Download Mineral Oil Metal Working Fluids (MWFs)– development of Practical

PII:
Ann. occup. Hyg., Vol. 44, No. 3, pp. 165±172, 2000
Crown Copyright 7 2000 Published by Elsevier Science Ltd.
All rights reserved
Printed in Great Britain.
S0003-4878(99)00085-X
0003±4878/00/$20.00
Mineral Oil Metal Working Fluids (MWFs)Ð
development of Practical Criteria for Mist Sampling
A. T. SIMPSON$*, J. A. GROVES$, J. UNWIN$ and M. PINEY%
$Health and Safety Laboratory, Broad Lane, Sheeld, S3 7HQ, UK; %Health and Safety
Executive, Haswell House, 54 Nicholas Street, Worcester WR1 1UW, UK
Not all mineral oil metalworking ¯uids (MWFs) in common use form stable airborne mists
which can be sampled quantitatively onto a ®lter. This much has been known for some time but
no simple method of identifying oils too volatile for customary ®lter sampling has been
developed. Past work was reviewed and experiments were done to select simple criteria which
would enable such oils to be identi®ed. The sampling eciency for a range of commercial
mineral oil MWF were assessed by drawing clean air through spiked ®lters at 2 l. minÿ1 for
periods up to 6 h before analysis. The physical properties of MWF are governed by their
composition and kinematic viscosity was found to be the most practical and easily available
index of the potential for sample loss from the ®lter. Oils with viscosities greater that 18 cSt
(at 408C) lost less than 5% of their weight, whereas those with viscosities less than 18 cSt gave
losses up to 71%. The losses from the MWF were mostly aliphatic hydrocarbons (C10±C18),
but additives such as alkyl benzenes, esters, phenols and terpene odorants were also lost. The
main recommendation to arise from the work is that ®lter sampling can be performed on
mineral oils with viscosities of 18 cSt (at 408C) or more with little evaporative losses from the
®lter. However, sampling oils with viscosities less than 18 cSt will produce results which may
signi®cantly underestimate the true value. Over a quarter of UK mineral oil MWFs are
formulated from mineral oils with viscosities less than 18 cSt (at 408C). The problem of
exposure under-estimation and inappropriate exposure sampling could be widespread. Further
work is being done on measurement of mixed phase mineral oil mist exposure. Crown
Copyright 7 2000 Published by Elsevier Science Ltd.. All rights reserved.
Keywords: mineral oil; oil mist; metalworking ¯uid; sampling
INTRODUCTION
Metal working ¯uids (MWF) is the general term
given to a range of products used as lubricants and
coolants during the machining or treatment of
metal components. These liquids consist of complex
mixtures of chemicals in either mineral oil or water
(known as water mix metal working ¯uids). Both
types contain a variety of chemical additives to
improve or maintain the performance of the MWF.
HSE's Engineering National Interest Group estimates that there are at least 50 000 businesses in
the UK using MWFs. These include not only engineering companies but the maintenance departments
Received 16 December 1998; in ®nal form 21 July 1999.
*Author to whom correspondence should be addressed.
Tel.: +44 114 289 2000; Fax: +44 114 289 2500.
165
of large companies and organisations outside of the
engineering sector. The number of employees
exposed could be anywhere between 100,000 and
200,000. Approximately 1300 tonnes of neat mineral
oil MWFs are used in the UK per month, and the
amount of water mix MWF concentrate used is approximately 600 tonnes per month (at an average
working strength of 5% this equates to roughly
12,000 tonnes). This paper is concerned with aspects
of sampling mineral oil MWF aerosol.
Mineral oil MWF generally consist of one or
more severely re®ned mineral base oils, together
with extreme pressure additives (such as esters,
chloroparans and sulphurised esters) and possibly
odorants, antimist and anti-corrosion additives.
