4
Air Monitoring Systems

SYSTEMS USED TO MONITOR AT THE 1988/1997 AELS

MINICAMS, a low-level, near-real-time (NRT) air monitor, and DAAMS, a manual historical monitoring system, are used for the detection of agents that may be present in the air at non-stockpile disposal sites, at stockpile disposal sites, and at agent storage facilities. MINICAMS, an automated, near-real-time (NRT) system, is used to monitor sulfur mustard (distilled) (HD), sarin (GB), VX, and other agents of concern in the non-stockpile program using the time-weighted-average (TWA) airborne exposure limits (AELs) and the immediately dangerous to life and health (IDLH) AELs (GB and VX only) set by the Centers for Disease Control and Prevention (CDC) in 1988 and the U.S. Army in 1997. MINICAMS typically reports the concentration of agent in air once every 3 to 10 minutes (U.S. Army, 2003a).

An IDLH AEL was only recently defined for HD (Federal Register, 2004). However, HD has been monitored for many years at stockpile sites by MINICAMS and by an automatic continuous agent monitoring system (ACAMS), the predecessor of MINICAMS, at concentrations much greater than the CDC’s 2004 IDLH value.1

If the agent concentration reported by a MINICAMS exceeds a preset alarm level, the MINICAMS displays audible and visible signals to alert an operator that the concentration of agent reported for the area being monitored has exceeded the set point.2 The operator then takes actions in response to the alarm. Alarm levels for MINICAMS used at non-stockpile sites are typically set at 0.70 of the AEL of concern for the agent of concern.

DAAMS, a manual monitoring system, is used at stockpile and non-stockpile sites to confirm or deny MINICAMS TWA alarms (that is, reports of the presence of agent at concentrations greater than the alarm level) and at stockpile sites to conduct historical monitoring at the CDC’s 1988 TWA and GPL AELs for HD, GB, and VX. Monitoring at GPL levels is not typically done at non-stockpile sites because non-stockpile operations involve only small quantities of agent (compared with stockpile operations) and are generally short term (U.S. Army, 2004g). Since it is highly unlikely for the general public to be exposed to agent for long periods, the general public is not considered at risk of long-term health problems from non-stockpile disposal operations.

Also, non-stockpile operations are often conducted within the perimeter of stockpile sites—for example, at the Newport Chemical Depot (NECD). In such instances, public access to non-stockpile sites is limited and the perimeter monitoring conducted in support of stockpile operations may be used to demonstrate that GPL levels are not exceeded at the perimeter, regardless of the source of agent (stockpile or non-stockpile operations).

In addition to continuous monitoring, at the time this report was written MINICAMS was used to verify decontamination to the 3X level—that is, to verify that the concentration of agent in the headspace air surrounding bagged items as a result of off-gassing does not exceed the CDC’s

1  

The stockpile program uses a totally encapsulated suit with a self-contained breathing apparatus (SCBA), known as the demilitarization protective ensemble (DPE)—up to the DPE use limit of 100 mg/m3 of airborne agent. ACAMS and MINICAMS have been used for more than 20 years to monitor for HD at concentrations ranging from 0.003 mg/m3 (the previous 8-hr TWA level) to 100 mg/m3 (the DPE limit). These NRT monitoring systems are able to monitor for HD over this wide concentration range simply by varying the volume of air from which agent is collected—through the adjustment of the sample flow rate and the duration of the sample period or through the use of an external fixed volume sampler (sample loop) connected to the inlet of the NRT monitor. Thus, it should be simple to monitor at the newly defined IDLH level for HD.

2  

Alarm level is a predetermined value for an NRT method that, when equaled or exceeded, will result in an audible and/or visual alarm at the NRT monitor. The alarm level must be set so that the statistical response rate is ≥95 percent. In other words, the probability, expressed as a percentage, that a 1.0-AEL first challenge to the NRT monitor will generate a response greater than or equal to the alarm level must be ≥95 percent (U.S. Army, 2004f).



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 43
Impact of Revised Airborne Exposure Limits on Non-Stockpile Chemical Materiel Program Activities 4 Air Monitoring Systems SYSTEMS USED TO MONITOR AT THE 1988/1997 AELS MINICAMS, a low-level, near-real-time (NRT) air monitor, and DAAMS, a manual historical monitoring system, are used for the detection of agents that may be present in the air at non-stockpile disposal sites, at stockpile disposal sites, and at agent storage facilities. MINICAMS, an automated, near-real-time (NRT) system, is used to monitor sulfur mustard (distilled) (HD), sarin (GB), VX, and other agents of concern in the non-stockpile program using the time-weighted-average (TWA) airborne exposure limits (AELs) and the immediately dangerous to life and health (IDLH) AELs (GB and VX only) set by the Centers for Disease Control and Prevention (CDC) in 1988 and the U.S. Army in 1997. MINICAMS typically reports the concentration of agent in air once every 3 to 10 minutes (U.S. Army, 2003a). An IDLH AEL was only recently defined for HD (Federal Register, 2004). However, HD has been monitored for many years at stockpile sites by MINICAMS and by an automatic continuous agent monitoring system (ACAMS), the predecessor of MINICAMS, at concentrations much greater than the CDC’s 2004 IDLH value.1 If the agent concentration reported by a MINICAMS exceeds a preset alarm level, the MINICAMS displays audible and visible signals to alert an operator that the concentration of agent reported for the area being monitored has exceeded the set point.2 The operator then takes actions in response to the alarm. Alarm levels for MINICAMS used at non-stockpile sites are typically set at 0.70 of the AEL of concern for the agent of concern. DAAMS, a manual monitoring system, is used at stockpile and non-stockpile sites to confirm or deny MINICAMS TWA alarms (that is, reports of the presence of agent at concentrations greater than the alarm level) and at stockpile sites to conduct historical monitoring at the CDC’s 1988 TWA and GPL AELs for HD, GB, and VX. Monitoring at GPL levels is not typically done at non-stockpile sites because non-stockpile operations involve only small quantities of agent (compared with stockpile operations) and are generally short term (U.S. Army, 2004g). Since it is highly unlikely for the general public to be exposed to agent for long periods, the general public is not considered at risk of long-term health problems from non-stockpile disposal operations. Also, non-stockpile operations are often conducted within the perimeter of stockpile sites—for example, at the Newport Chemical Depot (NECD). In such instances, public access to non-stockpile sites is limited and the perimeter monitoring conducted in support of stockpile operations may be used to demonstrate that GPL levels are not exceeded at the perimeter, regardless of the source of agent (stockpile or non-stockpile operations). In addition to continuous monitoring, at the time this report was written MINICAMS was used to verify decontamination to the 3X level—that is, to verify that the concentration of agent in the headspace air surrounding bagged items as a result of off-gassing does not exceed the CDC’s 1   The stockpile program uses a totally encapsulated suit with a self-contained breathing apparatus (SCBA), known as the demilitarization protective ensemble (DPE)—up to the DPE use limit of 100 mg/m3 of airborne agent. ACAMS and MINICAMS have been used for more than 20 years to monitor for HD at concentrations ranging from 0.003 mg/m3 (the previous 8-hr TWA level) to 100 mg/m3 (the DPE limit). These NRT monitoring systems are able to monitor for HD over this wide concentration range simply by varying the volume of air from which agent is collected—through the adjustment of the sample flow rate and the duration of the sample period or through the use of an external fixed volume sampler (sample loop) connected to the inlet of the NRT monitor. Thus, it should be simple to monitor at the newly defined IDLH level for HD. 2   Alarm level is a predetermined value for an NRT method that, when equaled or exceeded, will result in an audible and/or visual alarm at the NRT monitor. The alarm level must be set so that the statistical response rate is ≥95 percent. In other words, the probability, expressed as a percentage, that a 1.0-AEL first challenge to the NRT monitor will generate a response greater than or equal to the alarm level must be ≥95 percent (U.S. Army, 2004f).

OCR for page 43
Impact of Revised Airborne Exposure Limits on Non-Stockpile Chemical Materiel Program Activities 1988 TWA AELs.3 DAAMS is used to confirm or deny any MINICAMS alarms that occur during 3X monitoring. Both MINICAMS and DAAMS monitors are typically configured for sampling using glass tubes packed with a porous polymer. The sample is separated using temperature-programmed capillary gas chromatography, and detection is done using a flame photometric detector (FPD). The FPD may be operated in a phosphorus-specific mode for the detection of GB and VX or in a sulfur-specific mode for the detection of HD. The FPD in the MINICAMS may be replaced with either a pulsed flame photometric detector (PFPD), which may be operated in a phosphorus- or sulfur-specific mode, or with a halogen-selective detector (XSD), which is sensitive only to chlorinated and brominated compounds. A mass selective detector (MSD), in addition to the FPD, may be installed in the laboratory-grade gas chromatographs used in the DAAMS method. MINICAMS provides a more rapid warning, but there is generally a greater risk of false positives (for MINICAMS and NRT monitors in general) than there is with DAAMS. This is true because MINICAMS typically has poorer gas-chromatographic resolution than the more time-consuming and more sophisticated manual sampling, laboratory-based analysis, and reporting methods on which DAAMS is based.4 Also, DAAMS signal-to-noise ratios are typically greater than those for MINICAMS because the volume of air sampled by DAAMS tubes is greater, making the mass of agent collected for a given AEL setting greater as well. Other automated NRT systems that have been used or tested at various storage and disposal sites are essentially automated DAAMS, commonly known by the acronym A/DAM (Agilent/Dynatherm agent monitor). The A/DAM system consists of a Dynatherm ACEM 900 sorbent-based sampling system connected to an Agilent 6890 gas chromatograph or, in its latest configuration, a Dynatherm IACEM 980 sorbent-based sampling system connected to an Agilent 6852 gas chromatograph. Both A/DAM systems are configured for sampling using a glass tube packed with a porous polymer, separation using a temperature-programmed capillary gas chromatograph (GC), and flame photometric detection. Both the 6890- and the 6852-based A/DAM systems can be configured to achieve better chromatographic resolution than MINICAMS, so in certain situations, they may experience fewer false positives. In addition, the A/DAM system can be configured with two separate GC columns and with two separate FPDs, which improves selectivity with respect to chemical interferences. Monitoring systems (and their associated written methods) used at non-stockpile and stockpile disposal sites must be certified before use in accordance with requirements stated in the Chemical Materials Agency’s (CMA’s) Programmatic Laboratory and Monitoring Quality Assurance Plan (U.S. Army, 2004f). Certification generally includes passing a 4-day precision-and-accuracy (P&A) study using liquid standard solutions to estimate the performance of monitoring systems when they sample actual agent vapor. Note that P&A studies for DAAMS and MINICAMS are usually conducted over a relatively narrow concentration range, typically 0.20 to 1.50 AEL in the past and now 0.50 to 2.00 AEL (as presented in the latest version of the Programmatic Laboratory and Monitoring Quality Assurance Plan (U.S. Army, 2004f)). The goals of a P&A study are (1) to demonstrate that when used for the detection of a true agent concentration of 1.00 AEL, the monitoring system (and its associated written method) is predicted to report a found concentration in the range of 0.75 to 1.25 AEL (that is, 75 to 125 percent recovery) with a precision of ±25 percent with 95 percent confidence and (2) to document the precision and accuracy of the monitoring system at all concentrations used in the study (U.S. Army, 2001e). Monitoring systems and written methods are generally not tested formally outside the concentration range required for the P&A study (U.S. Army, 2004f). Thus, the accuracy of a given monitoring system for concentrations outside the range tested is generally uncertified. This fact is important to keep in mind when extrapolating the performance of monitoring systems and methods at the 1988 CDC 0.20 to 1.50 AEL concentration range to the 2003/2004 CDC 0.50 to 2.00 AEL concentration range. This chapter does four things: (1) it documents technologies used before 2005 in the non-stockpile program or available to monitor at the CDC’s 1988 AELs; (2) it assesses the ability of these monitoring systems and associated methods to monitor at the CDC’s 2003/2004 AELs; (3) it recommends upgrades to existing monitoring systems and identifies technologies recommended for further development; and (4) it addresses alarm levels and their relationship to AELs. MINICAMS MINICAMS is an automated NRT monitor for the detection of GB, HD, or VX that, as previously noted, is typically configured with a sampling tube, a capillary GC column, and an FPD. During the sampling period, chemicals present in the air stream pulled into the MINICAMS are trapped in the sampling tube, which is usually a glass tube packed with particles of HayeSep D (for G and VX agents) or Tenax-TA (for HD). After the sampling period, an inert carrier gas 3   Known as the X Classification System, this system, which is described in Chapter 5, defines levels of agent decontamination for materials and waste and defines subsequent management procedures (U.S. Army, 2002). 3X is applied to materials or waste that have been surface decontaminated such that they do not produce a vapor concentration in excess of the agent-specific AEL for an unmasked worker. 4   Letter from Vickie H. Paul, Dynatherm Sales Manager, CDS Analytical, Inc., to John Decker, CDC National Center for Environmental Health, June 28, 2002; Personal communication between Vickie H. Paul, Dynatherm Sales Manager, CDS Analytical, Inc., and Gary Sides, committee member, August 20, 2004.

