Annals of Occupational Hygiene Advance Access originally published online on August 3, 2004
Annals of Occupational Hygiene 2004 48(6):499-507; doi:10.1093/annhyg/meh048
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
© 2004 British Occupational Hygiene Society Published by Oxford University Press;
Evaluating Exposures to Complex Mixtures of Chemicals During a New Production Process in the Plastics Industry
1 Environmental and Occupational Health Division, Institute for Risk Assessment Sciences, Utrecht University, PO Box 80176, 3508 TD Utrecht, The Netherlands; 2 Department of Public Health Sciences, Faculty of Medicine and Dentistry, The University of Alberta, Edmonton, Alberta, Canada; 3 Department of Chemical Exposure Assessment, TNO Chemistry, Zeist, The Netherlands; 4 Laboratory of Organic Chemistry, Wageningen University, Wageningen, The Netherlands
* Author to whom correspondence should be addressed. Tel: +31 30 2539440; fax: +31 30 253 5077; e-mail: h.kromhout{at}iras.uu.nl
Received 7 March 2003; in final form 3 February 2004
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The goal of this study was to monitor emission of chemicals at a factory where plastics products were fabricated by a new robotic (impregnated tape winding) production process. Stationary and personal air measurements were taken to determine which chemicals were released and at what concentrations. Principal component analyses (PCA) and linear regression were used to determine the emission sources of different chemicals found in the air samples. We showed that complex mixtures of chemicals were released, but most concentrations were below Dutch exposure limits. Based on the results of the principal component analyses, the chemicals found were divided into three groups. The first group consisted of short chain aliphatic hydrocarbons (C2–C6). The second group included larger hydrocarbons (C9–C11) and some cyclic hydrocarbons. The third group contained all aromatic and two aliphatic hydrocarbons. Regression analyses showed that emission of the first group of chemicals was associated with cleaning activities and the use of epoxy resins. The second and third group showed strong association with the type of tape used in the new tape winding process. High levels of CO and HCN (above exposure limits) were measured on one occasion when a different brand of impregnated polypropylene sulphide tape was used in the tape winding process. Plans exist to drastically increase production with the new tape winding process. This will cause exposure levels to rise and therefore further control measures should be installed to reduce release of these chemicals.
Keywords: air sampling • exposure assessment • plastics production • principal component analysis • thermal degradation • thermoplastics
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The plastics industry has been constantly developing and growing for the last 20 years. The industry uses a wide variety of materials and production processes, each with its own particular techniques and working conditions. Many of the production processes are now fully automated continuous processes. However, manual labour is still extensively employed in small factories and when manufacturing specialized products. These small-scale processes also tend to be subject to more technical development, which can introduce new types of occupational exposure.
The company where the study was conducted specialized in the development of products made from high quality plastic materials (thermoplastics and thermo sets). Recently, a new production process has been developed, which enables the use of pre-impregnated fibre tapes for the production of different thermoplastic products. This process, called ‘impregnated filament or tape winding’, is an automatic process in which a robot winds the fibre tape on a rotating mandrel while heating the tape with a propane gas-fired burner. During our study this process was being tested in the production department of the company. The process was only used to produces pipes (Ø 10–15 cm) to test several physical properties (bending flexibility, etc.) of the pipes when produced with different tapes. The new process was not in constant use and only small amounts of tape were processed during a run (10–50 kg). During these short runs different tapes and heating temperatures were tested, as well as different ways of tape winding to achieve the required quality of product. One operator was constantly supervising the process and made adjustments during the runs where necessary. In the near future, there are plans to use the process for continuous production of thermoplastic products, possibly running multiple machines at the same time.
The heating and incidental burning of the tape can cause complex chemical fumes and vapours to be released. The main chemicals released when processing plastics are aliphatic hydrocarbons (Grote et al., 1984
; Forrest et al., 1994; Britton, 1998), aldehydes (Hoff et al., 1982; Frostling et al., 1984; Lewis, 1990), ketones and acids (Hoff et al., 1982; Frostling et al., 1984). Earlier studies also showed small concentrations of CO (Hoff et al., 1982; Lewis, 1990), SO2 (Nelson, 1973), benzene, styrene and xylene (Hoff et al., 1982; Forrest et al., 1994) being released during plastic processing. Although most of these chemicals are measured at low concentrations, generally a few µg/m3, little is known about the health effects of long-term exposure to low levels of mixtures of these chemicals. The material datasheets for the three thermoplastic impregnated tapes used indicated that some additional chemicals could be released during processing (Hoechst Celanese, 1997; BASF, 1999; General Electric, 2000).