During use, aerosols of the oil can be generated,
usually referred to as oil mist. These originate
from processes where the oil is subject to high
sheer forces (such as when it is applied to parts
166
A. T. Simpson et al.
rotating at high speeds) or where they are exposed
to excess heat (vaporised oil condenses into small
droplets as it cools). In some activities such as
metal forming (for example, shaping sheet metal
by pressing it into a die), the oil is applied in the
form of a mist using compressed air (nebulisation)
to coat areas quickly and evenly with a thin coat
of lubricant.
Occupational exposure to oil mist is associated
with eye, nose and throat complaints and dermatitis. In addition, there are concerns relating to occupational asthma, allergic alveolitis and other lung
diseases, as well as the possibility of cancer
(although this may relate to historical uses). Given
the complexity and changing nature of formulations
and likely contaminants it is unclear what the causative agent(s) might be. HSE is currently reviewing
the use patterns and available data for oil mists
with the aim of producing industry guidance. In the
UK the current occupational exposure limit for
mineral oil mist is an Occupational Exposure
Standard (OES) of 5 mg mÿ3 8-h time weighted
average (TWA), with a short-term exposure limit
(STEL) of 10 mg mÿ3 (15-min TWA) (HSE, 1999).
A number of methods have been described to
measure exposure to oil mist including gravimetric
methods
(CONCAWE,
1981;
HSE,
1991;
Menichini, 1986) and infrared and ultraviolet spectroscopic methods (CONCAWE, 1981; Menichini,
1986; NIOSH, 1994). In each case the sample is collected on a ®lter and the assumption is made that
there is no sample loss during the sampling period.
Mineral oil MWF can be made up of a variety
of base oils and other additives or diluents covering a range of volatilities. The aerosol formed will
consist of droplets and vapour, and will be
dynamic with material moving between the two
phases. The phase distribution of such aerosols is
vulnerable to changes in ambient conditions
(Soderholm, 1988). Variations in temperature and
vapour concentration as the newly formed aerosol
is diluted will cause further change in the phase
distribution. These changes will also a€ect the oil
on the ®lter during sampling. The mass of oil collected on the ®lter will continue to alter if it has
not reached equilibrium with the air being
sampled. If the person wearing the sampler moves
to an area which has air containing less vapour,
there will be evaporation from the sample on the
®lter. MWF which contain a high proportion of
more volatile components will be a€ected more
than MWF containing less volatile components. It
cannot therefore be assumed that simply collecting
the oil mist on a ®lter will give an accurate
measure of oil mist concentration. This problem
has been recognised by others: Menichini (1986)
found that when seven oils spiked on ®lters were
aspirated at 2±4 l. minÿ1 for 1 h with clean air,
most lost less than 1% mass, however a light oil,
viscosity 2.8 cSt, initial boiling point 2408C, lost
6%. It was recommended that recovery tests for
oils sampled should be performed. McAneny et al.
(1995) aspirated ®lters loaded with a new and used
oil aerosol sample with clear air at 1.5 l. minÿ1 for
4 h and found that the new oil lost an average of
35% of the collected oil, whilst the used oil lost
only an average of 12%, the di€erence being put
down to the previous loss of the more volatile components in the used oil during its use.
What is needed is some simple way of determining which mineral oil MWF can be measured by ®lter sampling without unacceptable sample losses. In
this work, methods are developed using easily available information which can be used to predict
which oil mists may be subject to such losses during
®lter sampling.
EXPERIMENTAL
To assess the scale and extent of the sample loss
from a mineral oil MWF trapped on a ®lter during
sampling, a variety of oils were examined under
simulated sampling conditions. Thirteen MWF and
one hydraulic oil were included covering a range of
viscosities (4±33 cSt (mm2 sÿ1) at 408C). The
hydraulic oil was included to ®ll a gap in the viscosity range.
Spiked ®lters were produced by pipetting neat oil
(1±2 mg) onto pre-weighed 25-mm glass ®bre ®lters.
The ®lters were placed in Gelman sampling heads
and clean air (0208C) was sucked through at 2 l.
minÿ1 to simulate collection of a sample. A guard
®lter was positioned in a second Gelman sampling
head fastened face to face with the ®rst to prevent
collection of airborne dust. The spiked ®lters were
re-weighed at intervals (0.5, 1, 3 and 6 h) during
aspiration to determine the weight loss. The ®lters
were analysed in groups of ®ve replicates, and
blank ®lters were run in parallel.
