentration signal from
the RAID-M was recorded. The RAID-M then sampled clean air for 30 minutes, and then the
AC mixture was sampled again. This procedure was repeated with the RAID-M operating
20
-------
continuously until the RAID-M no longer responded to the AC challenge. The total time of
operation was recorded as the measure of battery life.
3.3.10 Operational Characteristics
Key operational characteristics of the RAID-M were evaluated by means of the observations of
test operators and by inquiry to the RAID-M vendor. Ease of use was assessed by operator
observations, with particular attention to the conditions of use by emergency first responders.
Signal or data output capabilities were assessed by observations of the personnel who operated
the RAID-M during testing. The type of data that was output was noted on the data sheets (e.g.,
audio and/or visual alarm, bar graph, low/med/high indication, and/or quantitative measure of
concentration). In addition, the clarity and readability of the output were noted, especially in low
light conditions or when holding the RAID-M while walking, as in use by a first responder. The
availability of multiple forms of data output or display also was assessed (e.g., the availability of
both a visual display and an analog voltage output for recording purposes).
The vendor was asked for the purchase and operational costs of the RAID-M as tested. Estimates
for key maintenance items also were requested from the vendor.
21
-------
Chapter 4
Quality Assurance/Quality Control
QA/quality control (QC) procedures were performed in accordance with the ETV quality
management plan (QMP)(2) and the test/QA plan for this verification test.(1)
4.1 Equipment Calibration
4.1.1 Reference Methods
The reference methods used for determining TICs and chemical agents are summarized in
Section 3.2.5. The analytical equipment needed for these methods was calibrated, maintained,
and operated according to the quality requirements of the reference methods and Battelle's
normal documentation. Procedures for blank sampling during testing and for calibration of
reference methods are described below.
For AC, CK, and C12 testing, blank reference samples were run between each challenge
concentration. The sequence of reference sampling thus included establishing the concentration
prior to testing the RAID-Ms, running a blank on clean air, switching to challenge gas and
taking a reference sample immediately prior to challenging the RAID-Ms with the challenge gas,
and again running a blank when the RAID-Ms were once more sampling clean air. Blank
reference samples were taken before and after, but not during, SA and CG testing, because the
blank methods used were not conducive to running the large number of reference samples
analyzed during the other TIC tests. In testing with GB and HD, blank sorbent tubes were run at
the start of each test day.
Calibration procedures for the reference and other analyses were as follows:
Hydrogen cyanide (AC) and cyanogen chloride (CK)—The GC reference method for AC and
CK was calibrated by preparing gas mixtures in 1-L gas sampling bags. For AC, calibration
standards were prepared by diluting 0.5 to 4 mL of commercial concentrated AC gas standards
(e.g., 10,000 ppm AC in N2) in 800 mL of clean air in a bag. Three samples from each bag were
injected by syringe into the GC, and the peak area was recorded. Several such calibration
standards ranging from 1.25 to 50 ppm HCN were prepared and analyzed over a three-day
period. The regression of peak area versus AC standard concentration had the form Peak Area =
3.0671 x (AC, ppm), with an r2 value of 0.9984. For CK, 800-mL bag standards were prepared
in a similar manner; but, since CK was available as a neat gas, a two-stage dilution was needed.
22
-------
An initial bag dilution of the neat material was made, and then a small volume of that mixture
was diluted further in a second bag, producing CK standards of about 6 to 25 ppm. Again, three
samples from each final bag were analyzed, and the peak areas were used to generate a
regression line, which had the form Peak Area = 1.8374 x (CK, ppm), with an r2 value of
0.9987.
Phosgene (CG)—For calibrating the impinger-based method for CG, 25-L Tedlar bags were
filled with a known amount (-24 L) of zero air spiked with known amounts of CG from a high
concentration commercial standard. Final concentrations in the bags ranged from 0 (method
blank) to 2 ppm. Approximately 23 L of the phosgene/air mixture in each bag was then drawn
through a midget impinger containing the indicating solution. A multi-point calibration at the
beginning of the study showed linear response and an r2 value of 0.98. Before proceeding with
the analysis of samples on each test day, a one-point calibration check was conducted, typically
at a phosgene concentration of 1 ppm, by the procedures outlined above. Inclusion of each day's
calibration check data in the calibration curve showed correlation coefficients of 0.96 to 0.98,
which were deemed acceptable.
Due to the complexity of the impinger method for CG, a Draeger Pac HI Single Gas Monitor also
was used for the monitoring of phosgene. This device was new, and its factory calibration was
used in all monitoring.
Chlorine (C12)— C12 was monitored with a new Draeger MiniWarn Multi-Gas Monitor, which
was factory calibrated. The C12 calibration was checked by injecting a known amount of high
concentration gas into a known volume of clean air in a Tedlar gas sampling bag, to give a
concentration of a few ppm. The MiniWarn readings in this check were within 5% of the
expected standard concentrations.
Arsine (SA)—SA was quantified by GC using its relative response compared with xenon as the
internal standard. Xenon is a ubiquitous component of ambient air with a concentration of
90 ppb. Triplicate injections of SA standards at 0.3, 3, and 50 ppm and one method blank
yielded a linear calibration plot having an r2 value of 0.996.
A new Draeger MiniWarn Multi-Gas Monitor also was used for monitoring SA. This monitor
was factory calibrated, but the SA calibration was checked by injecting a known amount of high
concentration gas into a known volume of clean air in a Tedlar gas sampling bag, to give a
concentration of a few ppm. The MiniWarn readings in this check were within 5% of the
expected standard concentrations.
Sarin (GB) and sulfur mustard (HD)—Calibration standards for these CW agents were prepared
by diluting stock agent to 50 nanograms/|iL, and then injecting jiL volumes of that standard
onto sorbent tubes used for the agent sampling. These tubes were thermally desorbed in the same
manner as for all sample tubes, and a regression of peak area versus amount of agent was
prepared.
Total hydrocarbons—The THC analyzer used to document the interferent levels provided in
testing was calibrated by filling a 25-L Tedlar bag with 33 ppm of propane in air from a com-
mercial gas standard. Since propane is a three-carbon molecule, this standard constitutes a THC
23
-------
concentration of 99 ppmC. This standard was used for calibrating the THC analyzer throughout
the verification. Clean air from the room was used for zeroing.
4.1.2 Instrument Checks
The RAID-Ms were operated and maintained according to the vendor's instructions throughout
the verification test. Maintenance was performed according to predefined RAID-M diagnostics.
Daily operational check procedures for the RAID-Ms were performed with a vendor-supplied
simulant tube. Proper response of the RAID-Ms to the simulant was required before testing
could proceed.