OCR for page 43
Impact of Revised Airborne Exposure Limits on Non-Stockpile Chemical Materiel Program Activities (helium or nitrogen) is allowed to flow through the sampling tube and into the GC column. The sorbent bed in the tube is then heated to desorb the collected chemicals, which are swept into the GC column by the flow of carrier gas. The chemicals are then separated on the GC column, which is temperature programmed. Ideally, the agent of interest exits the GC column into the detector at a time, known as the retention time, when no other chemical is entering the detector. That is, the agent of interest should have a GC retention time that does not overlap the retention time for any other chemical exiting the GC column. The agent can then be detected without interference by measuring light emitted from the species HPO* (phosphorus emission) for G and V agents or by measuring light emitted from the species S2* (sulfur emission) for HD.5 Before use, each MINICAMS must be calibrated. Calibration consists of injecting known masses of agent into the inlet of the MINICAMS during successive instrument cycles—specifically, microliter volumes of a dilute solution of agent are injected. Thus, the response (detector signal) versus mass of agent ratio can be determined. After calibration, the responses obtained during subsequent MINICAMS cycles can be converted to detected masses and to detected concentration readings, which are then reported by the MINICAMS. The calibration procedure is covered in greater detail later in this section. GB and HD are sampled and detected directly by MINICAMS. Because of its low volatility and high affinity for irreversible adsorption on surfaces, however, VX is first derivatized by reaction with silver fluoride to yield the more volatile and less reactive G-analog of VX, which is then sampled and detected by MINICAMS. The derivatization of VX is accomplished in real time by installing a V-to-G conversion pad on the inlet of the MINICAMS or on the distal end of a heated sample line connected to the inlet. The conversion pad consists of a polyester felt pad impregnated with potassium fluoride and silver nitrate. As noted previously, the FPD may be operated in a phosphorus-specific mode (by monitoring HPO* emissions through a 525-nm, narrow-bandpass optical filter) or in a sulfur-specific mode (by monitoring S2* emissions through a 396-nm, narrow-bandpass optical filter). In the phosphorus-specific mode, the FPD is about 10,000 times more sensitive to phosphorus than to carbon on the basis of signal per unit mass. In the sulfur-specific mode, the FPD is also about 10,000 times more sensitive to sulfur than to carbon for the mass range of interest when monitoring for HD at its STEL (Thurbide and Aue, 1994). The FPD is less selective for phosphorus versus sulfur than for phosphorus versus carbon or phosphorus versus hydrocarbons. Thus, despite the selectivity of the FPD, sulfur emissions (resulting from the formation of S2* in the FPD) and hydrocarbon emissions (resulting, for example, from the formation of CH*) can cause interference (false positives) in the phosphorus-specific mode. Also, two kinds of organophosphorus compounds can cause false positives in the phosphorus-specific mode: (1) organophosphorus compounds that are not agents but have the same retention time as GB or the G-analog of VX and (2) organophosphorus compounds other than VX—for example, certain pesticides—that also undergo V-to-G conversion to yield the G-analog of VX. In addition, hydrocarbons can quench (reduce) sulfur and phosphorus emissions, causing false negatives. For example, a concentration of only a few parts per million of a hydrocarbon, present in an area sampled by MINICAMS and with the same retention time as the agent being monitored, can result in the quenching of phosphorus or sulfur emissions by about 50 percent (Aue and Sun, 1993). During one recent 22-month period at the former production facility at NECD (March 2002 through December 2003), about 1.5 percent of the VX readings reported by MINICAMS, corresponding to 80 different events, were greater than or equal to the alarm level set point (0.70 TWA). DAAMS samples collected to confirm or deny the MINICAMS alarm events showed that all but one of these events were due to false positives.6 At least some of the false positives were thought to have been caused by O,S-diethyl methyl-phosphonothioate (O,S-DMP), diethyl methylphosphonate (TRO), or related phosphorus-containing compounds in the air at the facility. In addition, false positives caused by plasticizers (hydrocarbons) have occurred at the NECD former production facility.7 Recent changes in operations at the NECD facility, especially better ventilation (that is, a higher rate of air exchange) in Building 143, greatly reduced the rate of false positives.8 5   The asterisks in HPO* and S2* refer to the electronically excited states for HPO and S2. Light is emitted from these excited states and detected by the photomultiplier tube in the FPD as follows: where hv represents a photon of light with a wavelength centered at about 526 nm for the HPO* emission and at about 396 nm for the S2* emission. 6   Tom Hoff, NECD Project Manager, and William Rogers, Tennessee Valley Authority (TVA) Quality Officer, Briefing to the Department of Health and Human Services, March 11, 2004; William Rogers, TVA Quality Officer, Briefing to the committee, August 3, 2004. 7   Terry Frederick, TVA, Briefing to the committee, September 14, 2004. 8   Terry Frederick, TVA, Briefing to the committee, September 14, 2004.

OCR for page 43
Impact of Revised Airborne Exposure Limits on Non-Stockpile Chemical Materiel Program Activities False positives and false negatives caused by chemicals with the same retention time as the agent being monitored can sometimes be eliminated by installing a detector not subject to the same interferences. For example, false positives caused by the presence of sulfur-containing compounds with the same retention time as HD can be eliminated by replacing the FPD with an XSD. The XSD detects the chlorine in HD, but it does not detect the sulfur present in HD or in other sulfur-containing compounds that may interfere with the detection of HD when using the FPD. Of course, a MINICAMS configured with an XSD in place of the FPD may then be subject to false positives caused by the presence of chlorine-containing compounds other than HD in the area being monitored. The XSD also has the advantage of a linear response for HD, in contrast to the FPD’s quadratic response, and the XSD can detect lower levels of HD than the FPD. The response of the XSD, however, is less stable than that of the FPD, and the XSD requires more frequent maintenance and service than the FPD. For example, the XSD for MINICAMS is usually sold with a spare reactor probe assembly, which must be replaced after a few months of operation. By comparison, the FPD may be operated for years without requiring maintenance or repair. Again, only one detector at a time (FPD, PFPD, or XSD) can be installed in the MINICAMS. As noted previously, MINICAMS may also be configured with a pulsed flame photometric detector (PFPD). The PFPD is more selective than the FPD for phosphorus emissions than for sulfur and hydrocarbon emissions (Cheskis et al., 1993). Thus, the PFPD operated in the phosphorus-specific mode can result in fewer false positives caused by organosulfur compounds and by hydrocarbons than the FPD operated in the phosphorus-specific mode. The PFPD, however, is still susceptible to false positives caused by organophosphorus compounds and to false negatives (quenching) caused by hydrocarbons (Cheskis et al., 1993). In addition, the PFPD is more complex and more costly than the FPD (FOCIS, 2003a). To certify MINICAMS as a monitor for a given agent at a specific AEL (TWA or IDLH limit), a method that describes the proper analytical use of the monitor must first be written, reviewed, and approved. At least two MINICAMS must then pass a 4-day precision-and-accuracy (P&A) study, during which the monitors are operated and maintained by two different trained operators in accordance with the written method. A P&A study consists of first calibrating the MINICAMS and then conducting two series of challenges of the monitor each day over a 4-day period using dilute solutions of the agent. Each series of challenges consists of injecting microliter volumes of the agent solution into the inlet of the MINICAMS so that the mass of agent injected corresponds to the mass of agent that would be collected if the monitor were sampling air containing agent at the following concentrations: 0.00 (blank), 0.20, 0.50, 0.80, 1.00, and 1.50 AEL (or, as stated in Table 10-2 of the latest version of the Programmatic Laboratory and Monitoring Quality Assurance Plan, 0.00, 0.50, 0.75, 1.00, 1.50, and 2.00 AEL) (U.S. Army, 2004f). At the end of the 4-day test period, the concentrations reported by the MINICAMS are analyzed statistically to determine whether the monitor passed the certification test. As noted previously, they are analyzed (1) to demonstrate that, when used for the detection of a true agent concentration of 1.00 AEL, the monitoring system (and associated written method) will report a found concentration in the range of 0.75 to 1.25 AEL (that is, 75 to 125 percent recovery) with a precision of ±25 percent and 95 percent confidence and (2) to document the precision and accuracy of the monitoring system at all of the concentrations used in the P&A study. Many P&A studies of MINICAMS have been conducted successfully by non-stockpile and stockpile staff during the past 10 years.9 These studies have shown that MINICAMS are capable of reporting GB, VX, and HD concentrations with an accuracy of ±25 percent and a precision of ±25 percent with 95 percent confidence at true concentrations of 1.00 TWA and 1.00 IDLH for each agent and have documented the precision and accuracy of the monitoring system at the concentrations used in the P&A studies in the ranges 0.20 to 1.50 TWA and 0.20 to 1.50 IDLH. Although MINICAMS will report agent concentration readings below 0.20 AEL and above 1.50 AEL, these concentration reports are not “certified.” That is, MINICAMS used in the field typically have not passed P&A studies that encompass agent challenge concentrations below 0.20 AEL or above 1.50 AEL. However, over the last 10 years, P&A studies and baseline studies—another type of certification test defined in the CMA Programmatic Laboratory and Monitoring Quality Assurance Plan (U.S. Army, 2004f)—have shown that if a MINICAMS reports an agent concentration greater than the P&A study range, there is confidence that the true concentration of agent is greater than 1.50 AEL. Thus, MINICAMS reliably warns of agent concentrations greater than 1.50 AEL even though it is generally not certified for accuracy at concentrations greater than 1.50 AEL. During the past 10 years, hundreds of MINICAMS have been certified for operation in the range 0.20 to 1.50 TWA for HD, GB, and VX and in the range 0.20 to 1.50 IDLH limit for GB and VX. MINICAMS have also been certified and used successfully to monitor at HD concentrations greater than the CDC’s IDLH limit for this agent. Successful operation for these agents and in these concentration ranges has been demonstrated repeatedly during the past 10 years for a wide variety of environments and facilities. It is thus reasonable to assume that the MINICAMS has been proved reliable during field operation. This is illustrated in Figures 4-1 and 4-2, where it is labeled “1988/1997 MINICAMS.” 9   In this report, “precision-and-accuracy study” refers to the practice of using liquid standard solutions to estimate the performance of monitoring systems when the systems are used to sample actual agent vapor.