In addition to examining commercial mineral oil
MWF products, the losses from a light base oil
(mineral seal oil) and deodorised kerosene were
also determined. Mineral seal oil is a base oil in
many of the low viscosity MWF, and kerosene is
reported to be in some products (CONCAWE,
1986). To identify the components lost during
aspiration, a Tenax sorbent tube was positioned
between the spiked ®lter and the sample pump to
collect any vapours released. Air was pumped
through at a ¯ow rate of 2 l. minÿ1 for a period
of 5 min, and samples were collected on tubes
loaded with 55 mg of Tenax TA (a ¯ow of 2 l.
minÿ1 could not be achieved with larger amounts of
Tenax). The Tenax tubes were analysed on a Perkin
Elmer ATD 400 automated thermal desorber connected to a Hewlett Packard 5970 Mass Selective
Detector (ATD-GC-MS).
MWFsÐdevelopment of practical criteria for mist sampling
167
Table 1. Results of evaporative weight loss experiments
Analyte
Viscosity (cSt at 408C) Flashpoint (8C) Boiling point range (8C)a
Kerosene
02
Mineral seal oil
3.7
MWF A
4.4
MWF B
5
MWF C
7
MWF D
9.2
MWF E
11.1
MWF F
13
MWF G
13.2
MWF Hb
15
MWF I
15.5
MWF J
18.3
MWF K
19.0
MWF L
19
MWF M
23
MWF N
33
65±85
122
115
115
160
152
135
140
138
180
> 180
168
200
185
202
190
197±236
258±330
205±400
260±330
IBP 350
IBP 260
284±449
205± > 400
337±439
IBP >320
% Cumulative weight loss at:
30 min 60 min 180 min 360 min
98.0
9.2
22.0
4.5
1.4
3.1
2.4
2.8
1.3
0.8
0.2
0.7
0.3
0.1
0.1
ÿ0.1
99.6
16.8
34.3
7.9
2.5
5.6
4.2
4.6
2.4
1.4
0.2
1.1
0.4
0.4
0.3
0.3
100.0
36.7
56.9
17.2
5.4
11.5
9.7
8.8
5.4
3.4
0.4
1.9
0.9
0.8
0.5
0.4
99.7
57.2
71.2
27.5
8.9
16.9
16.0
13.2
9.1
5.7
0.6
2.7
1.5
1.4
0.6
0.4
a
Estimated from data on base oils obtained from oil producers: IBP=initial boiling point.
Hydraulic oil.
b
RESULTS
The results of the weight loss experiments are
shown in Table 1. The data are presented with the
manufacturers' ®gures on the viscosities and ¯ash
points of the mineral oil MWFs tested. In addition,
where obtainable, an estimated boiling point range
is included. This is not the measured boiling point
range for the MWF, but is based upon information
obtained via manufacturers on the base oils known
to be present, and contains no allowance for other
constituents. An example of the change in weight
with time during aspiration is shown in Fig. 1
(MWF G).
The compounds identi®ed on the Tenax back up
tubes are shown in Table 2. Only prominent peaks
in the sample chromatograms were identi®ed. There
was often a hydrocarbon envelope of unresolved
peaks, predominantly branched aliphatic hydrocarbons, which were not investigated.
DISCUSSION
All the oils tested showed some weight loss after
6 h of aspiration. For those oils showing signi®cant
weight loss it was apparent that material continued
to evaporate from the samples throughout the 6 h.
The loss measured varied from less than 1% for the
heavier oils (MWF M and MWF N), to 71% for
the lighter oils (MWF A).
A simple way of identifying those mineral oil
MWF unsuitable for ®lter sampling was sought.
Such a method should link the weight loss of the
Fig. 1. The rate of loss of material from MWF G during aspiration: weight loss (%), time aspirated (min).