4.2 Audits
4.2.1 Performance Evaluation Audit
As described in Section 3.2.7.5, a PE audit was conducted to assess the quality of reference
measurements made in the verification test. For TICs, the PE audit was performed once during
the verification test by analyzing gaseous standards that were obtained from different suppliers
than those providing the standards used during the testing. The acceptable tolerance for each
target TIC was ±20%. For the two CW agents, the PE audit was conducted by an analyst other
than the person who prepared the normal calibration standards, using a sorbent trap spiked with
one of the target agents. This spiked trap was prepared using a different batch of agent than that
used to prepare calibration standards. The expected tolerance on the CW agent PE audit was
±30%. Table 4-1 shows that the results of the PE audits were all within the target tolerances.
Table 4-1. Summary of Performance Evaluation Audit Results(a)
TIC or
CW Agent
AC
CG
C12
GB
HD
Sample
Standard (Cylinder 7035)
PE Audit Std (Cyl. SD- 10271)
Standard (Cylinder RR37345)
PE Audit Std (Cyl. NA021189)
Standard (Cylinder RA64239)
PE Audit Std (Cyl. 1C 1857)
Standard (GB50 4-16-04 PR)
PE Audit Std (GB50 4-18-04
FM)
Standard (1000 4-16-11 PR)
PE Audit Std (1000 4-19-04 FM)
Concentration
10,000 ppm
967 ppm
990.7 ppm
997 ppm
5,811 ppm
5,830 ppm
50 ng/tube
50 ng/tube
1,000 ng/tube
1,000 ng/tube
Agreement
Result (%)
42.2 ppm
4.1 ppm
1.92 ppm
1.93 ppm
9.90 ppm
8.95 ppm
49.9 ng(b)
43.2 ng
966 ppm
1,034 ppm
0.5
0.1
9.9
13.4
7.0
(a) As noted in Section 3.2.7.5, PE audits were not performed for SA (audit standard had incorrect concentration) or CK (no
independent standard available).
-------
4.2.2 Technical Systems Audit
The Battelle Quality Manager also conducted a technical systems audit (TSA) to ensure that the
verification test was performed in accordance with the test/QA plan(1) and the ETV QMP.(2) As
part of the audit, the Battelle Quality Manager reviewed the reference sampling and analysis
methods used, compared actual test procedures with those specified in the test/QA plan,(1) and
reviewed data acquisition and handling procedures. Observations and findings from this audit
were documented and submitted to the Battelle Verification Test Coordinator for response. The
only technical finding of the TSA concerned carrying out test procedures for CK while reference
analyses for that compound were still in progress. A deviation form was prepared and approved,
describing that departure from the test/QA plan.(1) This deviation had no impact on data quality,
as all CK concentrations were confirmed to be within the target specification. The records
concerning the TSA are permanently stored with the Battelle Quality Manager.
4.2.3 Data Quality Audit
All of the data acquired during the verification test were audited, not merely the 10% required
by the test/QA plan.(1) Battelle's Quality 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.
The data quality audit disclosed three items requiring the preparation and approval of deviation
forms:
• The use of the impinger/visible absorption method for CG, rather than the planned GC/FID
method, which proved ineffective
• The failure to reverse the order of low and high concentrations of HD in the oscillating
concentration test with that CW agent (low/high was always done, and not high/low)
• The lack of completion of one-over-one review of all test data sheets within two weeks after
data collection.
The first two of these had a negligible impact on the test data. The impact of the third deviation
was minimized by the fact that all data sheets were scrutinized in compiling electronic data files
for statistical evaluation. In the end, only a handful of small corrections to the test data were
needed as a result of this deviation, with minimal impact on the verification results.
4.3 QA/QC Reporting
Each assessment and audit was documented in accordance with the test/QA plan.(1) Once the
assessment report was prepared, the Battelle Verification Test Coordinator ensured that a
response was provided for each adverse finding or potential problem and implemented any
necessary follow-up corrective action. The Battelle Quality Manager ensured that follow-up
corrective action was taken. The results of the TSA were sent to the EPA.
25
-------
Chapter 5
Statistical Methods
To extract the most information about RAID-M performance from the test procedures, a
statistical analysis of the test results was performed whenever appropriate. Such an analysis used
all available data to explore the impact of test parameters on RAID-M performance. Section 5.1
summarizes the statistical approaches and the parameters tested. The performance parameters of
response threshold and battery life were assessed with simple comparisons that did not require
statistical analysis. Section 5.2 describes the analyses used for these performance parameters.
5.1 Statistical Analyses
Several successive readings of the RAID-M responses, alternating with RAID-M baseline
readings, were recorded in each step of the test procedure. That is, for each RAID-M and each
TIC or CW agent, such readings were recorded with each concentration, at a range of
temperature and RH conditions (Table 3-3), both with and without each interferent. In addition,
readings were obtained at the normal temperature and RH conditions at different RAID-M
starting conditions. These data were the basis for the statistical analysis of RAID-M
performance.
The statistical analyses focused on the following performance parameters:
Response time
Recovery time
Accuracy
Repeatability
False positives/false negatives
by considering the following explanatory variables:
Identity of the target TIC or CW agent
Concentration of the target TIC or CW agent
Temperature
Humidity
Start state (i.e., warmed up, cold start, etc.)
Identity and presence/absence of interferent.
26
-------
The RAID-M does not provide quantitative concentration readings, but instead provides various
qualitative outputs, including an intensity of response (0 to 8 bar scale), alarm/no alarm state,
and identification of the TIC or CW agent that is detected. As a result, the statistical analysis
often relied on methods for categorical rather than continuously variable data.
Appendix A of this report provides a detailed description of the statistical procedures used for
the verification of the performance parameters listed above. In brief, a cumulative logit model
was used for analysis of the effects of TIC or CW agent identity and concentration, temperature,
RH, and RAID-M start state on RAID-M response. Response time and recovery time were
evaluated using an analysis of variance (ANOVA) model. The treatment of recovery times took
into account the presence of censored data for recovery time (i.e., RAID-M response that did not
return to baseline within 600 seconds and was therefore truncated at a recovery time of
600 seconds). Accuracy and repeatability were evaluated using a binomial logit model, and false
negatives and false positive were evaluated using a cumulative logit model. The mathematical
forms of the models used are given in Appendix A.
5.2 Other Analyses
The data used to evaluate the response threshold were the five replicate RAID-M readings
obtained at each succeeding TIC or CW agent concentration. These data were tabulated, along
with the corresponding reference method data that established the challenge concentration. The
response threshold was determined by inspection as the lowest reference method concentration
that produced a positive RAID-M response in all replicate runs. In this evaluation, any positive
RAID-M response was taken as detection of the target TIC or CW agent, i.e., RAID-M response
of 1 bar was sufficient in terms of the response threshold evaluation.
Both battery life and the effectiveness of battery operation were assessed. Battery life is reported
as the total time from start-up to battery exhaustion when a RAID-M is warmed up and operated
continuously solely on battery power at room temperature and 50% RH. This time was measured
from initial start-up to the point in time when the RAID-M no longer responded to a challenge
mixture of AC in air. Battery effectiveness was evaluated by comparing responses obtained
while operating on battery power to those obtained under identical sampling conditions
immediately before switching from alternating current power to battery power.
