Introduction

The general procedure for the air sample collection and analysis of ammonia is described in OSHA Method No. ID-188 (10.1.). The validation of this method examines the use of a glass sampling tube containing 500 mg of carbon bead impregnated with sulfuric acid (CISA). Sampling tubes were made in-house and also obtained from Supelco Inc. (Bellefonte, PA). During the validation these commercial tubes became available and were used for experiments not completed. The in-house and commercially prepared tubes are identical except the tube manufactured by Supelco has a 250-mg backup section. The in-house tubes were prepared using the method of Bishop, Belkin and Gaffney (10.2.) with a minor modification. A dilute sulfuric acid pre-wash of the carbon bead sorbent was performed when preparing the in-house tubes.

This method was evaluated when the OSHA Time Weighted Average (TWA) Permissible Exposure Limit (PEL) for ammonia was 50 ppm. As of this writing, the OSHA Final Rule Limit for exposure to ammonia is now a Short-Term Exposure Limit (STEL) of 35 ppm.

The method has been validated for TWA samples using sampling flow rates of about 0.10 liter per minute (L/min) and sampling times of 3 to 4 h. An evaluation of four types of ammonia detector tubes was carried out simultaneously with the method validation. The results of the detector tube study are reported in OSHA Product Evaluation No. 7 (10.3.). A preliminary test was also conducted on ammonia dosimeter tubes obtained from Wilson Safety Products (Reading, PA). The testing of the dosimeter tubes was discontinued since these tubes could not measure concentrations above 50 ppm ammonia during the experiments.

The validation of OSHA Method No. ID-188 (10.1.) consisted of the following experimental protocol:

1. Analysis of spiked samples.
   
2. Analysis of dynamically generated samples having concentrations of approximately 0.5, 1, and 2 times the TWA PEL. Test atmospheres were generated using 50% RH and 25°C.
   
3. Collection efficiency and breakthrough studies of the CISA sampling tubes.
   
4. Determination of the storage stability of ammonia collected with CISA tubes.
   
5. Determination of any variation in results when sampling at high and low RH.
   
6. Comparison of the ion chromatographic method with the ISE method for the determination of ammonia in workplace atmospheres.
   
7. Determination of the qualitative and quantitative detection limits.
   
8. Collection and analysis of ammonia samples employing Supelco tubes and comparison with tubes produced in-house.

A generation system, shown in Figure 1, was used for simultaneous testing of detector tubes and continual sampling devices. All generations of ammonia test atmospheres, and hence all experiments, with two exceptions, were performed using the apparatus shown in Figure 1. The analysis and detection limit studies did not use test atmospheres; these samples were spiked with solutions of ammonium sulfate.

Ammonia Gas Concentration

Ammonia in nitrogen (certified standard, 0.99% ammonia, Air Products Co., Long Beach, CA) was used as the contaminant source. The concentration of ammonia in the cylinder was verified by the following technique:

1. The undiluted gas stream from the cylinder was sampled with a gas syringe and then injected into 0.1 N sulfuric acid contained in septum-capped vials.
   
2. These samples were analyzed for ammonia with an Orion Model 9512 Ammonia Ion Specific Electrode (ISE) and an Orion Model EA940 IonAnalyzer (10.4.). The results are shown below:

The ammonia aliquot results were compared to the manufacturer's stated concentration value. The average spiked sample recovery was 91.3% of the stated value and the CV was 0.073. The manufacturer's stated value was used for all calculations. The lower recoveries were considered to be due to variability in analysis and not due to a change in the stated cylinder concentration. The ISE method has displayed a significant amount of variability in results and slightly lower than expected recoveries. Internal quality control data and results of ISE samples generated during this evaluation are evidence of this variability and decreased recovery when using the ISE method.