The goals of this study were: (i) to identify chemicals emitted during the production of high quality plastics products with the new robotic ‘tape winding’ process; (ii) to quantify composition of fumes and vapours released and compare them with occupational exposure limits; (iii) to identify determinants of exposure with the aim of suggesting exposure controls.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Air monitoring
Stationary sample sites were chosen in such a way that the entire production hall and all potential exposure sources present were covered (Fig. 1).
|
Generally, three people worked in the hall, one operator at the tape winding unit and two at other production processes. The operator of the tape winding process was measured at every run, while the two others workers took turns during the measurements (most of the time they performed the same task). Although measurements were done at several sites in the production hall, the main focus was on the tape winding process. The other processes were monitored to adjust measured concentrations at the new tape winding unit. Samples were taken during runs with three different types of tapes, polypropylene (PP), polyethylimide (PEI) polypropylene sulphide (PPS). Four runs per tape were sampled.
Sampling with charcoal and Tenax tubes
Samples were collected with active charcoal tubes (SKC) through which air was drawn at 
200 ml/min using a Gillian GilAir 5 sampling pump. In addition, two simultaneous measurements with Tenax tubes (supplied by analysis lab), which could be analysed using gas chromatography–mass spectrometry (GC-MS) to detect a wider variety of chemicals at low concentrations, were taken for each of the three tape types on the tape winding robot, with air drawn through it at 50 ml/min. The duration of the production runs using tape winding determined the duration of the measurement, which was stopped at the end of each run. Because of the experimental status of the tape winding, the process runs, and therefore the measurements, lasted between 0.5 and 1.5 h.
One field blank (charcoal) was taken for every measurement day (on site), for a total of 6 days. They were used in combination with analytical limits of detection to estimate detection limits. For the Tenax tubes, no field blanks were collected. The detection limits were estimated by multiplying the analytical limits of detection by a factor of two for uncertainty, due to the fact that no calibration curves were constructed to certify the exact uncertainty of the GC-MS analysis.
Colour diffusion tubes
In addition to the active air measurements, passive semi-quantitative measurements for CO, H2S, NH3, HCN were taken in the fume cloud of the tape winding robot, with Draeger colour diffusion tubes.
The colour diffusion tubes were read directly after the measurements. Three runs were sampled with the colour diffusion tubes for every tape.
Chemical analyses
To analyse the charcoal tubes, the charcoal was removed and extracted with CS2. Analyses were done using gas chromatography with flame ionization detection (GC-FID) on a 60 m capillary column containing different phases (Katholieke Universiteit Leuven, 2002).
The Tenax tubes were mounted in a thermodesorption unit (Chrompack, model 16200). Thermodesorption took place at 250° over 8 min, with a helium flow through the tube of 13 ml/min. The volatiles released were trapped in a deactivated 0.5 mm i.d. capillary kept at –95°C prior to analysis on a connected DB5ms capillary column (60 m x 0.25 mm i.d., 0.25 µm film thickness). The initial column temperature was 40°C for 4 min and then programmed to 270°C at 4°C/min.
The components were detected with a Finnigan MAT 95 mass spectrometer operating in the 70 eV EI mode and scanning from mass 24 to 300 a.m.u. with a cycle time of 0.65 s. Identification of the chemicals was based on searches against the NIST98 (US Environmental Protection Agency, 1998) and Wiley (McLafferty and Stauffen, 1989) mass spectral databases and the retention indices.
Ventilation rate
In order to be able to model the exposure to volatile chemicals in the production hall, it was considered essential to estimate the ventilation rate. The ventilation rate was determined on three consecutive days, using the method of decline of a tracer gas (N2O) concentration (Boleij et al., 1995).
Data processing
The data analysis used SAS 6.12 (SAS Institute, Cary, NC). Descriptive statistics were used to examine the data and it was determined that logarithmic (base e) transformation was needed prior to statistical modelling. General linear principal component analysis (PCA) was used to examine linear relationships between the quantitative variables and to look for common sources among the measured chemicals.
Principal components significant for interpretation of the data were selected upon examination of a scree plot (principal component number versus eigenvalue). In order to identify the determinants of each ‘significant’ principal component (PC) group, new exposure variables were created for these groups of chemicals by first multiplying the eigenvectors for each principal component by the concentrations of the chemicals and then summing the products (i.e. generating principal component scores). The logarithms of principal component scores were used as dependent variables in a multiple linear regression model.