27
-------
Chapter 6
Test Results
As discussed in Chapter 5 and detailed in Appendix A, cumulative logit models and other
approaches were used to test for the effects of different conditions on RAID-M performance.
The following sections present summaries of the statistical model results from these tests. An
extensive and detailed presentation of the modeled statistical results is included as Appendix B
of this report. In addition, a tabulation of all test data is included as Appendix C to this report.
Each section of Appendix B indicates which test identifications (IDs) in Appendix C were used
for the evaluation. Also note that the statistical evaluations reported in this chapter were not
applied to SA, since the RAID-Ms did not respond to that TIC.
Two CW agents were tested, GB and HD. In contrast to the TIC testing, in most testing each
CW agent was tested at only one concentration level, and the levels were different for each
agent: GB was tested at about 0.4 IDLH and HD at about 0.7 of its AEGL-2 level. Because of
the concentration difference across agents, the effects of concentration and agent on the behavior
of the RAID-M could not be separated. For this reason, the CW agent statistical analyses
described below and in Appendix B do not include agent or concentration effects. In descriptive
comments about figures that summarize results, differences may be noted that appear across the
agents. It should be kept in mind that such differences may be the result of a difference in
agents, a difference in concentration, or a mixture of both effects.
6.1 Response Time
Results of the response time analysis are summarized here and detailed in Appendix B,
Section B.I. Table 6-1 summarizes data used for the analysis of response time and other
performance parameters. This table is for illustration purposes, and for brevity the TIC results
shown are drawn only from data obtained at the IDLH concentration.
The RAID-M units produced an alarm and a visual display of response typically within a few
seconds after the initiation of sampling of one of the TICs. For CG, CK, and AC, the modeled
geometric mean response times were all between 3 and 5 seconds. Response for C12 was slightly
slower, with a modeled geometric mean response time of about 9 seconds. Neither temperature
nor relative humidity had any significant effect on the TIC response times. The effect of start
state (cold start, hot start) was tested with AC only, and no effect was found of start state on the
response time for that TIC.
28
-------
Table 6-1.
TIC/CWA(i
AC
CK
CG
C12
GB
HD
Summary Illustrative Data from RAID-M Verification Test
Environmental
0 Conditions
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
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
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
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
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
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
Response
Range
(bars)
8
8
8
8
8
7-8
5-8
8
7-8
4-8
4-8
8
8
8
8
8
6-8
8
8
8
6-8
7-8
5-6
(1 run - NR)
7-8
3
3
3
1-2
(2 runs - NR)
2
2
3-8
4-8
3-8
2-7
7
4-8
(1 run - NR)
Response
Time Range
(s)
2-4
2-5
3-5
2-6
1-4
3-6
2-8
2-4
1-4
3-10
3-4
2-4
2-5
2-5
2-5
2-4
2-4
2-4
3-15
2-23
3-15
5-16
4-31
10-17
7-14
10-13
7-12
8-13
8-14
19-26
5-9
4-8
3-7
3-10
5-7
7-9
Recovery
Time Range
(s)
185-299
112-323
70-201
51-82
25-36
>600
14-24
17-39
13-36
1-32
8-29
15-18
7-9
6-9
7-9
6-9
6-9
6-8
24-35
21-34
0-39
21-58
0-35
12-35
42-55
66-73
36-54
20-23
17-26
47-67
24-44
29-73
34-105
11-22
19-22
34-41
Alarms
(Indicated
Chemical)
10/10 (CY)
10/10 (CY)
10/10 (CY)
10/10 (CY)
10/10 (CY)
10/10 (CY)
6/10 (CY)
4/10 (CL2)
9/10 (CY)
1/10 (CL2)
5/10 (CY)
5/10 (CL2)
9/10 (CY)
1/10 (CL2)
6/10 (CY)
4/10 (CL2)
5/10 (CY)
5/10 (CL2)
10/10 (CLX)
10/10 (CLX)
10/10 (CLX)
10/10 (CLX)
10/10 (CLX)
10/10 (CLX)
10/10 (CL2)
10/10 (CL2)
10/10 (CL2)
10/10 (CL2)
5/6 (CL2)
10/10 (CL2)
5/5 (GB)
5/5 (GB)
5/5 (GB)
5/7 (GB)
5/5 (GB)
5/5 (GB)
10/10 (HD)
10/10 (HD)
10/10 (HD)
10/10 (HD)
5/5 (HD)
4/5 (HD)
(a) Data shown are for illustration, TIC results shown are from data at IDLH level only.
NR - indicates "No Response"
29
-------
For the CW agents, response times were about 5 to 8 seconds for HD under most conditions, and
about 10 seconds for GB under most conditions. The only exception was that at the lowest
temperature (approximately 5°C) the modeled response time for GB was approximately
20 seconds. Relative humidity had no effect on the response times for either GB or HD.
6.2 Recovery Time
Results of the recovery time analysis are summarized here and detailed in Appendix B,
Section B.2. Recovery time results also are illustrated in Table 6-1. In general, recovery times
differed considerably from one TIC to another. For CG, recovery times were always within
10 seconds, whereas those for CK and C12 ranged from about 10 to 40 seconds under nearly all
conditions. Recovery times for AC were longer and more variable, ranging from about
15 seconds to 600 seconds (the maximum value recorded) over all tests. AC recovery times were
shorter at higher temperatures, with model-estimated mean values of about 280 seconds at the
lowest temperature (5°C), about 100 seconds at room temperature, and about 35 seconds at the
highest temperature (35°C). Recovery times for AC also were shorter at lower concentrations,
with model-estimated mean values of less than 50 seconds at 0.1 IDLH and about 210 seconds
at IDLH concentrations. Relative humidity did not have a consistent effect on TIC recovery
times.
The effect of the start state of the RAID-Ms was tested with AC at the IDLH level and was found
to have a strong effect on the recovery times. Under fully warmed-up conditions, the model-
estimated geometric mean recovery time for AC was about 240 seconds; under cold start
conditions after an overnight hot soak (at 40°C), the recovery time was only slightly longer, with
a geometric mean of 286 seconds. However, after a cold soak (5 to 8°C) or with cold start-up
from room temperature, modeled geometric mean recovery times for AC were at least
600 seconds (recovery times were not recorded beyond that time).
Model-estimated geometric mean recovery times for the CW agents under normal conditions
were about 50 seconds for GB and about 34 seconds for HD. For both agents, higher tempera-
tures led to shorter recovery times, with estimated recovery times of about 24 seconds for GB
and 16 seconds for HD at the highest temperature (35°C). Relative humidity did not have a
consistent effect on recovery time for HD, but for GB recovery time increased as relative
humidity decreased. GB model-estimated recovery time ranged from 69 seconds at the lowest
humidity to 46 seconds at the highest humidity.
6.3 Accuracy
Results of the accuracy analysis are summarized here and detailed in Appendix B, Section B.3.