Generation System Components

The ammonia gas was mixed, using a glass mixing chamber, with filtered, tempered air. A flow, temperature and humidity control system (Miller-Nelson Research Inc., Model HCS-301) was used to condition the diluent air for mixing. Moisture and other contaminants were removed from the diluent air by using a charcoal, molecular sieve and Drierite filtering system. A Teflon sampling manifold and connections were attached to the mixing chamber. Diluent air flow was determined using a dry test meter (Singer Co., Model No. DTM 115) before, during, and after each experiment. Ammonia flow rates were controlled by a mass flow controller (Tylan Model FC 260), and were measured before and after each experiment with a soap bubble flow-meter. During each experiment the flow rate was monitored using the readout (LED display) for the mass flow controller.

Sample Collection

Air samples were collected from the manifold using Du Pont Model P125 pumps (Du Pont Co. Wilmington, DE) calibrated at flow rates of about 0.1 L/min for all generation experiments. Specific sampling times are mentioned in the procedures for each experiment.

Sample Analysis

Samples, blanks and prepared standards were analyzed for ammonium ion by ion chromatography (10.1.). Analyses and data reduction were performed using a Dionex Model 2010i Ion Chromatograph interfaced to a Hewlett-Packard 3357 Laboratory Automation System. An IBM AT Personal Computer with AutoIon 400 software was used during later analyses. All sample results were calculated using a concentration-response curve with peak areas used for signal measurement. Sample results were statistically examined for outliers and homogeneous variance. Possible outliers were determined using the American Society for Testing and Materials (ASTM) test for outliers (10.5.). Homogeneity of the coefficients of variation was determined using the Bartlett's test (10.6.).

1. Analysis (Spiked Samples)

Procedure: Twenty-one spiked samples (7 samples at each test level) were prepared and analyzed. The spiked concentrations correspond to approximately 0.5, 1 and 2 times the TWA PEL when using a 24-L air volume. Recoveries represent the desorption efficiency and also provide precision and accuracy data for the analytical portion of the method.

1.1. Three sets of spiked samples were prepared using the following procedure:

1.1.1. A concentrated solution of ammonium sulfate [(NH₄)₂SO₄] was prepared by dissolving a weighed amount of (NH₄)₂SO₄ ("Baker Analyzed" Reagent, ACS) in deionized water (DI H₂O).

1.1.2. Aliquots of this solution were injected into in-house prepared sorbent tubes using a calibrated micropipette.

1.1.3. Sufficient (NH₄)₂SO₄ solution was added so that the tubes in each set would contain an amount of ammonia expected after collection from atmospheres at approximately 0.5, 1, and 2 times the PEL, respectively.

1.1.4. The spiked tubes were allowed to stand overnight at room temperature.

1.2. The carbon beads were removed from each tube and desorbed with DI H₂O and analyzed as mentioned in the method (10.1.). The glass tube was also rinsed with DI H₂O. Blanks were also prepared in the same fashion.

Results: The results are given in Table 1. All data were used except for one sample (2 × TWA PEL) which appeared contaminated with an unknown substance and gave baseline irregularities during analysis. All results passed the outlier and Bartlett's tests and were pooled. The analytical coefficient of variation (CV₁) was 0.031 and recovery was 99.7%.

2. Sampling and Analysis

Procedure: Twenty-four samples were taken from the dynamic generation system mentioned in the Introduction. Sample results from the dynamic system provide the overall error and precision of the sampling and analytical method. Overall error (OE) should be within ±25% and is calculated using the following equation (10.7.):

Overall Error = ± [|mean bias| + 2CV_T] × 100%

2.1. Samples were collected from gas streams (50% RH and 25°C) containing ammonia at approximately 0.5, 1, and 2 times the TWA PEL.

2.2. In-house prepared tubes containing beaded activated carbon with 5% (by weight) sulfuric acid as described in the method (10.1.) and similar tubes prepared by Supelco were taken side-by-side.

2.3. Two to four samples were collected during each generation experiment. Three to four samples were desired, but in several cases pump failure ended the determination.

