5.1 Commercial trace detectors are used by first responders, security screeners, the military, and law enforcement to detect and identify explosive threats and drugs of interest quickly. These trace detectors typically operate by detecting chemical agents in residues and particles sampled from surfaces and can have detection limits for some compounds extending below 1 ng. A trace detector is set to alarm when its response to any target analyte exceeds a programmed threshold level for that analyte. Factory settings of such levels typically balance sensitivity and selectivity assuming standard operating and deployment conditions.

5.2 The LOD for a substance is commonly accepted as the smallest amount of that substance that can be reliably detected in a given type of medium by a specific measurement process (2). The analytical signal from this amount shall be high enough above ambient background variation to give statistical confidence that the signal is real. Methods for determining nominal LOD values are well known but pitfalls exist in specific applications. Vendors of trace detectors often report detection limits for only a single compound without defining the meaning of terms or reference to the method of determination.

Note 2: There are several different “detection limits” that can be determined for analytical procedures. These include the minimum detectable value, the instrument detection limit, the method detection limit, the limit of recognition, the limit of quantitation, and the minimum consistently detectable amount. Even when the same terminology is used, there can be differences in the LOD according to nuances in the definition used, the assumed response model, and the type of noise contributing to the measurement.

5.3 When deployed, the individual performance of a trace detector (for example, realistic LODs) is influenced by: (1) manufacturing differences, history, and maintenance; (2) operating configurations (for example, thermal desorption temperature, analyzer temperature, and type of swab); and (3) environmental conditions (for example, ambient humidity and temperature and chemical background). As a result, realistic LOD values for a trace detector may be poorly estimated by the factory specifications. These fundamental measures of performance are critically important for assessing the ability of an instrument to detect trace levels of particular compounds in a particular setting, so a reliable and accessible method is needed to estimate realistic LOD values, especially in the field.

5.4 Technical Challenges and Pitfalls to the Estimation of LOD Values in Trace Detectors and the Setting of Optimal Alarm Thresholds: 

5.4.1 Scope—The U.S. Department of Justice lists over 230 explosive materials and over 270 controlled drugs having a high potential for abuse.4 There are many technologies used for trace detection, and instrument manufacturers design their systems and balance operating conditions to provide detection capabilities across as many analytes as possible. However, a very limited subset of analytes is normally used to test and verify detector performance. Therefore, default operating conditions and alarm thresholds may not be optimally set to detect reliably certain compounds deemed important in particular scenarios.

5.4.2 Environment—Ambient conditions and chemical background vary with the deployment location, which would influence response sensitivities and LOD values.

5.4.3 Risk Tolerance and Balance—Values of alpha risk (false positive probability of process blanks) and beta risk (false nondetection probability of analytes at the detection limit) should be balanced and set according to security priorities (for example, alert level, probable threat compounds, throughput requirements, human factors, and risk tolerance). The default risk balance in a trace detector may not be adequate for the deployment situation.

5.4.4 Signal Variability (Heteroskedasticity)—The variance in instrument response may not be consistent across analyte mass levels introduced into the trace detector. In ion mobility spectrometry (IMS)-based technologies, the physicochemical mechanisms underlying atmospheric pressure ionization (with a finite number of available reactant ions) and ion mobility separation may be non-uniform across the response regions. Typical methods of LOD estimation usually assume constant variance.

5.4.5 Proprietary Signal Processing—Typical LOD determinations assume Gaussian distributions and use background variation as an important parameter. Unfortunately, alarm decisions in trace detectors are rarely based on raw measurement signals; rather, proprietary algorithms are used to process the raw measurements. This processing may attempt to minimize alpha risk by truncating or dampening background signals, so background signals may be absent or the true distribution in these processed signals may be non-Gaussian, confounding the calculation of an accurate LOD.

5.4.6 Multivariate Considerations—To improve selectivity and decrease alpha risk, alarm decisions in trace detectors may be based on multiple-peak responses rather than a single-peak amplitude measurement. Efforts to recognize and quantify unique ion fragmentation patterns across both the thermal desorption and drift-time domains are being developed for next-generation detectors.

5.4.7 Diversity of Technologies—The wide variety of trace detectors and technologies on the market and those under development challenge general response models for accurate estimation of LOD.

5.4.8 Security—LOD values for explosives in trace detectors may not be openly published because of security and classification issues.

1.1 In harmony with the Joint Committee for Guides in Metrology (JCGM) and detection concepts of the International Union of Pure and Applied Chemistry (IUPAC) (1, 2)2, this test method uses a series of replicated measurements of an analyte at dosage levels giving instrumental responses that bracket the critical value, a truncated normal distribution model, and confidence bounds to establish a standard for estimating practical and statistically robust limits of detection.

Note 1: Other standards are available that evaluate the general performance of detection technologies for various analytes in complex matrices (for example, Practice E2520).

1.2 Here, the limit of detection (LOD90) for a compound is defined to be the lowest mass of that compound deposited on a sampling swab for which there is 90 % confidence that a single measurement in a particular trace detector will have a true detection probability of at least 90 % and a true nondetection probability of at least 90 % when measuring a process blank sample.

1.3 This particular test method was chosen on the basis of reliability, practicability, and comprehensiveness across tested trace detectors, analytes, and deployment conditions. The calculations involved in this test method are published elsewhere (3), and are performed through an interactive web-based calculator available on the National Institute of Standards and Technology (NIST) site: https://www-s.nist.gov/loda.

1.4 Intended Users—Trace detector developers and manufacturers, vendors, testing laboratories, and agencies responsible for public safety and enabling effective deterrents to terrorism.

1.5 While this test method may be applied to any detection technology that produces numerical output, the method is especially applicable to measurement systems influenced by heterogeneous error sources that lead to non-linear and heteroskedastic dose/response relationships and truncated or censored response distributions at low analyte levels. The procedures have been tested using explosive and drug compounds in trace detectors based on ion mobility spectrometry, gas chromatography, and mass spectrometry (4). Compounds are deposited as liquid solutions on swabs and dried before use. Background interferences introduced to the test samples were representative of a variety of conditions expected during deployment, but these conditions were not intended as comprehensive in representing all possible scenarios. The user should be aware of the possibility that untested scenarios may lead to failure in the estimation of a reliable LOD90 value.

1.6 Units—The values stated in SI units are to be regarded as the standard. No other units of measurement are included in this standard.

1.7 This standard does not purport to address all of the safety concerns, if any, associated with its use. It is the responsibility of the user of this standard to establish appropriate safety, health, and environmental practices and determine the applicability of regulatory limitations prior to use. Some specific hazards statements are given in Section 8 on Hazards.

1.8 This international standard was developed in accordance with internationally recognized principles on standardization established in the Decision on Principles for the Development of International Standards, Guides and Recommendations issued by the World Trade Organization Technical Barriers to Trade (TBT) Committee.