OCR for page 43
Impact of Revised Airborne Exposure Limits on Non-Stockpile Chemical Materiel Program Activities FIGURE 4-1 MINICAMS and DAAMS operating ranges for the 1988/1997 GB AELs and required ranges for the CDC’s 2003 GB AELs. NOTES: (1) The 1988/1997 AEL concentration ranges for GB over which MINICAMS and DAAMS have been certified and operated for many years at various sites are indicated by horizontal lines that end in arrows. (2) The “1988/1997 MINICAMS” line includes a dotted line to its left. Although MINICAMS has not been certified or used to monitor for GB in the concentration range represented by the dotted line, its performance when monitoring for VX in the range 0.20 to 1.50 TWA indicates that it should be possible to calibrate and certify MINICAMS for monitoring GB at phosphorus-equivalent concentrations corresponding to 0.20 to 1.50 TWA of VX; (3) The “1988 DAAMS” line also has a dotted line to its left. Although DAAMS has not been certified or used to monitor for GB in the concentration range represented by the dotted line, its performance when monitoring for VX in the range 0.20 to 1.50 of the CDC’s 1988 GPL for VX indicates that it should be possible to calibrate and certify DAAMS for monitoring GB at phosphorus-equivalent concentrations corresponding to 0.20 to 1.50 of the 1988 GPL for VX; (4) The IDLH AELs are represented by diamonds. The IDLH concentration range used in the past when certifying MINICAMS, 0.20 to 1.50, and the concentration range required in the future, 0.50 to 2.00, are represented by range bars on the diamonds; (5) The CDC’s 1988 TWA AEL and the CDC’s 2003 15-minute STEL for GB (numerically equivalent to the 1988 TWA AEL) are represented by triangles with range bars; (6) The CDC’s 2003 WPL for GB is represented by a square with a range bar, and the 1988 and 2003 GPLs are represented by circles with range bars; (7) The GB concentration limit above which negative-pressure respirators, such as the M40 mask, may not be worn is marked by an asterisk and labeled “50 WPL (M40 limit)”; and (8) Limits for supplied-air respirators (1,000 WPL) and SCBA (10,000 WPL) are not shown.

OCR for page 43
Impact of Revised Airborne Exposure Limits on Non-Stockpile Chemical Materiel Program Activities FIGURE 4-2. MINICAMS and DAAMS operating ranges for the 1988/1997 VX AELs and required ranges for the CDC’s 2003 VX AELs. NOTES: (1) The 1988/1997 AEL concentration ranges for VX over which MINICAMS and DAAMS have been certified and operated for many years at various sites are indicated by horizontal lines that end in arrows. (2) The IDLH AELs are represented by diamonds. The IDLH concentration range used in the past when certifying MINICAMS, 0.20 to 1.50, and the concentration range required in the future, 0.50 to 2.00, are represented by range bars on the diamonds; (3) The CDC’s 1988 TWA AEL and the CDC’s 2003 15-minute STEL for VX (numerically equivalent to the 1988 TWA AEL) are represented by triangles with range bars; (4) The CDC’s 2003 WPL for VX is represented by a square with a range bar, and the 1988 and 2003 GPLs are represented by circles with range bars; (5) The VX concentration limit above which negative-pressure respirators, such as the M40 mask, may not be worn is marked by an asterisk; and (6) Limits for supplied-air respirators (1,000 WPL) and SCBA (10,000 WPL) are not shown. The 1988/1997 MINICAMS and current DAAMS concentration ranges shown in Figures 4-1 through 4-3 reflect the performance of these systems when monitoring at the CDC’s 1988 AELs (and at the Army’s 1997 IDLH limits for GB and VX), performance proven at many different non-stockpile and stockpile sites during the past 10-15 years. The CDC’s 2003/2004 AELs and required operating ranges for GB, VX, and HD are also presented in Figures 4-1 through 4-3. The projected and actual performance (as of the date of preparation of this report) for MINICAMS and DAAMS when monitoring at the CDC’s 2003/2004 AELs is discussed in the next section. Note that VX is detected as its G-analog, which differs from GB only by the presence of an ethyl group in place of an isopropyl group. Thus, it is likely that MINICAMS could be successfully certified for GB at concentrations at least an order of magnitude less than the lower limit shown for the 1988/1997 MINICAMS range in Figure 4-1. (This extension of the GB lower detection limit vis-à-vis the 1988/1997 MINICAMS range for VX is shown by the dotted line to the left of the 1988/1997 MINICAMS line at the top of Figure 4-1.) DAAMS Manual DAAMS samples are collected by pulling air through glass sampling tubes packed with a porous polymer for periods of time ranging from a few minutes for NRT confirmation samples to as long as 12 hours for GPL historical monitoring. Sampling tubes are placed at various locations within a given site. Most of the sampling tubes are co-located with NRT monitors such as MINICAMS. DAAMS samples

OCR for page 43
Impact of Revised Airborne Exposure Limits on Non-Stockpile Chemical Materiel Program Activities FIGURE 4-3 MINICAMS and DAAMS operating ranges for the 1988 HD AELs and required ranges for the CDC’s 2004 HD AELs. NOTES: (1) The 1988 AEL concentration ranges for HD over which MINICAMS and DAAMS have been certified and operated for many years at various sites are indicated by horizontal lines that end in arrows. (2) The “1988 MINICAMS” line includes a dotted line to its right. This dotted line represents the fact that even though an IDLH AEL for HD was not defined until recently, MINICAMS has been used for many years to monitor for HD at concentrations up to and far above the CDC’s 2004 IDLH AEL for HD; (3) The CDC’s 2004 IDLH AEL for HD is represented by a diamond. The IDLH concentration range required in the future when certifying MINICAMS, 0.50 to 2.00, is represented by range bars on the diamond; (4) The CDC’s 1988 TWA AEL and the CDC’s 2004 15-minute STEL for HD (numerically equivalent to the 1988 TWA AEL) are represented by triangles with range bars; (5) The CDC’s 2004 WPL is represented by a square with a range bar, and the 1988 and 2004 GPLs are represented by circles with range bars; and (6) Limits for supplied-air respirators (1,000 WPL) and SCBA (10,000 WPL) are not shown. are currently analyzed for agent to provide historical monitoring data for TWA AELs and, where applicable, for GPL AELs. DAAMS samples are also analyzed to confirm or deny TWA alarms sounded by NRT monitoring systems. The analysis of DAAMS samples is accomplished using an Agilent Model 6890 GC connected to a Dynatherm ACEM 900 or a Dynatherm IACEM 980 thermal desorption system, which is configured to receive and desorb manually collected samples. For samples collected and analyzed in support of historical monitoring, the DAAMS GC is usually configured with an FPD. DAAMS GCs used to confirm or deny NRT monitoring alarms may also be configured with a mass spectroscopy detector (MSD) or with an FPD and an MSD. Because the analysis of DAAMS samples is based on the use of laboratory-grade GCs, which may be configured in many different ways, the configuration of DAAMS GCs may vary widely. Most DAAMS GC systems in use at stockpile and non-stockpile sites include a precolumn and an analytical column in series. In this configuration, low-boiling compounds and agents are first allowed to pass from the precolumn into the analytical column. The flow of carrier gas within the precolumn is then reversed to allow high-boiling compounds to be backflushed from the precolumn to clean and ready it for the next sample to be analyzed. While the precolumn is being backflushed, carrier gas continues to flow through the analytical column and into the detector, allowing the detection of agents of interest. The initial step in the analysis of a DAAMS sample consists of inserting the sampling tube into a port on the Dynatherm thermal desorption unit. Agent desorbed from the sampling tube is first collected on a sorbent bed in a