C10±C18 Alkanes (C12±C16 n-alkanes major components)
n-Octane
1,2,3,4-tetra-hydro-naphthalene
C13±C18 Alkanes (C13±C15 n-alkanes major components)
C4 Alkyl benzenes
C5 Alkyl benzenes
C11±C18 Alkanes (C13±C15 n-Alkanes major components)
Trimethyl benzene tr
Decane tr
Ethyl, methyl benzene tr
6-methyl,-1,2,3,4-tetrahydro naphthalene
C13±C18 Alkanes (C18 trace amounts)
Unidenti®ed aromatics
Diethylene glycol
2,6-bis(1,1-Dimethyl ethyl)-4-methyl phenol
Methyl dodecanoate
Methyl tetradecanoate
Menthyl hexadecanoate
C14±C19 Alkanes (C14 and C19 trace amounts)
C4 Alkyl benzene
C5 Alkyl benzenes
C10±C18 Alkanes (C13±C15 n-alkanes major components)
MWF A
MWF B
Mineral seal oil
Kerosene
MWF M
MWF N
MWF L
MWF K
MWF H
MWF I
MWF J
MWF G
Analyte
Major component: signi®cantly above instrument detection limit; tr: trace levels detected, close to instrument detection limit.
a
MWF F
MWF E
MWF D
MWF C
Possible identi®cation of components
Analyte
Table 2. Identity of the materials evaporated from the MWFa
Decahydro naphthalene
Methyl decahydro naphthalene
C11±C18 Alkanes (C12±C17 n-alkanes major components)
C5 Benzenes
C14±C18 Alkanes (C18 at trace level)
3-Methyl, butyl acetate tr
1-methyl, butyl acetate tr
Decane tr
1,8 Cineole tr
Undecane tr
Bornyl acetate tr
C10±C16 Alkanes (C11±C15 n-alkanes major components)
C6 Alkyl cyclohexane
C10±C18 Alkanes
Hydrocarbon envelope with no outstanding feature
C13±C18 Alkanes (C13, C14, C15 and C18 at trace level)
C12±C17 Alkanes (C12±C15 n-alkanes major components)
Unidenti®ed aromatics
C15±C18 Alkanes tr
Unidenti®ed phenol compound tr
±
Possible identi®cation of components
168
A. T. Simpson et al.
MWFsÐdevelopment of practical criteria for mist sampling
oil during aspiration with a physical property of the
oil, such as boiling point range, vapour pressure,
viscosity or ¯ashpoint. The data must be easily
obtainable so that analysis of the mineral oil MWF
is not required before sampling of the oil mist can
begin. The most obvious sources of this kind of information are the manufacturers' product information sheets.
The boiling point range and vapour pressure of
the mineral oil MWF were not used since neither
property is usually measured by manufacturers. The
information which is available on boiling points
generally refers to measurements on the base oils by
the base oil producers, and no allowance is made
for the e€ect of blending base oils and additives;
over half of a mineral oil MWF can consist of additives. The boiling point range of the base oils could
be used as a guide, but the information is hard to
obtain as it is not always known by the MWF manufacturer, and it is questionable how applicable it
would be for mineral oil MWF which contain a
considerable portion of additives. A similar situation exists for vapour pressure measurements; any
data available refers to the base oil, not the MWF,
and additives can change the vapour pressure of the
MWF. Generally, when information is quoted it is
in the form of an all encompassing maximum
vapour pressure (for example, <0.1 kPa at 208C).
The kinematic viscosity (at 408C) and the ¯ashpoint of the mineral oil MWF are usually quoted
on the product data sheets, and there is a clear relationship between both parameters and the ®lter
weight loss for the oils in this study, illustrated in
Fig. 2 and 3. The error bars represent +/ÿ1.96
times the standard error of the data.
