Respiratory exposures associated with silicon carbide production: estimation of cumulative exposures for an epidemiological study

ABSTRACT

Silicon carbide is produced by heating a mixture of petroleum coke and silica sand to approximately 2000 degrees C in an electric furnace for 36 hours. During heating, large amounts of carbon monoxide are released, sulphur dioxide is produced from residual sulphur in the coke, and hydrocarbon fume is produced by pyrolysis of the coke. Loading and unloading furnaces causes exposures to respirable dust containing crystalline silica, silicon carbide, and hydrocarbons. In the autumn of 1980 extensive measurements were made of personal exposures to air contaminants. Eight hour time weighted exposures to sulphur dioxide ranged from less than 0.1 ppm to 1.5 ppm and respirable participate exposures ranged from 0.01 mg/m3 to 9.0 mg/m3. Geometric mean particulate exposures for jobs ranged from 0.1 mg/m3 to 1.46 mg/m3. The particulate contained varying amounts of alpha-quartz, ranging from less than 1% to 17%, and most quartz exposures were substantially below the threshold limit value of 100 micrograms/m3. Only traces of cristobalite (less than 1%) were found in the particulate. Median exposures to air contaminants in each job were estimated. Since the operations at the plant had been stable over the past 30 years, it was possible to estimate long term exposures of workers to sulphur dioxide, respirable particulate, quartz, total inorganic material, and extractable organic material. Cumulative exposure (average concentration times exposure duration) for each of the air contaminants was estimated for each worker using his job history. There was sufficient independent variability in the sulphur dioxide and respirable particulate cumulative exposures to make an assessment of their independent effects feasible. The theoretical basis for using the cumulative exposure index and its shortcomings for epidemiological applications were presented.

METHODS AND MATERIALS

Exposures were measured by personal sampling, and sample collection was stratified by job category and work area as shown in table 2. Emphasis was placed on evaluation of furnace area exposures. Samples from two to four full shifts were collected on some individuals to determine their day to day variation in exposure. A total of 182 personal samples were collected. Fifteen samples were collected to measure S02 exposure. A direct reading instrument (Ecolyzer) was used to measure the concentration of carbon monoxide in several locations within the furnace area.

Respirable particulate was collected on a tared filter, which was preceded by a 10 mm nylon cyclone to remove non-respirable particles. Two types of filters were used: a 5.0 µm pore diameter, PVC membrane filter (type #WS-B, Mine Safety Appliance Co, Pittsburgh, PA) and a Teflon coated glass fibre filter (type T60 A20 Pallflex Products Corp, Putnam, CT). Half the samples in each job or work area were obtained with each filter type: the PVC for x ray diffraction analysis and the glass fibre for hydrocarbon analysis. The glass fibre filters were extracted with methylene chloride before use to remove contamination. All samples were refrigerated after sampling to minimise losses of volatile hydrocarbons.

⍺-Quartz was measured by x ray diffraction after oxidising the PVC filter matrix and sample in a low temperature radio frequency combuster and then depositing the unoxidised residue on a silver membrane filter. The mass of unoxidised residue was reported as inorganic matter. Because of the relatively low sensitivity of the x ray diffraction technique, all samples from a given job category were composited and analysed as a single sample. Semiquantitative analyses were also performed to determine the approximate amounts of crystobalite and silicon carbide present on the x ray diffraction tracings.

Total extractable organic compounds were determined gravimetrically on an aliquot of a methylene chloride extract of the sample, which was evaporated to dryness. Samples were extracted for 18 hours in a Soxhlet extractor. The polycyclic aromatic hydrocarbon content of some of the extracts was measured semiqualitatively by a fluorescent spot test.

The nicotine content of the methylene extracts was measured by gas chromatography using a nitrogen/phosphorus sensitive flame ionisation detector and 1 m column of 10% Carbowax 20M pretreated with 3% potassium hydroxide.

