Diffusive sampling of methyl isocyanate using 4-nitro-7-piperazinobenzo-2-oxa-1,3-diazole (NBDPZ) as derivatizing agent

Abstract

A diffusive sampling method for the determination of methyl isocyanate (MIC) in air is introduced. MIC is collected using a glass fiber filter impregnated with 4-nitro-7-piperazinobenzo-2-oxa-1,3-diazole (NBDPZ). The urea derivative formed is desorbed from the filter with acetonitrile and analyzed by means of high-performance liquid chromatography (HPLC) using fluorescence detection (FLD) with λ_ex = 471 nm and λ_em = 540 nm. Additionally, a method was developed using tandem mass spectrometric (MS-MS) detection, which was performed as selected reaction monitoring (SRM) on the transition [MIC–NBDPZ + H]⁺ (m/z 307) to [NBDPZ + H]⁺ (m/z 250). The diffusive sampler was tested with MIC concentrations between 1 and 35 µg m⁻³. The sampling periods varied from 15 min to 8 h, and the relative humidity (RH) was set from 20% up to 80%. The sampling rate for all 15 min experiments was determined to be 15.0 mL min⁻¹ (using HPLC-FLD) with a relative standard deviation of 9.9% for 56 experiments. At 80% RH, only 15 min sampling gave acceptable results. Further experiments revealed that humidity did not affect the MIC derivative but the reagent on the filter prior to and during sampling. The sampling rate for all experiments (including long term sampling) performed at 20% RH was found to be 15.0 mL min⁻¹ with a relative standard deviation of 6.3% (N = 42). The limit of quantification was 3 µg m⁻³ (LC-MS-MS: 1.3 µg m⁻³) for 15 min sampling periods and 0.2 µg m⁻³ (LC-MS-MS: 0.15 µg m⁻³) for 8 h sampling runs applying fluorescence detection.

---

Introduction

Monoisocyanates are widely used as intermediates during the manufacture of various pharmaceutical and agricultural products. The widespread application of isocyanates is mainly based on their high reactivity towards nucleophilic agents, e.g., alcohols or amines, often showing quantitative reaction yields without any side reactions.¹ Methyl isocyanate (MIC) is primarily used in the production of carbamate-based pesticides. The toxicological effects of exposure to methyl isocyanate were well explored after the Bhopal accident in 1984 and are associated with irritation of the respiratory system, the mucous membranes and the eyes.^2–6 Isocyanates in general have strong sensitizing properties and are supposed to be the most common inducers of occupational asthma.⁷ Exposure can occur while handling the native compounds or heating and processing polyurethane (PUR) or other isocyanate-based products.⁸ The low-molecular and highly volatile methyl isocyanate and isocyanic acid were found as degradation products resulting from PUR products that neither originally contained these substances nor have been used during manufacture.⁹ In order to comply with the 5 ppb Threshold Limit Value (TLV) and to detect hazardous conditions it is important to monitor isocyanate concentrations in workplace atmospheres. Known procedures involve pumped sampling in combination with impingers, reagent-coated filters or impregnated sorbent tubes for collection of the analytes. Compounds containing amino groups, such as 1-(2-methoxyphenyl)piperazine (2-MP), dibutylamine (DBA) and 1-(2-pyridyl)piperazine (2-PP) serve as derivatizing reagents.^10–13 Further known reagents are 9-(N-methylaminomethyl)anthracene (MAMA) or 1-(9-anthracenylmethyl)piperazine (MAP).^14,15 Most of these methods are well developed and sufficiently sensitive. However, they are very complicated, expensive and require a high degree of technical competence. Therefore, active sampling methods are less suitable for personal sampling to monitor occupational exposures. In order to perform sampling close to the breathing zone, it cannot be recommended to use impingers with possibly toxic solvents, which would represent an additional potential health risk to the performing person. In contrast, diffusive sampling offers several advantages when compared with pumped sampling. Diffusive samplers are easier to use than active sampling methods. No specially trained personnel are needed, and the sampling is much more comfortable for the performing person during work time. Eight hours sampling periods can be done without inconvenience and no pumps (that have to be calibrated prior to sampling) are needed. However, regarding quantification, especially for short period sampling often going together with very low analyte concentrations, highly sensitive analytical methods are necessary, such as LC-MS-MS. Recently, Zweigbergk et al. presented a diffusive sampling method for the determination of methyl isocyanate in air based on the derivatization with 2-MP and subsequent analysis by means of LC-MS-MS.¹⁶ Also recently, 4-nitro-7-piperazinobenzo-2-oxa-1,3-diazole (NBDPZ) has been introduced as a derivatizing reagent for isocyanates.^17,18 On the basis of NBDPZ, methyl isocyanate can be determined by means of liquid chromatography with subsequent fluorescence, photometric or even mass spectrometric detection. The aim of this work was to develop a new user-friendly diffusive sampling method for the determination of airborne methyl isocyanate. The analysis should be performed using HPLC with fluorescence detection (LC-FLD), but a method using the more sensitive tandem mass spectrometric detection should also be developed.

