A diVusive sampling device for the determination of formaldehyde in air using N-methyl-4-hydrazino-7-nitrobenzofurazan (MNBDH) as reagent Andrea Bu�ldt,a Roger Lindahl,b Jan-Olof Levinb and Uwe Karst*a aWestfa�lische Wilhelms-Universita�t Mu� nster, Anorganisch-Chemisches Institut, Abteilung Analytische Chemie, Wilhelm-Klemm-Str. 8, 48149 Mu�nster, Germany bNational Institute for Working Life, Department of Chemistry, P.O. Box 7654, 90713 Umea° , Sweden Received 1st October 1998, Accepted 2nd December 1998 A new method utilizing the diVusive sampling of formaldehyde in air has been developed. Formaldehyde is sampled with the use of a glass fiber filter impregnated with N-methyl-4-hydrazino-7-nitrobenzofurazan (MNBDH) and phosphoric acid. The formaldehyde hydrazone formed is desorbed from the filter with acetonitrile and determined by high-performance liquid chromatography (HPLC) with UV/visible detection at 474 nm.The sampling rate was determined to be 24.7 mL min-1 with a relative standard deviation of 7% for 48 experiments. The measured sampling rates were not dependent on the formaldehyde concentration (0.1–1.0 mg m-3), sampling time (15–482 min) or relative humidity (20–85%).The detection limit was 70 mg m-3 for a 15 min sampling period and 2 mg m-3 for an 8 h sampling period. Therefore, a diVusive sampler containing a filter coated with Introduction a reagent which reacts as fast as DNPH with formaldehyde Formaldehyde is an important industrial chemical, a well- and is less susceptible to interferences by oxidants would be a known irritant, and a suspected carcinogen.1 Occupational great improvement.N-Methyl-4-hydrazino-7-nitrobenzoexposure occurs in the range from 0.1 to 5 ppm in air,1 and furazan (MNBDH) has been studied recently as a reagent levels from 0.01 to 0.1 ppm are often found in oYces and with reduced interferences by ozone and nitrogen dioxide.15 It homes.2 A very low occupational threshold value for formal- reacts with these oxidants to give only one product, N-methyldehyde is recommended in most countries [e.g., 0.75 ppm in 4-amino-7-nitrobenzofurazan (MNBDA).In this work, the USA (value given by the Occupational Safety and Health MNBDH is investigated as a potential alternative to DNPH Administration, OSHA) and 0.5 ppm in Germany3]. in diVusive sampling devices.Therefore, reliable and accurate methods for the determination of formaldehyde are needed. Experimental section A large number of methods for the determination of formaldehyde in air are already known. Many of these methods are Chemicals based on direct photometric4,5 or fluorimetric6,7 measurements. Solvents used for HPLC analysis were acetonitrile (HPLC However, chromatographic methods with UV/visible detecgrade, Rathburn, Walkerburn, UK), water (purified by use of tion8 or fluorescence detection9 are advantageous for the Milli-RQ systems, Millipore, Bedford, MA, USA), acetic acid identification of individual aldehydes.One of the most widely (Merck, Darmstadt, Germany) and triethylamine (Fluka, Neu- used methods is high-performance liquid chromatographic Ulm, Germany).Formaldehyde 2,4-dinitrophenylhydrazone (HPLC) determination using 2,4-dinitrophenylhydrazine for calibration was prepared from formaldehyde (37%, p.a., (DNPH) as derivatization reagent for formaldehyde. Sampling Merck), 2,4-dinitrophenylhydrazine (p.a., Fluka) and concen- of formaldehyde in air can be performed using solutions of trated HCl (Fluka) and recrystallized twice from ethanol DNPH in impingers8 or solid sorbents coated with DNPH, (Merck).Formaldehyde MNBDhydrazone for calibration was including test tubes for pumped sampling.10 However, prepared from formaldehyde (35%, p.a., Merck) and N- impingers are not convenient for personal monitoring. The methyl-4-hydrazino-7-nitrobenzofurazan, which was prepared use of solvent-free test tubes constitutes a great improvement, from 4-chloro-7-nitrobenzofurazan (Fluka) and methylhydra- but these methods also require pumped sampling.zine (Aldrich Chemie, Steinheim, Germany).15 