Figure 2 shows that the oils at the top end of the
viscosity range (23 and 33 cSt) lost very little material during aspiration, while those in the middle
169
and bottom end of the range show an increase in
sample loss with decreasing viscosity. Those at very
low viscosity (MWF A and B) had very high evaporative losses from the ®lter. The mineral seal oil
sample, which has a similar viscosity to those very
low viscosity mineral oil MWF and is probably a
base oil present in the MWF formulation, had a
similar weight loss. MWF C does not appear to ®t
the trend, having a lower than expected weight loss.
This MWF is exceptional in that 75% of it is additive and the remaining 25% is made up of two base
oils. Unsurprisingly the kerosene sample was fully
evaporated in the ®rst half hour. There should be
very little contribution to gravimetric results from
any kerosene present in samples trapped on glass
®bre ®lters. Figure 3 shows similar behaviour when
the weight losses were plotted against ¯ashpoint.
Oils with high ¯ashpoints unsurprisingly lost little
material during aspiration, whilst oils with lower
¯ashpoints showed increasing sample losses.
There is little di€erence between the trend within
the graphs, with each showing a similar pattern.
Although a degree of caution should be taken in
analysing the data (each MWF is relatively unique,
and data at levels of sample loss near to zero could
be expected to skew some regression models), a line
of best ®t was determined for both graphs. Using
an exponential model, the viscosity data gave the
line y = e4.24x ÿ 0.183, correlation coecient ÿ0.88
(Fig. 2), and the ¯ashpoint data gave the line y =
e9.43x ÿ 0.0488, correlation coecient ÿ0.91 (Fig. 3).
The graphs were used to try and estimate a cut-o€
point, below which oils would be liable to unacceptable losses. From inspection of the graphs in Fig. 2
and 3, the oils which had less than 5% losses were
generally those with viscosities greater than 18 cSt
and ¯ashpoints of 1808C and above. If losses of 5%
after 6 h of aspiration are deemed acceptable then
Fig. 2. The relationship between weight loss and MWF viscosity: weight loss after 6 h (%), MWF viscosity (cSt at 408C).
170
A. T. Simpson et al.
these results would suggest that 8-h sampling of oil
mist onto glass ®bre ®lters is a suitable sampling
method for mineral oil MWF with viscosities of 18
cSt and above or ¯ashpoints of 1808C and above.
Although both ¯ashpoint and viscosity are capable
of identifying these oils, it was considered that the 18
cSt viscosity criterion was a clearer cut-o€ point and
could be more usefully applied.
Mists from oils below these limits may be liable
to su€er losses during sampling, giving erroneous
oil mist measurements. An enquiry to the British
Lubricant Federation revealed that analysis of neat
cutting oil sales shows that 38% by number, and
28% by volume are less than 18 cSt at 408C. This
represents a signi®cant proportion of oils used
where occupational exposures may, to some extent,
be underestimated.
The evaporative losses from ®lters when sampling
the aerosol from very light mineral oil MWFs (for
example, <5 cSt) could be large, with losses of up
to 71.2% recorded here. In these cases collecting
the oil mist on a ®lter could give very misleading
results since the sample collected on the ®lter could
evaporate easily during sampling. Oil mist from
these oils could be associated with very high levels
of oil in vapour form. Although the vapour is considered less harmful than the mist, it may become a
signi®cant form of exposure at high levels. It should
be noted that these results could be used as an indicator of the potential for error during sampling, but
should not be used to back-calculate the true mist
concentration from existing data.
It has been reported that sampling oil mist using
personal electrostatic precipitators will retain more
sample than ®lter sampling (Leith et al., 1996),
however, such devices are highly specialised and not
widely available. Filter sampling using inhalable
samplers such as those recommended by HSE
(1997) is currently the only feasible method to collect the particle size fraction of the mist required
for comparison with the OES, but as can be seen,
®lter samples can be subject to evaporative losses.
The experimental methods used here do not fully
model the conditions during normal sampling in the
workplace. The ®lters were spiked by pipetting neat
mineral oil onto the ®lter; real oil mist samples
would have covered a larger surface area of the ®lter which would have increased the rate of evaporation, thus the results here should be seen as the
minimum losses under the test conditions used.