In these models, the type of tape used (PP, PPS or PEI) in the tape winding process, cleaning activities and activities using epoxy resins (production activities other than tape winding) were entered as predictor variables. Time of day was also used in the models, since two measurement sessions were done each day (one in the morning and one in the afternoon).
The models were run separately for three sites in the hall (around the tape winding unit, around the other production processes and background) in order to look at the distribution of released chemicals throughout the hall. Each site in its turn consisted of multiple measuring points (Fig. 1). For statistical power the personal measurements were grouped together with stationary measurements.
The variables epoxy activity and polypropylene (one of the tapes used) were highly correlated (r > 0.8, P < 0.0001). It was therefore decided to run separate models, one containing the variables connected with the tape winding process (PP, PPS and PEI) and one containing other possible exposure variables (cleaning activities and activities using epoxy resins) for all significant principal components.
The fit of the linear model was checked through plots of residuals.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Air monitoring
Qualitative assessment for hydrocarbons
The results of the GC-MS analyses of the duplicate Tenax air measurements were very similar. Therefore, we decided to analyse only one of two samples in detail. A great number of chemicals were present (150 for PP, 154 for PPS and 159 for PEI). In general, the chemicals identified were to a large extent similar for all tapes; however, some chemicals were only detected when specific tapes were used (e.g. sulphur-containing PPS and sulphur dioxide) (Table 1).
|
Quantitative assessment for hydrocarbons
Air concentrations measured with the charcoal tubes are summarized in Table 2. Acetone and ethanol were the two major chemicals present. Seven aliphatic hydrocarbons were found in 15–30% of all samples. The rest of the chemicals were only found in a small fraction of the samples. None of the observed concentrations exceeded the Dutch occupational exposure limits.
|
Exposure to other gases
The results from the colour diffusion tubes are presented in Table 3. The measurements in the fume cloud at the tape winding machine showed that neither NHx (0.25 p.p.m.) nor H2S (0.2 p.p.m.) could be detected at the limit of detection of the colour diffusion tube for these chemicals. High concentrations of HCN were found on two occasions while using PPS (>30 p.p.m.), during the third and fourth runs of the tape winding process. These concentrations exceeded the Dutch limit value of 10 p.p.m. (Ministerie voor Sociale Zaken en Werkgelegenheid, 2001
). All but one measurements of CO were found to be between 10 and 30% of the Dutch limit value (25 p.p.m.) (Ministerie voor Sociale Zaken en Werkgelegenheid, 2001). The Dutch OEL for CO was exceeded during the fourth run with PPS.
|
Ventilation
On all three occassions that ventilation rates were measured, hardly any decline in N2O concentration was observed. This indicates that the ventilation rate was negligible: it was estimated to be between 0.05 and 0.35 air changes/h.
Statistical modelling
PCA revealed that three PC accounted for 67% of multiple correlations among measured chemicals. Other PC with eigenvalues <2 accounted for <5% of the correlation and were left out of the analyses. A chemical was included in a PC when its eigenvector was >0.2 and Pearson correlations of the chemical with the PC were calculated. The first PC (PC1), with an eigenvalue of 9.95, consisted mainly of short chain hydrocarbons (C2–C6). The second group (PC2), eigenvalue 3.04, contained larger hydrocarbons (C9–C11) and some cyclic hydrocarbons. The third group (PC3), eigenvalue 2.43, contained all aromatic and two aliphatic hydrocarbons (n-hexane and cyclohexane) hydrocarbons. Styrene and ethanol appeared to be unrelated to the other chemicals. Table 4 shows the measured compounds with the respective eigenvectors and Pearson correlation coefficients between PC scores and individual chemicals.
|
The fact that the three PC each consisted of chemicals with a distinct molecular structure for that particular PC suggests different sources of chemicals.
Regression analyses showed an association of PC1 with cleaning activities and activities using epoxy resins (P < 0.07) (Table 5). Cleaning led to an 2-fold increase in concentrations, while use of epoxy led to an 1.5-fold increase. The fact that this increase is comparable for all three locations shows that the chemicals disperse throughout the production hall. Statistically significant higher afternoon than morning concentrations indicated a build-up during the working day caused by evaporation from epoxy products drying in the production hall.
|
PC2 showed an association with the use of PP and PPS in the tape winding process. The association was strongest for PPS (ßPPS > ßPP), which was significant for both the tape winding process and hand lay-up locations in the hall. The association with PP was significant for tape winding location only.