Results of tests that involved alternating different challenge concentrations, as opposed to
alternating clean air and a single challenge concentration, are detailed in Appendix B,
Section B.8. Accuracy results also are illustrated in Table 6-1. The RAID-M units were 100%
accurate in identifying the target TICs and CW agents under the majority of test conditions.
Accuracy below 100% was observed primarily for CK. Estimated mean accuracy for CK ranged
30
-------
from about 50% to nearly 100% at both 0.1 and 1IDLH, with the lowest accuracy occurring at
the lowest temperature and at the highest relative humidity. As Table 6-1 illustrates, the
inaccurate responses for CK occurred when the RAID-Ms identified CK as chlorine gas (C12).
Accuracy was less than 100% for CG only at 0.1 IDLH and the highest temperature tested, and
for C12 only at IDLH and the highest relative humidity. TIC accuracy was 100% in all other tests
and conditions.
Neither temperature nor relative humidity had a significant effect on accuracy for the CW
agents. Excluding those interferences that suppressed GB response, overall accuracy for GB was
97.5%, based on 81 total readings. Overall accuracy for HD was 99.4%, based on 161 readings
in 17 total tests, including all interferent tests. In both cases, inaccuracies were in the form of
absence of response to the agent. It should be noted that, in addition to correctly identifying GB
or HD, in most tests with the CW agents, the RAID-Ms also gave an indication of "HN."
The accuracy of RAID-M response when alternating between different TIC or CW agent
challenge concentrations also was essentially the same as that when alternating between clean air
and a challenge concentration.
6.4 Repeatability
Results of the repeatability analysis are summarized here and detailed in Appendix B,
Section B.4. Repeatability of response was always perfect for AC, as the RAID-M units always
gave a full-scale reading of 8 bars at both 0.1 and 1 IDLH for this TIC. Concentration had a
marked effect on response repeatability for CG, with a model-estimated mean repeatability of
50% at 0.1 IDLH, but with perfect repeatability at 1 IDLH. Concentration had minimal effect on
response repeatability for CK and C12, with modeled repeatability values near or above 90% for
both these TICs. On the other hand, temperature had no effect on repeatability for CG, but
strongly affected repeatability for CK and C12. Repeatability was 100% for these TICs at the
lowest temperature (5°C), but the model-estimated mean dropped to about 60% at the highest
temperature (35°C). Humidity had a strong effect on repeatability of response for C12, with better
repeatability at lower humidity. Repeatability of response also was reduced to a model-estimated
mean of about 50% for CK at medium and high humidities.
The modeled repeatability of response times varied among the TICs, from 24 to 28 %RSD for
AC to 70 to 75 %RSD for CK, but neither temperature nor relative humidity affected the
repeatability of response times.
The repeatability of TIC recovery times also varied from 11 %RSD for AC to 47 %RSD for CK
and was unaffected by temperature. However, for CG and C12, repeatability of recovery times
was better at 1 IDLH than at 0.1 IDLH; for C12, repeatability of recovery times was very
sensitive to humidity. Model-estimated mean %RSD for C12 was 20 to 30% at low and medium
humidities, but at high humidity was 110%.
The RAID-M response data for the CW agents showed no significant effect of relative humidity
on the repeatability of response for either GB or HD. However, the repeatability of RAID-M
31
-------
response for HD was highly dependent on temperature. For that agent, the model-estimated
mean response repeatability was very low at the lowest temperature (5°C), but increased to about
70% at the highest temperature (35°C). In contrast, GB mean response repeatability was 100%
at low and medium temperatures, but fell below 60% at the highest temperature.
The analysis of response time repeatability for GB and HD suggested that HD response times
became less repeatable (more variable) with increasing temperature (observed means of 12, 21,
and 36 %RSD at low, medium, and high temperatures, respectively), whereas for GB response
time variability peaked at 27 %RSD at the medium temperature (22°C). HD response time
repeatability showed little dependence on relative humidity, and for GB the response time was
least repeatable (most variable) at medium humidity (27 %RSD).
The analysis of recovery time repeatability for the CW agents showed similar results to that of
response time repeatability. HD recovery times became less repeatable as temperature increased
(observed means of 9, 19, and 27 %RSD at low, medium, and high temperatures, respectively),
whereas temperature had little effect on the repeatability of recovery times for GB. For both GB
and HD, the repeatability of recovery times was lowest at the highest humidity (means of 14 and
44 %RSD for GB and HD, respectively), though only GB showed a consistent trend of recovery
time repeatability with humidity.
6.5 Response Threshold
The response threshold for each TIC was determined by challenging the RAID-Ms with
successively lower concentrations until both RAID-Ms no longer responded. Table 6-2 provides
the results for each TIC for the response threshold test. The concentrations used in each of these
tests are given in the table and are well below the IDLH or other target concentration used in the
other challenge/clean air tests. The responses listed in the table give the results for three
successive challenge/clean air cycles.
For AC and CK, the response threshold could be determined only as an upper limit of
< 0.06 ppm and < 0.6 ppm, respectively. Both RAID-Ms were responding with readings of 4 to
6 bars at the lowest concentration obtainable. Thus, the actual threshold response levels of the
RAID-M for AC and CK are well below these upper limits. For CG, the response threshold was
different for the two RAID-Ms tested. RAID-M BW000701 had a response threshold between
0.08 and 0.15 ppm. RAID-M BW01335 had a response threshold between 0.15 and 0.33 ppm.
For C12, both RAID-Ms had a response threshold between 0.25 and 0.50 ppm. No response
threshold test was conducted for SA because the RAID-Ms did not respond to SA at the IDLH
concentration level. The response threshold for GB was 0.0035 to 0.007 ppm (0.02 to
0.04 mg/m3) and for HD was 0.01 to 0.02 ppm (0.07 to 0.13 mg/m3).
6.6 Temperature and Humidity Effects on Response
The results of investigating temperature and relative humidity effects on RAID-M response are
summarized here and are detailed in Appendix B, Section B.5. Table 6-1 also illustrates T/RH
32
-------
effect data. In general, the effects of temperature and relative humidity on RAID-M response
were small. Considering the TICs, full-scale (i.e., 8 bars) responses to AC, CK, and C12 were
observed in most tests at both 0.1 and 1 IDLH. Responses to CG varied more widely than those
for the other TICs, and responses to 0.1 IDLH of CG were clearly lower than corresponding
responses to 1 IDLH levels.