OCR for page 43
Impact of Revised Airborne Exposure Limits on Non-Stockpile Chemical Materiel Program Activities small-bore focusing tube within the Dynatherm unit. Agent is then desorbed from the small-bore tube into the precolumn within a narrow band. For this reason, the DAAMS typically achieves better chromatographic resolution than MINICAMS and thus is more selective and experiences fewer false positives (interferences) than MINICAMS. This enables DAAMS monitors to be used effectively to confirm or deny the presence of agent in areas monitored by MINICAMS. As with MINICAMS, each DAAMS system must be calibrated before use. Calibration consists of injecting known masses of agent in dilute microliter volumes of solution into the inlets of DAAMS sampling tubes while ambient air is flowing through the tubes. Each tube is then analyzed for agent to yield the response (detector signal) versus the mass of agent. After calibration, the responses obtained for subsequent DAAMS samples can be converted to detected masses and detected concentrations, which are then reported by the DAAMS system. At most sites, DAAMS relies on the direct detection of GB and HD and the detection of VX as its G-analog. A DAAMS configured only with an FPD, however, is susceptible to the same types of false positives (interferences) as MINICAMS. For example, in the phosphorus-specific mode of operation, false positives may be caused by any hydrocarbon, organosulfur compound, or organophosphorus compound that has the same retention time as GB or the G-analog of VX. However, DAAMS generally experiences far fewer false positives for a given AEL than MINICAMS. This is due to the superior (relative to MINICAMS) chromatographic resolution possible using laboratory-grade gas chromatographs and the larger mass of agent that can be collected using DAAMS methods, which results in a greater signal-to-noise ratio. For example, as noted previously, during one 22-month period at the NECD FPF, about 1.5 percent of the VX readings reported by MINICAMS, corresponding to 80 different events, were greater than or equal to the alarm level set point (0.70 TWA). DAAMS samples collected to confirm or deny the MINICAMS alarms showed that all but one of these events were due to false positives caused by chemical interferences. Despite the better resolution of DAAMS compared with MINICAMS, DAAMS is susceptible to false positives caused by organophosphorus compounds that undergo V-to-G conversion to yield the G-analog of VX, whether the DAAMS GC is configured with an FPD, an MSD, or both. For example, O,S-DMP will interfere with the DAAMS determination of VX when the V-to-G conversion method is used.10 At NECD, a DAAMS method that collects and analyzes VX directly has proven effective in eliminating false positives caused by O,S-DMP and similar compounds. At first, this method could only be used for sampling times up to about 1 hour because of poor VX recoveries observed for longer sample periods.11 Recent work at NECD, however, has resulted in a direct VX DAAMS method for the CDC’s 2003 VX WPL that has passed P&A studies with a sample period of 6 hours.12 Like MINICAMS, DAAMS instruments and methods must be certified before use by conducting 4-day P&A studies. At the end of the 4-day test period, the concentrations detected using DAAMS are analyzed statistically to determine whether a given instrument or method passed the certification test—that is, whether the instrument or method will report a found concentration in the range of 0.75 to 1.25 AEL (that is, 75 to 125 percent recovery) with a precision of ±25 percent and 95 percent confidence—and, also, to document the precision and accuracy of the DAAMS method at all concentrations used in the study (typically in the range 0.20 to 1.50 AEL). During the past 20 years, hundreds of DAAMS systems have been certified for operation and used successfully to monitor GB, VX, and HD manually in the range 0.20 to 1.50 TWA and 0.20 to 1.50 GPL. Successful operation for these agents and in these concentration ranges has been demonstrated repeatedly during this period in a wide variety of environments and facilities. It is thus reasonable to assume that the DAAMS has been field proven in the concentration noted above. This is illustrated in Figures 4-1 through 4-3, where these ranges are labeled “1988 DAAMS.” Other Monitoring Systems (A/DAM) Other automated NRT monitors that have been used or tested at various storage and disposal sites include a monitor based on the Dynatherm ACEM 900 thermal desorption unit connected to an Agilent 6890 GC and a newer, improved NRT monitor based on the Dynatherm IACEM 980 unit and the Agilent 6852 GC. Both monitoring systems are known by the same acronym, A/DAM. Both are typically configured for sampling using a glass tube packed with a porous polymer, separation using a temperature-programmed capillary GC, and flame photometric detection. Both the 6890- and the 6852-based monitoring systems can be configured to achieve better chromatographic resolution (and thus better selectivity) than MINICAMS and may therefore experience fewer false positives for phosphorus-containing compounds and other compounds with retention times similar to that of the G-analog of VX (for phosphorus-containing compounds that do not undergo conversion to yield the G-analog of VX). 10   Tom Hoff, NECD Project Manager, and William Rogers, TVA Quality Assurance Officer, Briefing to the Department of Health and Human Services, March 11, 2004; William Rogers, TVA Quality Assurance Officer, Briefing to the committee, August 3, 2004. 11   Tom Hoff, NECD Project Manager, and William Rogers, TVA Quality Assurance Officer, Briefing to the Department of Health and Human Services, March 11, 2004; William Rogers, TVA Quality Assurance Officer, Briefing to the committee, August 3, 2004. 12   Terry Frederick, TVA, Briefing to the committee, September 14, 2004.

OCR for page 43
Impact of Revised Airborne Exposure Limits on Non-Stockpile Chemical Materiel Program Activities In addition, the 6852-based A/DAM system can be configured with two separate GC columns and two separate FPDs. Gas streams exiting the IACEM 980 can be split between the two analytical channels. If the two columns have liquid phases that are sufficiently dissimilar (for example, nonpolar and polar), a given agent will be detected at two different retention times on the two different analytical channels. Other potential interferences (for example, hydrocarbons) are unlikely to exhibit the same retention times as the agent on the two different columns. In this manner, a much higher degree of selectivity is possible than for MINICAMS. Agilent recently introduced enhancements to the FPD that resulted in a twofold to fivefold improvement in the signal-to-noise ratio.13 In addition, Agilent has developed a method for the IACEM 980/Agilent 6852 system that allows determining VX directly, without derivatization. Although often difficult to implement for routine use, if successful, a direct method for VX should result in fewer interferences (false positives) in comparison with the V-to-G conversion method for VX. The ACEM 900/Agilent 6890 and the IACEM 980/Agilent 6852 A/DAM systems have been proven capable of monitoring at the CDC’s 1988 TWA levels for GB, HD, and VX at several sites during the past 10 years. The primary barriers to the more widespread use of these systems have been their greater cost, size, weight, and complexity compared with the MINICAMS. However, their greater analytical flexibility might offset these barriers in the future, especially when monitoring sites or operations where MINICAMS has produced numerous false positives. ABILITY OF SYSTEMS USED FOR MONITORING AT THE 1988/1997 AELS TO MONITOR AT THE 2003/2004 AELS The Army plans to use MINICAMS to monitor at non-stockpile sites for GB, HD, and VX at the 2003/2004 STELs and, when necessary, at the IDLH levels. DAAMS will be used at non-stockpile sites for historical monitoring at the CDC’s 2003/2004 WPLs and to confirm or deny MINICAMS alarms at the CDC’s 2003/2004 STELs (equal to the CDC’s 1988 TWA AELs). (Since the Army does not currently conduct GPL monitoring at non-stockpile sites, the reductions in the GPL levels are not expected to impact non-stockpile operations.) One other issue that must be considered is the protection factor of 50 that is assigned for negative-pressure respirators, such as the M40 mask. This means the M40 can be used in environments with GB or VX present at concentrations up to 50 times their WPLs for 8 hours and in environments with agent present at concentrations up to 50 times the STELs for not more than 15 minutes. Because HD is suspected to be a carcinogen, the concentration limit for the use of negative-pressure respirators for this agent is 1.00 STEL (equal to 7.50 WPL). MINICAMS The use of MINICAMS to monitor for GB, HD, and VX at the 2003/2004 IDLH levels and at the 2003/2004 STELs (numerically equal to the CDC’s 1988 TWAs) should be straightforward in view of its performance in monitoring at the 1988 AELs. It will only be necessary to develop and test a method for the 2004 HD IDLH level, to make minor modifications in operating parameters for existing IDLH methods for GB and VX, and to test the modified methods. The main problem for MINICAMS will continue to be monitoring at the STEL for VX (equal to the CDC’s 1988 TWA value)—especially at the NECD former production facility. False positives for VX at the TWA level at the NECD caused by phosphorus-containing compounds and other compounds with elution times similar to that of the G-analog of VX may be reduced by reconfiguring or upgrading the MINICAMS to improve its chromatographic resolution (for phosphorus-containing compounds that do not undergo conversion to yield the G-analog of VX, that is, O-ethyl methylphosphonofluoridate). False positives for VX at the TWA level caused by phosphorus-containing compounds other than VX that undergo conversion to yield the G-analog of VX may be eliminated by developing a MINICAMS method that can detect VX directly rather than as the G-analog. Both techniques will decrease interferences when monitoring for VX, so it would be preferable to monitor for VX directly and to improve chromatographic resolution. The dual-tube sampling system now available as an accessory for the MINICAMS results in a larger sample volume and, accordingly, the collection of a larger mass of agent for a given AEL, improving the signal-to-noise ratio but not the chromatographic resolution of the MINICAMS. As noted previously, the use of the PFPD in place of the FPD in MINICAMS results in improved selectivity for phosphorus-containing agents (GB and VX) versus hydrocarbons and sulfur-containing compounds (which may cause false positives). The PFPD, however, is more costly and complex to operate and maintain than the FPD (FOCIS, 2003a). Moreover, it would not solve the main problem with MINICAMS at the NECD former production facility—namely, the false positives caused by the presence of phosphorus-containing compounds, especially compounds that may undergo conversion to yield the G-analog of VX. In fact, a MINICAMS with a PFPD was recently tested at NECD, and there was no reduction in false positives.14 13   Letter from Wayne Abrams, Senior Consultant, Agilent Technologies, to John Decker, CDC National Center for Environmental Health, May 31, 2002. 14   Tom Hoff, NECD Project Manager, and William Rogers, TVA Quality Assurance Officer, Briefing to the Department of Health and Human Services, March 11, 2004; William Rogers, TVA Quality Assurance Officer, Briefing to the committee, August 3, 2004.