For coating In recent years, diVusive (passive) sampling has been filters, phosphoric acid (p.a., Merck), glycerol (p.a., May and recognized as an eYcient alternative to pumped sampling.11 Baker, Dagenham, UK), ethanol (99.99%, Merck) and aceto- Of the diVerent constructions proposed for the diVusive sam- nitrile (HPLC Grade S, Rathburn) were used.pling device, the sampler investigated in ref. 12 with a DNPH coated filter has been established as that with the best proper- Synthesis ties for collecting formaldehyde. However, problems occur when formaldehyde has to be determined in air which also N-Methyl-4-hydrazino-7-nitrobenzofurazan (MNBDH).15 contains oxidants such as nitrogen dioxide or ozone.10,13,14 Methylhydrazine (2.24 mL) (0.04 mol ) in 100 mL methanol Both reagent and formaldehyde hydrazone are oxidatively was added dropwise to a solution of 1 g (0.005 mol) 4-chloro- 7-nitrobenzofurazan in 80 mL chloroform.After heating to decomposed.J. Environ. Monit., 1999, 1, 39–43 39reflux for 20 min, the solution was cooled and MNBDH was precipitated as a red crystalline material. The precipitate was filtered oV and washed with methanol. The yield was 48%. The purity of the product was examined by means of HPLC. The product was fully characterized by means of 1H NMR, MS, IR, UV and elemental analysis: 1H NMR (CDCl3) d 3.89 (3 H, s, N-CH3), 4.70 (2 H, s, NH2), 6.63 (1 H, d, J=9.4 Hz, C5-H), 8.46 (1 H, J=8.9 Hz, C6-H);MS m/z 209 (M+, 100%), 194 (M+-CH3, 63%), 164 (194-NO, 41%), 132 (88%), 118 (48%); IR (KBr pellet) 3334, 1639, 1555, 1459, 1426, 1322, 1295, 1279 cm-1; UV (CH3CN) lmax 486 nm, e (lmax) 21 600 Fig. 1 DiVusive sampler for formaldehyde (GMD sampler).L mol-1 cm-1.Analysis: calculated for C7H7N5O3: C, 40.20%; H, 3.37%; N, 33.48%; found: C, 40.13%; H, 3.60%; N, 33.51%. diameter of 1.0 mm. The filter beneath the holes is used for Formaldehyde MNBDhydrazone.15 N-Methyl-4-hydrazino- sampling (sampling filter), and the second is used to quantify 7-nitrobenzofurazan (100 mg) (4.8×10-4 mol) was dissolved the formaldehyde blank of the filter (control filter).A sliding in 0.7 mL water, 0.5 mL sulfuric acid and 2.5 mL 95% ethanol. cover is used to seal the holes when the sampler is not in use. A 50% molar excess of formaldehyde (7.2×10-4 mol) was The sampler is available from GMD Systems, Inc., added and the hydrazone was precipitated as a reddish mate- Hendersonville, PA, USA). The sampling rate (SR) for the rial. The precipitate was filtered oV and washed first with a 5 diVusive samplers was calculated according to Fick’s law,17 mass% aqueous solution of sodium bicarbonate until no from the cross-sectional area (A) in cm2 and the diVusion path further development of carbon dioxide was observed, and then length (L ) in cm: with distilled water.The product was recrystallized from SR (mL min-1)=D (cm2 s-1)×A (cm2)×L-1 (cm-1)×60 ethanol. The yield was 70%.The reaction product was fully characterized by means of 1H NMR, MS, IR, UV and As A and L are physical parameters associated with the elemental analysis: 1H NMR (CDCl3) d 4.03 (3 H, s, N-CH3), sampler construction, the sampling rate SR is constant for a 6.86 (d) and 7.04 (d) (2 H, J=9.7 and 9.4 Hz, NCH2), 7.40 selected analyte and diVusive sampler.This calculated value (1 H, d, J=8.9 Hz, C5-H), 8.53 (1 H, d, J=8.8 Hz, C6-H); diVers from experimentally determined sampling rates where MS m/z 221 (M+, 81%), 145 (51%), 117 (100%); IR (KBr diVusive samplers have been exposed in atmospheres of known pellet) 3023, 2969, 1616, 1541, 1496, 1320, 1186, 1079, analyte concentrations. An explanation for this is given in 1001 cm-1; UV (CH3CN) lmax 474 nm, e (lmax) ref. 18. 