Table 6-2. Response Threshold Data for the TIC and CW Agent Testing
TIC (Concentration)
AC (0.50 ppm)
AC (0.25 ppm)
AC (0.1 3 ppm)
AC (0.06 ppm)
CK (2.50 ppm)
CK (1.25 ppm)
CK (0.63 ppm)
CG (0.33 ppm)
CG (0.15 ppm)
CG (0.08 ppm)
C12 (0.50 ppm)
C12 (0.25 ppm)
GB (0.007 ppm) (0.04 mg/m3)
GB (0.0035 ppm) (0.02 mg/m3)
HD (0.02 ppm) (0.13 mg/m3)
HD (0.01 ppm) (0.07 mg/m3)
RAID-M
BW000701
8 bars
8 bars
7 bars
4-5 bars
8 bars
8 bars
5-6 bars
2 bars
Ibar
No Response
7-8 bars
No Response
1 bar (8 - 16 s)
No Response
NA
NA
Identification
BW001001
8 bars
8 bars
8 bars
6 bars
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
Number
BW01335
NA
NA
NA
NA
7-8 bars
5 -7 bars
4 bars
Ibar
No Response
No Response
6 bars
No Response
1 bar (17 - 28 s)
No Response
1 bar (8 s)
No Response
The effect of temperature on RAID-M response was most noticeable for CK, with higher
predicted RAID-M readings at low temperature. A significant effect of relative humidity also
was found. For CK and C12, higher predicted RAID-M readings were seen at lower humidity.
Considering the CW agents, relative humidity did not have a significant effect on RAID-M
response to either GB or HD. A significant effect of temperature on response was found, with
the highest temperature (35°C) producing slightly lower readings for both GB and HD.
6.7 Interference Effects
The results of investigating interference effects on RAID-M response are summarized here and
are detailed in Appendix B, Section B.6. Table 6-3 summarizes data used for the analysis of
33
-------
interference effects. This table is for illustration purposes, and for brevity the TIC results shown
are drawn only from data obtained at the IDLH concentration.
Table 6-3. Summary Data Illustrating Interference Effects
TIC/CWA(a) Interfered
AC Control
Paint Fumes
Floor Cleaner
DEAF
Gasoline Engine
Exhaust
Air Freshener
CK Control
Paint Fumes
Floor Cleaner
DEAE
Gasoline Engine
Exhaust
Air Freshener
CG Control
Paint Fumes
Floor Cleaner
DEAE
Gasoline Engine
Exhaust
Air Freshener
C12 Control
Paint Fumes
Floor Cleaner
DEAE
Gasoline Engine
Exhaust
Air Freshener
GB Control
Paint Fumes
Floor Cleaner
DEAE
Gasoline Engine
Exhaust
Air Freshener
HD Control
Paint Fumes
Floor Cleaner
DEAE
Gasoline Engine
Exhaust
Air Freshener
(a) Data shown are for illustration, TIC results
Response
Range
(bars)
8
8
8
8
8
8
5-8
8
8
8
8
8
8
8
8
8
8
4-8
8
4-5
(7 runs - NR)
10 runs - NR
7-8
8
7-8
3
1-5
(3 runs - NR)
6 runs - NR
4
5
5 runs - NR
3-8
3-4
6-7
3-4
3-4
2-4
shown are from data
Response
Time Range
(s)
2-4
2-4
2-4
1-4
1-4
2-4
2-8
2-6
2-9
2-6
2-9
2-5
2-5
1-5
3-5
2-4
2-5
1-5
3-15
6-17
NA
3-14
5-14
3-16
7-14
18-31
NA
8-11
8-15
NA
5-9
3-6
5-6
4-8
5-10
3-18
at IDLH level only.
Recovery
Time Range
(s)
185-299
347-590
141-342
123-448
100-201
142-522
14-24
12-37
14-39
12-38
13-40
13-35
7-9
5-9
6-10
6-9
6-9
6-9
24-35
0-24
NA
31->600
18-60
18-35
42-55
30->600
NA
25-47
36-49
NA
24-44
8-25
26-38
20-27
26-35
27->600
Alarms
(Indicated
Chemical)
10/10 (CY)
10/10 (CY)
10/10 (CY)
10/10 (CY)
10/10 (CY)
10/10 (CY)
6/10 (CY)
4/10 (CL2)
10/10 (CY)
10/10 (CY)
10/10 (CY)
10/10 (CY)
10/10 (CY)
10/10 (CLX)
10/10 (CLX)
10/10 (CLX)
10/10 (CLX)
10/10 (CLX)
10/10 (CLX)
10/10 (CL2)
3/10 (CL2)
0/10
10/10 (CL2)
10/10 (CL2)
10/10 (CL2)
5/5 (GB)
2/5 (GB)
0/6
5/5 (GB)
5/5 (GB)
0/5
10/10 (HD)
10/10 (HD)
5/5 (HD)
5/5 (HD)
5/5 (HD)
10/10 (HD)
NR - indicates "No Response"
34
-------
The interference effects varied widely among the different TICs. For AC and CG, response was
essentially unaffected by the presence of interferents along with the TIC. For CK, the RAID-M
response was generally higher when both CK and an interferent were present than when CK
alone was present. This was true for all five interferents tested. On the other hand, RAID-M
response for C12 was strongly suppressed by the presence of latex paint fumes and floor cleaner
vapors. In fact, no C12 response was seen at all when floor cleaner vapors and C12 were present
together. These interference effects for CK constitute a form of false positive, in that CK
response is exaggerated due to the presence of an interferent. The interference effects for C12
constitute false negatives, since the indication of C12 by the RAID-M is suppressed partly or
completely by the presence of an interferent.
RAID-M response times for the TICs were not affected to any practical extent by the presence of
the interferents, with the exception that C12 response was completely eliminated by the floor
cleaner vapors.
The interferents also had relatively little impact on RAID-M recovery times for the TICs, with
the exception that C12 response was completely eliminated by the floor cleaner vapors. At the
IDLH concentrations used in interferent testing, the estimated mean recovery time for AC was
about 230 seconds with no interferents present, and ranged from 140 to 270 seconds with all
interferents except latex paint fumes. With paint fumes, the RAID-M estimated recovery time
for AC was lengthened to nearly 500 seconds. Also, the estimated recovery times for C12 were all
less than about 30 seconds, except with DEAF present, when estimated recovery time was
lengthened to nearly 100 seconds. Other than these two observations, the interferents had no
effect on TIC recovery times.
Substantial interference effects were observed in detection of the CW agents. RAID-M response
to GB was sharply reduced by the presence of air freshener vapors, floor cleaner vapors, and
latex paint fumes. In fact, no response at all was observed to GB in the presence of air freshener
vapors and floor cleaner vapors. However, the presence of those interferents caused the RAID-M
to indicate the presence of other CW agents, specifically VX and GA. For HD, RAID-M
response was reduced by about half by the presence of air freshener vapors, latex paint fumes,
DEAF, and gasoline engine exhaust hydrocarbons. However, floor cleaner vapors had no effect
on RAID-M response to HD. The presence of any of the interferents caused the RAID-Ms to
indicate the presence of VX and, occasionally, GA or GB. These indications were in addition to
the indications of HN, which often occurred in the GB and HD tests, with or without
interferents.
The RAID-M response time for HD was little affected by the presence of the interferents. For
GB, response time was unaffected by DEAF or gasoline engine exhaust; however, the presence
of latex paint fumes lengthened the estimated response time (from about 9 to about 24 seconds).
Note that no response at all was observed to GB in the presence of air freshener vapors and floor
cleaner vapors.