PC3 also showed a significant association with the tape winding process. The model showed concentrations increased by 60% when using PP or PPS instead of PEI at the tape winding location. The association here was strongest for PP, being significant at all measurement sites. Morning concentrations were found to be 80% of concentrations measured in the afternoon.
In general, no obvious build-up occurred for chemicals associated with the tape winding process, which were only emitted for a short period of time during the day.
Models were also run for ethanol, which did not contribute to any of the principal components. Ethanol exposure was strongly associated with cleaning activities (ß > 1.6, P << 0.05). Levels of ethanol were found to be nearly a factor of 5 higher when cleaning activities occurred during the day.
Styrene showed an association with epoxy activities (ß = 0.8 – 0.6, P << 0.05) for all measured sites.
Examining residuals showed neither a large deviation from the assumption of the regression analysis nor indicated a poor model fit.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Limited knowledge of emissions in the plastics industry was available prior to this study. The available reports provided only limited insight into type and quantity of emitted chemicals. Consequently, we chose sampling and analytical methods that could detect a great variety of hydrocarbons, since in most studies (Grote et al., 1984
; Forrest et al., 1994; Frostling et al., 1984) this group of chemicals was most prevalent. Furthermore, some specific chemicals [based on available material data sheets (Hoechst Celanese, 1997; BASF, 1999; General Electric, 2000)] were semi-quantitatively sampled with colour diffusion tubes. Finally, duplicate samples were taken for each tape with Tenax tubes and extensively analysed for qualitative information on variety and number of chemicals released. In doing so, a detailed quantitative as well as qualitative picture of released chemicals was obtained for the processes studied. Three principal sources of hydrocarbons were identified through statistical modelling. The smaller hydrocarbons were strongly associated with cleaning and activities where epoxy resins were used. This is in line with the fact that cleaning materials and epoxy resins are rich in such volatile hydrocarbons. The association found was strongest with cleaning activities, implying that this was the major factor in the release.
Emission of the longer hydrocarbons (C9–C11) and some aromatic hydrocarbons seemed to be strongly associated with the tape winding process when PP or PPS was used. This is in accordance with the literature, which reports the release of C8–C12 hydrocarbon as well as aromatic hydrocarbons during polypropylene processing (Grote et al., 1984; Britton, 1998; BASF, 1999). Statistical analyses showed that the released concentrations were related more to the use of PPS than PP for the chemicals in PC2. This is probably because the tape speed was slower when PPS was processed, which caused longer heating of the tape that might have led to increased generation of fumes and vapours containing these larger hydrocarbons.
For the PC3 group associations were also found with the tape winding process. In this case the association was stronger for PP than for PPS.
No distinction between personal and stationary measurement results were made in the data analyses, because insufficient numbers of measurements were made to analyse the data by type of measurement. This was, however, believed to be a valid approach, since workers hardly moved during their activities.
It is unclear whether differences in measured chemicals and concentrations per tape resulted from physical/chemical differences between the tapes or were influenced by characteristics of a specific production run such as production volume and process variables (such as heating temperature, tape angle, winding speed). Within this study, it was not possible to gather enough data on production variables to answer this question in a satisfactory manner and, since the tape winding process is very much in an experimental phase, the process variables were often changed ad hoc during the runs, which made them very difficult to document.
It should also be stated that as far as the measured concentrations of CO and HCN are concerned, the measurements were done in the fume clouds at the tape winding machine. Therefore, specific personal measurements for these chemicals are needed to investigate whether concentrations of CO and HCN exceed exposure levels in the breathing zone.
Because two different processes were in use during the measurements, it was difficult to attribute emissions to any specific source. Nevertheless, the statistical analysis techniques employed clearly indicated to what extent each process contributed to the identified chemicals in the workplace atmosphere.
This study also showed that a combination of different qualitative and quantitative methods could give a good insight into chemicals emitted during these processes and the concentrations at which they can be expected.
Recommendations for decreasing exposure at the ‘tape winding’ unit took into account that the process is expected to be scaled-up in the near future. The first recommendation that follows from this study is to decrease the release of chemicals at the source. For the ‘tape winding’ process this would mean preventing tape burning by changing the heating source for a more controllable one (e.g. infrared or laser). It also seems that the use of one of the brands of PPS caused high levels of CO and HCN to be released and it may be necessary to stop the use of this specific brand of PPS. If, rather than the tape, other process variables (like the propane burner) caused the high CO and HCN levels, local air exhaust ventilation might be placed on the ‘tape winding’ machine as close as possible to the spot where fumes and vapours are released.