Increasing the air temperature would also have the
e€ect of increasing the rate of evaporation.
Conversely, the clean laboratory air used for aspiration did not contain any signi®cant amount of hydrocarbon vapour which would have reduced or
prevented evaporation. In the workplace, sta€ may
enter areas of clean air during the sampling period,
but when operating the relevant machinery will be
exposed to air containing either high or even saturated levels of vapour from the oil. In addition, the
spiked ®lters tested were aspirated for 6 h, but
during actual sampling, oil would accumulate on
the ®lter over the whole sampling period, and only
a small portion of the ®nal sample will have been
aspirated for longer than 6 h; most would have
been aspirated for a shorter period. Nevertheless,
the experimental method was thought to represent
the best practical way of estimating the losses.
Because there are many mineral oil MWF formulations available, the results reported here are
necessarily based upon a limited number of products, but these are thought to be representative of
the range available.
All the oils used were new oils, but it is possible
Fig. 3. The relationship between weight loss and MWF ¯ashpoint: weight loss after 6 h (%), MWF ¯ashpoint (8C).
MWFsÐdevelopment of practical criteria for mist sampling
that the composition of oils which have been used
for a period of time may change. They could contain fewer volatiles because of preferential evaporation of lighter components, or become
contaminated by tramp oil (for example, lubricating
and hydraulic oil). The loss of volatiles would raise
the ¯ashpoint, increase viscosity and reduce the
amount of material liable to loss from the ®lter
during sampling. The hydraulic and lubricating
¯uids used for metalworking machines are predominantly neat oils with higher viscosities than MWF.
Machine lubricating oil is typically 68 cSt and
hydraulic oil 32 cSt. These oils contain heavier, less
volatile components than MWF, and would generally increase viscosity, raise the ¯ashpoint and
reduce the volatility of the oil.
A range of compounds from the base oils and
additives in the mineral oil MWF were identi®ed
amongst the material lost from the ®lters during
sampling aspiration. It is dicult to estimate the
amount of material trapped on the Tenax tubes
without doing quantitative work, but from comparing chromatographic peak heights it was apparent
that more material was detected in the analyses of
the lower viscosity MWF than in those of the
higher viscosity MWF. This is not unexpected considering the results in Table 1. Aliphatic hydrocarbons in the range C10±C19 were detected; 10 of the
14 samples contained C14±C18 n-alkanes. This is
consistent with other research into the evaporation
of oil droplets, where Raynor et al. (1996) found
C14±C18 n-alkanes in vapour originating from an oil
containing C14±C20 alkanes. Little material was
detected from the four most viscous mineral oil
MWF (MWF K, L, M and N). `MWF' H (the
hydraulic oil), despite having a hydrocarbon envelope in the chromatogram, did not contain any prominent peaks. The base oil was not a paranic oil
unlike the MWF analysed. Paranic base oils
would appear to be favoured over naphthenic base
oils for use in MWF because of their greater stability at high temperatures and because they are
cheaper. Some compounds were identi®ed which
may be additives to the MWF. These included alkyl
benzenes, diethylene glycol, an alkylated phenol,
fatty acid methyl esters, iso pentyl acetate, bornyl
acetate and 1,8 cineole. The last three listed are
most probably odorants and were detected at low
levels.
During the Tenax back up tube analysis, about
25% of the kerosene sample was lost during the
®rst 5 min; the mineral seal oil sample lost less than
5% in the ®rst 5 min (but 57% over 8 h). All of the
components of kerosene were identi®ed, and most
of the components of mineral seal oil were identi®ed (C19±C23 were not detected). The rate of
sample loss re¯ects the volatility of the components
present on the ®lters; kerosene lost each of its components quickly, whilst mineral seal oil lost most
171
components at a slower rate. Given that mineral
seal oil or similar oil is the main base oil in very
low viscosity mineral oil MWF, it is not surprising
that MWF of this type lost a large proportion of
sample during aspiration.
Little material was lost from mineral oil MWF
with viscosities around 20 cSt during aspiration.