OCR for page 43
Impact of Revised Airborne Exposure Limits on Non-Stockpile Chemical Materiel Program Activities Because MINICAMS will be used in the non-stockpile program to monitor at the 2003/2004 STELs, which are numerically equivalent to the 1988 TWA levels, there is no need to improve the sensitivity of MINICAMS, at least based on the Army’s near-term monitoring requirements. Nonetheless, in anticipation of a possible future need for NRT monitoring at the 2003/2004 WPLs, the Army recently completed a laboratory study of the performance of MINICAMS and A/DAM systems modified to monitor for GB, HD, and VX at the 2003/2004 WPLs (FOCIS, 2003a). MINICAMS modified to include a PFPD and an external dual-tube sampler passed 4-day laboratory P&A studies for GB, VX, and HD. It passed 20-day laboratory baseline studies for GB and HD but failed the baseline study for VX because chemical interferences were present in the atmosphere being sampled during part of the baseline test period. A 4-day field P&A study and a field baseline study for all three agents at the 2003/2004 WPLs was recently conducted using MINICAMS and A/DAM systems at the Anniston, Alabama, stockpile site, and a report on this test is currently in preparation.15 Finding 4-1: MINICAMS, with only minor modifications to methods used to monitor at the 1988/1997 AELs, is capable of monitoring for GB, HD, and VX at the CDC’s 2003/2004 IDLH values and at the CDC’s 2003/2004 STELs, which are numerically equivalent to the CDC’s 1988 TWA AELs. Interference problems (false positives), especially those caused by organophosphorus compounds and plasticizers when monitoring for VX, will continue to occur for the current low-resolution MINICAMS configuration, especially when using the V-to-G conversion method. The A/DAM system can be configured to achieve better chromatographic resolution than the MINICAMS, to confirm agent detection automatically (using two independent analytical channels), and, although difficult to implement routinely, to detect VX directly (that is, without conversion of VX to the G-analog). Recommendation 4-1: To reduce false positives when monitoring at critical locations susceptible to chemical interferences, the Army should explore ways to improve the gas-chromatographic resolution of the MINICAMS. As an alternative, at critical locations, the Army should consider using the A/DAM system, which can be configured to achieve better chromatographic resolution than the MINICAMS. DAAMS Confirming or denying MINICAMS alarms at the 2003/2004 STELs will be no more difficult for the DAAMS than confirming or denying alarms at the CDC’s 1988 TWA levels. The CDC’s 2003 WPL for GB is greater than the CDC’s 1988 GPL for GB and greater than the CDC’s 1988 TWA level for VX, concentrations that have been monitored using DAAMS for many years. Thus, from the standpoint of sensitivity, it should be possible to use DAAMS techniques for monitoring at CDC’s 1988 AELs with only minor modifications for monitoring at CDC’s 2003 WPL for GB. Similarly, because the CDC’s 2004 WPL for HD is greater than the CDC’s 1988 GPL for HD, which has been monitored using DAAMS for many years, only relatively minor changes should be necessary with respect to sensitivity to allow DAAMS to monitor for HD at the new level. These statements assume that the sampling period for DAAMS, when sampling at the WPLs, is no less than 8 hours. (The typical DAAMS sample period used to monitor at the 1988 GPL for HD is 12 hours.) The response of the DAAMS FPD to sulfur-containing compounds, such as HD, is approximately quadratic. Thus, although DAAMS configured with an FPD appears to have the sensitivity to monitor at the CDC’s 2004 WPL for HD, the signal obtained will be about 50 times weaker than the signal at the CDC’s 1988 TWA AEL (for the same sample volume). Because the concentrations of potential chemical interferences in the environments being sampled will, of course, be unaffected by changes in the regulatory limits for HD, it is likely that a much higher rate of false positives will be observed when using DAAMS to monitor for HD at the CDC’s 2004 WPL rather than at the CDC’s 1988 TWA AEL. If interferences (false positives) increase for the CDC’s 2004 HD WPL and HD GPL—compared with the CDC’s 1988 AELs—installing an XSD in DAAMS in place of or in addition to the FPD may be a solution. The XSD has a linear response to HD, is more sensitive than the FPD, and is less susceptible to false positives from hydrocarbons than the FPD. Of course, the XSD is susceptible to false positives caused by chlorine-containing compounds. Almost all sites use the V-to-G conversion method when sampling for VX. Concentration ranges for DAAMS methods used to monitor at 1988/1997 AELs for VX are shown in Figure 4-2. The CDC’s 2003 WPL value for VX is less than the CDC’s 1988 GPL for VX, but it falls within the concentration range over which current DAAMS methods must be certified for monitoring at the CDC’s 1988 VX GPL. Detection limits for VX for current DAAMS have been reported to be as low as 30 picograms.16 For a sample flow rate of 1 liter per minute, a sample period of 8 hours, and a concentration of 1.00 WPL for VX (1 picogram per liter), 480 picograms of VX would be sampled. This mass is about 16 times greater than the lowest detection limit reported. MINICAMS and A/DAM systems have demonstrated the ability to detect VX at the CDC’s 2003 WPL with instrument cycles of 15 minutes or less. Both MINICAMS and the A/DAM are based on the same technologies as the historical 15   Personal communication between Rob O’Neil, FOCIS Associates, Inc., and Gary Sides, committee member, August 24, 2004. 16   Letter from Michael McNaughton, Southwest Research Institute, to John Decker, Centers for Disease Control and Prevention, June 28, 2002.

OCR for page 43
Impact of Revised Airborne Exposure Limits on Non-Stockpile Chemical Materiel Program Activities DAAMS method. Because sample periods for DAAMS may be 8 hours or more, DAAMS should collect a greater mass of agent when sampling at the WPLs and thus be able to achieve much higher signal-to-noise ratios than MINICAMS or the A/DAM system. From the standpoint of instrument sensitivity, therefore, it appears that DAAMS methods used to monitor at the 1988 AELs can be modified to determine VX at the CDC’s 2003 WPL. It is anticipated, however, that interference problems are likely to be much greater when using DAAMS based on V-to-G conversion to monitor at the CDC’s 2003 VX WPL than at the CDC’s 1988 VX TWA level, which is 10 times greater. The types of interference expected are those caused by hydrocarbons, sulfur-containing compounds, and phosphorus-containing precursors, impurities, and breakdown or decontamination products. At some sites, interferences caused by compounds such as O,S-DMP, which undergo V-to-G conversion to yield the G-analog of VX, may be a serious problem. For example, interferences from March 2002 through December 2003 resulted in MINICAMS readings >0.10 TWA (equal to 1.0 WPL) 24 percent of the time in Building 143 at the NECD former production facility. Non-stockpile staff anticipate that chemical interferences will be a serious problem at this facility if a V-to-G-based DAAMS method must be used for routine monitoring at the CDC’s 2003 VX WPL.17 Staff at the NECD former production facility have developed a DAAMS method that allows VX to be detected directly (without V-to-G conversion), as long as the sample period is less than about 6 hours. (Poor recoveries of VX were obtained for sample periods greater than 6 hours.) If successful in routine use, this direct DAAMS method for VX is expected to result in improved selectivity and fewer false positives. In anticipation of the implementation of the CDC’s 2003/2004 AELs, the Army has conducted tests of modified DAAMS methods at various sites. For example, at the NECD stockpile facility, a DAAMS configured with an IACEM 980 connected to an Agilent 6890 GC configured with a precolumn and an analytical column, to an Agilent heart-cut system, and to an FPD has been used successfully to determine VX at the CDC’s 2003 WPL and GPL.18 A DAAMS configured with a heart-cut valve is usually more selective than systems configured using only a backflush valve; that is, interferences caused by chemical compounds other than agents are greatly reduced with the heart-cut approach. The heart-cut-based DAAMS/FPD system at the NECD stockpile facility, which relies on V-to-G conversion for the detection of VX, has passed 4-day P&A studies and baseline studies for VX WPL and GPL methods (with a sample period of 12 hours for both methods). In addition to developing DAAMS/FPD-based methods, the staff at the NECD stockpile facility has successfully developed confirmation methods for the CDC’s 2003 VX WPL and its 2003 VX GPL using a DAAMS configured with an FPD and an MSD. Although DAAMS/FPD and DAAMS/FPD/MSD methods for VX have been certified at NECD, staff at the NECD stockpile facility expect more false positives at the CDC’s 2003 VX WPL and GPL than have been observed for VX in the past. Staff at the TOCDF stockpile site (Tooele, Utah) have also developed DAAMS methods to monitor at the CDC’s 2003 VX and GB WPLs. These methods have been submitted to the CDC for review and approval. In addition to the recent work at the Tooele and NECD stockpile sites, the Army’s CMA has undertaken a study with the goal of modifying DAAMS methods to meet the requirements of monitoring at the CDC’s 2003 WPLs and GPLs for GB and VX (FOCIS, 2003b). The study was recently expanded to include HD. It aimed not only to achieve the sensitivities necessary to detect the agents at the various AELs but also to reduce the potential for interferences at the CDC’s 2003/2004 AELs by improving the selectivity of the DAAMS. This work addressed both FPD-based screening systems used for routine DAAMS monitoring and FPD/MSD-based DAAMS, which are typically used to confirm the detection of agent by other DAAMS or by NRT monitors. Various technologies have been incorporated into the development work on DAAMS methods begun by the CMA about 2 years ago. These technologies and upgrades include the following: More extensive use of heart-cut methods; Upgrades of the backflush technique; Cryogenic cooling to narrow chromatographic peaks to improve chromatographic resolution (using a tank of compressed, liquid carbon dioxide connected to a cryotrap surrounding a short length of the GC column); Use of a convex lens to increase the signal-to-noise ratio for the FPD; and Faster sample flow, made possible by using 8-mm-diameter sampling tubes in place of the 6-mm-diameter tubes that had been used at most sites. The new DAAMs methods developed in the effort noted above have successfully passed laboratory P&A and baseline 17   Tom Hoff, NECD Project Manager, and William Rogers, TVA Quality Assurance Officer, Briefing to the Department of Health and Human Services, March 11, 2004; William Rogers, TVA Quality Assurance Officer, Briefing to the committee, August 3, 2004. 18   The key component of the heart-cut system is a Dean switch, which allows the effluent of the precolumn to be directed into the analytical column only during a short period, from the time just before the agent of interest begins to exit the precolumn to the time just after the agent has exited the precolumn. At all other times, before and after this agent window, the carrier-gas effluent from the precolumn vents through a restrictor column to the atmosphere (or to a second detector installed to monitor the effluent from the precolumn). The liquid phase of the analytical column generally differs greatly in polarity from the liquid phase of the precolumn.

