Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

Atmospheric EnvironmentVol. 27A,No. 15, pp. 2441 2444,1993. 0004-6981/93$6.00+0.

00
Printed in Great Britain. © 1993PergamonPress Ltd

TECHNICAL NOTE

AN IMPROVED PROCEDURE FOR ANALYSIS OF PMi0 FILTERS BY X-


RAY POWDER DIFFRACTION

BRIANT L. DAVIS a n d H o N o CHEN


X-ray Diffraction Laboratory, South Dakota School of Mines and Technology, Rapid City,
SD 57701-3995, U.S.A.

(First received 2 February 1993 and in final form 8 June 1993)

Abstract--Speciation results from earlier methods of quantitative analysis of TSP, dichotomous, and PM lo
filters for compounds by XRD were often marginal to unacceptable because of very light loadings and the
interfering effects of substrate fibres in cases of ultrasonic extraction and concentration. A method for
analysis of very lightly loaded PM to filters has been developed, using ultrasonic particle stripping, X-ray
transmission of unloaded blank filters, substrate diffraction X-ray analysis, and mass absorption balance,
along with the usual X-ray diffraction scanning. Analysis of a set of filters having less than 105 #gm crn-2
sample load is used to illustrate the improved procedure.

Key word index: Atmospheric aerosols, X-ray diffraction.

1. INTRODUCTION AND BACKGROUND 2. DESCRIPTION OF THE NEW PROCEDURE

The X-ray reference intensity method (RIM) of X-ray pow- Complete quantitative evaluation of both crystalline and
der diffraction analysis is potentially a powerful procedure amorphous components of a filter load may be completed
for determination of species of atmospheric pollutants. providing the following conditions exist.
Speciation of ambient aerosols, combined with similar quan-
(a) Accurate tare and load weights are available for the
titation of source materials, can be used in source apportion-
whole filter (assumed to be 8" by 10" [20.34 x 25.4 cm],
ment (Johnson et al., 1984; Gorden et al., 1984; Davis, 1984a)
although the principle applies to any size filter).
either alone or in combination with elemental data. The
(b) A blank filter is available for direct mass absorption
major drawback of X-ray diffraction analysis is its relative
measurement of the glass material comprising the filter.
insensitivity to mass compared to X-ray fluorescence. Never-
theless, compounds are more diagnostic as signature compo- If these conditions exist then it is, in principle, possible to
nents than elements; for example the element calcium may ultrasonically remove the sample, along with contaminating
have many sources but CaSiO 3 has only two sources; a com- fibres of the filter substrate, and still be able to both correct
ponent of portland cement, and as a rare mineral component the intensitites for full quantitative analysis and complete
of metamorphic rocks. a mass absorption balance to determine the weight fractions
In the Quail Roost study (Stevens and Pace, 1984; of one or (with additional optical information) two amor-
Johnson et al., 1984) XRD analysis proved adequate to phous components in this sample.
determine the major components of the aerosol, but minor The details of this procedure (except for the RIM analysis
components were undetected or unquantifiable, and which is summarized in Davis, 1984b, 1992) are presented here
amorphous components could not be quantified because of in order that applications may be readily made to real samples.
insufficient material for mass absorption analysis. In the Equations required in each step are noted as (A-l), etc., and are
Quail Roost study loads of only 50-200 #g cm-2 were avail- stated in the Appendix. The procedure is accomplished ac-
able on Teflon (PTFE) filters; such load levels are marginal cording to the following step-by-step instructions:
at best for quantitative XRD analysis.
In recent fugitive dust source apportionment studies (un- 1. From two to four circles are cut from the blank filter for
published) of Whatman QMA "Quartz" PM l o filters, speci- either direct-beam transmission (equation (A-l)) or substrate
fic load levels of 100-300/~g cm - 2 on the original filter were diffraction analysis (equation (A-2)). This provides the mass
increased by ultrasonic extraction in methanol and the sus- absorption coefficient of the glass material of the filter (#o).
pension concentrated by liquid filtration onto 25-mm VM-1 2. The loaded filter is cut into large (we suggest 6-8 cm)
filters for XRD analysis. The resulting specific masses of the circles and weighed (Mr... M,3 using a sensitive micro-
XRD sample "disk" were found to lie somewhere in the balance. The total filter area AT to be stripped so that
range of 1000-5000 # g c m - 2, however, a significant adequate specific mass can be concentrated onto a 2.5-ern
proportion of the sample consisted of glass fibres pulled diameter load area can be estimated from the equation
down from the filter during extraction. These loads provided Ar(cm 2) = 30exp [ --0.0024571] (#g cm- 2)
sufficient mass to permit evalution of minor clay components
(very weak diffracting materials), but only by rather inaccur- where AT/is the specific mass of the load over the ambient
ate and very tedious polarizing optical analysis of portions of fiter.
the composite sample could the amount of amorphous (car- 3. The cut circles are stripped (face down) in methanol for
bonaceous) component be estimated. Ordinarily routine about 2 rain. Longer stripping brings down more substrate
XRT and SDIF analysis could not be performed on these contamination and should be avoided. It is not necessary to
samples because of the glass fibre contamination. completely remove all ambient particles from the filter. Turn