When a base oil of similar viscosity (known as `100
solvent neutral') was analysed by GC±MS it was
found to contain very little material below C19
(nonadecane), the heaviest component lost from the
MWF in detectable quantities. This is consistent
with a CONCAWE (the oil companies internal
study group for conservation of clean air and
waterÐEurope) report (1986) which states that oil
mists from lubricants containing hydrocarbons
greater than C20 are not normally associated with
signi®cant levels of vapour. If the proportion of a
mineral oil MWF more volatile than nonadecane is
very low, then it appears that little material will be
lost from the ®lter during the sampling of its aerosol. In the NIOSH mineral oil mist sampling
method (NIOSH, 1994) the sampling procedure for
mineral oil mist is said to be applicable to mineral
oils with the formula Cn H2n‡2 , where n is greater
than or equal to 16. This approach may identify
those oils which can be sampled without signi®cant
losses from the ®lter, but would be dicult to apply
in practice because of the data being unavailable. In
1975 the Institute of Petroleum Occupational
Hygiene Subcommittee stated that the threshold
limit value (TLV) used as a guide at the time only
referred to oil mist and not vapour, and warned
that accurate sampling of particulate mist from oils
containing a signi®cant proportion of relatively volatile components was dicult, but did not o€er any
further guidance or criteria on this problem (Turner
et al., 1975). CONCAWE warn that the methods
described in their report may not be applicable to
more volatile products, and cite low boiling range
oils and products containing a signi®cant proportion
of
relatively
volatile
components
(CONCAWE, 1981), but do not attempt to nominate a limit. They suggest that in such products
vapour concentrations are more likely to be the
appropriate criteria for assessing exposures.
Little has been reported on the concentrations
and e€ects of the vapour found with oil mists, but
it is considered to be present in potentially much
higher concentrations than the particles, and also to
be much less toxic (CONCAWE, 1981). It was considered less toxic because it was thought that,
although the vapour penetrates the lungs to the
alveoli and a portion will dissolve in the mucous
membrane, it is then exhaled, whilst the aerosol will
be deposited on the walls of the lungs as droplets
(Beviz, 1975). Once deposited, oil particles may be
removed by clearance mechanisms or absorbed into
body ¯uids and therefore are capable of causing
172
A. T. Simpson et al.
both local and systemic toxic e€ects (CONCAWE,
1986). As a consequence it was considered that
measured oil mist concentrations should refer to
particles alone, and contain no contribution from
the vapour. The problem of which phase to sample
and which standard to apply is not particular to
airborne mineral oil but applies to other relatively
high boiling liquids. Thus, separate mist and vapour
standards (Occupational Exposure Standards, 10
and 60 mg mÿ3 respectively) are listed for ethane1,2-diol (boiling point 196±1988C), whereas diethylene glycol (boiling point 2458C) has only a single
limit (23 ppm or 101 mg mÿ3).
The case of oil is complicated by it being a variable mixture of compounds which cover a range of
volatilities, resulting in signi®cant variation in the
distributions between liquid and vapour phases
encountered. The traditional method of estimating
exposure is via sampling of the mist phase onto a
®lter, with no account taken of the vapour phase. If
both vapour and particulate were to be taken into
account then they could be sampled together simultaneously and expressed as exposure to total airborne oil, avoiding the problem of evaporation of
mist trapped on the ®lter. However it seems likely,
given the di€erences in deposition and uptake and
the potential di€erences in health e€ects which may
result, that the measurement of mineral oil exposure
should involve the separate (possibly still simultaneous) measurement of aerosol and vapour. From
a practical point of view, distinguishing between the
mist and vapour may be dicult, especially for the
very low viscosity oils.
CONCLUSION
Mineral oil mist samples lose material during
sampling because of evaporation of volatile components. The longer a sample is subject to aspiration the more material is likely to be lost. The
material lost is mostly aliphatic hydrocarbons in the
range C10±C19, but can include additives in the oil.