Experimental

Chemicals

The solvents used for HPLC analysis were methanol (HPLC gradient grade, J. T. Baker, Deventer, The Netherlands), acetonitrile (ultra-gradient HPLC grade, J. T. Baker), water (purified using Milli-RQ systems, Millipore, Bedford, MA, USA), formic acid (98%, p.a., J. T. Baker) and ammonium acetate (98%, p.a., Merck, Darmstadt, Germany). For synthesis and filter coating, dichloromethane (p.a., J. T. Baker) and acetonitrile (HPLC grade S, Rathburn, Walkerburn, UK) were used. MIC-NBDPZ for calibration was prepared from methyl isocyanate (99%, Chem Service, West Chester, UK) and 4-nitro-7-piperazinobenzo-2-oxa-1,3-diazole which was prepared from 4-chloro-7-nitrobenzo-2-oxa-1,3-diazole (98%, Fluka, Neu-Ulm, Germany) and piperazine (99%, Sigma–Aldrich, Steinheim, Germany) according to literature.^17,18 Toluene (J. T. Baker) used as solvent for pure MIC was dried with calcium hydride (Fluka, Buchs, Switzerland), distilled and stored over molecular sieves. 1-(2-Methoxyphenyl)piperazine was obtained from Sigma (St. Louis, USA) and trideuterated 1-(2-methoxyphenyl)piperazine was delivered by Synthelec (Lund, Sweden). Acetic anhydride (99.5%, p.a.), propionic anhydride (98%), butyric anhydride (97%) and isopropyl isocyanate (99%) were obtained from Fluka (Neu-Ulm, Germany).

Synthesis

4-Nitro-7-piperazinobenzo-2-oxa-1,3-diazole (NBDPZ). 4-Chloro-7-nitrobenzo-2-oxa-1,3-diazole (2.0 g, 10 mmol) in 120 mL of dichloromethane was added dropwise to a stirred solution of 3.6 g (40 mmol) piperazine in 160 mL of methanol. The NBDPZ precipitated as red crystals. It was filtered off and washed with cold water and methanol. The yield was 65%. The product was characterized by means of ¹H-NMR, IR, UV/Vis and fluorescence spectroscopy, mass spectrometry and elemental analysis.¹⁷

Methyl isocyanate–NBDPZ urea derivative. 4-Nitro-7-piperazinobenzo-2-oxa-1,3-diazole (0.7 g, 3 mmol) was dissolved in 200 mL of dichloromethane. 500 µL methyl isocyanate (9 mmol) were added rapidly with stirring. The MIC–NBDPZ precipitated as an orange solid. Two mL of methanol were added to remove excess MIC. The product was filtered off, washed with cold methanol and dried in a vacuum desiccator. The yield was 73%. The product was characterized by means of ¹H-NMR, IR, UV/Vis and fluorescence spectroscopy, mass spectrometry and elemental analysis.¹⁷

Isopropyl isocyanate–NBDPZ urea derivative (iPIC–NBDPZ). NBDPZ (0.17 g, 0.7 mmol) was dissolved in 60 mL of dichloromethane. 75 µL of isopropyl isocyanate (0.8 mmol) were added rapidly with stirring. 40 mL n-hexane were added and the iPIC–NBDPZ precipitated as an orange solid. Two mL of methanol were added to remove excess isocyanate. The product was filtered off, washed with cold methanol and dried under reduced pressure. The yield was 49%.

