Annals of Occupational Hygiene Advance Access originally published online on July 7, 2004
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Ann. occup. Hyg., Vol. 48, No. 5, pp. 415-424, 2004
© 2004 British Occupational Hygiene Society
Published by Oxford University Press
Determination of Isocyanates in Air Using 1-(2-Methoxyphenyl)piperazine-impregnated Filters: Long-term Sampling Performance and Field Comparison with Impingers with Dibutylamine
1 Department of Occupational and Environmental Medicine, University Hospital, SE-221 85 Lund, Sweden; 2 National Institute for Working Life North, Department of Work and the Physical Environment, PO Box 7654, SE-907 13 Umeå, Sweden
Received 15 September 2003; in final form 22 December 2003; published online on 7 July 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT DISCUSSIONINTRODUCTIONMATERIALS AND METHODSRESULTSREFERENCES |
|---|
Isocyanates may be harmful to workers and methods for monitoring air exposure in the field are necessary. The main aim of this study was to study the field performance of a method using 1-(2-methoxyphenyl)piperazine (2MP)-impregnated filters, by side-by-side comparison of long-term sampling with consecutive short-term samplings and also by short-term comparisons with other methods. Apart from using 2MP-impregnated filters, air monitoring was also performed by a modified 2MP method (FINMP) and by an impinger method using dibutylamine (DBA), which was the reference method. For short-term sampling the compared methods performed equally well for 2,6-toluenediisocyanate (2,6-TDI) and for isocyanic acid. For 2,4-toluenediisocyanate (2,4-TDI), the DBA method gave
10% higher results according to linear regression than the 2MP method and for phenyl isocyanate, the DBA method gave significantly higher results than both the 2MP and FINMP methods. During long-term sampling (2–4 h) of TDI with the 2MP method, significantly lower levels were found compared with parallel sampling with consecutive short-term samplings. A time-dependent correction factor for long-term sampling was calculated to be 1.7 for 2,4-TDI and 1.5 for 2,6-TDI for 4 h sampling. The long-term sampling performance for other isocyanates was not studied. In conclusion, short-term monitoring shows that the 2MP method slightly underestimates the true air concentration for some of the isocyanates studied, but the error is relatively small considering the variation in exposure. For long-term monitoring the 2MP method can be applied for TDI but, since the method underestimates the concentrations, a correction factor is needed which needs to be corroborated further.
Keywords: diisocyanates; isocyanic acid; occupational exposure; toluenediisocyanate
|
DISCUSSION |
|---|
|
TOPABSTRACTDISCUSSIONINTRODUCTIONMATERIALS AND METHODSRESULTSREFERENCES |
|---|
In this study we have compared three methods for the determination of airborne 2,4-TDI, 2,6-TDI, NDI, MDI, IPDI, PI and ICA in different work environments by comparison of parallel samples, with emphasis on evaluation of the 2MP method. We have also evaluated the long-term sampling performance (2–4 h) of the 2MP method.
When comparing the two filter methods using 2MP with the DBA impinger method, the DBA method was considered the independent method. This might be controversial, as this method is also subject to errors, but there are several findings that argue for this. Firstly, DBA-isocyanate derivatives have been shown to be stable in toluene at room temperature for 3 weeks (Spanne et al., 1996), whereas 2MP-isocyanate derivatives have been shown to have limited stability (67–78% recovery) under these conditions (National Institute for Occupational Safety and Health, 1994). Secondly, the reaction rate with isocyanates has been shown to be faster for DBA compared with 2MP (Karlsson et al., 1998; Tremblay et al., 2003), even though Kuck et al. (1999) observed the opposite. Finally, DBA is used at a much higher concentration in the sampling devices than is 2MP on filters, thus avoiding side reactions of the isocyanates to a greater degree. For long-term sampling, though, the DBA method or other methods using solvents are not applicable. Instead, it may be used as a reference control for the long-term sampling method applied, by side-by-side short-term sampling, as performed in this study, which was carried out in parallel with a survey study (Sennbro et al., 2004) where personal exposure of workers was monitored by the 2MP method.
The comparative studies showed that the three methods performed more or less equally well when measuring 2,6-TDI. For 2,4-TDI, the correlation and consensus was poorer. The lower correlation was mainly due to five sample sets collected in one moulding plant inside an enclosed area, which were sampled on the same day with the same equipment, where the 2MP method gave considerably higher values than the DBA method. Either there might be something in this environment that causes this method difference or there were unobserved problems with the equipment that was used. If these values were excluded, the correlation coefficient for 2,4-TDI increased to 0.973 and the linear regression curve was calculated as y = 0.70 x x – 0.02, indicating a bigger difference between the two methods. These sample sets did not have the same influence for 2,6-TDI, as the measured concentrations were higher for 2,4-TDI than for 2,6-TDI. The reason for the less exact agreement for 2,4-TDI than 2,6-TDI is not obvious, but one possible explanation is that the higher reactivity of 2,4-TDI leads to more side reactions on the filters.