Sulphur dioxide was measured by absorption in hydrogen peroxide and titration of the resulting sulphuric acid solution. Air was drawn at 0.9 l/min through two midget impingers in series (used as gas scrubbers), each containing hydrogen peroxide. Because of the inconvenience of the sampling system, it was only possible to collect a few personal samples.

Distributions of the exposures were described by log-normal statistics, the geometric mean, and geometric standard deviation because the observations were positively skewed and personal exposure measurements are commonly log-normal. Analyses of variance were performed on the logarithms of the exposures using the statistical analysis system (SAS).

Estimation of cumulative dose

The rate an air contaminant deposited in a worker's lungs depends on the air concentration (x), his inhalation rate (r), and the fraction of the inhaled material that is deposited (f). Over a single work shift the amount deposited, the worker's dose (Di), is the integral of the deposition rate:

Di = f 'xrfdt.

If the inhalation rate and fraction deposited are approximately constant then his dose on day i may be estimated by the time weighted average air concentration (xi) in his breathing zone times his average inhalation rate (r), times the fraction of the substance deposited in his lungs (f). Or,

Di = xirf

For a long term exposure, the total amount of material deposited in the lungs is the sum of each of these daily doses. Thus the cumulative dose, D, over a long period, N days, is

N

D = Σ x x1rf


Since the arithmetic mean of the daily exposures is given by

X=1/N Σ Xi.X

the cumulative dose can be written as

D = NXrf.

Since the cumulative exposure is the product of average concentration times the duration of exposure (N z), if the inhalation rate and fraction deposited in the lungs are approximately constant then the cumulative dose is proportional to the cumulative exposure. Thus if a worker is exposed to different levels of the same agent in several jobs, the sum of his cumulative exposures in each of the jobs will be proportional to the total amount of the agent deposited in his lungs.

Although the arithmetic mean of an individual's daily exposures was the desired index of his dose level, it was not possible to measure the average exposure of each individual in the study cohort. As an alternative, we used the geometric mean of the average exposures of workers doing a given job. The geometric mean was used because the distribution of average exposures of individuals performing the same job is approximately log-normal. If the distribution of exposures is truly log-normal then the median of the exposures is estimated by the geometric mean. Thus the exposure for each job was estimated by the geometric mean of the individual arithmetic mean exposures measured for each job in the 1980 survey. A detailed derivation of this approach and its implications and statistical characteristics is in preparation (T J Smith).

Since workers frequently had a large number of different jobs, many of which had the same levels of exposure, the job histories were condensed into 10 job categories with different exposures for pre-1962 and 1962-80. The duration of exposure in each job category and each period was determined for each subject from his work history at the company. An exposure level was assigned to each of the 10 job categories based on the 1980 data and on information about the nature of the exposures. Cumulative exposures for sulphur dioxide, respirable particulate, ⍺-quartz, inorganic material, and extractable organic matter were calculated for each subject using the condensed job histories and the exposure assignments. These cumulative exposures are proportional to the total amount of each substance deposited or absorbed in the subject's respiratory tracts during their total exposures.
RESULTS

Carbon monoxide (CO) levels were measured in several locations on several days. Average levels ranged from 10 ppm to 25 ppm, although brief peaks were seen in the crane cabs as high as 160 ppm for a fraction of a minute, and as high as 80 ppm for several minutes. The operation of removing the furnace sides produced a large emission of CO. In some locations near the furnaces levels ranging from 100 ppm to 180 ppm were observed for several hours during startup. These areas were not usually active work sites.

The duration of many of the personal exposures was less than eight hours because the workers completed their assigned tasks and returned to the main lunchroom, where they waited for the remainder of the shift (table 3). Thus the eight hour time weighted exposures shown in table 3 do not represent the average air concentration inhaled by the workers in the work area.

Virtually all the workers had some exposure to airborne respirable particulates as shown in table 3, but the level of the exposure and composition of the dust varied widely as shown in tables 3 and 4. The outdoor and furnace area payloader operators had high geometric mean, time weighted average exposures to respirable particulate (1.66 and 1.46 mg/m3 respectively), although less than 10% of the exposures of either exceeded the 5 mg/m3 permissible exposure for respirable nuisance dust. The remainder of the job categories had moderate or low exposures to respirable dust.

