Audit of EPA Protocol Gas Suppliers

Protocol gas standards are employed in measurements made for compliance purposes; one of the major uses is for measurements of regulated gaseous species from stationary source emitters, i.e., stack gases. Many of these stationary sources are regulated through the Cap and Trade program; therefore, accurate measurements are critical due to the economic and health consequences. EPA protocol gas mixture production is governed by Document EPA/600/R-12/531. Subsection 2.1.13 states that the EPA will periodically conduct audits of these gas mixtures, with a selection of at least two samples from every US manufacturing site. The samples are to consist of NO (25 µmol/mol to 1000 µmol/mol), SO₂ (25 µmol/mol to 1000 µmol/mol), and CO₂ (5 % mol/mol to 20 % mol/mol) in balance N₂.

The objective of the audit is to determine whether the concentrations of the components of the gas mixtures are within the limits stated on the certificate provided by the vendor. The resulting data can be used to facilitate accurate reporting of continuous emissions monitoring (CEM) and stack measurements, to provide an independent quality check of the manufacturers, and to connect end users with reputable vendors. NIST has served as the independent analytical laboratory for audits since 2006. NIST has a stake in this process, as the gases and the measurements made using them are claimed to be NIST traceable.

In the 2006 PGVP audit, only 11 samples were received, and only 3 out of 17 vendors (6 of 37 manufacturing sites) were represented. In 2008, 87 samples were received and all 11 vendors (18 manufacturing sites) were represented. (The number of sites decreased due to industry consolidation.) In 2010, 57 samples were tested from 10 vendors (17 manufacturing sites).  In 2013, 108 samples were tested from 11 vendors (22 manufacturing sites).  In 2015, 116 samples were tested from 14 vendors (26 manufacturing sites).  And in 2018, 66 samples were tested from 11 vendors (22 manufacturing sites).

The audit gas mixtures have three components: NO, SO₂ and CO₂, The presence of % mol/mol levels of CO₂ changes the viscosity and molar mass of the audit gas, compared to single-component standards. It is important to ensure that such a change does not cause an unexpected bias in the analytical results. This bias is dependent on the analytical technique chosen and could also be subject to interference from one component to another.

NIST developed a variety of methods that achieved an analytical uncertainty of ≤0.5 % for NO, SO₂, and CO₂. The expected CO₂ interference on chemiluminescence analysis was accurately modeled, as was the unexpected interference of pressure broadening (due to CO₂) on nondispersive infrared (NDIR) and Fourier-transform infrared (FTIR) measurements.

The figure illustrates the relationship between NIST and the Protocol gas vendors, with the uncertainty expressed as k = 1. If the analytical uncertainty claims of NIST (≤0.5 %) and the gas vendors (≤1.0 %) are valid, and either there is no bias (unexpected or expected) or the bias has been accurately corrected, then the difference between the NIST analysis and the vendor certified concentrations of the audit mixture should be ideally ≤1.0 % relative, and as a worst case no more than 2 % relative (2 % Tag Rule). The audit therefore validates the claims of traceability of the Protocol mixtures to NIST.