For NDI, there was a good correlation and consensus between the 2MP and DBA methods up to levels of 1 p.p.b., but very large deviations for two samples with higher levels.
In this study, the 2MP and DBA methods were only compared in two sample sets for determination of MDI, but the methods have previously been compared by Ekman et al. (2000), who observed no significant difference between the methods, and by Henriks-Eckerman et al. (2002), who concluded that the methods performed equally well in the field.
The most striking differences between the methods were observed for PI. The DBA method gave two and three times higher air levels than the FINMP and 2MP methods, respectively. The reason for this discrepancy could obviously be inter-laboratory systematic errors, since the DBA samples were analysed in a different laboratory from the other samples. However, since there was also a large difference (2-fold) between the PI concentrations measured using the 2MP and FINMP methods analysed in the same laboratory with the same standards, there is a strong indication that the methods differ. One probable reason for this is that PI may undergo side reactions on the filters to a greater degree than the other isocyanates measured, since it is more reactive (Ulrich, 1996). Also, since PI is more volatile than TDI, recovery on the filters is probably lower. Hence, the real exposure to PI could be strongly underestimated if monitored by the 2MP method. Ekman et al. (2000) found that the DBA method gave 70% higher air levels than the 2MP method for PI in MDI painting operations.
In recent years there has been growing interest in exposure to thermal degradation products from PUR (Henriks-Eckerman et al., 2000; Karlsson et al., 2001). Two of the most interesting isocyanates in this matter are ICA and MIC. The number of measurements performed in this study for this comparison was relatively few (n = 13), but the correlation and the agreement for ICA between the DBA and FINMP methods are good (r = 0.952, slope = 1.065). Since the results were obtained by two independent analysis methods, it would appear that both sampling methods measure ICA levels well. The exposure to MIC was low and could not be quantified. A previous comparison of the 2MP and DBA methods for determination of MIC in the work environment (Henriks-Eckerman et al., 2000) observed >25% higher levels found by the DBA method. For ICA, no previous comparisons of methods in the field have been reported.
Since instability of isocyanate-2MP derivatives has been observed (National Institute for Occupational Safety and Health, 1994) and performance in long-term sampling has been insufficiently described, field experiments with parallel sampling with different sampling times was performed. In these experiments significantly lower air levels were observed for the long-term samplings compared with the sum of the short-term samplings. The reason for this is not obvious. One reason could be that the isocyanate-2MP derivatives formed degrade on the filters during sampling and therefore we developed a model for this time-dependent loss. Earlier studies concerning the stability of isocyanate-2MP derivatives on filters have shown conflicting results, but in general, good stability. However, in these studies the samples had been stored in the dark in refrigerators or freezers (Robert and Simon, 1987; Henriks-Eckerman et al., 2000; Kääriä et al., 2001). Another possible explanation relates to the reaction kinetics of isocyanates and 2MP on the filters. During long-term sampling a slow reaction rate may lead to side reactions on the filters to a greater extent than for a short-term sampling, since in the latter case the filters are desorbed into 2MP solution earlier, in which the reaction may become more complete. It has previously been observed for field sampling of MDI by Karoly (1998) that samples desorbed in the field gave significantly higher results than parallel samples desorbed after transport to the laboratory. However, in that study the airborne MDI existed mainly as an aerosol, while in this study the samples were collected in mouldings where the TDI is believed to mainly exist as a vapour. Vapour phase isocyanates should not be physically hindered in reacting with 2MP reagent on the filters to the same extent as aerosol. We tried to understand this further by checking the short-term sampling comparisons between sampling with DBA and 2MP and plotted the ratio between the methods versus time and concentration, respectively, but we found no correlation (results not shown). Regardless of what reason there is for the long-term underestimations, a correction factor for the results obtained with 2MP is needed to obtain a better estimate of the true exposure levels. It seems that the correction factor is sampling time dependent and the model presented may be useful in long-term field measurements. In Health and Safety Laboratory (1999), a sampling time of up to 7.5 h is recommended, which in our model would result in a 2.5-fold underestimation of the true exposure levels. However, in our study there was a large standard deviation for the calculated loss and the model needs further elaboration and corroboration.
We have evaluated the short-term field performance of the 2MP method for the determination of airborne isocyanates by comparison with the DBA method. The largest difference in performance was observed for the determination of PI, but was also high for 2,4-TDI, which was underestimated by the 2MP method compared with the DBA method, while for 2,6-TDI there was a good correlation and consensus. For the determination of ICA there was good agreement between the FINMP and DBA methods.