Generation of standard atmospheres of methyl isocyanate

Methyl isocyanate spiked air was dynamically generated and controlled with respect to relative humidity (RH), temperature and concentration. A solution of pure MIC in toluene was injected into an evaporation chamber using a syringe pump (CMA/100, Carnegie Medicine, Stockholm, Sweden). The injection was performed through a nebulizer (J. E. Meinhard, Santa Ana, CA, USA). The airflow through the nebulizer was 0.5 L min⁻¹, and the aerosol from the nebulizer was mixed with air (4.5 L min⁻¹). After evaporation, the mixture was further diluted with humidified air and delivered to the exposure chamber with a total flow of 40 L min⁻¹. The airflows were controlled by a mass flow meter (Bronkhorst Hi-Tec, Ruurlo, The Netherlands) and the relative humidity was measured at the end of the exposure chamber (RH & T Indicator HMI 14, Vaisala, Helsinki, Finland). The concentration was always checked with a reference method described below, using six parallel ports which were mounted to the exposure chamber, thus allowing to take reference samples while diffusive sampling experiments were carried out. These experimentally determined values were always taken to calculate the uptake rate. The generation equipment with exposure chamber has been described previously.^19,20

Coated filters for diffusive sampling

NBDPZ (100 mg) was dissolved in 150 mL of acetonitrile. Glass fiber filters (20 × 20 mm) were cut from round filters (Type A/E, diameter 37 mm, SKC, Eighty Four, PA, USA), put onto a glass surface and impregnated twice with 100 µL of the reagent solution. The filters were subsequently allowed to dry for 20 min under reduced pressure. One filter was placed under the sampling part of the sampler and another under the control part.

Coated filters for pumped sampling (reference method)

A pumped filter method described by Henriks-Eckerman et al.²¹ was applied to verify the MIC concentrations of generated test atmospheres. Round glass fiber filters (GFB, diameter 25 mm, Whatman, Maidstone, Kent, UK) were placed on a glass surface and impregnated twice with 200 µL of a solution containing 500 mg 2-MP in 50 mL of toluene (52 mmol L⁻¹). Afterwards, the filters were dried in a gentle stream of filtered air and finally stored in a refrigerator. Two filters were placed on top of each other in a Swinnex 25 filter cassette (Millipore, Bedford, MA, USA). Six filter cassettes were connected to the ports on the exposure chamber. For a period of 15 min, samples were taken at a sampling rate of 0.2–0.3 L min⁻¹. For long-term experiments, this reference method was carried out twice, once in the beginning and once at the end of the test period, respectively.

Diffusive sampling

The diffusive sampler is schematically shown in Fig. 1. The housing, with dimensions of 86 × 28 × 9 mm, is made of polypropylene. Two impregnated filters are placed beneath a 2.9 mm thick screen. The part of the screen covering the sampling filter comprises 112 holes within a total area of 20 × 20 mm and with a diameter of 1.0 mm for each hole. A sliding cover was used to seal the holes when the sampler was not in use. The second filter (control filter) was used to quantify the methyl isocyanate blank. The sampler is commercially available as UMEx 100 (with coated filters for sampling of formaldehyde) from SKC.

Fig. 1 Set-up of the UMEx diffusive sampler.

Diffusive sampling experiments were performed mainly according to EN 838²² in the low concentration range of approximately 0.1–3 times the Swedish Time Weighted Averages (TWA) limit value, which is approximately 12 µg m⁻³ MIC (5 ppb). The Swedish Short-Term Exposure Limit (STEL) is set to 10 ppb MIC (23 µg m⁻³) and equals the German TWA (MAK) value.^23,24

Laboratory validation

HPLC instrumentation. The chromatographic system for LC-MS-MS analysis was supplied by PerkinElmer (PE Series 200, Norwalk, CT, USA) and consisted of an autosampler and two micro-pumps. The column outlet was coupled to a triple quadrupole mass spectrometer (API 2000, Applied Biosystems, Foster City, CA, USA) with ESI (for the reference method) or APCI interface. Data were collected and analyzed using Analyst 1.1 software (Applied Biosystems). The columns used were two GROM-SIL 80 ODS-7 pH columns with 4 µm particle size and 80 Å pore diameter (GROM, Herrenberg, Germany), 200 × 3 mm (for the reference method) and 60 × 4 mm for the MIC–NBDPZ method.

The chromatographic system used in combination with fluorescence detection was delivered by Waters and consisted of two Millipore Model 510 pumps, a 717plus autosampler, a SAT/IN communications bus module, a 474 scanning fluorescence detector and an ERC 3415 degasser (Scantec, Gothenburg, Sweden). Data were collected using Millenium 32 Chromatography Manager Version 3.05. The column used was an YMC-Pack Pro-C18 (YMC Co. Ltd., Kyoto, Japan), 150 × 4.6 mm, with 5 µm particle size and a pore diameter of 120 Å.