OCR for page 43
Impact of Revised Airborne Exposure Limits on Non-Stockpile Chemical Materiel Program Activities studies. They are now undergoing P&A and baseline studies at the Umatilla stockpile site (UMCDF) (FOCIS, 2004). Finding 4-2: Work is currently under way or has been completed at several stockpile and non-stockpile sites to modify DAAMS methods to meet the requirements of monitoring at the CDC’s 2003/2004 AELs. The DAAMS methods and equipment configurations used to enable monitoring at the CDC’s 2003/2004 AELs vary widely from site to site, however. Also, the methods that are being developed at those sites appear to be focusing on achieving adequate sensitivities to monitor at the new AELs. Although it is likely that agents can be detected at the CDC’s 2003/2004 WPLs (and GPLs) using DAAMS, it is also likely that interference will be a bigger problem than it was for DAAMS in the past. Recommendation 4-2: The Army should immediately convene a workshop of non-stockpile and stockpile personnel working on DAAMS methods from each site to allow them to exchange written procedures, test data, and other information regarding the CDC’s 2003/2004 AELs. This workshop should also offer presentations by knowledgeable technical personnel involved in the recent CMA-sponsored effort to develop more selective DAAMS methods. Also, the Army should continue to work on improving the selectivity of DAAMS methods, especially FPD-based methods, to further reduce the number of false positive alarms. A/DAM A/DAM has been used routinely at a few sites to monitor for GB, HD, and VX at CDC’s 1988 TWA AELs and for GB and VX at the IDLH levels. Thus, the use of A/DAM to monitor for GB, HD, and VX at the CDC’s 2003/2004 IDLH levels and STELs (equal to the CDC’s 1988 TWAs) should be straightforward. However, the use of the A/DAM system as an NRT monitor has not been widespread because it is more expensive, larger, heavier, and more complex than MINICAMS. Because no IDLH level had been defined for HD before May 2004, a method must be developed and certified for the determination of the IDLH levels of this agent by the A/DAM system. This task should be straightforward using an A/DAM configured with an external loop sampler. Testing of the A/DAM for monitoring at the STEL for VX (equal to CDC’s 1988 TWA level) at the NECD former production facility is planned.19 The A/DAM will determine VX directly (instead of as the G-analog). It is anticipated that the A/DAM system, configured for the direct determination of VX, will not experience the relatively high rate of false positives that has sometimes plagued MINICAMS at the NECD former production facility, which was thought to be caused by phosphorus-containing compounds (TRO or related compounds, O,S-DMP, etc.). As it did for MINICAMS, the Army recently completed laboratory studies of the performance of A/DAM at the CDC’s 2003/2004 WPLs for GB, HD, and VX (FOCIS, 2003a). The A/DAM (6852-based) system passed 4-day P&A studies and baseline studies for each agent during laboratory tests, with VX determined using the V-to-G conversion. Finally, A/DAM systems recently underwent both 4-day P&A and baseline studies at the Anniston (ANCDF) stockpile facility. A report describing the results of these tests is in preparation.20 ALTERNATIVE TECHNOLOGIES FOR MONITORING AT THE 2003/2004 AELS The CDC’s 2003 STEL level for VX, 1 × 10−5 mg/m3, corresponds to a concentration of about one part per trillion by volume. Not only must NRT monitoring systems be capable of detecting VX at this concentration, but NRT systems used prior to 2005 also had to be capable of meeting quality assurance/quality control (QA/QC) requirements for concentrations as low as about 0.50 parts per trillion (equal to 0.50 STEL, the lowest level—other than the blank—used during P&A studies). In other words, automated detection systems used in the non-stockpile program are actually automated analytical instrument systems. The CDC’s 2003 WPL for VX is 1 × 10−6 mg/m3, or about 0.1 parts per trillion. The DAAMS method used to monitor at this concentration must also be capable of meeting stringent QA/QC requirements, including those of P&A studies, where the lowest test concentration is about 0.05 parts per trillion. In addition to measuring VX at concentrations of less than one part per trillion and meeting QA/QC requirements, automated and manual methods must be amenable to reliable, long-term operation by personnel with minimal technical skills. For VX concentrations as low as 0.05 parts per trillion, the only technologies mature enough to be considered for use in the non-stockpile program in the next 3 years are sorbent-based sampling, temperature-programmed capillary gas chromatography, and detection using the FPD, PFPD, XSD, MSD, or FPD/MSD. Such technology has also proved capable of meeting the requirement for reliable long-term operation by relatively nontechnical personnel. Given these discriminators, the near-term choice in NRT systems is the MINICAMS (configured with an FPD, a PFPD, or an XSD) or a 6852-based A/DAM system (configured with an FPD).21 19   William Brankowitz, PMNSCM, Information provided to the committee, May 6, 2004. 20   Personal communication between Rob O’Neil, FOCIS Associates, Inc., and Gary Sides, committee member, August 24, 2004. 21   It should be possible to configure the A/DAM system with a PFPD or an XSD, but—to the best of the committee’s knowledge—these configurations have not been tested. It should also be possible to configure the A/DAM system with an MSD or with an FPD and an MSD, but these configurations are not practical for routine NRT monitoring, primarily because of cost and complexity.

OCR for page 43
Impact of Revised Airborne Exposure Limits on Non-Stockpile Chemical Materiel Program Activities Some of the advantages of the MINICAMS compared with the A/DAM are the availability of more selective detectors for routine monitoring (PFPD and XSD) and MINICAMS’s greater simplicity, lower cost, lighter weight, and smaller size. Some of the advantages of the A/DAM system compared with the MINICAMS are dual columns/dual detectors to enhance selectivity, a more flexible analytical system, heart-cut capability (using the Dean switch to enhance selectivity), and FPD enhancements for a greater signal-to-noise ratio. MINICAMS, compared with the A/DAM, is severely limited by its relatively rigid analytical system. For example, it is not possible to use backflush, heart-cut, or dual columns/detectors with MINICAMS—all techniques that would improve selectivity. Poor chromatographic resolution is the main disadvantage of MINICAMS, and this makes it more susceptible to false positives than the A/DAM system. Repackaging of MINICAMS or A/DAM technology to eliminate their “faults” would require at least 3 years to complete engineering, laboratory testing, field testing, production, and integration of the changes into the non-stockpile program. Although DAAMS configured with FPDs appear adequate for detecting VX and HD at the CDC’s 2003/2004 AELs, they are not expected to be sufficiently selective to pass published P&A requirements, to pass initial and continuing baseline certification requirements, and to achieve the required statistical response rate at reportable limits when they are used to monitor at the 2003/2004 WPLs for these agents.22 Thus, although sensitive enough to detect the WPLs and pass P&A requirements for the agents alone, the presence of chemical interferences (and associated false positives) will make it difficult to pass P&A certification requirements. The main problem anticipated by the committee is the high false positives rates for primary DAAMS tubes analyzed using a GC configured with an FPD and the resulting need to analyze backup DAAMS tubes by alternative methods (for example, a separate GC configured with a column of significantly different polarity or configured with a different detector). Work is currently under way to improve the selectivity of DAAMS systems used for historical monitoring and systems used to confirm or deny alarms reported by NRT monitors. To fully implement the DAAMS modifications now being developed to improve selectivity will take 2 or 3 years from the time that field tests of these modifications are completed. Finding 4-3: The committee observes that although DAAMS methods used to monitor at the 1988 AELs may currently be capable of monitoring at the 2003/2004 WPLs for VX and HD, improvements in the sensitivity and selectivity of the DAAMS would make it easier to pass published precision and accuracy certification requirements, to pass initial and continuing baseline certification requirements, and to achieve the required statistical response rate at reportable limits. Recommendation 4-3: PMNSCM should take advantage of research and development being funded by the stockpile program to develop more selective and more sensitive DAAMS methods for monitoring VX and HD at the 2003/2004 WPLs. During the past year, several technical meetings have included or have been dedicated to the detection of chemical agents. These meetings covered the following technologies: Conventional Fourier-transform ion mobility spectrometry (IMS), differential IMS, and dual-cell IMS; Cylindrical ion-trap mass spectrometry and IMS/time-of-flight mass spectrometry; Surface-enhanced Raman microwave spectroscopy, terahertz and millimeter-wave microwave spectroscopy, and Fourier-transform microwave spectroscopy; Cavitands and liquid crystals; Ceramic-metallic (cermet) solid state sensors, surface-acoustic-wave (SAW) solid state sensors, and metal-insulator/metal-ensemble (MIME) solid state sensors; Fluorescent indicating chromophores (fluorescent reporters); and Enzyme-based methods. The developers of most of these technologies are focusing on homeland security applications, which require the detection of agents at concentrations like the IDLH AELs or, perhaps, an order of magnitude less. Thus, the concentrations of interest to most developers are several orders of magnitude greater than the STEL and WPL AELs of interest in the non-stockpile program. Although a few of the technologies presented at the meetings listed might be able to detect agents at concentrations of parts per trillion, this capability has not been demonstrated. Also, most of these technologies are in the research and development phase, and monitoring systems suitable for use in the non-stockpile program generally are at least 3-5 years from being commercially available. Finally, many of the researchers involved in these technologies are focusing on automated point-detection systems that are simply meant to sound an alarm when agent is detected as opposed to the automated analytical instruments needed by the non-stockpile program, which will accurately determine and report the concentrations of agents, meet stringent QA/QC requirements, and activate alarms. 22   The Army defines reportable limit as “a predetermined value for historical method, that when equaled or exceeded will be reported as chemical materiel that may have exceeded the monitoring level” (U.S. Army, 2004f, p. B-9). For a Class I historical method (that is, a manual method such as DAAMS), the reportable limit must be set so that the statistical response rate at the reportable limit is greater than or equal to 95 percent—that is, the probability, expressed as a percentage, that a 1.0-Z QP challenge will generate a response greater than or equal to the reportable limit must be equal to or greater than 95.

OCR for page 43
Impact of Revised Airborne Exposure Limits on Non-Stockpile Chemical Materiel Program Activities Although the detection limits and stringent QA/QC requirements for NRT monitoring systems and for manual DAAMS methods seem to preclude modifying other technologies for use in the non-stockpile program in the near term, rapid advances in the miniaturization of mass spectrometers may allow them to meet non-stockpile program requirements for NRT monitors and manual historical/confirmation methods at a reasonable cost within about 5 years. It should be emphasized that the NRT monitoring systems used in the non-stockpile program to monitor at the 1988/1997 AELs generally have adequate sensitivity but need much better selectivity. Mass spectrometers are currently the holy grail in the verification of agent alarms at stockpile and non-stockpile facilities. One existing technology that may enable the development of small, affordable mass spectrometers is an instrument based on a simple cylindrical ion trap that is capable of the direct detection of toluene in air at a concentration as low as 17 parts per billion by volume (Griffin Analytical Technologies, West Lafayette, Indiana). Membrane-based concentration methods currently enable such mass spectrometers to detect methyl salicylate at about 300 parts per trillion. The use of a porous-polymer-based trap on the front end of the mass spectrometer should enable the detection of toluene at concentrations approaching 20-30 parts per trillion. Further improvements might allow cylindrical ion traps to detect less than one part per billion. Finding 4-4: The CDC’s 2003 STEL for VX corresponds to a concentration of less than one part per trillion by volume. DAAMS systems utilizing mass selective detectors with chemical ionization sources are currently capable of detecting VX at these levels. Other technologies, especially miniature mass spectrometers, might be able to meet the requirements of the non-stockpile program for parts-pertrillion sensitivity and improved selectivity at a reasonable cost within a 5 years. Recommendation 4-4: PMNSCM should conduct a paper study of the state of miniature mass spectrometer technologies and, if warranted, support the development of near-real-time (NRT) systems based on the best available technology. The paper study should be done by technical personnel with extensive hands-on experience in air monitoring at the 1988 AELs, who—along with personnel involved in the manufacture of miniature mass spectrometers—should also conduct the effort to develop or modify mass spectrometer systems for NRT monitoring. ALARM LEVELS FOR NEAR-REAL-TIME MONITORS NRT monitors have been used in the non-stockpile program for many years to detect agent at true concentrations that equal or exceed 1.00 TWA and then to sound an alarm that warns workers to take action in a timely manner. The CDC’s 1988 TWA airborne exposure limit for each agent is intended to ensure the absence of long-term health effects, even for workers not wearing respiratory protection, for exposures up to 8 hours per day (Federal Register, 1988). The same numerical values defined by the CDC in 1988 as TWA levels were recently renamed “short-term exposure limits” (STELs) (Federal Register, 2003a, 2004). It is desirable that an alarm sound each and every time the true concentration of agent in an area being monitored equals or exceeds 1.00 STEL and never when the true concentration is less than 1.00 STEL.23 Because of measurement errors, however, the concentration of agent reported by an NRT system is typically not the same as the true concentration of the agent in the area being monitored. In fact, the most stringent certification requirement for NRT systems published by the Army requires an accuracy of only ±25 percent and a precision of ±25 percent with 95 percent confidence for challenges at 1.00 STEL (U.S. Army, 2004f). As an example, a certified NRT system may report an agent concentration of 0.75 STEL even though the true agent concentration in the area being monitored is 1.00 STEL or greater. Thus, it is clearly not possible to set the alarm level for an NRT monitor to sound an alarm only when the true concentration of agent equals or exceeds 1.00 STEL and to avoid sounding an alarm whenever the true concentration is less than 1.00 STEL. Currently, in the non-stockpile program, the statistical uncertainty of NRT systems is usually accounted for by setting the alarm level at 0.70 STEL. Thus, an alarm is sounded and required actions are taken any time the concentration of agent reported by an NRT monitor exceeds 70 percent of the STEL. Past experience with NRT systems and the statistical analysis of data from numerous certification studies during the past 20 years have shown that any time the true concentration of agent in the area being monitored equals or exceeds 1.00 STEL, NRT systems in use by the Army (that is, ACAMS and MINICAMS) typically have at least an 80 percent probability of reporting a concentration ≥0.70 STEL and sounding an alarm. The Army’s stockpile program has long used (and in some states the non-stockpile program may also be required to use) alarm levels as low as 0.20 STEL. At this alarm level, NRT systems in use by the Army typically have at least a 99 percent probability of reporting a detected concentration greater than or equal to 0.20 STEL and of sounding an alarm any time the true concentration of agent in the area being monitored equals or exceeds 1.00 STEL. 23   The discussion in this section focuses on alarm levels for NRT monitoring systems used for monitoring at the 2003/2004 STELs. It should be noted, however, that an NRT monitoring system may report an air agent concentration above the 2003/2004 WPL but below the STEL alarm level. For this reason, a STEL concentration reading ≥0.30 STEL for GB, ≥0.10 STEL for VX, or, ≥0.13 STEL for HD may indicate the presence of agent at a concentration ≥1.00 WPL and may indicate the need to use DAAMS to monitor the area at the WPL level.