For long-term sampling performance of the 2MP method in monitoring TDI, we observed significantly lower values compared with parallel consecutive sampling with shorter sampling times. From the results we have calculated a time-dependent correction factor for error adjustment. These results show that the 2MP method underestimates the true air concentration of TDI during long-term sampling.
In conclusion, short-term monitoring shows that the 2MP method slightly underestimates the true isocyanate air concentration. However, the error is relatively small considering the variation in exposure to isocyanates over a working day. The 2MP method can be applied for long-term monitoring, but since the method underestimates the concentration a correction factor is needed. A correction factor has been calculated in the present work, but needs to be corroborated further.
Acknowledgements—Thanks are due to Jonas Björk and Ulf Strömberg for statistical and mathematical consultations.The authors thank Arbetsmarknadens Försäkringsaktiebolag (AFA Foundation), Sweden and Forskningsrådet för Arbetsliv och Socialvetenskap (Swedish Council for Working Life and Social Research) for financial support of this work.
|
INTRODUCTION |
|---|
|
TOPABSTRACTDISCUSSIONINTRODUCTIONMATERIALS AND METHODSRESULTSREFERENCES |
|---|
It is essential to assess the occupational exposure to isocyanates since these substances may be harmful to workers, causing disorders mainly in the airways, such as occupational asthma or asthma-like symptoms (Vandenplas et al., 1993a; Baur et al., 1994). It has been shown that the pulmonary response to isocyanates is dependent on the total dose (Vandenplas et al., 1993b). Thus, it is important to assess whole day exposure. Exposure occurs in working environments where isocyanates or polyurethanes (PUR) are manufactured or processed. Different sampling methods and the challenges and complexity of exposure assessment of isocyanates have been reviewed by Streicher et al. (1994) and Levine (2002).
The present sampling techniques generally use a derivatization reagent to stabilize the sampled isocyanates and the sampling is usually performed with impregnated filters, impingers or paper tape instruments. After sampling, the isocyanate derivatives are analysed and quantified with appropriate calibration standards on chromatographic systems. An exception is when direct-reading paper tape methods are used where the derivatives are quantified according to their absorbance without separation. Apart from different sampling techniques, several different derivatization reagents are also used, most being secondary amines. In several studies, parallel sampling has been performed in order to compare different techniques and derivatization reagents for air monitoring of isocyanates, both on the laboratory scale and in the field (Streicher et al., 1994). In this study we have performed parallel sampling of airborne 2,4-toluene diisocyanate (2,4-TDI), 2,6-toluene diisocyanate (2,6-TDI), 1,5-naphthalene diisocyanate (NDI), 4,4'-methylenediphenyl diisocyanate (MDI), isophorone diisocyanate (IPDI), phenyl isocyanate (PI), methyl isocyanate (MIC) and isocyanic acid (ICA) in various work environments, using both different sampling techniques and derivatization agents. Three air monitoring methods were used: a filter method using 1-(2-methoxyphenyl)piperazine (2MP), referred to as the 2MP method (Health and Safety Laboratory, 1999; Östin et al., 2002), a modified filter method with 2MP, referred to as the FINMP method (Henriks-Eckerman et al., 2000), and an impinger method with dibutylamine (DBA) in toluene, referred to as the DBA method (Spanne et al., 1996; Karlsson et al., 2000). The DBA method was used as the reference method.
This study was also carried out in parallel with a survey study (Sennbro et al., 2004) where the personal 8 h time-weighted average (TWA) exposure was monitored for workers by the 2MP method. This method was used earlier in a study by Maitre et al. (1993), where the 8 h TWA exposures for workers exposed to TDI were monitored, and it is also within the recommendations of the Health and Safety Laboratory (1999) to use this method for long-term monitoring.
However, the long-term performance of the 2MP method is sparsely described. To evaluate long-term performance, parallel samplings using only the 2MP method with different sampling times were performed.
The aims of this study were to evaluate the field performance of this method and to test its accuracy by comparison with other methods.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTDISCUSSIONINTRODUCTIONMATERIALS AND METHODSRESULTSREFERENCES |
|---|
Study design
The short-term field performance of the 2MP and FINMP methods were evaluated by comparison with the DBA method, which was used as an independent method. The long-term field performance of the 2MP method was evaluated by parallel sampling with consecutive samples with shorter sampling times using the same method. Air sampling was performed in 13 different plants, described in detail by Sennbro et al. (2004). The plants included were two moulding plants using NDI and MDI, five moulding plants using TDI, two continuous foaming plants using TDI, two flame lamination plants using TDI PUR and two plants with low heating processes only using cold TDI or MDI. One of the moulding plants also used IPDI and at the flame lamination plants, where exposure arose from thermal decomposition of PUR, exposure to PI, MIC and ICA also occurred.