Filter work-up and HPLC analysis.
Reference method. Exposed filters from one filter cassette were placed together in 4 mL HPLC vials containing 3 mL of 1-(2-methoxyphenyl)-piperazine (2-MP) in toluene (0.26 mmol L⁻¹). Unreacted 2-MP reagent was acetylated with acetic anhydride according to the procedure described in MDHS 25/3,¹³ and trideuterated MIC-2-MP was added as internal standard. The samples were evaporated to dryness in a vacuum centrifuge (Speed-Vac 290, Savant, Farmingdale, NY, USA), redissolved in acetonitrile, centrifuged for 5 min at 5000 rpm (Uniequip UEC 13, Martinsried, Germany) and analyzed by means of LC-MS-MS. Later experiments showed that the work-up procedure could be simplified by leaving out the acetylating step, which did not affect the results. This was due to the fact that in this case only MIC was analyzed. If a more complex mixture of isocyanates had to be determined, this step could not be left out. By using 3 mL of a 2-MP solution (0.26 mmol L⁻¹) in acetonitrile instead of toluene for desorption, the LC-MS-MS analysis could be performed directly without evaporating and redissolving the samples.

The injection volume was 3 µL and the flow rate was set to 400 µL min⁻¹ applying a water–acetonitrile gradient (with 2 mmol L⁻¹ NH₄Ac in both eluents). After 5 min of isocratic elution at 60% acetonitrile, a linear gradient for 2 min to 95% acetonitrile was run subsequently.

The ESI capillary voltage was set to 5.5 kV, and the added drying gas was heated to 320 °C. Selected reaction monitoring (SRM) was performed on the transition [M + H]⁺ to [2-MP + H]⁺, and data were collected with a dwell time of 200 ms. The analyte was quantified referring to the ratio between analyte and internal standard.

Diffusive samplers.
Fluorescence detection. Sample and control filters of exposed diffusive samplers were transferred into separate HPLC vials and eluted with 3 mL of acetonitrile. Subsequently, 150 µL aliquots were transferred into 200 µL vials and a standard solution of NBDPZ derivative of isopropyl isocyanate (iPIC–NBDPZ) in acetonitrile was added as internal standard. These vials were retained for later MS-MS analysis.

The samples in the HPLC vials were injected directly into the HPLC system for analysis with fluorescence detection. The injection volume was 10 µL, and the sample was eluted with 1.0 mL min⁻¹ in a water–methanol gradient (with 0.25% (v/v) formic acid in both components). After 12 min of isocratic elution at 35% methanol, a linear gradient for 2 min to 100% methanol followed. The total time for the analysis was 30 min, including re-equilibration. Conditions for fluorescence detection were λ_ex = 471 nm and λ_em = 540 nm.

Additionally, it was investigated whether the separation of NBDPZ from analyte derivative could be improved by addition of acid anhydrides to the sample. The anhydride was made to react with excess NBDPZ from exposed diffusive samplers in order to elute after the MIC–NBDPZ. Therefore, an excess of appropriate anhydride was added to the vials containing the eluted filters in acetonitrile. The tested anhydrides were acetic, propionic and butyric anhydride. Propionic anhydride (PrAn) was found to be most suitable as the acetylated NBDPZ eluted just in front of the MIC–NBDPZ while the retention time of the butyric anhydride adduct was much longer than necessary. Thus, following the first analysis with LC-FLD, 1 µL of propionic anhydride was added to the sample solution to propionylate the NBDPZ reagent, and the analysis was repeated.

Mass spectrometric detection. The solutions that were transferred to the 200 µL vials were taken for LC-MS-MS analysis. The injection volume was 10 µL and the sample was eluted with 1.0 mL min⁻¹ in a water–methanol gradient (with 0.5% (v/v) formic acid in both). After 5 min of isocratic elution at 20% methanol, a linear gradient for 2 min to 90% methanol followed. The total analysis time was 10 min, including re-equilibration. The quadrupole was operated in the APCI(+) mode, the ionization voltage was set to 2.5 kV, and the APCI temperature was set to 500 °C. Detection was performed as selected reaction monitoring (SRM) on the transition [MIC–NBDPZ + H]⁺ (m/z 307) to [NBDPZ + H]⁺ (m/z 250), and data were collected with a dwell time of 200 ms. The analyte was quantified referring to the ratio between analyte and internal standard (iPIC-NBDPZ). For the internal standard, SRM was performed on the transition [iPIC–NBDPZ + H]⁺ (m/z 335) to [NBDPZ + H]⁺ (m/z 250).