2.4. Collection rates were about 0.1 L/min. Collection times were 3 to 4 h.

2.5. Samples were desorbed with 50 or 100 mL of DI H₂O and analyzed.

Results: Sample results are listed in Table 2 and are only for the in-house tubes. Supelco tube results are similar. The in-house tube samples collected at 50% RH (Table 2) show good precision and accuracy. The total coefficient of variation (CV_T) was 0.050 and the OE (total) was ±10.9%. Bias was less than 1%.

3. Collection Efficiency and Breakthrough Studies

3.1. Collection Efficiency:

Procedure: Test atmospheres were generated and samples were taken to measure the sorbent collection efficiency at the upper concentration limit of the validation.

3.1.1. A determination of the collection efficiency was performed using five Supelco tubes in which ammonia at 2 times the TWA PEL was collected. Two samples were collected at 30% and three samples at 80% RH. Samples were taken for approximately 200 min. The amounts of ammonia collected in the first and second sections of the tubes were determined.

3.1.2. The collection efficiency was calculated by dividing the amount collected in the first section by the total amount of ammonia collected in the first and second sections.

Results: The results are given in Table 3. The collection efficiency was 100%. No ammonia was detected in the backup sections. The results indicate that the collection efficiency is excellent.

3.2. Breakthrough:

Procedure: Samples were generated at a concentration greater than the validation level to determine the extent of breakthrough from the first solid sorbent section into a second section.

3.2.1. The possibility of breakthrough of ammonia during sampling was examined by collecting six samples at approximately 5 times the TWA PEL for 335 min, using Supelco tubes. Generation conditions were 50% RH and 25°C.

3.2.2. The main sections were each desorbed in 250 mL of DI H₂O. To facilitate detection of any breakthrough, the backup sections were each desorbed in 10 mL of DI H₂O.

Results: The results given in Table 3 show the overall recovery was 94.6% and the CV was 0.032. Ammonia was not detected in the backup sections, indicating no evidence of breakthrough.

4. Storage Stability of Ammonia Samples Collected on CISA

Procedure: A long-term evaluation of sample media stability was done to assess any potential problems if delays in sample analyses occur.

4.1. Supelco tube samples were collected from a gas stream adjusted to 50% RH, 25°C and an ammonia concentration of approximately 50 ppm.

4.2. A collection time of about 3.5 h was used.

4.3. Samples were capped and stored for various periods up to 29 days before preparation and analysis. Samples were stored in a desk drawer.

Results: The results given in Table 4 indicate a slight loss of ammonia (<10%) during approximately one month of storage. The slight decrease in recovery does not appear detrimental to the overall accuracy of the method. Samples can be stored for at least 29 days before analysis.

5. Humidity Tests

Procedure: Samples were collected at high (80%) and low (30%) RH using similar conditions mentioned in Section 2, where samples were collected at 50% RH. Supelco and in-house tubes were taken side-by-side. Sample recoveries for these three humidity levels were examined to determine any significant differences.

Results: The low and high RH results are given in Table 5; the 50% RH results are shown in Table 2. Results listed are for in-house tubes only. Supelco tubes gave similar results. Sampling at different humidities displayed no apparent effect on recovery. An analysis of variance (F test) was performed on the data to determine any significant difference among or within the different humidity groups. Variance at each concentration level (0.5, 1, and 2 times the TWA PEL) was compared across the 3 humidity levels (25-30, 50, and 80% RH). The variance among and within the different concentration groups gave acceptable calculated F values with the exception of the data at the TWA PEL. At this concentration only three samples were collected at the low humidity level; these recoveries were lower than expected and appeared to be due to a dilution problem during ammonia generation. The reference method bubbler samples taken side-by-side also gave lower than expected results. The test at low RH was one of the first experiments conducted. This anomalous behavior was not observed for any of the other data.

The recoveries across the 3 different humidity levels were also considered. No evidence of any significant constant increase or decrease in average recovery was apparent when generation data are compared at different humidity levels. Therefore, the data indicate no apparent humidity effect on recovery.