Chemicals
The chemicals used for analysis of the 2MP and the FINMP samples are described in Östin et al. (2002) except for tri-deuterated 1-(2-methoxyphenyl)piperazine (d3-2MP), which were purchased from Synthelec (Lund, Sweden).
The chemicals used for analysis of the DBA samples were toluene (HPLC grade) and acetonitrile (HPLC grade) from LabScan (Dublin, Ireland). Acetic acid (HOAc) and PI were from Merck (Darmstadt, Germany), while 2,6-TDI, IPDI and DBA were from Aldrich (Gillingham, UK). NDI was from Pfaltz & Bauer (Waterbury, CT), while 2,4-TDI and MDI were from Acros (Geel, Belgium). Octa-decadeuterated DBA (d18-DBA) was purchased from Cambridge Isotope Laboratories Inc. (Andover, MA). The DBA conjugates of MIC and ICA (DBA-MIC and DBA-ICA) and the d18-DBA conjugate of MIC were purchased from Synthelec (Lund, Sweden). All chemicals were of analytical grade, except where otherwise indicated.
Air sampling
Three methods were used for air monitoring of isocyanates: a filter method using 2MP, a modified filter method with 2MP and an impinger method using DBA. The first method used, referred to as the 2MP method, used single filters impregnated with 2MP. The filters used were glassfibre, 25 mm diameter, 1 µm pore size, from SKC Inc. (Eighty Four, PA), which were fixed in a 25 mm monitor case from Millipore Corp. (Bedford, MA). This sampling method has been described by the Health and Safety Laboratory (1999). The second method, referred to as the FINMP method, used double filters impregnated with 2MP. The filters used were GF/B glassfibre filters, 25 mm diameter, pore size 1.3 µm, from Whatman International Ltd (Maidstone, UK), which were fixed in 25 mm Swinnex filter holders from Millipore Corp. (Bedford, MA). Immediately after sampling by the 2MP or FINMP methods, the filters were transferred to glass vials containing 2MP reagent in toluene. This method has been described by Henriks-Eckerman et al. (2000). The third method, referred to as the DBA method, used an impinger flask containing 0.01 M DBA in toluene, in series with a filter. Midget glass impingers (30 ml) (Werner-Glas & Instrument, Stockholm, Sweden) and glassfibre filters (13 mm diameter, pore size 0.3 µm) (SKC Inc., Eighty Four, PA), fixed in a 13 mm Swinnex filter holder from Millipore Corp. (Bedford, MA), were used. Immediately after sampling, the impinger solution was transferred to test tubes with PFTE screw caps and the filter was transferred to a test tube of the same kind, containing 0.01 M DBA in toluene. This method has been described by Spanne et al. (1996) and Karlsson et al. (2000). After transport to the laboratory all samples were then stored at 4°C until analysis.
The pumps used for drawing air through the sampling units were Gilair Personal Air Samplers from Gilian Instruments Corp. (West Caldwell, NJ) or Escort Elf Pumps from Mine Safety Appliance Co. (Pittsburgh, PA). Calibrated GPE Meterate rotameters from Jencons-PLS (Leighton Buzzard, UK) were used for air flow rate measurements. The air flow rates through the sampling units were calibrated prior to sampling and checked after sampling. The pump flow rates were 1 l/min for all methods. The 2MP method was used in all plants, the DBA method at all plants except one moulding plant and the FINMP method was mainly used in the two flame lamination plants but also in one moulding plant and one plant using a low heating process.
In total 237 air samples were collected, of which 124, 86 and 27 were sampled using the 2MP, DBA and FINMP method, respectively. The sampling times ranged from 3 to 264 min (median 21 min). The samples were divided in 109 sample sets; in each sample set parallel sampling had been performed. When performing parallel sampling, the inlets of the sampling devices were placed side-by-side 
5 cm apart and the sampling times were equal for all methods. The samples and sample sets are summarized in Fig. 1.
|
Analysis and quantification
The air levels of 2,4-TDI, 2,6-TDI, IPDI, MDI, NDI and PI were analysed by all three methods. MIC and ICA were included with the DBA and FINMP methods.