For the traditional processes, when working with very volatile chemicals it may be prudent to use personal protection in order to prevent occasional high peak exposures. When volatile chemicals are handled manually, these activities should take place in a fume cupboard as much as possible. Since there seems to be no general ventilation, it should be introduced. Additional measurements are clearly needed to test the effectiveness of the suggested exposure control measures.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The most important conclusion is that there are two sources at this factory that are responsible for the release of a variety of chemicals. The smaller more volatile chemicals are most likely to be associated with cleaning and epoxy activities. The concentrations in which these chemicals were found do not exceed exposure limits. On the other hand, some of these very volatile chemicals might lead to high peak concentrations that could result in hazardous situations when no personal protection is used.
For the larger chemicals associated with the tape winding process, although most concentrations stay below the OEL, it could be that when production increases (for example continuous production with more than one machine) limits could be exceeded, leading to adverse health effects (Lees, 1995; Lewis, 1999; Mikov et al., 2000). Furthermore, little is known about possible cumulative effects for the complex mixtures identified.
Since the plan for the tape winding process is to use it on a much larger scale in the future, release of potential harmful chemicals should be reduced as much as possible. It is likely that engineering controls at the winding machine itself (e.g. a more specific heating source and better process control) will decrease the release of organic chemicals. We further recommend that investigations similar to ours be undertaken at a very early stage whenever significant changes in technology are expected in a workplace.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
BASF. (1999) Material safety and data sheets: polypropylene (PP). BASF.
Boleij J, Buringh E, Heederik D, Kromhout H. (1995) Occupational hygiene of chemicals and biological agents. Elsevier, Amsterdam, The Netherlands.
Britton TJ. (1998) Encyclopedia of occupational health and safety, 4th edn. International Labour Office, Geneva, Switzerland, 77.22–77.29.
Forrest MJ, Jolly AM, Holding SR, Richards SJ. (1994) Emissions from processing thermoplastics. Ann Occup Hyg; 39: 35–53.
Frostling H, Hoff A, Jacobsson S, Pfaffli P, Vainiotalo S, Zitting A. (1984) Analytical, occupational and toxicological aspects of the degradation products of polypropylene plastics. Scand J Work Environ Health; 10: 163–9.[Medline]
General Electric. (2000) Material safety and data sheets: polyethylimide (PEI). General Electric.
Grote AA, Kim WS, Kupel RE. (1984) Establishing a protocol from laboratory studies to be used in field sampling operations. Am Ind Hyg Assoc; 39: 880–4.
Hoechst Celanese. (1997) Material safety and data sheets: polypropylene-sulphide (PPS). Hoechst Celanese.
Hoff A, Jacobsson S, Pfaffli P, Zitting A, Frostling H. (1982) Degradation products of plastics. Polyethylene and styrene-containing thermoplastics—analytical, occupational and toxicologic aspects. Scand J Work Environ Health; 8: 1–60.[Web of Science][Medline]
Katholieke Universiteit Leuven. (2002) Informatiebrochure lucht- en productanalyse. Leuven: Faculteit Geneeskunde, Katholieke Universiteit.
Lees PS. (1995) Combustion products and other fire-fighter exposures. Occup Med; 10: 691–706.[Medline]
Lewis R. (1990) Plastics. J Occup Med; 32: 387–96.
Lewis R. (1999) Health issues in plastics production and processing. Occup Med; 14; 777–95.[Medline]
McLafferty F, Stauffen DB. (1989) Registry of mass spectral database, 5th edn. New York, NY: Wiley.
Mikov I, Draskovic D, Savic, M, Arsic M, Todorovski Z, Glavaski M. (2000) Occupational intoxication with carbon monoxide. Arch Environ Health; 14: 455–6.
Ministerie voor Sociale Zaken en Werkgelegenheid. (2001) Nationale MAC-lijst 2001. Ministerie voor Sociale Zaken en Werkgelegenheid.
Nelson GL. (1973) Smoke evaluation: thermoplastics. J Fire Flammability; 5: 125–35.
US Environmental Protection Agency (1998) Mass spectral library NIST, Gaithersburg, MD: National Institutes of Health.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  T. OGDEN Annals of Occupational Hygiene at Volume 50: Many Achievements, a Few Mistakes, and an Interesting Future Ann. Hyg., November 1, 2006; 50(8): 751 - 764. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
I. BURSTYN Principal Component Analysis is a Powerful Instrument in Occupational Hygiene Inquiries Ann. Hyg., November 1, 2004; 48(8): 655 - 661. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||