Dust in buildings with man-made mineral fiber ceiling boards

SCHNEIDER T, NIELSEN 0, BREDSDORFFP, LINDE P. Dust in buildings with man-made mineral fiber ceilingboards. Scand J Work Environ Health 1990;16:434-9. Man-made mineral fibers (MMMF) and other airborne dusts were measured in 105rooms in a representativesample of public buildings,ex cludingrooms with physicallydamagedboards or buildingswith notable indoor climateproblems. There wereno differencesin the MMMF concentrations with respectto the type of binder. The averageconcen trations ranged from 17to 210 respirableMMMF/m'. The averageconcentrations of the referencegroup was intermediate and therefore indicated that sources other than ceilingboards contributed to the ob tained values. No grouping by concentration of MMMFon cupboards waspossible. Airborne concentra tions of respirableMMMFwere2.7 timeslowerin rooms with mechanicalventilation than in rooms with natural ventilation. For nonrespirable MMMF, the most important factor wasthe quality of the cleaning. The concentration in poorly cleanedrooms was5.5timesthat of well-cleaned rooms. Ventilation, quality of cleaning, and number of persons affected the non-MMMF and total dust concentrations.

fibers were known to be present in concentrations that were much higher than those of MMMF (2). There were also indications that MMMF on surfaces may be responsible for some cases of skin itching and irritation of the upper airways and eyes (2). Therefore we decided to measure the MMMF concentrations in air and on surfaces, non-MMMF fibers in air, and total dust in air in a random sample of Danish buildings with MMMF ceiling boards. Preliminary results of this study have been presented (6), and the results have been quoted in reference 1. This paper presents results from the final report (7) with additional data and analyses.

Materials and methods
A sample of rooms with MMMF ceiling boards of types covering the Danish market and building habits was selected from nurseries, kindergartens, schools, and offices listed by the municipalities in and around Greater Copenhagen. Rooms having physically damaged boards or notable indoor climate problems were excluded. As a result, nine different types of MMMF ceiling board and one type without MMMF (reference group) were included in the study (table 1). The following building/room-related background variables were used: (i) type of room, (ii) number of persons in the room during the measurement, (iii) quality of cleaning subjectively ranked as good, medium, and poor with the number of weekly cleaning hours relative to the building area as the main criteria, (iv) floor number, (v) air temperature, (vi) air humidity.
No changes in the usual activities were introduced. There was no information on smoking, but smoking would not have occurred in nurseries, kindergartens, and schools.
Airborne dust was sampled at three stationary sites in each room, 1.1 m above the floor. Two of the sites were for fibers and one was for total dust.
The sampling duration was typically 1-2 h, and 37-mm filter cassettes were used. Fibers were sampled on 1.2 urncellulose acetate filters in open cassettes with a protective aluminum cowl, facing upward, and a face velocity of 0.04 m/s, Total dust was sampled on 3-J.lm Fluoropore'" (Millipore) filters in closed-face cassettes with an entry velocity of 1.25 m/s. The total dust filters were weighed on a microbalance in a room at a constant temperature and relative humidity (50 ± 3 070), with a precision of 20 ug.
MMMF and other fibers were analyzed with standard positive phase contrast microscopy under an objective with a magnification of 40 x . A polarizer and an analyzer were added to test for birefringence as described in reference 2. The composition of the non-MMMF was not determined, but their visual appearance led us to believe that the majority were organic. The analytical sensitivity (ie, concentration corresponding to one fiber detected) was 70 MMMF/m' and 230 non-MMMF/m'. The same criterion for respirability (diameter less than 3 urn) was used for non-MMMF. The counts were corrected for blank values: Dust settled on horizontal nontextile surfaces was sampled with sticky gelatinous foils (2). The sampling sites were stratified. A table was considered to represent surfaces regularly cleaned and the top of a cupboard to represent surfaces not regularly cleaned. No attempt was made to determine the elapsed time since the last cleaning of the surface. The measurements thus reflect the state of contamination of the surface at the time of the sampling.
Settled MMMF were analyzed under a microscope with a transmitted light objective (magnification 20 x ) with a polarizer and an analyzer and with projection onto a digitizer for size determination. About 70 % of all the fibers detectable in a 40 x phase contrast microscope were seen with this method. All fibers longer than 50 urnor thicker than 3 urn were detected. This method is thus suitable for settled fibers, since the majority are larger than these sizes (table 10). The analytical sensitivity was 0.6 fibers/em.'. Blank foils gave no MMMF counts.
The distributions of the airborne fiber concentrations were very skew and with a large number of 0 fibers counted. On the average, 54 (range 0-84) 070 of all the samples in the groups defined by ceiling type had no detected fibers. The summary statistics were thus based on simple arithmetic averages. The result for each room was obtained as an average of the two parallel samples. Since the concentrations were the results of fiber counts and thus, ideally, would be Poisson-distributed, the data were transformed by square root as this procedure renders a Poisson distribution that is approximately normal. The transformed distribution was still skew, however. Average values of the data transformed by square root for the ceiling types were grouped homogeneously with Duncan's method of multiple comparisons of means (8). A homogeneous group was a group with no significant differences between the means of the members of the group.
The square-root data on airborne fiber concentration were analyzed and related to the background variables with a multidimensional analysis of variance of the SAS (statistical analysis system) (9). First, all nine variables were included. Then the least significant was omitted and the procedure was repeated until only variables having a level of significance of at most 5 % were left.
We decided to concentrate the statistical analysis of the MMMF on surfaces to total MMMF/cm' on cupboards. It was assumed that these surfaces would reflect possible differences between ceiling types better than the table surfaces. The square-root data on the total MMMF/cm' on cupboards were subjected to a two-dimensional analysis of variance. The first dimension was ceiling types (on 10 levels), and the second dimension was the quality of cleaning (on two levels) (table 5).