The relationship between the amount of material
which could be lost during sampling and MWF
composition can be predicted from the MWF viscosity or ¯ashpoint data regularly found on the
product data sheet provided by the manufacturers.
Viscosity was the favoured method as it was considered to be a clearer and more useful indicator of
whether the mist from an oil was suitable for ®lter
sampling. Oil mist samples of mineral oil MWF
with viscosities greater than about 18 cSt at 408C
(or ¯ashpoints greater than 01808C) can be
expected to lose less than 5% during 8-h sampling
because of evaporation from the ®lter. Below 18
cSt, losses are more variable and generally greater.
It is proposed that 18 cSt could be taken as a guide
value below which there may be losses during
sampling for oil mist. Over a quarter of MWF for-
mulations in the UK have viscosities below 18 cSt.
Filter sampling may underestimate mist exposure
from such MWFs. For very low viscosity MWF
(<5 cSt) any occupational exposure measurements
taken may seriously underestimate mist concentrations and will also not indicate the high levels of
oil vapour likely to be present. The current UK
mineral oil OES applies to exposure to oil mist and
takes no account of simultaneous exposure to hydrocarbon vapour from the mineral oil. For low
viscosity oils it may be appropriate to measure both
the mist and vapour. Currently there is no validated
method of measuring mixed phase exposure to mineral oil mist but further experimental work is
planned.
AcknowledgementsÐWe acknowledge the comments and
suggestions of Dr R. Gardner, HSE O€shore Safety
Division.
REFERENCES
Beviz A. (1975) Oljedimma: EgenskaperÐMaÈtningÐ
Skydd
(Oil
Mist:
PropertiesÐMeasurementÐ
Protection). Arbetsmiljo 8/75 (available from HSE
Translation Services, HSE transl. no. 15128).
CONCAWE. (1981) Guidelines for the Determination of
Atmospheric Concentrations of Oil Mists. Report no. 1/
81, CONCAWE, Madouplein 1, B-1030, Brussels,
Belgium.
CONCAWE. (1986) Health Aspects of Worker Exposure
to Oil Mists. Report no. 86/69, CONCAWE,
Madouplein 1, B-1030, Brussels, Belgium.
HSE (1991) Metalworking Fluids: Health Precautions,
Guidance Note EH 62. HSE Books, ISBN 0 11 885667
7.
HSE (1997) General Methods for Sampling and Gravimetric
Analysis of Respirable and Total Inhalable Dust,
Methods for the Determination of Hazardous Substances
MDHS 14/2. HSE Books, ISBN 0 7176 1295 3.
HSE (1999) Occupational Exposure Limits 1999, Guidance
Note EH 40/99. HSE Books, ISBN 0 7176 1660 6.
Leith, D., Leith, F. A. and Boundy, M. G. (1996)
Laboratory measurements of oil mist concentrations
using ®lters and an electrostatic precipitator. American
Industrial Hygiene Association Journal 57, 1137±1141.
McAneny, J. J., Leith, D. and Boundy, G. B. (1995)
Volatilization of mineral oil mist collected on sampling
®lters. Applied Occupational and Environmental Hygiene
10, 783±787.
Menichini, E. (1986) Sampling and analytical methods for
determining oil mist concentrations. Annals of
Occupational Hygiene 30, 335±348.
NIOSH (1994) Mineral oil mist. In Manual of Analytical
Methods, Method 5026, 4th ed. National Institute of
Occupational Safety and Health, Cincinatti, OH.
Raynor, P. C., Cooper, S. and Leith, D. (1996)
Evaporation of polydisperse multicomponent oil droplets. American Industrial Hygiene Association Journal
57, 1128±1136.
Soderholm, S. C. (1988) Aerosol instabilities. Applied
Industrial Hygiene 3, 35±40.
Turner, D., Baxter, H. G., Davies, G. M., Eyres, A. R.,
Lees, G. L., Mitchell, J., Sanderson, J. T., Steel, J. and
Moore, J. P. (1975) Methods for the determination of
the atmospheric concentration of oil mist. Annals of
Occupational Hygiene 18, 293±297.