In all cases but one, the RAID-M estimated recovery times for GB and HD were actually shorter
in the presence of interferents than with the agent alone. Estimated mean recovery times for GB
were about 50 seconds without interferents and about 30 to 42 seconds with DEAE, latex paint
35
-------
fumes, or gasoline exhaust hydrocarbons present. Note that no response at all was observed to
GB in the presence of air freshener vapors and floor cleaner vapors. Estimated recovery times for
HD were about 33 seconds without interferents and about 19 to 30 seconds with DEAE, latex
paint fumes, floor cleaner vapors, or gasoline exhaust hydrocarbons present. The only exception
was that estimated recovery time for HD was 57 seconds in the presence of air freshener vapors.
False positives were observed only for floor cleaner vapors and DEAE. In 16 trials with floor
cleaner vapors, a false positive response was observed five times, for a 31% false positive rate.
In eight trials with DEAE, a false positive response was observed three times, for a 38% false
positive rate. The RAID-M responses in these cases were relatively low, never exceeding 3 bars
on the display. When the false positives occurred, the RAID-Ms erroneously identified the
interferent as being the nerve agent VX.
6.8 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 here and detailed in Appendix B, Section B.7.
Table 6-4 illustrates the data obtained in testing for cold-/hot-start effects, showing the RAID-M
unit used, the start condition, sequential experiment number, delay time, response reading,
response and recovery times, and indicated chemical. Such testing was conducted only with AC,
at the IDLH concentration.
In all cases of start-up without sufficient warm-up, whether from cold, room temperature, or hot
conditions, the RAID-M response initially gave readings that were lower than those observed
under fully warmed-up operation. This difference was most noticeable in the first few RAID-M
readings and diminished as measurements continued, as the RAID-M warmed up during
operation. The delay times of the RAID-Ms (i.e., the time after being powered on before the
RAID-Ms were ready to give a reading) varied with start condition and from one RAID-M unit
to another. Starting cold at room temperature, two RAID-M units each showed delay times of
about 40 seconds. Starting from a cold-soak condition, the delay times of both units were longer,
at 1 minute 2 seconds and 2 minutes 59 seconds, respectively. The longest delay time for RAID-
M BW000701 occurred when starting after a hot soak, at 2 minutes 5 seconds; and the longest
delay time of all was observed for RAID-M BW01335 after a hot soak, at 13 minutes 52
seconds. Recovery times were lengthened to over 600 seconds in operation from a cold
temperature or room temperature cold start, but response times were not affected by start-up
conditions. The RAID-Ms always correctly identified AC with the "CY" designation, regardless
of start condition.
6.9 Battery Life
The battery life test was conducted by placing fully charged batteries in the RAID-Ms. The
RAID-Ms were then powered on and allowed to warm up fully according to the manufacturer's
directions. The delay or standby time before the RAID-M was ready to give a reading was
31 seconds and 24 seconds for the RAID-M BW000701 and BW01335 units, respectively. An
initial response time test was conducted with AC at the IDLH concentration level. The
36
-------
Table 6-4. Summary of Cold-/Hot-Start Test Data
RAID-M Unit
BW000701
BW001001
BW01335
Start
Condition
Control
(Fully
Warmed Up)
Room
Temperature
(Cold Start)
Cold
Temperature
(Cold Start)
Hot
Temperature
(Cold Start)
Control
(Fully
Warmed Up)
Room
Temperature
(Cold Start)
Cold
Temperature
(Cold Start)
Hot
Temperature
(Cold Start)
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
Delay
Time
(mm:ss)
00:40
01:02
02:05
00:44
02:59
13:52
Response
(bars)
8
8
8
8
8
2
1
7
7
8
4
7
8
8
8
6
8
8
8
8
8
8
8
8
8
3
6
7
8
8
2
7
8
8
8
5
7
8
8
8
Response
Time (s)
4
4
3
4
2
3
2
3
5
4
2
5
5
4
2
4
3
6
5
2
4
4
4
4
4
5
6
5
4
6
5
5
5
2
2
5
3
4
4
3
Recovery
Time (s)
185
203
227
229
234
445
>600
>600
>600
>600
440
>600
>600
>600
>600
>600
402
413
424
435
220
256
294
299
297
>600
>600
>600
>600
>600
>600
>600
>600
>600
>600
321
161
142
150
156
Alarm(a)
(Indicated
Chemical)
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
CY
(a) All tests conducted with AC as the chaUenge TIC, at the IDLH level.
37
-------
RAID-Ms then sampled clean air for approximately 30 minutes, and then the AC mixture was
sampled again. This procedure was repeated with the RAID-Ms operating continuously until the
batteries were depleted and the RAID-Ms no longer responded to the presence of the AC.
Table 6-5 provides the results for the battery life test. The elapsed time and the response (and
response time) for each test are provided in the table. Both units of the RAID-M gave a response
of 8 bars and showed response times of a few seconds every time the AC challenge was sampled.
RAID-M BW000701 responded in a normal fashion until the "EMPTY" battery indicator was
given, which was quickly followed by the "ERROR E070" and finally the powering off of the
detector. The battery life for this detector was 6 hours and 29 minutes. RAID-M BW01335 also
responded in a normal fashion until the same battery indicator and error message were given.
The battery life for this detector was 7 hours and 52 minutes. There was no degradation of
response or response time with either RAID-M as the batteries approached
Table 6-5. Responses Recorded from the RAID-Ms in Battery Life Testing
(a)
Test Time (from start-up)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
3 minutes
33 minutes
1 hour 4 minutes
1 hour 34 minutes
2 hours 5 minutes
2 hours 35 minutes
3 hours 6 minutes
3 hours 36 minutes
4 hours 7 minutes
4 hours 37 minutes
5 hours 8 minutes
5 hours 37 minutes
6 hours 8 minutes
6 hours 24 minutes
6 hours 28 minutes
6 hours 29 minutes
6 hours 39 minutes
7 hours 9 minutes
7 hours 41 minutes
7 hours 47 minutes
7 hours 50 minutes
7 hours 52 minutes
RAID-M Identification Number
BW000701
Response
(Response Time)
8 bars (2 s)
8 bars (2 s)
8 bars (4 s)
8 bars (2 s)
8 bars (2 s)
8 bars (4 s)
8 bars (3 s)
8 bars (3 s)
8 bars (3 s)
8 bars (5 s)
8 bars (3 s)
8 bars (5 s)
8 bars (3 s)
Battery
Indicator
Full
Full
Full
Full
Full
Full
Full
Full
Full
Full
Full
Half
Half
Empty
Error E070
Power Off
BW01335
Response
(Response
Time)
8 bars (5 s)
8 bars (3 s)
8 bars (2 s)
8 bars (2 s)
8 bars (3 s)
8 bars (4 s)
8 bars (3 s)
8 bars (3 s)
8 bars (4 s)
8 bars (2 s)
8 bars (3 s)
8 bars (3 s)
8 bars (5 s)
8 bars (2 s)
8 bars (4 s)
8 bars (4 s)
Battery
Indicator
Full
Full
Full
Full
Full
Full
Full
Full
Full
Full
Full
Full
Half
Half
Half
Half
Empty
Error E070
Power Off
All battery life tests were conducted with AC as the challenge TIC at the IDLH concentration of 50 ppm
(50 mg/m3).