6. Method Comparison

Procedure: The CISA sampling and ion chromatography (IC) analytical method was compared to a reference method to determine if any significant disagreement existed between the two methods. The bubbler sampling and ISE analytical method (10.4.) was chosen as the reference method. The comparison was conducted as described below:

6.1. Side-by-side samples of CISA tubes (both commercial and in-house tubes) and bubblers containing 0.1 N sulfuric acid were collected at 25°C and 25-30%, 50%, and 80% RH. Du Pont Model P4000 pumps were used to collect bubbler samples. Bubbler collection rates were 1 L/min and collection times were 3 to 4 h. The bubbler solutions were analyzed using an Orion Model 9512 Ammonia ISE and an Orion Model EA940 IonAnalyzer (10.4.).

6.2. The CISA tubes were sampled and analyzed using the procedures described in the Introduction.

Results: The results are given in Table 6. Each CISA and bubbler sample value listed is an average value of two to four samples. Each comparison is for samples collected side-by-side. Statistical treatments are also given. Each sample comparison is based on the averages of the CISA and bubbler samples collected side-by-side. Linear regression comparison calculations are also given. The results of the two methods show the CISA sampling and corresponding IC analytical method has an overall positive bias of approximately 7% (slope = 1.07 ± 0.0387) when compared to the bubbler method. Complete listings of individual CISA and bubbler sample results are given in the Appendix.

7. Analytical Detection Limits

Procedure: The qualitative detection limit for the analysis of ammonia by IC was calculated using the Rank Sum Test (10.8.). A modification or derivation of the International Union of Pure and Applied Chemistry (IUPAC) detection limit equation (10.9.) was used to determine the quantitative detection limit. At the sensitivity level tested, blank readings and the standard deviation of the blank were equal to zero. The lack of a blank signal does not satisfy a strict interpretation of the IUPAC detection limit when using the equation shown in Table 7. The quantitative detection limit for this method is calculated using the standard deviation of a standard below the range of the expected detection limit as a substitute for the blank readings. The procedure used for sample preparation to determine detection limits is discussed below:

7.1. Low concentration ammonia samples were prepared from an ammonium chloride solution (1,000 µg/mL as NH₃). The sample concentrations were 0.05, 0.1, 0.2, and 0.5 µg/mL ammonia.

7.2. All solutions were made in 0.005 N sulfuric acid. This is the concentration expected when a typical 500-mg section of acid-treated carbon beads from a sample tube is desorbed with 100 mL of DI H₂O.

7.3. A blank sample was prepared containing the same sulfuric acid concentration as the ammonia samples.

7.4. Samples were analyzed by IC with a 50-µL sample loop and 30-microsiemens detector setting.

Results: The qualitative and quantitative detection limits (Table 7) are 0.20 and 0.50 µg NH₃/mL solution, respectively. Using a 24-L air volume and a 50-mL sample volume, the qualitative limit is 0.60 ppm and the quantitative limit is 1.5 ppm ammonia in air. For a 7.5-L air sample, these limits are 1.9 and 4.8 ppm, respectively.

8. Collection of Ammonia Samples - Supelco Tubes

Procedure: Experiments were done with in-house sampling tubes and tubes prepared by Supelco. Side-by-side sampling at varied humidity and concentration levels was performed to compare both tubes. Recoveries were compared and a t-test was used to determine if any significant difference in results existed between the two tubes (10.10.).

Results: The results are shown in Table 8. Separate Supelco tube results are also shown in Table 8. The results indicate the two different sets of tubes compare well with each other in their ability to collect and retain ammonia. A t-test (10.10.) showed no significant difference between recoveries for the two different sets of tubes.

The backup sections of the Supelco tubes were also analyzed. All backup results were non-detected when the samples were analyzed shortly after desorption. A very small peak eluting at the same time as ammonia developed after solutions stood for 1 week.