Standards for quantification of the samples using the 2MP method were prepared according to the procedure described in Health and Safety Laboratory (1999). The deuterated internal standards were prepared in a similar way to their non-deuterated homologues but by small scale synthesis (Ekman et al., 2002). Analysis was performed by liquid chromatography combined with tandem mass spectrometry (LC-MS/MS) as described by Östin et al. (2002), except for the applied internal standard. The mobile phase started with a 4 min isocratic mode (60:40 acetonitrile/water) followed by a 10 min gradient to 95% acetonitrile. The column was a Grom-sil 80 ODS-7 (200 x 3 mm, 4 µm particle size) from Grom Analytik+HPLC GmbH (Herrenberg, Germany) and the flow rate was 0.4 ml/min. The analytes were then analysed in an API 2000 triple quadropole instrument from PE Biosystems (Foster City, CA) by electrospray ionization in positive mode with an ion spray voltage of 5500 V at 320°C at atmospheric pressure. As internal standards, d3-2MP bound to hexamethylene diisocyanate was used for 2,4-TDI, 2,6-TDI, NDI and PI, while d3-2MP bound to MDI was used for MDI and IPDI. For ICA and MIC, d3-2MP bound to ICA and MIC, respectively, were used as internal standards. For the FINMP samples, the same analysis procedure as for the 2MP samples was applied. The limit of quantification (LOQ) was 20 ng/sample for each isocyanate.
Standards for quantification of the samples using the DBA method were prepared as described by Karlsson et al. (1998). The deuterated internal standard was synthesized in the same way except that d18-DBA was used instead of DBA. Analysis was performed by LC-MS/MS using an isocratic method with a 47/53 mixture of water with 0.5% HOAc/acetonitrile as the mobile phase. The column was a HypurityTM Advance (150 x 3 mm, 5 µm particle size) from ThermoQuest (Runcorn, UK) and the flow rate was 0.5 ml/min. The analytes were then analysed in an API 3000 triple quadropole instrument from PE Biosystem (Foster City, CA) by electrospray ionization in positive mode with an ion spray voltage of 3500 V at 375°C at atmospheric pressure. For quantification, the internal standards used were d18-DBA-MIC for ICA and MIC, di-(d18-DBA)-2,6-TDI for 2,6-TDI and di-(d18-DBA)-2,4-TDI for the other isocyanates. The LOQ was 20 ng/sample for ICA and 2 ng/sample for the other isocyanates. The results for the DBA samples for each isocyanate are given as the sum of the levels in the impinger and on the filter.
Comparison of methods by short-term sampling
For 96 of the sample sets, consisting in total of 209 samples, parallel sampling was performed using two or three different methods. Only sample sets, where quantitative data was obtained for each method and isocyanate, were used for comparison of methods. The agreement of the methods was analysed by paired t-tests, Wilcoxon signed ranks test and linear regression.
Long-term sampling
In order to study the long-term sampling performance of the 2MP method, a total of 13 sample sets, each consisting of one long-term sample and between two and four consecutive short-term samples, were collected at four moulding plants where TDI was used. Of a total of 34 2MP samples in these sample sets, six were also used for the comparison of methods. The 13 different sample sets are described in Table 1 and a schematic description of the experimental design is shown in Fig. 2. The sum of amounts of TDI sampled by the consecutive short-term samples [mS(total)] and the amount of TDI sampled by the long-term sample (mL) were compared by a paired t-test and Wilcoxon signed ranks test.
|
|
Two different models for the loss of TDI were fitted to the obtained field data, assuming that the losses per unit time were the same in both the long- and short-term samples.
In the first model, assuming that loss was by a first order process, the relation between mL and the amount of TDI sampled by the consecutive short-term samplings (mS(i)), as shown in Fig. 2, is
where k is the first order constant, ti is the time from the end of the short-term sampling to the end of the long-term sampling, mS(i) is the amount sampled for the short-term sampling between time ti and t(i – 1) and n is the number of short-term samples in the sample set. The constant e–k was determined for the 13 observations and set to the average e–k.
In the second model, assuming that loss followed a zero order process, the relation between mL and mS(i), as shown in Fig. 2, is
where 
m is the zero order constant and ti is the time from the end of the short-term sampling to the end of the long-term sampling. The constant m was determined for the 13 observations and set to the average m.
By comparing the observed mL with the back-calculated values for mL for each model, the best model was chosen. From the chosen model, the time-dependent correction factor was calculated for long-term sampling of TDI.
Correction factor for long-term sampling of TDI
Assuming an environment with constant air concentrations of TDI, the true amount of TDI [mL(TRUE)] collected on a filter without correction for losses will increase linearly with time according to
where F is the mass flow of airborne TDI drawn through the filter and t is the sampling time.