Results
The results for the airborne MMMF concentrations averaged for each ceiling type are shown in table I . The standard deviations in table 1 are measures of the total variation, which consisted of two independent sources, namely, the variation between two parallel samples and the variation between averages for rooms. The standard deviation due to parallel samples was proportional to the average concentration and was of the order of 100 %. This large value can be explained by the low fiber concentrations in comparison with the analytical sensitivity.
For both respirable and nonrespirable airborne MMMF, the ceiling boards could be subdivided into three homogeneous groups. Table 2 shows the grouping according to respirable fibers. In the same table, the concentration and the groups corresponding to nonrespirable fibers are shown.
For the respirable fibers that were airborne, the only significant background variables were type of ceiling and ventilation. For the airborne nonrespirable fibers the only significant background variables were type of ceiling and quality of cleaning. Since ventilation had an effect on the concentration of respirable fibers and since, in the various groups in table 2, there was a vari-    able prop ort ion of buildings with different types of ventilation, we checked whether these factors would affec t the gro uping. It was found that correction for the ventilation effect would have strengthened the difference between the three groups.
As shown in table 3, there were 65 rooms in which no airborne nonrespirable MMMF were counted among the 95 roo ms measured , but in most of these roo ms nonr espirable MMMF were found on cupboard s. In onl y eight of the room s were no fibers  counted. On the whole, there were 10 room s in which no nonr espirable MMM F were cou nted on the cupboard s. In eight of these 10 room~no airborne nonrespira ble MMMF were count ed either. It can thu s be concluded that measurement of the surface concentration has a greater ability to detect the presence of fibers than measurement of the air concentration. Ta ble 4 shows the surface concentration of total MMMF for the tables and the cupboards. Table 5 shows the surface con centration of total MMMF accordin g to the quality of the cleaning. Since the mean for medium quality was a little less than the mean for high qua lity, these data were pooled.
The two-dimensional an alysis o f varia nce for total MMMF on cupboards showed that type of ceiling could not be shown to be a significant factor in explaining the variation in the data, whereas quality of cleanin g was significant at a 3 lJIo level.
No significant subdivision of the averages over ceiling types into homogenous groups was found by Duncan's method , even if the level of significance was raised from 5 to 10 lJIo.
Concen trations are given according to the type of building in table 6 for non-MMMF and for total dust. It should be not ed tha t both conta minants ranked the building types in the same way.
Th e number of persons present influenced the MMMF, non-MMMF, and total du st con centrations as shown in tabl e 7.
Type of ventilation and quality of cleaning influenced the airborn e concentration s as shown in tables 8 and 9.