24 700 L mol-1 cm-1. Analysis calculated for C8H7N5O3: C, 43.44%; H, 3.19%; N, 31.66%; found: C, 43.29%; H, 3.24%; Generation of standard atmospheres of formaldehyde N, 30.80%. Gaseous formaldehyde was generated by decomposition of paraformaldehyde and permeation through a silicone tube (id Coated filters for pumped sampling 10 mm, od 14 mm, length 41 mm) as shown in Fig. 2.The DNPH (300 mg) was dissoheating in 0.5 mL concen- temperature of the permeation device was controlled and trated phosphoric acid (85%), 1.5 mL ethanol containing 20% varied between 54 and 83 °C for diVerent runs. The weight glycerol and 9 mL acetonitrile. Glass fiber filters (13 mm in loss of the permeation device was used to calculate the diameter, (type AE, 0.3 mm pore size, SKC, Inc., PA, USA) concentration in the exposure chamber.The air flow of were immersed in the solution and immediately removed again. 0.4 L min-1 passing the permeation device was further mixed The filters were then allowed to dry on a glass surface at room with controlled humidified air. The total air flow in the temperature and stored in a refrigerator.exposure chamber was 40 L min-1, which corresponds to an air velocity of 0.3 m s-1. The Teflon exposure chamber, con- Coated filters for diVusive sampling structed for sampling of 6 diVusive samplers and with ports for pumped sampling with the reference method, has been MNBDH (180 mg) was dissolved with heating in 0.3 mL described previously.16,18 concentrated phosphoric acid (85%), 0.7 mL ethanol containing 20% glycerol and 20 mL acetonitrile.Glass fiber filters Laboratory validation of the samplers (2×2 cm) were cut from round filters (type AE, 0.3 mm pore size, diameter 25 mm, SKC, Inc., PA, USA). These were then For all experiments regarding the recovery rate of formaldipped into the coating solution and allowed to dry on a glass dehyde on the MNBDH diVusive sampler, the samplers were surface at room temperature.One filter was placed under the exposed in a sampling chamber,16 six at a time, to formalsampling part of the sampler and another one under the dehyde levels from 0.1 to 1.0 mg m-3, with sampling times control part. between 15 min and approximately 8 h. The relative humidity was varied between 20% and 85%.The samplers were oriented Pumped sampling parallel to the air stream. Three pumped samples were taken For pumped sampling, the 13 mm diameter DNPH coated filters were used in two-section polypropylene filter holders (No. 225–32, SKC, Inc., PA, USA).16 For all measurements, collecting and control filter cassettes were connected in series to identify incomplete recovery on the collecting cassette.DiVusive sampling The diVusive sampler is shown in Fig. 1. The housing, measuring 60×30×5 mm, is made of polypropylene. Two impregnated filters are placed beneath a 2.9 mm thick screen. The Fig. 2 Formaldehyde generation device and exposure chamber. screen has, within an area of 20×20 mm, 112 holes with a 40 J. Environ. Monit., 1999, 1, 39–43simultaneously from the sampling chamber with a flow of 100 mL min-1.Exposure of the diVusive samplers to ozone Six diVusive samplers were first exposed to formaldehyde (1 mg m-3) for one hour. Afterwards, three of these were exposed to an ozone atmosphere (1.8 mg m-3) for two hours. HPLC instrumentation A Waters HPLC system (Milford, MA, USA), consisting of two pumps (model 510), a WISP-710B autosampler, a Waters 486 UV/visible detector and the Millennium Software Version 2.15 was used.The column material was Nova Pak RP-18 (Waters): particle size, 4 mm; column dimensions, 250 mm×4.6 mm. HPLC analysis The formaldehyde DNPhydrazone and the formaldehyde MNBDhydrazone were eluted from the filter by shaking with 3 mL of acetonitrile in a 4 mL glass vial; 10 mL of this solution were injected into the HPLC.The analysis of the DNPH coated filters was carried out with an isocratic mobile phase consisting of 60% acetonitrile and 40% water with a flow rate Fig. 3 Chromatograms of desorbed MNBDH filters of a diVusive of 1 mL min-1. The detection wavelength was 365 nm. The sampler: blank (full line) and an air sample containing 0.98 mg m-3 MNBDH coated filters were analyzed with an isocratic mobile formaldehyde (broken line).Duration of exposure was 481 min. phase consisting of 45% acetonitrile and 55% buVered water [composition: 500 mL water with 2415 mL triethylamine and Eight diVerent experiments