Introducing the first order loss process model, mL is related to the sum of a number (n) of short-term samples times according to equation (1). If n is increased towards infinity and the time intervals for short-term sampling are equal, the losses during each short-term sampling are negligible and the amount sampled for each short-term sampling will be mS(i) = F x dt and the observed mL will equal
or
Inserting the sampling time t into equation (3) and substituting F in equation (5) gives
Dividing both sides by the sampling volume V gives
where C(TRUE) is the true air concentration of TDI, C is the observed air concentration of TDI and (k x t)/(1 – e–kt) is a time-dependent correction factor for long-term sampling.
|
RESULTS |
|---|
|
TOPABSTRACTDISCUSSIONINTRODUCTIONMATERIALS AND METHODSRESULTSREFERENCES |
|---|
Comparison of methods by short-term sampling
Nineteen of the 96 sample sets for method comparison were excluded from the statistical analysis. In 15 of these cases the sampled amounts using the 2MP and FINMP methods were below the LOQ for all isocyanates, in one case the sampled amounts using the DBA method were below the LOQ for all isocyanates, in two cases the sampling times were different due to pump failure and in one case the levels were too high for determination by both the DBA and 2MP methods.
Systematic errors
The results of the statistical tests are shown in Table 2. There were significant differences (P < 0.05) between the 2MP method and the other two methods for 2,4-TDI and PI. For 2,6-TDI, NDI, IPDI and ICA no significant differences were observed. Only two sample sets with quantifiable levels of MDI were obtained, in which the levels determined by the 2MP method were 157 and 99% of the levels determined by the DBA method. Sampling of MIC could not be evaluated since the levels were below the LOQ in the plants where the DBA and FINMP methods were used in parallel.
|
Linear regression
When included in the linear regression analysis, the DBA method result was set as the independent variable (x). When the 2MP and FINMP method results were compared with each other, the 2MP method was set as the independent variable.
2,4-TDI. For 2,4-TDI, the comparisons of the 2MP method versus the DBA method are shown in Fig. 3a. For comparisons of the FINMP method versus the DBA method, 12 sample sets were used and the range of 2,4-TDI levels was 0.09–13 p.p.b. The calculated regression curve was y = 0.83 x x + 0.5 and the correlation was significant (P < 0.001) with R = 0.91. For comparisons of the FINMP method versus the 2MP method, 25 sample sets were used and the range of 2,4-TDI levels was 0.01–9 p.p.b. The calculated regression curve was y = 1.1 x x + 0.1 and the correlation was significant (P < 0.001) with R = 0.95.
|
2,6-TDI. For 2,6-TDI, the comparisons of the 2MP method versus the DBA method are shown in Fig. 3b. For comparisons of the FINMP method versus the DBA method, 12 sample sets were used and the range of 2,6-TDI levels was 0.4–15 p.p.b. The calculated regression curve was y = 1.0 x x + 0.2 and the correlation was significant (P < 0.001) with R = 0.98. For comparisons of the FINMP method versus the 2MP method, 26 sample sets were used and the range of 2,6-TDI levels was 0.02–15 p.p.b. The calculated regression curve was y = 0.99 x x – 0.1 and the correlation was significant (P < 0.001) with R = 0.98.
IPDI. For IPDI, only comparisons of the 2MP method versus the DBA method were performed. Seven sample sets were used and the range of IPDI levels was 0.05–2 p.p.b. The calculated linear regression curve was y = 0.68 x x – 0.001 and the correlation was significant (P < 0.001) with R = 0.99.
NDI. For NDI, only comparisons of the 2MP method versus the DBA method were performed. Seven sample sets were used and the range of NDI levels was 0.2–24 p.p.b. The calculated regression curve was y = 0.22 x x + 0.7 and the correlation was significant (P = 0.027) with R = 0.81. If the two samples with highest levels of NDI (>4 p.p.b.) were excluded, the range of the remaining five samples was 0.2–1.7 p.p.b. Separate analysis of these samplings resulted in a regression curve of y = 1.1 x x + 0.1 and the correlation was significant (P = 0.003) with R = 0.98.
MDI. Since only two sample sets with quantifiable levels of MDI were obtained, no linear regression analysis was performed.
PI. For PI, the comparisons of the 2MP and FINMP methods versus the DBA method are shown in Fig. 4. For comparisons of the FINMP method versus the 2MP method, 15 sample sets were used and the range in PI levels was 0.03–2.5 p.p.b. The calculated regression curve was y = 1.7 x x + 0.03 and the correlation was significant (P < 0.001) with R = 0.97.
|
ICA and MIC. For ICA, the comparisons of the FINMP method versus the DBA method are shown in Fig. 5. No sample sets with quantifiable levels of MIC were obtained in the plants where the DBA and FINMP methods were used in parallel and, hence, no comparisons could be performed for this compound. ICA and MIC could not be sampled by the 2MP method and, hence, no comparisons could be performed with this method.
