Sulfur-Sulfate History for IMPROVE

Author:  Robert Eldred, CNL, University of California, Davis (raeldred@ucdavis.edu)

Date:  June 15, 2001

The purpose of this history is to document changes in the IMPROVE system that might affect the sulfur or sulfate concentrations.  There were three issues that could produce invalid data.  Since the two most important issues primarily affected eastern sites, the history will concentrate on eastern sites.  The history is shown in Table 1. 

Issue 1.  Sulfate on nylon, all sites.  Before 3/15/89, a significant number of sulfate concentrations were unrealistically low.  This was an analytical error with IC that was corrected.  The nitrate concentrations were retained.  The primary season involved was Winter 1988, when 43% of the sulfate concentrations were invalid.  This issue is not significant for trends beginning in 1990.  The issue is documented in Appendix I. 

Issue 2.  Sulfur on Teflon, Eastern sites only.  X-ray analysis on Teflon filters can underestimate ambient sulfur under certain conditions—very high RH (approaching 100%) during a large fraction of the collection period, high sulfate acidity, and high face velocity.  It is hypothesized that with these conditions sulfate may migrate away from the center of the filter and perhaps even be lost from the filter.  The result is much lower sulfate (sulfur times 3) on Teflon than sulfate on nylon.  The loss was never observed at western sites.  The loss at eastern sites was predominantly between June and August.  Since face velocity is one of the factors, decreasing the face velocity by removing the collection mask will minimize occurrences.  However, it will not eliminate them.  Several significant differences were observed at Washington DC even with unmasked filters.  On 8/16/95, significant differences were observed at four sites after the mask was removed.  The effect was observed on a few samples before the summer of 1992.  The number of events significantly increased in summers 1993 and 1994.  In May 1995, the mask was removed at the eastern sites with highest ambient concentrations of PM_2.5.  Since that time, the loss was observed at these sites on only one day, 8/16/95.  Four other eastern sites in the far north and far south did not have the mask removed until midsummer 1998.  A few events were observed in 1993 and 1994.  Except for 8/16/95, there were no significant losses during 1995-1997 despite the presence of the mask.  This again shows that face velocity is only part of the cause.  The issue is documented in Appendix II. 

In should be noted the mask was removed from Module A Teflon filters for two reasons:  (1) to minimize clogging of Teflon filters at sites with high mass loading, and (2) to decrease the face velocity, thereby decreasing the loss of sulfur from Teflon A filters in conditions of high RH and possibly high acidity.  Thus, the mask was removed at sites with potentially high mass loadings, even if there were no significant sulfur loses.   

Issue 3.  Sulfate on nylon, mostly eastern sites.  In 1998, the manufacturer of the nylon material (MSI) inadvertently changed the specifications, increasing the pressure drop.  The material from Whatman had an even higher pressure drop, so it would not have helped to change manufacturer.  (Gelman had stopped producing nylon filters for legal reasons.)  With the high pressure drop, the 25mm filter was just at the edge of acceptable flow rate.  With high ambient concentrations, the new filters had a pressure drop that significantly reduced the flow rate during the period of sample collection.  The loss was primarily at eastern sites, but did include several events at western sites.  The special difficulty is that the loss is biased toward samples with high mass and high reconstructed extinction.  The issue is documented in Appendix III. 

Table 1.  History of changes.  The dates are for sample collection.

Appendix 1.  Sulfate Analytical Problem before March 18, 1989

Because of analytical problems, the sulfate concentration was occasionally underreported by RTI during the initial five quarters.  Figure 1 shows the fraction of samples invalidated for each sampling period during the five quarters.  Figure 2 shows that the problem was not associated with the concentration of sulfate; the distribution of sulfur on Teflon was the same for valid and invalid sulfate.  The sulfate values in the database after validation are probably acceptable.  The last data with this problem was 3/15/89.

Figure 1.  Fraction of  samples with invalid (low) sulfate on the nylon filter and apparently normal nitrate. 

Figure 2.  Distributions of sulfur for samples collected between March 1988 and February 1989.  One distribution is for samples with valid sulfate measurements and the other for invalid sulfate.

Appendix II.  Sulfur Loss on Teflon Filters.

Normally, the sulfur on Teflon and the sulfate on nylon have correlated extremely well.  A comparison of 27,274 points in the western US is shown in Figure 3. 

Figure 3.  Comparison of sulfur and sulfate at all western sites except five in California.  Adding these five would have made no difference.  There are 27,274 points in the plot.

However, there was poor agreement summer (June-August) at some Eastern sites.  The problem was most acute during the summers of 1992-94.  For a large number of samples at Eastern sites in summer, the sulfur on Teflon was much lower than sulfate on nylon.  The phenomenon was never observed in the West, and rarely in other months in the East.  The mechanism has never been fully understood, but probably is associated with very high relative humidity, acidic sulfate particles, and high face velocities.  One hypothesis is that under certain conditions, the sulfuric acid becomes saturated and migrates outside the x-ray analysis area.  When the collection area was changed from 2.2 cm² to 3.5 cm² (the full 25 mm filter) in 1995, thereby reducing the face velocity from 170 cm/s to 110 cm/s, the discrepancy disappeared, except for one day in August 1995. 