were carried out to investigate 975 mL acetic acid (pH#7.5)] with a flow rate of 1 mL min-1. the recovery of formaldehyde.The conditions chosen for the The detection wavelength was 474 nm. individual measurements are described in Table 1. The diVusive sampler was partially tested according to EN 838.19 The Results formaldehyde concentrations in the exposure chamber were gravimetrically determined by measuring the weight loss of MNBDH was synthesized in a nucleophilic substitution reacthe permeation device and the air flow in the exposure chamber tion of N-methylhydrazine and 4-chloro-7-nitrobenzofurazan: (a). The values obtained with the pumped reference method (b) were corrected for a 95% recovery, found in previous studies.16,20 The reference values were mainly within±10%, as can be seen in Table 1, which verifies the gravimetrically determined values used for all calculations.The MNBDH coated diVusive samplers were exposed to the standard formaldehyde concentrations simultaneously with the DNPH pumped samplers. Table 2 gives the experimentally The corresponding formaldehyde MNBDhydrazone was pre- determined sampling rates calculated with the known analyte pared by reacting MNBDH with formaldehyde in the presence concentrations obtained gravimetrically.The results were of acid as catalyst: obtained from a series of experiments with diVerent formaldehyde concentrations, sampling times and relative humidities. The collecting filter and the control filter of each diVusive sampling device were eluted separately and injected into the HPLC system. In every case, blank values could be found on the control filter.Two reasons are responsible for this fact: firstly, the GMD sampler is not completely tight and leakage may occur beneath the sliding cover and between the screen and the badge housing; the leakage during sampling to the control filter is about 5% according to ref. 18; secondly, some This reaction can be utilized to collect formaldehyde on diVusive sampling devices. As the collection of formaldehyde formaldehyde MNBDhydrazone is formed during the coating of the filters due to the ubiquitous occurrence of formaldehyde.is well known using DNPH, a pumped sampling method with DNPH coated filters is used as reference method to validate The leakage is incorporated in the experimentally determined sampling rate since the amount on the control filter is always the new MNBDH diVusive sampler.Fig. 3 shows a chromatogram of the eluate of a MNBDH coated diVusive filter after subtracted from the amount on the sampling filter. Due to the blanks, problems may occur in experiments with low formal- exposure to formaldehyde containing air and the corresponding eluate of the control filter. As can be seen, the formaldehyde dehyde concentrations and short sampling times (measurement 1).In this case, the formaldehyde MNBDhydrazone on hydrazone is separated from the reagent very well. To investigate the reproducibility of the chromatographic analysis, the the control filter was about 50% of the value found on the sampling filter. This resulted in a high relative standard eluate of one filter was injected into the HPLC system ten times.The evaluation via the peak areas results in a relative deviation (13%). All other measurements show that the sampling rate was not dependent on the relative humidity, sampling standard deviation of 0.8%. This value proves the very good reproducibility of the chromatographic analysis. time or formaldehyde concentration. The mean sampling rate J. Environ. Monit., 1999, 1, 39–43 41Table 1 Recovery of formaldehyde as 2,4-dinitrophenylhydrazone obtained by the gravimetrically determined formaldehyde concentration (a) in comparison with the pumped sampling method (b) in eight diVerent experiments Determined HCHO HCHO concentration/mg m-3 concentration/ using the DNPH pumped mg m-3 obtained method (b) (corrected Recovery RSD (%) Measurement RH (%) ST/min gravimetrically (a) value) (%) (n=3) 1 20 15 0.116 0.100 86 8 2 20 478 0.116 0.111 96 3 3 85 60 0.116 