Only the furnace loaders had time weighted average exposures to ⍺-quartz that frequently exceeded the permissible level allowed by the Province of Quebec, 100 µg/m3; 50% of their ⍺-quartz exposures were greater than 100 ug/m3. Only four samples were collected, however, so the level of exposure was not well defined. The mixers, payloader operators, old mix operators, and carboselectors had geometric mean exposures to ⍺-quartz that were about half of the 100 µg/m3 limit. The remainder of the job categories exposures were well below the permissible limit.

⍺-quartz was the predominant crystalline silica mineral found in the samples. Small amounts of crystobalite were observed in some of the samples, but none exceeded approximately 1% of the total particulate.

SiC was presumed to form most of the inorganic portion of the particulate. Large amounts of inorganic material were found in most of the furnace area samples and in all of the product area samples (table 4). The payloader operators had the highest time weighted exposures to inorganic matter.

Large amounts of extractable organic compounds were found in the cranemen's particulate samples. This is consistent with their work location; the open windowed crane cabs are frequently within the stream of hot gases and smoke rising from the furnaces. Other workers generally had low percentages of extractable hydrocarbons in their particulate. The maintenance workers had low levels of extractable hydrocarbons in their samples, but they periodically work in the upper parts of the furnace building (repairing cranes, conveyor belt systems, or other equipment) and during this time their particulate exposures would closely resemble the cranemen's exposure.

The relatively high level of extractable matenal in the carboselectors' samples (20% of the particulate mass) was probably an artifact caused by the collection of cigarette smoke from the air of their small lunchroom next to the cleaning floor where they wait during their frequent breaks in work. Analyses of the nicotine content of the extractable organic materials showed that the carboselector samples contained an average of 0-049% nicotine. This was substantially more nicotine than samples from the other seven jobs in the furnace area: only three samples of nine contained any detectable nicotine content (>0.001 %) and all contained less than 0-015%. The total extractables content of these samples was not, however, correlated with the nicotine content.

Sulphur dioxide exposures were associated with the furnace off-gases, and were highest immediately around the furnaces. One stationary sample indicated a four hour average of 7.3 ppm (18-9 mg/m3) near one of the furnaces. Thus the cranemen, furnace attendants, and the payloader operators were likely to have the highest exposures. Personal, time weighted average samples were all much lower than the stationary samples. Other furnace area workers had low level exposures, and those outside this area had little or no exposure. The assignment of sulphur dioxide exposures in table 4 is tentative because only a few samples were collected.

Analyses of variance (ANOVA) were performed to determine if there were differences in exposure between the morning and afternoon work shifts and among the days of the week. Although there were differences, none was statistically significant (p > 0.1). Subtle differences may exist that might be detectable with larger sample sizes, but they represent differences of less than a factor of two and in some cases less than 30% between the geometric means, which are small differences relative to the variability of the measurements.

The collection of multiple samples on some individuals permitted the determination of day to day variation in individual exposures and a comparison of variability among individuals doing the same job. Table 5 shows a summary of the results for four job categories that had sufficient observations for this analysis. Even though these four jobs had a wide range of mean exposures, the day to day variability was larger than the between individual variability for all of them. The number of samples was too small for these differences to be statistically significant.

The cumulative exposures calculated for each subject in the epidemiological study covered a wide range for each of the air contaminants, as shown in table 6 The results are expressed in concentration times hours times years; for example, the median respirable particulate exposure was 56-4 mg/m3 x h year, which might represent four hours a day exposure to 0.7 mg/m3 for 20 years, or eight hours a day exposure to 0.47 mg/m3 for 15 years. Table 7 shows the correlation coefficients of the cumulative exposure variables with each other. The inorganic and quartz content dust variables were both highly correlated with the total respirable particulate. Sulphur dioxide exposures showed less correlation with the dust variables. The extractable hydrocarbon content of the dust also behaved somewhat independently of the other air contaminants.