The effect is shown in Figure 4 to Figure 8.  The first two sets (Figure 4 and Figure 5) show the comparison for each year at the 11 eastern sites that had the mask removed in 1995.  These are Acadia, Lye Brook, Brigantine, Shenandoah, Dolly Sods, Jefferson (James River Face), Mammoth Cave, Great Smoky Mountains, Shining Rock, Upper Buffalo, and Sipsey.  In 1988 to 1991, there were only a few samples in which sulfur was significantly low.  In 1992 to 1994, the fraction and severity of events increased significantly.  Perhaps there was a change in ambient conditions, with more days of near 100% RH.  Perhaps, the Teflon filter material was slightly different.  In May 1995, the mask was removed from the Teflon filters at all of these sites.  After that, the losses were observed on only one day, 8/16/95. 

Figure 4. Sulfur vs sulfate at 11 Eastern sites, 1988-93.  The masks were removed in May 1995.  Thus all samples had the collection mask. 

Figure 5.  Sulfur vs sulfate at 11 Eastern sites, 1994-99.  The masks were removed in May 1995.  Thus all summer 1995 samples were collected with no mask.

Figure 6 and Figure 7 show the annual comparisons at the four sites whose masks were removed in 1998.  There were no events in the first year at Okefenokee.  In 1993-94, these were two events at Okefenokee and two at Chassahowitzka.  Note that three of the events occurred during the same period (8/18/93 at two sites, and 8/21 at one site).  In 1995, two events were observed on 8/16/95, the same day are the events at four other eastern sites.  From these plots, it does not appear that removing the masks in 1998 was a deciding factor. 

Figure 6. Sulfur vs sulfate at 4 Eastern sites, 1994-95.  The masks were removed in 1998.

Figure 7.  Sulfur vs sulfate at 4 Eastern sites, 1996-99.  The masks were removed in 1998.

Finally, Figure 8 shows the comparison at Washington DC, which had the mask removed in January 1991, between the first and second plots.  This was much earlier than at other sites.  What is observed is that these are a few events between 1991 and 1994 and no events between 1995 and 1999.  There was no change in the mask size between these two periods.  The conclusion is that removing the mask at the 11 Eastern sites may not have been the sole reason for reducing the number of events.

Figure 8.  Sulfur vs sulfate at Washington DC.  The mask was removed in January 1991, between the first and second plots.

Possible Causes of Sulfur Loss

Sulfur is measured on the Teflon filter by both X-Ray Fluorescence (XRF) and Particle Induced X-ray Emission (PIXE).  The values in the data base are from PIXE.  Both methods analyze the central section of the deposit with an area of 1.5 cm².  (IC, one the other hand desorbs sample from the entire filter.)   Several results have led to the hypothesis that RH, acidity, and mask size are the crucial factors. 

Relative Humidity.  Examination of the relative humidity at Shenandoah during the summer of 1993 showed that the samples with differences over 1 µg/m3 occurred during periods of higher RH:  an average of 5 of the 24 hours had RH at 100%, compared to 2 hours for samples with smaller differences.

Acidity.  Unfortunately aerosol acidity is not directly measured.  There is an indirect method by comparing hydrogen and sulfur (measured on the quartz filter), and organic carbon (measured on the quartz filter).  If organic hydrogen is estimated from organic carbon using ratios in the western US, and the remaining hydrogen is assumed to be from either H₂SO₄ or (NH₄)₂SO₄, then it is possible to obtain a rough estimate of acidity.  This rough estimate indicated that the sulfur remaining in the analytical area of the Teflon filter was more acidic for samples with major losses than for normal samples.  The second argument is that the phenomenon is not observed in the west, even though some samples may be collected during elevated RH.  One of the main differences between East and West is sulfate acidity. 

Mask size.  Tests conducted in Dolly Sods in summer 1997 showed that we could observe the effect with a 2.2 cm² mask but not with no mask.  (At this time, the normal sampler had no mask.)  These tests showed that removing the mask did indeed help.  Since removing the mask reduces the face velocity from 170 cm/s to 106 cm/s, we presume that the reduced face velocity removed the effect for these samples.  We assume that the other factors were sufficiently large on 8/16/95 that even with the lower face velocity, the effect was observed at 3 or 4 sites. 

Other field tests.   