Discussion
A t-test on the data transformed by square-root showed that there was no significant difference in airborne MMMF between th e two binder types, neither for respirable (P =0.19) nor for nonrespirable (P =0.78) airborn e MMMF. As has alread y been stat ed , only roo ms with und ama ged ceilings were included.
Ceiling boards with no surface treatment and ceiling board s of an old type with water-soluble binder (taken out of production several years ago) were both placed in the high concentration group with regard to respirable and nonrespirable airborne MMMF concentrat ions. The reference gro up was placed in the medium group for both respirable and nonrespirable airborne MMMF. Therefore there were evidently other MMMF sources than ceilings. Infiltration from outdoor air could not be evaluated since no outdoor measurements were mad e, but it is unlik ely that infiltration would cont ribut e with nonre spirable fibers. Actually, all rooms without MMMF ceiling board s were on the top floor or in on e-floor houses and thus in close pro ximity with the massive amount of MMMF present in Table 6. Airborne nonrn an-rnade min eral fiber (non·MMMF) and total dust concentrati o ns.   the thermal insulation in the roof, which is thu s believed to be one of the sour ces. In mechani cally ventilated rooms (suppl y air) MMM F in silencers or ducts could be another source.
No group ing of ceilings was possible by surface concentration of MMMF on cupboards, even if the level of significance was raised to 0.1. ActualI y, no statistically significant correlation between the concentrations of total airborne and total settled fibers was found. A normalization of the surface concentra tion with respect to number of days since last cleaning would probably have improved the correlation. This information was not available, and reliable information may prove difficult to obtain. As demonstrated previously by others (10), sur face sampling was more sensitive than 438 air sampling for detecting the presence of MMMF in the indoo r environment.
The background variables affected the concentrations as expected . It is intere sting to note that the only stat istically significant influ ence on the respirable MMMF concentra tio n was ventilation, and for the nonr espirable MMMF the only statistically significant influence was the quality o f the cleaning. These findings support theoretical calculations (II ) showing that , whereas even low ventilat ion rates are efficient in removing respirable fibers, a ventilation rate of one air change per hour has very little effect on nonrespirable fibers. Settlin g is more efficient in removing nonrespirable fibers from the air. Table 10. Data on airborne and settled fiber size fitted with a bivariate log-normal distribution. (MMMF = man-made mineral fibers, CM = count median, GSD = geometric standard deviation, D= diameter, L= length, 'DL= correlation between log. D and log. Size distributions of MMMF can be described by the bivariate log-normal distribution (12). The parameters ch aracterizing the distribution have been calculated from the pooled and grouped data according to a procedure given in reference 13. The results, including data from two other studies to which the authors had acces s, are shown in table 10. Notice that optical microscopy was used for the airborne fibers and thus the length (5 urn) and diameter (0.2 urn) were truncated (14). Since , in our study, the size of settled fibers was determined with a digitizer, more accurate data are available for these fibers.
The number of days since the last cleaning would not affe ct the size distribution of the set tled fibers. If it is assumed that there is no other size-selective redistribution of sett led fibers, one can calculate the size distribu tion on the cupboard as the size di stribution in the air , weighted by the settling speed, or vice vers a . By us ing the theoretical expression given in reference 15, we calculated the di stribution for the airborne MMMF given in table 10 from the size distribution of the settled fibers. There was very good agreement with the measured size distribution .
The number of non-MMMF wa s orders o f magnitude gr eater than that of MMMF and correlated strongly with the total dust concentration. Therefore, the mass concentration of non-MMMF was estimated. This was done in the following way: the average volume of a fiber in each size class was calculated under the assumption that the fibers follow th e size distribution given in table 10. The value was not very dependent on the actual size distribution . Each non-MMMF fiber was then given thi s volume. Sub sequently, the total volume of all fibers was correlated with th e total du st concentration determined in the sa me room . Assuming an a verage density of I g/cm' , we found that the non-MMMF fibers constituted 15-20 % of the weight of the total dust.
Thi s study shows that it is possible to maintain MMMF co ncentrations below about 200 respirable fibers /m'. Such concentrations present a very low risk, if an y, to the general population (I) . It can be concluded th at me chanical ventila tio n, in combination with go od and frequent cleaning, ca n be used to control the gen eral dust level in room a ir.