9. Conclusions

The sample determinations shown in Tables 1 and 2 are well within NIOSH or OSHA accuracy and precision guidelines (10.6., 10.7.). Collection efficiency, breakthrough, and storage stability are adequate. A humidity effect was not noted. The comparison of the bubbler with the IC method showed a positive linear regression slope of 1.07. This is to be expected since overall recoveries of the bubbler method were about 9% lower than expected (theoretical) and IC sample recoveries exhibited minimal bias. The sampling and analysis data for bubbler samples (three different RH and concentration levels) gave an overall recovery of 91.0% and a CV of 0.091. In general, the lowest recoveries for the bubbler samples were measured during preliminary experiments before improvements were made in the ISE procedure. Results for all samples taken and analyzed by IC or ISE are shown in the Appendix.

Detection limits are adequate for 4-h exposure determinations. For STEL measurements, larger sample volumes will need to be taken over the 15-min sampling period. It is recommended to use 0.5 L/min for STEL assessments (total air volume = 7.5 L). Although no STEL experiments were performed during this evaluation, a literature reference (10.2.) experiment conducted at a sampling rate of 0.5 L/min for 12 min at a concentration of 25 ppm indicated no detectable breakthrough. The experimenters also indicated the theoretical capacity of the sampling tube is 8.5 mg ammonia.

The method for collecting ammonia with CISA tubes and analysis by IC is a precise and accurate method for the determination of ammonia in workplace atmospheres.

10. References

10.1. Occupational Safety and Health Administration Technical Center: Ammonia in Workplace Atmospheres - Solid Sorbent by R.G. Adler (OSHA-SLTC Method No. ID-188). Salt Lake City, UT. Revised 1991.

10.2. Bishop, R.W., F. Belkin and R. Gaffney: Evaluation of a New Ammonia Sampling and Analytical Procedure. Am. Ind. Hyg. Assoc. J. 47: 135-137 (1986).

10.3. Occupational Safety and Health Administration Analytical Laboratory: Ammonia Detector Tubes (PE-7). Salt Lake City, UT. 1987.

10.4. Occupational Safety and Health Administration Analytical Laboratory: OSHA Analytical Methods Manual (USDOL/OSHA-SLCAL Method No. ID-164). Cincinnati, OH: American Conference of Governmental Industrial Hygienists (Pub. No. ISBN: 0-936712-66-X), 1985.

10.5. Mandel, J.: Accuracy and Precision, Evaluation and Interpretation of Analytical Results, The Treatment of Outliers. In Treatise On Analytical Chemistry, 2nd Ed., edited by I. M. Kolthoff and P. J. Elving. New York: John Wiley and Sons, 1978. pp. 282-285.

10.6. National Institute for Occupational Safety and Health: Documentation of the NIOSH Validation Tests by D. Taylor, R. Kupel and J. Bryant (DHEW/NIOSH Pub. No. 77-185). Washington, D.C.: U.S. Government Printing Office, 1977. pp. 1-12.

10.7. Occupational Safety and Health Administration Analytical Laboratory: Precision and Accuracy Data Protocol for Laboratory Validations. In OSHA Analytical Methods Manual. Cincinnati, OH: American Conference of Governmental Industrial Hygienists (Pub. No. ISBN: 0-936712-66-X), 1985.

10.8. National Bureau of Standards: Experimental Statistics by M.G. Natrella (U.S. Dept. of Commerce/NBS Handbook 91). Washington, D.C.: U.S. Government Printing Office, 1966. Chapter 16, pp 8-14.

10.9. Long, G.L. and J.D. Windfordner: Limit of Detection, A Closer Look at the IUPAC Definition. Anal. Chem. 55: 712a-724a (1983).

10.10. Osol, Arthur, ed.: Remington's Pharmaceutical Sciences. Easton, PA: Mack Publishing Co., 1980. pp. 113-114.