OCR for page 43
Impact of Revised Airborne Exposure Limits on Non-Stockpile Chemical Materiel Program Activities Note that the target concentration of concern when using an NRT monitor is 1.00 STEL, no matter the set point for the alarm level. Also, the accuracy of an NRT monitor, its minimum detection limits, its certification requirements, and its operation are not affected in any way by the choice of alarm level. That is, NRT monitors used in the non-stockpile program are certified to demonstrate that they meet QA/QC requirements for 1.00 STEL; are typically calibrated only at 1.00 STEL; and are challenged at least once a day at 1.00 STEL to make certain that they respond properly—again, no matter whether the alarm level is set at 0.20 STEL or 0.70 STEL. The alarm level, in effect, simply defines the probability that an alarm will be sounded if the true concentration of agent in an area being monitored is 1.00 STEL or greater—that is, it is an indication of how certain the Army is that an agent excursion above 1.00 STEL will be detected and an alarm sounded. The CDC’s 2003 Federal Register announcement regarding AELs for G and V agents states as follows: 24 In implementing the WPLs, STELs, and GPLs, specific reduction factors for statistical assurance of action at the exposure limits are not needed because of safety factors already built into the derivation of the exposure limit. This recommendation assumes that the sampling and analytical methods are measuring within ±25 percent of the true concentration 95 percent of the time. If this criterion is not met, an alarm level or action level below the exposure limit may be required. (Federal Register, 2003a, p. 58349) Furthermore, written clarification received by the committee from the CDC on August 3, 2004, makes it clear that, for GB and VX, the CDC recommends that the alarm level for an NRT system be set at 1.00 AEL as long as “the sampling and analytical methods are measuring within ±25 percent of the true concentration 95 percent of the time.” If this condition cannot be met, then the CDC says that an alarm level below the AEL may be required, but the CDC did not provide any guidance on how to determine that level. With regard to HD, the CDC announcement (Federal Register, 2004, p. 24167) reads as follows: “Although the CDC does not specifically recommend additional reduction factors for statistical assurance of action at the exposure limit, exposures to sulfur mustard should be minimized given the uncertainties in risk assessment, particularly as related to characterizing carcinogenic potency.” There, the CDC seems to say it is acceptable to set the alarm level at 1.00 AEL, but at the same time the CDC requires procedures to minimize exposures to mustard—two directives that seem somewhat in conflict. The Army apparently used the CDC recommendations for guidance and states in the most recent Programmatic Laboratory and Monitoring Quality Assurance Program (U.S. Army, 2004f, Table 10-3) that the alarm levels for GB, VX, and HD can be set one of two ways: If the first-challenge pass rate is ≥95 percent for the NRT monitor, the alarm level may be set at 1.00 AEL. If a first-challenge pass rate of ≥95 percent cannot be achieved, the alarm level must be set to a value that results in a statistical response rate of ≥95 percent, and a first-challenge pass rate of ≥75 percent must be maintained. In other words, for a given NRT monitor, according to the CDC, the alarm level may be set at 1.00 AEL if at least 95 percent of the first 1.00-AEL daily quality control (QC) challenges of the monitor over a defined period of time result in concentration readings between 0.75 and 1.25 AEL, which corresponds to ±25 percent accuracy. The requirement for a first-challenge pass rate of ≥95 percent must be met during an initial 28-day baseline study and during successive measured operational intervals (weekly, monthly, etc.), which were not defined in the Programmatic Laboratory and Monitoring Quality Assurance Program (U.S. Army, 2004f). If a first-challenge pass rate of ≥95 percent cannot be achieved during the initial baseline study or during continuing operations for a given NRT monitor, then (1) the alarm level for the monitor must be set so that there is a ≥95 percent probability (statistical response rate) that a 1.00-AEL challenge results in an alarm and (2) a first-challenge pass rate of ≥75 percent must be maintained. Based on past performance data for NRT monitoring systems, if a first-challenge pass rate of ≥95 percent cannot be achieved, it is likely that the alarm level will have to be set to a value well below 0.70 AEL to achieve a statistical response rate of ≥95 percent. It is important to understand that the operation of NRT systems is not affected by the alarm level selected. The probability of sounding an alarm in response to a true agent concentration at or above 1.00 AEL, however, is clearly affected by the choice of the alarm level. As an example, for an unbiased, normal distribution (that is, a bell-shaped distribution with the maximum at 1.00 AEL) and with the alarm level set at 1.00 AEL, an alarm would be sounded only 50 percent of the time that the true agent concentration in the area being monitored is at or just above 1.00 AEL. In reality, distributions of agent concentration are not perfectly unbiased or normal. A series of 1.00-AEL challenges made 24   The Army defines an action level as a predetermined value, usually for an NRT method, that, when equaled or exceeded, indicates the need to conduct a series of required actions in response to the apparent detection of agent. An action level is typically less than the alarm level for an NRT monitor. Actions taken when the action level is exceeded (but the alarm level is not exceeded) may include checking to ensure that the NRT monitor is functioning properly, locating and correcting a leak before the concentration of agent at the location being sampled exceeds the alarm level, etc. (Personal communication between Robert Durgin, Chief, Monitoring Team, Office of the Program Manager for CMA; Jeff Kiley, Monitoring Office, Risk Management Directorate, CMA; and Gary Sides, committee member, November 30, 2004)

OCR for page 43
Impact of Revised Airborne Exposure Limits on Non-Stockpile Chemical Materiel Program Activities just after calibration and in subsequent days would often result in all reported agent concentration readings being less than 1.00 AEL. In this case, no alarms would be sounded in response to 1.00-AEL QC challenges conducted over several days. The instrument would then appear to be malfunctioning—that is, challenge after challenge at a true concentration of 1.00 AEL would result in no alarms. Because of variations in the distribution of agent concentration readings, if the alarm levels are set to 1.00 AEL, the percentage of 1.00-AEL challenges that result in an alarm will vary widely from instrument to instrument and from day to day. This is illustrated by the 1.00-TWA challenge data shown in Table 4-1, which were generated during 4 weeks of operation of two different MINICAMS at the Center for Domestic Preparedness’ COBRA Training Facility, Anniston, Alabama. During this period, each MINICAMS was calibrated about once a week and used to monitor for GB and VX continuously at the CDC’s 1988 TWA levels for these agents (except when 1.00-TWA test challenges were made). For an alarm level of 1.00 TWA, the 4 weeks of 1.00-TWA challenges for VX would have resulted in alarms 24 percent of the time for Instrument 1 and 28 percent of the time for Instrument 2. For GB, Instrument 1 would have alarmed 9 percent of the time and Instrument 2 would have alarmed 16 percent of the time. On this basis, it appears that the NRT monitors performed poorly. If the alarm level had been set to 0.70 TWA, however, the 1.00-TWA challenges would have resulted in alarms 100 percent of the time for both agents for both instruments, which would correctly reflect the excellent performance of the MINICAMS during the 4-week test period. The only possible benefit to be gained by raising the alarm level from 0.70 to 1.00 AEL for the non-stockpile program is a reduction in the rate of false positives, which can be achieved only at the expense of increasing the rate of false negatives (that is, failure to sound an alarm even though the true agent concentration equals or exceeds 1.00 AEL). As an example of the impact of raising the alarm level, the TABLE 4-1 TWA Concentrations Reported by Two Different MINICAMS for 1.00-TWA Challenges Made During 4 Weeks of Operation (August 2004) VX Challenge Data − Instrument 1 1.03 0.87 0.96 0.97 0.90 0.91 0.82 0.78 0.88 0.83 1.02 0.94 0.87 0.84 0.77 0.71 0.82 0.97 1.10 1.00 0.95 0.91 0.88 0.83 0.83 1.09 1.05 0.97 0.96 0.87 1.02 0.91 1.01 0.92     24 percent of challenges result in alarm with alarm level at 1.00 TWA 100 percent of challenges result in alarm with alarm level at 0.70 TWA GB Challenge Data − Instrument 1 0.99 0.84 0.93 0.89 0.86 0.85 0.78 0.73 0.80 0.75 0.98 0.90 0.85 0.84 0.76 0.76 0.85 0.93 1.00 0.94 0.90 0.85 0.83 0.81 0.80 1.05 1.00 0.93 0.92 0.86 0.96 0.90 0.97 0.88     9 percent of challenges result in alarm with alarm level at 1.00 TWA 100 percent of challenges result in alarm with alarm level at 0.70 TWA VX Challenge Data − Instrument 2 0.96 1.00 1.06 1.02 1.05 1.03 1.04 1.03 0.90 1.00 0.92 0.97 0.99 0.93 0.90 0.89 0.98 1.02 0.95 0.97 0.98 0.83 0.97 0.98 0.96 0.96 0.94 0.86 0.98 0.93 0.98 0.94         28 percent of challenges result in alarm with alarm level at 1.00 TWA 100 percent of challenges result in alarm with alarm level at 0.70 TWA GB Challenge Data − Instrument 2 0.94 0.97 1.01 1.00 1.01 1.01 1.02 0.94 0.94 0.94 0.92 0.93 0.91 0.88 0.86 0.87 0.93 0.97 0.95 0.92 0.90 0.78 0.93 0.95 0.94 0.94 0.93 0.87 0.93 0.90 0.98 0.94         16 percent of challenges result in alarm with alarm level at 1.00 TWA 100 percent of challenges result in alarm with alarm level at 0.70 TWA SOURCE: Information provided to the committee by the Department of Homeland Security, Center for Domestic Preparedness, COBRA Training Facility, Anniston, Alabama, October 2004.