38
-------
depletion. However, the RAID-M battery indicator readings did not change from "Full" to
"Half until the useful battery life was nearly gone. Thus, this indicator should not be taken as a
measure of the battery life remaining.
6.10 Operational Characteristics
General performance observations noted during verification testing:
• Two Libraries—The RAID-M internal software consists of a Library A and a Library B.
Library A is the CW agent library and opens automatically when powering up the RAID-M.
Library B is the TIC library. Use of the RAID-M to detect unknown contaminants would
require switching between Libraries A and B to assure monitoring of both CW agents and
TICs. A few compounds, such as hydrogen cyanide, are detectable by both libraries;
however, they do not have the same alarm identification in both libraries. Hydrogen cyanide
alarms as AC in Library A and as CY (cyanides) in Library B. Also, the simulants utilized to
ensure that the RAID-M is functioning properly are identified with different acronyms in
each library.
• Simulants—Two simulants are used to test the RAID-M prior to any challenge test. The
RAID-M gives a response to these simulants, which allows the user to determine whether the
RAID-M is functioning properly. Testing with simulant prior to use provides the only indica-
tion that the detector is functioning properly, unless an error message is provided by the
instrument.
• Display and Alarms—The liquid crystal display (LCD) visual display can be backlit, which
allows easy reading of the display. The visual alarms include a strip of flashing red lights
above the display window and a bar reading on the display window with an acronym
identifying the chemical detected. The red lights flash slowly for an alarm reading of 1 to
4 bars and quickly for an alarm reading of 5 to 8 bars. The identification of the chemical is
easy to read, but distinguishing the bars corresponding to the concentration of the chemical
is more difficult, especially between 4 and 8 bars. The audible alarm is very loud and
obvious. The audible alarm is slower for an alarm reading of 1 to 4 bars and more rapid for
an alarm reading of 5 to 8 bars. The LCD display can provide information on up to three
compounds at one time, with the compounds grouped by agent types in the display. Given
that the RAID-M operates in both positive and negative ion modes essentially
simultaneously, this ability to display multiple compound IDs is valuable.
• Push Switches—Two push switches on the RAID-M allow the operator to easily change
parameters. The parameters that can be changed include turning the audible alarm on/off,
changing from one library to the other, and resetting the RAID-M after a consumable has
been changed.
• Consumables—Three consumables are used when operating the RAID-M: a backflush filter
(carbon), a drying tube, and an ammonia dopant. All of these consumables, when depleted,
affected the performance of the RAID-M, causing behavior such as longer clear-down times
and lower concentration responses. The RAID-M provided error messages when these
39
-------
consumables should be changed; however, these messages are based on metered time of use,
not on the actual state of the consumable. Verification testing took place over a five-month
period; the RAID-Ms were operating continuously (24 hours/day) for those five months,
with challenge testing five days a week for the five-month period. The ammonia dopant was
changed once during this five-month period. The backflush filter and carbon trap were
changed much more frequently as needed. Changing consumables involved powering the
detector off, unscrewing a cap, removing the expired consumable, replacing with a new
consumable, replacing the cap, powering the detector on, and resetting the time on the
display.
Instrument Warm-Up—It was observed during testing that the detector response (bar reading
shown) was strongly dependent on the length of time the RAID-M was allowed to warm up
prior to use. The RAID-M did not give as high a concentration reading for a challenge when
initially powered on as when the RAID-M was allowed to fully warm up.
Positive/Negative Ion Identification—The RAID-M continuously switches between modes
that detect and identify either positive or negative ions, at intervals of a couple of seconds.
That is, the RAID-M will detect positive ions for 2 to 3 seconds and then will switch and
detect negative ions. This is a continuous cycle that allows the rapid identification of
chemicals producing both positive and negative ions.
Double Alarms—If the RAID-M produced two simultaneous identifications of different
chemicals during a verification test procedure, two possible responses were provided. If one
of the chemicals provided a positive ion and the other a negative ion, both chemicals would
be identified in the display window. For example, if both GB and HD were present, a
response would be provided for both in the display window. On the other hand, if both
chemicals present provided a positive (or negative) ion, the chemical with the higher alarm
concentration (i.e., more bars indicated) would be identified in the display window, and the
other chemical would not be identified.
Errors—Several error readings were provided by the RAID-Ms over the course of
verification testing. Some of the errors encountered included indications that the backflush
filter or carbon trap needed to be changed, that the battery was empty, and that an electrical
fault had occurred.
Instrument Failure—A total of three RAID-Ms were used in the verification test. Two of the
three RAID-Ms failed at some point in the test. One RAID-M gave an electrical fault error
message and was unusable. The other RAID-M displayed an error message and could not be
recovered without connecting the instrument to a laptop computer and overriding the fault.
Verification testing took place over five months; the RAID-Ms were operated continuously,
with challenge testing five days a week.
Vendor Support— Before the verification, a vendor representative trained three Battelle
employees to operate the RAID-Ms. Testing proceeded according to the representative's
recommendations on how to operate the RAID-M for testing. The vendor supplied all of the
consumables necessary for verification testing and responded promptly when information
was needed or an instrument needed to be replaced.
40
-------
Back-Flush—The RAID-M went into a "back-flush" mode on certain chemicals when the
instrument reading reached 8 bars. These chemicals included the two simulants used to
verify instrument operation and the CW agents HD and GB. In this mode, the intake airflow
is reversed to flush the detected chemical out of the RAID-M. The "back-flush" mode does
not apply when a TIC is detected.
6.11 Cost
The purchase price of the RAID-M portable IMS, as used in this verification test, is
approximately $13,000.
41
-------
Chapter 7
Performance Summary
This chapter summarizes the overall performance results found in testing of the RAID-M
portable IMS with both TICs and CW agents. This summary focuses on aspects of the
performance that are most important in field use of the RAID-M by first responders. Consistent
with that use, most test procedures were conducted with challenge levels of the TIC or CW agent
that were at or near IDLH concentrations. Section 3.2.1 describes the challenge levels used.
Response thresholds were tested by stepping downard in concentration from these challenge
levels. A general observation is that the RAID-Ms were not programmed to respond to arsine, so
no results are reported for that TIC.
RAID-M response to AC, CK, and C12 was very sensitive. Full-scale readings were obtained
even at concentrations of a few percent or less of the IDLH concentrations for these TICs.