Results and discussion

The experiments performed are based on the derivatization reaction of NBDPZ with methyl isocyanate (Fig. 2). A reliable pumped sampling method with 2-MP coated filters was used as a reference method to validate the new NBDPZ diffusive sampler. The 2-MP reference method itself was validated using an independent DBA impinger method. This work is described in more detail in the paper of von Zweigbergk et al.¹⁶

Fig. 2 Scheme of the derivatization reaction of NBDPZ with MIC.

Using NBDPZ as the derivatizing reagent, fluorescence and mass spectrometric detection of isocyanate derivatives are possible as well as UV/Vis detection.¹⁸ In Table 1, the instrumental limits of detection of different detection methods are listed. While UV/Vis detection does not show sufficient sensitivity for the given analytical problem, fluorescence and triple quadrupole MS detection are working well in the required concentration range.

Table 1 Instrumental limits of detection (LOD) of MIC–NBDPZ determined with different detection methods

Detection method	LOD/10^−9 mol L^−1	LOD/pg (analyte)
Tandem-MS detection m/z (Q1) = 307; m/z (Q3) = 250	0.82	0.47
Fluorescence detection λex = 471 nm; λem = 540 nm	6.5	3.7
UV/Vis detection λ = 480 nm	35	20

---

Since fluorescence detection can be performed by less experienced personnel and requires less expensive equipment than MS-MS spectrometry, attention has been primarily focused on that method although the developed MS-MS method is very sensitive and highly selective. Nevertheless, most of the exposed and eluted diffusive samplers were analyzed applying both methods in order to obtain comparable results.

Fluorescence detection

In order to achieve baseline separation of NBDPZ and its corresponding derivatives, it was necessary to add formic acid to the mobile phase to protonate the NBDPZ. Two problems had to be solved to obtain peaks that were easy to quantify.

As NBDPZ shows a higher fluorescence intensity in acidic media than its urea derivatives do and as the reagent peak tails severely, also due to the fact that the eluted filter solution inherently contains a large excess of reagent, it was inevitable that an eluent mixture shifting the retention time of the MIC-peak well behind the reagent peak had to be chosen.

In addition, during storage of the loaded diffusive samplers some interfering peaks appeared in the chromatogram. As these peaks had not been present in the chromatogram of the solution used for impregnation, the corresponding compounds were yielded after the impregnation process. As can be seen from Fig. 3, their intensity is increasing with storage time of the samplers (which were loaded immediately after impregnation). These interferences are acceptable provided that the extent shown in Fig. 3 is not exceeded. For that purpose, it is important to strictly follow the impregnation procedure described above, since first attempts with longer drying periods in a gentle stream of filtered air led to interfering peaks of intensities that made quantification by means of fluorescence detection almost impossible.

Fig. 3 Chromatograms of eluted NBDPZ coated filters from diffusive samplers stored for different time periods at room temperature.

Although several attempts were performed to identify the character of the interfering peaks by means of LC-MS-MS, the unknown compounds could not be structurally elucidated yet. As the addition of propionic anhydride had no influence on the retention times, some unknown components present in air must have reacted with the piperazine functionality, blocking this position for propionylation or derivatization with isocyanates. Though only a small part of the reagent was affected in that way (the area of these disturbing peaks accounted for only about 0.01–0.1% of the reagent peak area) this was well in the range of the analytes' peak areas.

Obviously, the influence of the facts stated above increased with a larger amount of reagent on the filter. Therefore, the NBDPZ amount applied for impregnation was thoroughly optimized: a fifty-fold excess based on the complete filter area and on an uptake rate of 15 mL min⁻¹ during an 8 h experiment at 23 µg m⁻³ MIC was experimentally determined to be sufficient for quantitative collection of the analyte. This is equivalent to a range from LOD up to over 500 µg m⁻³ with 15 min sampling periods.

For the calculation of the minimum amount of reagent it must be considered that only one quarter of the filter surface is positioned beneath the holes and is thus accessible to the airborne analyte.