|
Long-term sampling
The results in the 13 sample sets collected for evaluation of the long-term sampling performance of the 2MP method for TDI are shown in Table 3. It was shown with both paired t-tests and Wilcoxon signed rank test that significantly higher amounts were sampled by consecutive short-term sampling compared with the corresponding long-term sample for both 2,4-TDI (P < 0.001 and P = 0.001, respectively) and 2,6-TDI (P = 0.001 and P = 0.004, respectively). The ratio of the sum of short-term samplings and the long-term sampling was on average 131% (range 102–174%) for 2,4-TDI and 124% (range 90–163%) for 2,6-TDI.
|
The calculated first order constant e–k, was on average 0.74 (SD = 0.2, n = 13) and 0.79 (SD = 0.2) for 2,4- and 2,6-TDI, respectively, which corresponds to k = 0.30 h–1 and 0.23 h–1. For the first order process model, the ratio between the back-calculated values of mL and the observed values was on average 105% (range 77–141%) for 2,4-TDI and 104% (range 83–137%) for 2,6-TDI.
The calculated zero order constant m was on average 0.04 µg/h (SD = 0.1) and 0.02 µg/h (SD = 0.03) for 2,4-TDI-2MP and 2,6-TDI-2MP derivatives, respectively. For the zero order process model, the ratio between the back-calculated values of mL and the observed values was on average 171% (range –90–590%) of the observed values for 2,4-TDI and 143% (range 91–460%) for 2,6-TDI.
Correction factor for long-term sampling
The first-order process model was chosen for calculation of a correction factor for long-term sampling of TDI by the 2MP method, since it was shown to best fit the observed data, having the best accuracy of back-calculated values of mL. The correction factor for long-term sampling by the 2MP method is, according to equation (7),
for 2,4-TDI, and
for 2,6-TDI.
This means that for a 4 h sampling, as performed in the survey study by Sennbro et al. (2004), the air levels are underestimated by a factor of 1.7 for 2,4-TDI and by a factor of 1.5 for 2,6-TDI.
|
FOOTNOTES |
|---|
* Author to whom correspondence should be addressed. E-mail: hakan.tinnerberg{at}ymed.lu.se
|
REFERENCES |
|---|
|
TOPABSTRACTDISCUSSIONINTRODUCTIONMATERIALS AND METHODSRESULTSREFERENCES |
|---|
Baur X, Marek W, Ammon J, et al. (1994) Respiratory and other hazards of isocyanates. Int Arch Occup Environ Health; 66: 141–52.[CrossRef][Web of Science][Medline]
Ekman J, Levin J-O, Lindahl R, et al. (2000) Comparison of isocyanate sampling methods. Abstract presented at 1st Isocyanate Conference 2000, Stockholm, Sweden.
Ekman J, Levin J-O, Lindahl R, Sundgren M, Östin A. (2002) Comparison of sampling methods for 1,6-hexamethylene diisocyanate (HDI), in a commercial box. Analyst; 127: 169–73.[Medline]
Health and Safety Laboratory. (1999) Organic isocyanates in air, MDHS 25/3. Sudbury: HSE Books.
Henriks-Eckerman M-J, Välimää J, Rosenberg C. (2000) Determination of airborne methyl isocyanate as dibutylamine or 1-(2-methoxyphyl)piperazine derivatives by liquid and gas chromatography. Analyst; 125: 1949–54.[Medline]
Henriks-Eckerman M-J, Välimää J, Rosenberg C, Peltonen K, Engström K. (2002) Exposure to airborne isocyanates and other thermal degradation products at polyurethane-processing workplaces. J Environ Monit; 4: 717–21.[CrossRef][Web of Science][Medline]
Kääriä K, Hirvonen A, Norppa H, Piirilä P, Vainio H, Rosenberg C. (2001) Exposure to 2,4- and 2,6-toluene diisocyanate (TDI) during production of flexible foam: determination of airborne TDI and urinary 2,4- and 2,6-toluenediamine (TDA). Analyst; 126: 1025–31.[CrossRef][Medline]
Karlsson D, Spanne M, Dalene M, Skarping G. (1998) Determination of complex mixtures of airborne isocyanates and amines. Part 4. Determination of aliphatic isocyanates as dibutylamine derivatives using liquid chromatography and mass spectrometry. Analyst; 123: 117–23.[CrossRef]
Karlsson D, Spanne, Dalene M, Skarping G. (2000) Airborne thermal degradation products of polyurethane coatings in car repair shops. J Environ Monit; 2: 462–9.[CrossRef][Web of Science][Medline]
Karlsson D, Dalene M, Skarping G, Marand A. (2001) Determination of isocyanic acid in air. J Environ Monit; 3: 432–6.[Medline]
Karoly WJ. (1998) Stability studies of diphenylmethane diisocyanate (MDI) on glass fiber filters. Am Ind Hyg Assoc J; 59: 645–7.