The RAID-Ms provided an audible and visual alarm within 3 to 5 seconds response time after
exposure to CG, CK, and AC, and within about 9 seconds for C12. Response times for GB and
HD were about 10 seconds and 5 to 8 seconds, respectively. Over the ranges of 5 to 35°C and
<20 to >80% relative humidity, temperature and RH had no effect on response time for any TIC
or CW agent, with the sole exception that at 5°C the response time for GB was lengthened to
about 20 seconds. Response times for AC also were unaffected by operating the RAID-M from a
cold start (i.e., with insufficient warm-up time).
RAID-M recovery times (i.e., the time needed for the RAID-M to return to baseline after the end
of exposure to a TIC or CW agent) varied widely, depending on the TIC or CW agent sampled
and also on the sampling conditions. Average recovery times for CG, CK, and C12 were
relatively consistent under all conditions and were always less than 10 seconds for CG and
generally 10 to 40 seconds for CK and C12. Recovery times for AC ranged from 15 seconds to
over 600 seconds, with the fastest recovery times occurring at low concentrations and high
temperatures. Recovery times for GB and HD averaged about 50 seconds and about 34 seconds,
respectively, at normal temperature, with average recovery times reduced by about half at higher
temperatures. The overall ranges of all observed recovery times were about 15 to 70 seconds for
GB and about 10 to 100 seconds for HD. Relative humidity had minimal effect on recovery
times. In operation from a cold start, the recovery time for AC was lengthened to at least 600
seconds.
The RAID-Ms were 100% accurate in identifying the TIC being sampled under almost all test
conditions. Accuracy in identifying the CW agents also was high: overall accuracy for GB was
97.5% (excluding data from interferences that suppressed GB response), and for HD was 99.4%,
42
-------
when all test data were included. In addition to correctly identifying GB and HD, the RAID-Ms
usually gave a secondary indication of "HN" when testing with these agents. Accuracy below
100% occurred primarily for CK, with the lowest accuracy (-50%) at high humidity and low
temperature. The inaccuracy for CK occurred in the form of misidentification of CK as chlorine
gas (C12).
The accuracy of RAID-M response when alternating between different challenge concentrations
also was essentially the same as that when alternating between clean air and a challenge
concentration.
The repeatability, or consistency, of RAID-M response, response times, and recovery times also
was evaluated. AC showed the most consistent responses and also the lowest %RSD of response
times and recovery times. In fact, repeatability of response for AC was perfect, as full-scale
readings consistently resulted at the test concentrations. The %RSD of recovery times was low
for AC primarily because of the long average recovery times for that TIC under many conditions
(see above). Response and recovery times were most variable for CK. RAID-M readings and
recovery times for C12 were strongly affected by relative humidity, with the most variability at
high humidity. For the CW agents, the repeatability of RAID-M response to HD improved as
temperature increased, but the repeatability of response and recovery times for HD lessened.
Repeatability of response for GB did not vary substantially with test conditions, and the only
effect on repeatability was that recovery times for GB were less repeatable at high humidity.
The response thresholds of the RAID-M were <0.06 ppm for AC, <0.6 ppm for CK, 0.08 to
0.33 ppm for CG, and 0.25 to 0.5 ppm for C12, where the ranges are for two RAID-M units. The
response threshold for GB was 0.0035 to 0.007 ppm, and for HD was 0.01 to 0.02 ppm.
Temperature and relative humidity had little effect on RAID-M response to the TICs and CW
agents. Higher readings for CK were generally found at lower temperatures, and higher readings
for CK and C12 were generally found at lower humidity. Slightly higher readings for both CW
agents also were found at lower temperatures.
Interferents likely to be present indoors had large effects with individual TICs and CW agents. In
terms of false negatives, RAID-M response for C12 was sharply reduced by latex paint fumes and
floor cleaner vapors; the floor cleaner vapors resulted in zero response for C12. Response to GB
was sharply reduced by latex paint fumes, floor cleaner vapors, and air freshener vapors; the
latter two interferents resulted in zero response for GB. Response for HD was reduced by about
half by latex paint fumes, air freshener vapors, DEAE, and gasoline engine exhaust hydro-
carbons. The interferents also caused the RAID-Ms to incorrectly report the presence of other
agents, such as VX or GA. False positive responses occurred only with floor cleaner vapors and
DEAE. Both of these interferents produced small positive responses in about one-third of the
trials; in those cases the RAID-Ms incorrectly identified the interferent as the nerve agent VX.
The interferents had little effect on response or recovery times for the TICs or CW agents.
Operating the RAID-M with insufficient warm-up time reduced the initial responses to AC,
regardless of whether the cold start occurred after storage at 5°C, at room temperature, or at
40°C. The response time for AC was not affected by operating from a cold start, but the recovery
time was lengthened in such operation. The delay time (time for the RAID-M to be ready for a
43
-------
first reading after start-up) ranged from 40 seconds to about 3 minutes under cold-start
conditions, except for one unit that showed a delay time of nearly 14 minutes after a 40°C soak
and cold start.
The useful operating lives for fully charged batteries in two RAID-M units in continuous
operation were found to be 6 hours 29 minutes and 7 hours 52 minutes, respectively.
Several operational characteristics of the RAID-M were noted during testing. In general, the
RAID-M was easy to use, gave clear alarms and a readable and informative display, and
provided error and diagnostic messages. The RAID-M automatically switched between positive
and negative ion detection modes at intervals of a few seconds, allowing detection of a wide
variety of chemicals. Among the most important other operational characteristics are
• The use in the RAID-M of two separate software libraries, one for TICs and one for CW
agents, necessitating switching between libraries to detect both types of chemicals.
• The need for three types of consumables (carbon backflush filter, drying tube, and ammonia
dopant), the first two of which needed to be replaced several times during the nearly five-
month test period. RAID-M error messages calling for replacement of consumables are based
on metered time of use, not on the actual state of the consumable.
• The need for proper warm-up of the RAID-M before use, to assure that full response is
achieved when monitoring starts.
• The failure during testing of two of the three RAID-Ms used in this verification, one due to
an electrical fault, and the other to an apparently incorrect error message that required
overriding the message by connection to a laptop computer.
44
-------
Chapter 8
References
1. Test/QA Plan for Verification of Portable Ion Mobility Spectrometers for Detection of
Chemicals and Chemical Agents in Buildings, Battelle, Columbus, Ohio, July 2003.
2. Environmental Technology Verification Program Quality Management Plan, Version 2.0,
December 2002.
3. "Determination of Phosgene in Air," in Methods of Air Sampling and Analysis, Third
Edition, J. P. Lodge, ed., Lewis Publishers, Chelsea, Michigan, 1989.
4. Spicer, C. W., Gordon, S. M., Holdren, M. W., Kelly, T. J., and Mukund, R., Hazardous Air
Pollutant Handbook: Measurements, Properties, and Fate in Ambient Air, ISBN 1-56670-
571-1, CRC Press, Boca Raton, Florida, 2002.
5. Turpin, J. "Direct Steam Injection Humidification: Is It Safe for Building Occupants?"
Engineered Systems, www.esmagazine.com/CDA/ArticleInformation/features/BNP
Features__Item/0,2503,23246,OO.html.
45
------- |