Based on the MIC concentration in the evaporation chamber determined by the reference method and on the results obtained by analyzing the diffusive samplers, an uptake rate was determined for all experiments. The amount of MIC collected on the control part always accounts for about 10% of the sample part's result, which is owing to a leakage into the diffusive sampler that cannot be avoided, as the sampler is not completely tight. However, this is not crucial as the amount found on the control part is always to be subtracted from the amount found on the sample filter, and the sampling rate is determined under these conditions. Fig. 4 shows chromatograms of a typical quantification of both filters from a diffusive sampler exposed to 19 µg m⁻³ of MIC for 6 h.

Fig. 4 Chromatograms of an eluted sample filter (dotted line) and an eluted control filter (continuous line) from a diffusive sampler exposed to 19 µg m⁻³ MIC (8 ppb) for 6 h.

The mean sampling rate for all 15 min experiments was determined to 15.0 mL min⁻¹ with a relative standard deviation of 9.9% for 56 experiments. Only 15 min experiments yielded acceptable results at 80% RH. Under these conditions, 8 h experiments led to sampling rates of ∼1 mL min⁻¹. All results of the LC-FLD analysis are listed in Table 2.

Table 2 MIC concentrations of test atmospheres obtained by the pumped reference method and the uptake rate of the NBDPZ diffusive samplers determined with fluorescence detection

Measurement	ST^a/min	RH^b (%)	Pumped reference method		Diffusive sampler
			c (MIC)/µg m^−3	RSD^c (N = 6) (%)	SR^d/mL min^−1	RSD (%)	N^e
a ST, sampling time.b RH, relative humidity.c RSD, relative standard deviation.d SR, sampling ratee N, number of experiments.
1	15	20	25.8	7.3	15.8	6.1	9
2	15	80	27.8	2.5	14.8	8.9	6
3	15	20	34.7	2.3	13.7	2.8	6
4	15	20	17.8	2.0	14.4	1.8	6
5	15	20	4.4	8.6	14.9	3.9	6
6	15	80	15.1	6.2	13.6	15.8	6
7	15	80	3.3	4.5	15.1	13.8	5
8	16	80	32.4	1.9	15.5	6.9	6
9	15	80	16.7	2.8	14.1	10.3	6
10	356	20	18.9	3.3	15.0	2.9	9
11	480	20	1.4	7.0	8.4	2.1	6
12	480	20	17.5	2.1	15.9	3.8	6
13	480	80	1.2	6.9	1.0	13.9	6
14	480	80	16.8	3.4	1.1	15.6	6

---

The sampling rate for all experiments (including long term sampling) performed at 20% RH was found to be 15.0 mL min⁻¹, with a relative standard deviation of 6.3% (N = 42) (Table 2). The limit of quantification (LOQ) was 3 µg m⁻³ for 15 min sampling periods and 0.2 µg m⁻³ for 8 h sampling periods using fluorescence detection. The LOQ was determined as ten times the mean standard deviation of six control filters exposed to MIC concentrations of 4.4 µg m⁻³ (for 15 min experiments) and 1.4 µg m⁻³ (for 8 h experiments).

It could be shown that humidity did not affect the stability of the formed MIC derivative but the reagent itself, thus leading to decreased sample capacity of the filters. This is likely to be a physical problem, as the NBDPZ might be flushed from the filter surface into the filter. This assumption is supported taking visual aspects into account: impregnated filters (orange from NBDPZ, originally white) that were exposed to humidified air showed the pattern of the samplers' holes mapped in white on the filter surface. Table 3 shows the results of the experiment to examine the behavior of coated filters towards humidity: The samplers of series A were exposed to MIC only, those of series B to MIC and subsequently to humidified air, those of series C first to humidified air and then to MIC, and those of series D again only to MIC. Each series comprised 4 samplers; MIC exposure was set to 30 µg m⁻³ for 30 min at 20% RH, and exposure to humidity was for 10 h at 80% RH. When the data sets were compared statistically, the Student t-test showed that the means of A, B and D were not significantly different at the 95% level.