Kuck M, Ballé G, Slawyk W. (1999) Sampling of diisocyanates (HDI, TDI) in air by derivatisation with secondary amines as reagents. Part 1. Partial rate factors (PRF) of reagents. Analyst; 124: 933–9.[CrossRef]
Levine SP. (2002) Critical review of methods for sampling, analysis, and monitoring of vapour-phase toluene diisocyanate. Appl Occup Environ Hyg; 17: 878–90.[Medline]
Maitre A, Berode M, Perdrix A, Romazini S, Savolainen H. (1993) Biological monitoring of occupational exposure to toluene diisocyanate. Int Arch Occup Environ Health; 65: 97–100.[CrossRef][Web of Science][Medline]
National Institute for Occupational Safety and Health. (1994) Manual of analytical methods, 4th edn, Method 5521, isocyanates, monomeric. Cincinnati, OH: NIOSH.
Östin A, Sundgren M, Ekman J, Lindahl R, Levin J-O. (2002) Analysis of isocyanates with LC-MS/MS. In Lesage J, DeGraff ID, Danchik RS, editors, Isocyanates: sampling, analysis and health effects, American Society for Testing and Materials Special Technical Publication 1408. West Conshohocken, PA: American Society for Testing and Materials.
Robert A, Simon P. (1987) A solvent-free sampling method for airborne toluene diisocyanate. Chromatographia; 23: 507–11.
Sennbro CJ, Lindh CH, Östin A, Welinder H, Jönsson BAG, Tinnerberg H. (2004) A survey of airborne isocyanate exposure in 13 Swedish polyurethane industries. Ann Occup Hyg; 48: 405–14.
Spanne M, Tinnerberg H, Dalene M, Skarping G. (1996) Determination of complex mixtures of airborne isocyanates and amines. Part 1. Liquid chromatography with ultraviolet detection of monomeric and polymeric isocyanates as their dibutylamine derivatives. Analyst; 121: 1095–9.[CrossRef]
Streicher RP, Kennedy ER, Lorberau CD. (1994) Strategies for the simultaneous collection of vapours and aerosols with emphasis on isocyanate sampling. Analyst; 119: 89–97.
Tinnerberg H, Dalene M, Skarping G. (1997) Air and biological monitoring of toluene diisocyanate (TDI) in a flexible foam plant. Am Ind Hyg Assoc J; 58: 229–35.[Medline]
Tremblay P, Lesage J, Ostiguy C, Van Tra H. (2003) Investigation of the competitive rate of derivatisation of several secondary amines with phenyl isocyanate (PHI), hexamethylene-1,6-diisocyanate (HDI), 4,4'-methylenebis(phenyl isocyanate) (MDI) and toluene diisocyanate (TDI) in liquid medium. Analyst; 128: 142–9.[Medline]
Ulrich H. (1996) Chemistry and technology of isocyanates. Chichester: John Wiley & Sons.
Vandenplas O, Malo J.-L, Saetta M, Mapp CE, Fabbri LM. (1993a) Occupational asthma and extrinsic alveolitis due to isocyanates: current status and perspectives. Br J Ind Med; 50: 213–28.[Web of Science][Medline]
Vandenplas O, Cartier A, Ghezzo H, Cloutier Y, Malo J-L. (1993b) Response to isocyanates: effect of concentration, duration of exposure, and dose. Am Rev Respir Dis; 147: 1287–90.[Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  K. W. Fent, L. G. T. Gaines, J. M. Thomasen, S. L. Flack, K. Ding, A. H. Herring, S. G. Whittaker, and L. A. Nylander-French Quantification and Statistical Modeling--Part I: Breathing-Zone Concentrations of Monomeric and Polymeric 1,6-Hexamethylene Diisocyanate Ann. Hyg., October 1, 2009; 53(7): 677 - 689. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
I. Liljelind, C. Norberg, L. Egelrud, H. Westberg, K. Eriksson, and L. A. Nylander-French Dermal and Inhalation Exposure to Methylene Bisphenyl Isocyanate (MDI) in Iron Foundry Workers Ann. Hyg., September 26, 2009; (2009) mep067v1. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Tinnerberg and C. Mattsson Usage of Air Monitoring and Biomarkers of Isocyanate Exposure to Assess the Effect of a Control Intervention Ann. Hyg., April 1, 2008; 52(3): 187 - 194. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. J. SENNBRO, C. H. LINDH, A. OSTIN, H. WELINDER, B. A. G. JONSSON, and H. TINNERBERG A Survey of Airborne Isocyanate Exposure in 13 Swedish Polyurethane Industries Ann. Hyg., July 1, 2004; 48(5): 405 - 414. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||