In a special study at Great Smoky Mountains in summer 1994, 4h and 12h measurements were made with Teflon and nylon.  All modules had a denuder except the 12h Teflon.  The 4h Teflon agreed with the 4h nylon and 12h nylon, but the 12h Teflon sometimes gave lower concentrations.  Since the face velocity, RH, and acidity were the same for both 12h and 4h Teflon filters, some other factor had to be involved.  There are two possible hypotheses.  One is that the migration takes several hours.  A more likely hypothesis is that the carbonate denuder moderated the RH at the filter.  The amount of water that the denuder could retain is finite, but the denuder might reduce the RH at the filter when the ambient RH is at 100% and later add some water when the ambient RH decreases.  The conclusion would then be that the RH must be almost 100% for the effect to occur. 

In the summer 1995 SEAVS study in Great Smoky Mountains, a series of tests were conducted to examine the phenomenon.  Unfortunately, there were no cases of the loss of sulfur from Teflon, so no information was obtained.  During this study, there were no days with extreme RH.

IC analysis of Teflon filters.  In order to answer the question of whether the sulfur remains on the filter but simply migrates outside of the analytical area of the PIXE and XRF analysis, selected Teflon filters were analyzed by IC after the XRF and PIXE analyses.  Teflon filters routinely reanalyzed by PIXE show that no sulfur is lost during the PIXE analysis.  Thus, the IC measurement on Teflon should give the sulfate on the entire Teflon filter.  The set included some samples with major losses and some samples with no losses.  The results are shown in Figure 9.  The first plot shows that there are several samples with major differences between SO₄ on nylon and S3 on Teflon.  The second plot compares both IC measurements on the two filter types, and the third plot compares the two measurement methods for the same Teflon filter.  The fourth plot compares some derived parameters.  The x-variable is the fraction difference between the IC on nylon and PIXE on Teflon, while the y-variable is the fraction of lost sulfate that could be explained by migration outside the x-ray analytical area to somewhere else on the filter.  A value of 0 indicates that the two Teflon measurements agree, while a value of 1 indicates that the two IC measurements agree.  Only samples where the loss exceeds 15% were included, in order to get away from statistical uncertainty. 

Figure 9.  Comparison of IC sulfate on nylon [SO4(N)], IC sulfate on Teflon [SO4(T)], and PIXE sulfur on Teflon [S3(T)].  The IC analysis on the Teflon filters was performed after the XRF and PIXE analyses.  The fourth plot compares the fraction of loss explained by migration with the fractional difference between SO4(n) and S3(T).  Only samples with a difference between IC sulfate on nylon and PIXE sulfur on Teflon greater that 15% are included.  The samples are differentiated as to whether the IC on Teflon agrees more with the IC on nylon or the PIXE on Teflon, as shown on the fourth plot.  The open squares are 8 samples in which the agreement is best with IC on nylon.  For the 10 solid squares, the agreement is better with PIXE on Teflon.  The agreement is about equal for the 9 diamonds. 

In all cases, the IC on Teflon was intermediate between the IC on nylon and the PIXE on Teflon.  There were 8 cases where the IC on Teflon agreed much better with the IC on nylon.  The conclusion here is straightforward.  The sulfur must have remained on the filter but simply migrated outside the view of PIXE.  These had migration parameters of 0.7 to 1.0.  There were 10 cases where the IC on Teflon agreed much better with the PIXE on Teflon. These had migration parameters of 0 to 0.3.  There most reasonable interpretation is that the sulfur somehow was completely removed from the Teflon filter.  Finally, there were 9 cases cases when the IC values on Teflon were approximately halfway between the PIXE on Teflon and the IC on nylon.  In these case, some of the sulfur remained on the filter but migrated further out, and some of the sulfur disappeared from the Teflon filter completely.  This mixture of effects was also obtained by scanning the entire filter using a focused PIXE beam.  On a few filters, sulfur was observed on the outer portions of the filter, but on most filters no additional sulfur was observed. 

Sampler artifact.  The phenomenon was not an artifact of the IMPROVE sampler.  Differences were observed in CASTNet samplers in the East using Teflon filters with very high face velocities.   The Teflon filters at that time were analyzed by UCD. 

Appendix III.  Clogging of nylon filters.

Beginning in spring 1998, MSI increased the pressure drop across the filters.  The pressure drop was even higher with filters manufactured by other companies.  The importance of this change was that it resulted in the invalidation of a significant number of nylon filters and the corresponding sulfate.  The Table 2 shows the fraction of samples in which the flow dropped 25% and the samples were invalidated.  (A value of 4% or 5% corresponds to one sample lost during the quarter.)  This category of major clogs accounted for 60% of the loses on the nylon filter when no other problem was reported.  Most of the sites with major losses are in the east.  Non-eastern sites with major loses were Sequoia and Seattle.  Lye Brook had 7 clogged filters, while the other New England sites (Acadia and Moosehorn) had one clogged filter each.  Great Gulf had none.  The loss has been minimized in the Version II samplers by increasing the nylon filter diameter from 25mm to 37mm.

Table 2.  Loss of nylon filters when the flow dropped more than 25% (major clogs).