OCR for page 43
Impact of Revised Airborne Exposure Limits on Non-Stockpile Chemical Materiel Program Activities MINICAMS TWA concentration data for Building 143 at the NECD former production facility for the period March 2002 through December 2003 showed that agent concentration exceeded the alarm level of 0.70 TWA during only 288 of 18,675 total cycles.25 It should be noted that the 288 cycles during which the NRT monitors alarmed represented 80 different apparent chemical events (periods of time), and that only one of these events was confirmed as caused by VX. If the alarm level had been set to 1.00 TWA during this period, 209 cycles out of 18,675 total cycles would have resulted in alarms. This seems to be a relatively minor reduction in the rate of false positives (false alarms), achieved by increasing the probability of a false negative for a true agent concentration of 1.00 TWA from about 20 percent to about 50 percent (for an unbiased, normal distribution). Based on the uncertainty in the toxicity and health effects data used by the CDC to establish the 2003/2004 AELs, no significant additional health risk would be created by increasing the alarm level from 0.70 AEL, currently used at most sites in the non-stockpile program, to 1.00 AEL. However, a number of potentially serious problems involving worker perception, public perception, and logistics or tracking could result from using an alarm level of 1.00 AEL in place of the previous value of 0.70 AEL. These problems include the following: The Army has renamed the CDC’s 1988 TWA, adopting the more traditional occupational safety terminology “short-term exposure limit (STEL),” kept the same numerical value, and changed the allowed exposure time from 8 hours to 15 minutes. With an alarm level of 0.70 TWA, there was at least an 80 percent probability of an alarm sounding when the true agent concentration in an area being monitored was at or just above 1.00 TWA. Now, Army documents (for example, U.S. Army, 2004f) allow alarm levels to be set at 1.00 STEL, which will cause NRT monitors to sound an alarm only 50 percent of the time at 1.00 STEL (for an unbiased, normal distribution). It makes little sense to reduce the exposure time, which implies that these concentrations are now greater hazards than previously thought, while changing the alarm setting from 0.70 to 1.00 STEL, which reduces the probability than an alarm will sound. The Army has specified actions that must be taken in response to an excursion of agent above a given AEL. The actions that must be taken were presumably based on the primary intent of the CDC’s recommendations—to define concentration boundaries above which workers needed added respiratory protection. With the alarm level set at 1.00 AEL, a given NRT system will sound an alarm only 50 percent of the time when the true concentration of agent in the area being monitored is at or just above 1.00 AEL (for an unbiased, normal distribution of reported concentrations), whereas set at 0.70 AEL, an alarm will sound at least 80 percent of the time. Thus, the probability of false negatives is much greater with the alarm level set at 1.00 AEL rather than 0.70 AEL. Because the distributions of concentrations reported by MINICAMS are typically biased and are not normal over an operating period of several days, the percentage of 1.00-AEL challenges that result in an alarm will vary widely from instrument to instrument, from day to day, and from week to week. It is entirely possible that if the alarm level is set to 1.00 AEL, the percentage of 1.00-AEL QC challenges that result in alarms will vary within the range 0 to 100 percent for a given group of instruments monitoring the same agent in the same facility. Thus, it will appear that some instruments work properly and that others do not. Workers have calibrated NRT monitoring systems at 1.00 TWA (or 1.00-IDLH limit) and then conducted daily challenges at 1.00 TWA (or 1.00-IDLH limit). Because the alarm levels used were 0.70 TWA and the instruments used for GB, VX, and HD were required to maintain an accuracy of ±25 percent with 95 percent confidence, almost every time an operator conducted a daily 1.00-TWA challenge, an alarm sounded. With the alarm level set at 1.00 STEL, the NRT monitoring system will alarm, at best (for an unbiased, normal distribution), 50 percent of the time—even if functioning perfectly—and often much less frequently (for other distributions) in response to 1.00-STEL challenges. This will be perceived by the worker as a reduction in worker safety. The two different ways that the Army allows the alarm level to be set for the CDC’s 2003/2004 AELs will be confusing at best. It is likely that some instruments will be able to achieve ±25 percent accuracy ≥95 percent of the time for 1.00-AEL challenges; the alarm level for these instruments will be set to 1.00 AEL. Other instruments will not be able to meet this requirement, and their alarm levels will have to be set so that ≥95 percent of 1.00-AEL challenges will result in an alarm. This may result in some instruments at a single site being set at an alarm level of 1.00 AEL and the remainder of the instruments being set at a level less than 0.70 AEL (for example, 0.50 AEL). In addition, because the technique required to determine the proper alarm level must be based on the value of the first-challenge pass rate achieved for each continuing baseline test period, the technique used may be different for a given monitor from week to week or month to month. Tracking alarm levels and challenge data that 25   Tom Hoff, NECD Project Manager, and William Rogers, TVA Quality Assurance Officer, Briefing to the Department of Health and Human Services, March 11, 2004; William Rogers, TVA Quality Assurance Officer, Briefing to the committee, August 3, 2004.

OCR for page 43
Impact of Revised Airborne Exposure Limits on Non-Stockpile Chemical Materiel Program Activities must be analyzed in two different ways to justify specific alarm levels from instrument to instrument and from baseline period to baseline period will be confusing to operators of NRT monitors. By stating that exposures to sulfur mustard, unlike to GB and VX, “should be minimized given the uncertainties in risk assessment, particularly as related to characterizing carcinogenic potency” (Federal Register, 2004, p. 24167), CDC would seem to imply that the alarm level should be set at the lowest practical value to obtain the maximum practical probability of sounding an alarm when the true concentration of HD in the area being monitored exceeds 1.00 AEL. Raising the alarm level from 0.70 to 1.00 AEL, thereby reducing the probability of sounding an alarm from 80 percent to 50 percent (for an unbiased, normal distribution), seems at odds with the CDC’s statements on minimizing exposure to HD. GB, VX, and HD have been monitored successfully by NRT monitoring systems at concentrations equal to the CDC’s 1988 TWA values and at the IDLH levels for more than 20 years—with alarm levels set at 0.70 AEL or less. The alarm levels used in the past ensured at least an 80 percent probability of sounding an alarm when the true concentration of agent in the area being monitored exceeded 1.00 AEL. The CDC’s 2003/2004 IDLH values are between one-half and one-sixth of the 1997 IDLH values, but they are still several orders of magnitude above the detection limits for NRT monitoring systems. There is no logical justification for or any significant benefit to raising the alarm level from 0.70 IDLH to 1.00 IDLH and thereby reducing the probability of sounding an alarm at a true concentration of 1.00 IDLH from 80 percent to 50 percent. The TWA concentrations are new in name only: STEL. The numerical values of the CDC’s 2003/2004 STELs are identical to the 1988 TWA limits. Again, there is no good reason for raising the alarm level when the numerical concentrations being monitored have not changed and given that the Army has successfully monitored at these levels for more than 20 years with the alarm level set at 0.70 TWA or less. The Army has relied on a statistical approach to setting alarm levels for more than 20 years. It has briefed the public many times during this period and repeatedly assured the public and workers that if an agent excursion above a given AEL value occurs, there is an 80 percent or greater probability of detecting the agent and sounding an alarm. If the Army’s statistical approach to setting alarm levels is abandoned and alarm levels are allowed to be set to 1.00 AEL, the Army will have to admit that it has adopted a policy that results in a 50 percent or greater chance that an agent excursion could occur without an alarm being sounded to alert workers or the general public. Finding 4-5: The Army’s plan to allow alarm levels to be set at 1.00 of the CDC-recommended AELs (especially for the CDC’s 2003/2004 STEL and IDLH levels) has the potential to be perceived by workers and the general public as significantly reducing worker safety, for four reasons: (1) the alarm levels will be higher than the alarm levels used historically, (2) the percentage of QC challenges that yield alarms will vary widely from instrument to instrument and from day to day and week to week, (3) the probability will increase that a worker might be exposed to unacceptable levels of the carcinogen HD, and (4) there is a greater likelihood that the Army will not respond properly or in a timely manner to the presence of agents at true concentrations above the AELs. The rationale for such a large change in the alarm level will be difficult to explain to regulators, auditors, judges, and the general public, especially because concentrations have not changed and remain orders of magnitude above detection limits for NRT monitors and because the Army has a sterling record monitoring at these levels during the past 20 years with the alarm level set to 0.70 AEL or less. The only perceived benefit of raising the alarm level to 1.00 AEL is a reduction in false positives, but this benefit is gained at the expense of a higher probability of false negatives, which is unacceptable. Recommendation 4-5: For near-real-time monitoring, the non-stockpile program should meet the 2003/2004 AELs promulgated by the CDC using an approach that establishes a sufficiently high confidence level (that is, a high statistical response rate) for the detection of excursions above 1.00 AEL. The alarm levels for near-real-time (NRT) monitors should then be set to achieve the required confidence. Finding 4-6: The purpose of adjusting alarm levels is to ensure a sufficiently high degree of confidence that an NRT monitoring system will sound an alarm any time that the true concentration of agent in the area being monitored exceeds 1.00 AEL. The non-stockpile program sometimes uses alarm levels that are greater than those used by the stockpile program, making it seem that the non-stockpile program is less likely to detect agent excursions above 1.00 AEL than the stockpile program. Recommendation 4-6: The non-stockpile program should justify sometimes using alarm levels for near-real-time monitoring systems that are different from those used by the stockpile program. The issues raised in this section of Chapter 4 regarding alarm levels for NRT monitoring systems also apply to reportable limits for manual monitoring methods such as DAAMS. However, the committee chose to limit the discussion here to NRT monitoring systems for the following reasons:

OCR for page 43
Impact of Revised Airborne Exposure Limits on Non-Stockpile Chemical Materiel Program Activities Typically, more than 90 percent of the agent concentration reports at non-stockpile sites are obtained using NRT monitoring systems. NRT monitoring systems provide an immediate warning to workers to allow them to take proper actions to protect themselves and to get the situation under control, and the setting of an alarm level is the key determinant of the probability of detecting and reporting true agent concentrations above 1.00 AEL. All DAAMS results are essentially historical; the event or situation that caused the detection of agent by DAAMS has likely been detected by other means and corrected by the time the DAAMS sample is analyzed.