0.108 93 3 4 85 475 0.116 0.101 87 4 5 20 16 1.130 1.051 93 5 6 20 481 1.030 1.032 100 1 7 85 15 1.030 1.061 103 1 8 85 482 1.030 1.084 105 1 RH, relative humidity; ST, sampling time; RSD, relative standard deviation; n, number of experiments. Table 3 Recovery of formaldehyde on the diVusive sampling devices Table 2 Experimentally determined sampling rates by means of the MNBDH coated diVusive sampling device RSD (%) Measurement Recovery (%) (n=6) RSD (%) Measurement SR/mL min-1 (n=6) 1 93 13 2 84 4 1 23.8 13 2 23.0 4 3 102 9 4 98 9 3 25.7 9 4 24.7 9 5 90 8 6 101 4 5 23.3 8 6 25.5 4 7 104 5 8 99 4 7 26.3 5 8 24.9 4 RSD, relative standard deviation; n, number of experiments.SR, sampling rate; RSD, relative standard deviation; n, number of experiments. was 24.7 mL min-1 with a relative standard deviation of 7% for 48 experiments. This value coincides well with older publications which reported a sampling rate of 25.2 mL min-1 12 obtained by means of DNPH coated diVusive samplers. The European Committee for Standardization (CEN) has published Workplace Atmospheres—General Requirements for the Performance of Procedures for the Measurement of Chemical Products (EN 482).21 This document defines the relative overall uncertainty (ROU) by combining bias and precision according to the formula: ROU= |x: -xref|+2s xref ×100 where x: is the mean value of repeated measurements, xref is the true or accepted reference value of the concentration and s is the standard deviation of the measurements.The ROU is used to specify the performance requirements of a measurement method. These requirements vary depending on the measurement task. The ROU must be 30% within the measuring range of 0.5 to 2 times the limit value, and 50% within 0.1 to 0.5 times the limit value. The ROU for the diVusive sampling of formaldehyde with MNBDH impreg- Fig. 4 Chromatogram of an eluted filter which was only exposed to nated filters was calculated to be 14%, which easily meets the formaldehyde (full line) and a chromatogram of an eluted filter which CEN requirements for a measurement method. was exposed to both formaldehyde and ozone (dotted line). Another way of interpreting the results involves the determination of the recovery rate of formaldehyde on the MNBDH coated diVusive sampling devices based on the supposition formaldehyde with the new reagent.Furthermore, the results prove that the sampling rate is not dependent on the reagent. that the sampling rate of formaldehyde must be 25.2 mL min-1, which was found by means of DNPH coated The use of a personal diVusive sampling device requires a high storage stability of the exposed samplers, because it is diVusive sampling devices.12 Table 3 shows the results of this method of interpretation.The mean recovery of formaldehyde not always possible to analyze the filters directly after exposure. To investigate this, the exposed samplers were stored in sealed on the diVusive sampler is 96% with a standard deviation of 7%.These results also show the possibility to determine bags made of laminated aluminum. Some were placed in a 42 J. Environ. Monit., 1999, 1, 39–43refrigerator, and others were stored at room temperature. The ferences when using MNBDH coated diVusive samplers compared with DNPH coated diVusive filters. analysis of the filters was carried out after five days.Once formed, the hydrazone is stable on the filter, since no decrease in recovery of formaldehyde hydrazone was noted for both Acknowledgements the cold stored filters and those stored at room temperature. Financial support of parts of this work by the European The storage stability of the diVusive samplers before expo- Commission (‘Aldehydes’ project, SMT4-CT97–2190) and by sure was investigated. Some samplers were stored in a refrigerthe Fonds der Chemischen Industrie (Frankfurt, Germany) is ator and some were stored at room temperature for two gratefully acknowledged.months and then exposed to formaldehyde. The analysis of these filters showed that the samplers stored in a refrigerator References gave a good recovery of formaldehyde. The sampling rate was determined to be 26.7 mL min-1 with a relative standard 1 R.W. Hart, A. Terturro and L. Neimeth, Environ. Health deviation of 5% for 6 experiments. The other samplers that Perspect., 1984, 58, 323. were stored at room temperature for a long time gave lower 2 T. H. Stock and S. R. Mendez, Am. Ind. Hyg. Assoc. J., 1985, 46, 313. sampling rates. This shows that the filters must be stored in a 3 Deutsche Forschungsgemeinschaft, MAK- und BAT-Werte-Liste refrigerator to give good results. 