Table 3 Diffusive samplers exposed to MIC atmospheres without (A, D), before (B) and after (C) zero exposure to humidified air (80% RH) for 10 h

Series (N = 4)	MIC–NBDPZ (peak area units)
A	66600 ± 4200
B	60500 ± 3500
C	18000 ± 1200
D	63500 ± 2700

---

The addition of propionic anhydride to the sample solution prior to analysis resolved the tailing problem, but the analysis time could not be reduced because the reaction products mentioned above were still present. The mean uptake rate was about the same as before, but the standard deviation was higher. The sampling rate was 14.8 mL min⁻¹, with an RSD of 10.6% for all 15 min experiments (N = 56) and 14.6 mL min⁻¹ with an RSD of 8.9% for all experiments performed at 20% RH (N = 42) (Table 4). It is not always reasonable to use this procedure, but in some cases it offers the possibility of improving the results: If the filter matrix is for any reason worse than is illustrated in Fig. 3, the analyte peak might not be baseline separated from the previous peak (or could lay within the tailing) and could thus be difficult to quantify. To create such a situation, some filters were dried overnight in a gentle stream of filtered air after impregnation. These filters were spiked with MIC–NBDPZ and analyzed in the same way as exposed diffusive samplers, first without and subsequently with addition of propionic anhydride. The standard concentrations were chosen such that numbers 3 and 4 equalled diffusive samplers exposed for 15 min to MIC concentrations at twice and exactly the TLV, respectively. As the results from Table 5 show, the recovery was much better after addition of PrAn, especially at low concentrations.

Table 4 Sampling rate of the NBDPZ diffusive samplers determined with fluorescence detection after addition of propionic anhydride

Measurement	Diffusive sampler
	SR/mL min^−1^a	RSD (%)^b
a SR, sampling rate.b RSD, relative standard deviation.
1	15.5	6.4
2	15.9	5.9
3	14.5	4.3
4	14.7	6.2
5	12.6	11.0
6	14.0	11.0
7	15.2	5.2
8	15.4	8.0
9	13.7	5.8
10	15.3	3.9
11	8.8	3.1
12	14.9	4.5

---

Table 5 Recovery of MIC-NBDPZ spiked samples (complex filter matrix) with and without addition of propionic anhydride (known concentration set to 100%)

Number	MIC-NBDPZ concentration/mol L^−1	Recovery (direct analysis)	Recovery (after PrAn was added)
1	8.9 × 10^−7	86.6%	95.6%
2	8.9 × 10^−8	58.5%	95.8%
3	3.6 × 10^−8	37.8%	98.2%
4	1.8 × 10^−8	14.3%	97.8%
5	7.1 × 10^−9	0%	91.1%

---

MS-MS detection

If tandem mass spectrometry is used for detection, the selectivity is much higher. In this case, there are no tailing or coelution problems, and the analysis time can be reduced to less than 10 min per run. The MS analysis was performed in the APCI(+)-mode as SRM on the transition [MIC-NBDPZ+H]⁺ (m/z 307) to [NBDPZ+H]⁺ (m/z 250) (Fig. 5).

Fig. 5 LC-MS-MS analysis (APCI(+)) of an eluted sample filter from a diffusive sampler exposed to 10 ppb MIC (24 µg m⁻³) for 15 min.

With MS-MS-detection, the sampling rate was determined to 11.8 mL min⁻¹ with an RSD of 15.3% for all MS experiments (N = 56) (Table 6). This was significantly lower than the uptake rate obtained on the basis of fluorescence detection. That behavior could not be explained as always the same samples were analyzed with both methods. Since the deviation always showed the same tendency, the results could be evaluated without problems. The LOQ was 1.3 µg m⁻³ for 15 min sampling periods and 0.15 µg m⁻³ for 8 h experiments.

Table 6 Sampling rate of the NBDPZ diffusive samplers determined by means of tandem mass spectrometry

Measurement	Diffusive sampler
	SR/mL min^−1^a	RSD (%)^b
a SR, sampling rate.b RSD, relative standard deviation.
1	11.8	12.2
3	12.0	6.2
4	10.8	7.3
5	11.3	5.4
6	10.0	6.4
7	10.0	4.6
8	14.9	8.5
9	11.4	5.6
11	7.1	5.8
12	13.8	10.9

---

Shelf life

The shelf life of coated filters was investigated by storing impregnated filters for different time periods at room temperature prior to sampling. The recovery was 83% after 5 days, 60% after 11 days and only 17% after 50 days, with the results obtained from immediately prior to sampling impregnated filters set to 100%. Therefore, the storage time of impregnated filters prior to sampling must be minimized. No negative effect, e.g. loss of analyte was observed while storing the exposed samplers for five days at room temperature after sampling. This was expected, as it is known from the literature and from own experiments not shown in this paper that the MIC-NBDPZ derivative is very stable in solution and as solid compound over the complete period observed (more than 40 days).¹⁸

Conclusion

The experiments performed showed that the derivatization reaction of NBDPZ with methyl isocyanate can be used for collection of MIC in diffusive sampling devices with reagent-coated filters. As mentioned above, there is no passive sampling method known in the literature, which allows the determination of airborne MIC by means of HPLC-MS-MS and HPLC-FLD. A great advantage of the NBDPZ method is therefore the possibility to apply fluorescence detection for analysis in addition to the use of a more complicated and costly LC-MS-MS system to obtain the required sensitivity. A problem is the strong dependence upon humidity, which inhibits long sampling periods if high relative humidity is present. The reduced shelf life limits the use of this diffusive sampler to applications that allow sampling within about one week from impregnation whereas the storage time from sampling to analysis is less problematic.

In summary, the developed method meets the requirements for a rapid determination if MIC concentrations in workplace atmospheres are in the range of the threshold values. Despite shelf-life problems, the NBDPZ diffusive sampler is well suitable for screening purposes.

References

1. H. Ulrich, Chemistry and Technology of Isocyanates, John Wiley & Sons, Chichester, 1996.
2. Office of Environmental Health Hazard Assessment, Chronic Toxicity Summary, Methyl Isocyanate, OEHHA, California, USA, 2001.
3. N. Andersson, M. K. Ajwani, S. Mahashabde, M. K. Tiwari, M. K. Muir, V. Mehra, K. Ashiru and D. C. Mackenzie, Br. J. Ind. Med., 1990, 47, 553–558.
4. B. K. Bhattacharya, S. K. Sharma and D. K. Jaiswal, Biochem. Pharmacol., 1988, 37, 2489–2493.
5. J. S. Ferguson, A. L. Kennedy, M. F. Stock, W. E. Brown and Y. Alarie, Toxicol. Appl. Pharmacol., 1988, 94, 104–117.
6. D. R. Varma, Environ. Health Perspect., 1987, 72, 153–157.
7. J. Elms, P. N. Beckett, P. Griffin and A. D. Curran, Toxicol. in Vitro, 2001, 15, 631–634.
8. D. Karlsson, J. Dahlin, G. Skarping and M. Dalene, J. Environ. Monit., 2002, 4, 216–222.
9. D. Karlsson, M. Spanne, M. Dalene and G. Skarping, J. Environ. Monit., 2000, 2, 462–469.
10. J. Ekman, J. O. Levin, R. Lindahl, M. Sundgren and A. Östin, Analyst, 2002, 127, 169–173.
11. D. Karlsson, M. Dalene and G. Skarping, Analyst, 1998, 123, 1507–1512.
12. OSHA Analytical Laboratory, Method No. 54: Methyl Isocyanate (MIC), 1985, OSHA, Salt Lake City.
13. Methods for the Determination of Hazardous Substances, MDHS 25/3: Organic Isocyanates in air 1999, Health and Safety Executive (HSE), London.
14. R. P. Streicher, J. E. Arnold, C. V. Cooper and T. J. Fischbach, Am. Ind. Hyg. Assoc. J., 1996, 57, 905–913.
15. C. Sangö and E. Zimerson, J. Liq. Chromatogr., 1980, 2, 971–990.
16. P. von Zweigbergk, R. Lindahl, A. Östin, J. Ekman and J. O. Levin, J. Environ. Monit., 2002, 4, 663–666.
17. M. Vogel and U. Karst, German Patent DE 199 36 731, 1999.
18. M. Vogel and U. Karst, Anal. Chem., 2002, 74, 6418–6426.
19. R. Lindahl, J. O. Levin and M. Mårtensson, Analyst, 1996, 121, 1177–1181.
20. J. O. Levin, R. Lindahl and K. Andersson, J. Environ. Sci. Technol., 1986, 20, 1273–1276.
21. M.-L. Henriks-Eckerman, J. Välimaa and C. Rosenberg, Analyst, 2000, 125, 1949–1954.
22. CEN, European Committee for Standardization, EN 838, Workplace Atmospheres–Diffusive Samplers for the Determination of Gases and Vapours–Requirements and Test Methods, CEN, Brussels, 1995.
23. Arbetarskyddsstyrelsen, A. F. S. 2000:3, Hygieniska gränsvärden och åtgärder mot luftföroreningar, ISBN 91-7930-357-9, 2000, Stockholm (in Swedish).
24. Deutsche Forschungsgemeinschaft, MAK- und BAT-Werte-Liste der Senatskommission zur Prüfung gesundheitsschädlicher Arbeitsstoffe, VCH, Weinheim, 1996 (in German).

---