1997,Wiley,Weinheim, 1997. The stability of formaldehyde MNBDhydrazone on the 4 R. R. Miksch and W. A. Douglas, Am. Ind. Hyg. Assoc. J., 1982, diVusive samplers in the presence of ozone was also investi- 43, 362. gated. For this purpose, six diVusive samplers were first 5 R.R. Miksch, D. W. Anthon, L. Z. Fanning, C. D. Hollowell, K. Revzan and J. Glanville, Anal. Chem., 1981, 53, 2118. exposed to formaldehyde. Afterwards, three of these were 6 I. Ahonen, E. Priha and M.-L. A� ija�la�, Chemosphere, 1984, 13, 521. exposed to ozone. All filters were eluted and the eluate was 7 P. Bisgaard, L. Molhave, B. Rietz and P. Wilhardt, Am. Ind. Hyg. injected into the HPLC system.Fig. 4 shows a chromatogram Assoc. J., 1984, 45(6), 425. of an eluted filter which was only exposed to formaldehyde 8 K. Fung and D. Grosjean, Anal. Chem., 1981, 53, 168. and a chromatogram of an eluted filter which was exposed to 9 W. Schmied, M. Przewosnik and K. Ba�chmann, Fresenius’ Z. both formaldehyde and ozone. As can be seen, only one Anal. Chem., 1989, 335, 464. 10 W. Po� tter and U. Karst, Anal. Chem., 1996, 68, 3354. product is formed when MNBDH reacts with ozone. This 11 J. O. Levin and R. Lindahl, Analyst, 1994, 119, 79. was identified as N-methyl-4-amino-7-nitrobenzofurazan 12 J. O. Levin, R. Lindahl and K. Andersson, Environ. Technol. Lett., (MNBDA).15 As the peak area of the hydrazone in the 1988, 9, 1423. chromatogram has not decreased, the formaldehyde 13 R.R. Arnts and S. B. Tejada, Environ. Sci. Technol., 1989, 23, MNBDhydrazone is considered to be stable in the presence of 1428. 14 D. F. Smith, T. U. Kleindienst and E. E. Hudgens, J. Chromatogr., ozone. The evaluation of the chromatograms led to the 1989, 483, 431. following sampling rates for formaldehyde: 24.3 mL min-1 15 A. Bu� ldt and U.Karst, German Patent, DE 198 00 537.7, 1998. with a relative standard deviation of 1% for the filters which 16 J. O. Levin, R. Lindahl and K. Andersson, J. Environ. Sci. were only exposed to formaldehyde and 24.6 mL min-1 with Technol., 1986, 20, 1273. a relative standard deviation of 3% for the filters which were 17 V. Rose and J. L. Perkins, Am. Ind. Hyg. Assoc. J., 1982, 43, 605. 18 R. Lindahl, PhD Thesis, Umea° University and National Institute exposed to both formaldehyde and ozone. for Working Life, Umea°, Sweden, 1997. These results prove that MNBDH can be used to collect 19 CEN, European Committee for Standardization, EN 838, formaldehyde on diVusive filters in the same way as DNPH. Workplace Atmospheres—DiVusive Samplers for the Determination This is especially advantageous when formaldehyde has to be of Gases and Vapours—Requirements and Test Methods, CEN, determined in matrices that also contain oxidizing substances. Brussels, 1995. Besides, the determination using MNBDH as derivatization 20 J. O. Levin, K. Andersson, R. Lindahl and C. A. Nilsson, Anal. Chem., 1985, 57, 1032. reagent is much more selective, because the absorption maxi- 21 CEN, European Committee for Standardization, EN 482, mum of formaldehyde MNBDhydrazone is located at longer Workplace Atmospheres—General Requirements for the wavelengths compared with the DNPH derivative (474 nm Performance of Procedures for the Measurement of Chemical compared with formaldehyde DNPhydrazone at 349 nm). Agents, CEN, Brussels, 1994. Formaldehyde can be collected and determined in matrices containing ozone and nitrogen dioxide with reduced inter- Paper 8/07631F J. Environ. Monit., 1999, 1, 39