2441
2442 Technical Note

each filter upside down in the methanol for 15 s at the end of The 10 steps presented above may, of course, be modified
the stripping to remove redeposited particles from the back in any manner necessary that will leave the analyst with the
side.
desired end result. For example, X-ray transmission may be
4. The methanol suspension is passed through an used in place of substrate diffraction to determine #no, pro-
aspirator filtration system, with collection onto a 2.5-cm viding that a device to hold small 2.5-cm loaded filters is
membrane filter of tare weight Mf (Metricel VM-I is recom- available. Our experience in attempting to suspend as
mended). The loaded filter is dried and weighed (Ms). With an aerosol the dried methanol suspension stripped from
a VM-1 filter area of 4.909 cmz and load coverage of 3.464 the original filters is not good; carbonaceous matter
cm 2 for our cassette the specific masses ~ f and ~ , ~ (g is sufficiently soluble so that self adhesion of particles pre-
cm -z) for blank filter and loaded filter may be computed. cludes use of aerosol suspension as a means for final sample
5. The stripped circles are dried and reweighed preparation.
(MI°... M°).
6. The reconcentrated VM-I load (with substrate fibre
contamination) is analysed by substrate diffraction to obtain 3. EXPERIMENTAL METHODS AND RESULTS FOR A PMto
the composite (sample + glass) mass absorption ~ao) using FILTER SET
(equation {A-3)).
7. The composite sample is scanned in the usual manner Three Whatman QMA "quartz" PMIo filters were
by X-ray diffraction, weighed and loaded at an outdoor site in the western U.S.
8. Quantitative RIM analysis (equation (A-4)) or other The original filter and loaded filter data are presented in
suitable method is applied to determine all crystalline com- Table 1. X-ray diffraction scans were completed with a Phil-
ponent weight fractions (Vii). lips Type 1205 generator, high-angle goniometer with scintil-
9. A mass absorption balance (equation (A-5)) is then lation detector, graphite crystal monochromator, automatic
applied to determine the true (uncontaminated) sample mass divergence slits, and circular sample spinner. The original
absorption coefficient (p.). filter circles were stripped with a Branson B-220H ultrasonic
I0. From the calculated mass absorption coefficient cleaner bath and filters were weighed on a Mettler H51AR
#Be obtained from the RIM analysis of crystalline compo- electronic balance readable to 10 #g. Extremely light load-
nents combined with #a from step (9), a second mass absorp- ings make the measurement of mass absorption by direct-
tion balance (equation (A-6)) is applied to determine the beam X-ray transmission subject to large experimental
amount of amorphous material (W.m) present and allow errors. Therefore the above procedure makes use of substrate
rcnormalization of the original crystalline weight fraction diffraction (SDIF) anaysis of the concentrated sample com-
W~ to yield the final set W[. bined with mass absorption balance to determine #,.

1200 EME$0638 .02/4 $/D 20000 IMX

M = MU$COUITF./ILLITE
110o
Q 0 = OLIGOCLI:ISE
I~ = QUARTZ
1000,
C = CRLCITE

900,

800.

¢0
tQ 7 0 0 •

600
0
f ORIGIHRL PH-IO
500
C
H Q HO 0 O
400

.. o IAI DO
300

o HO
200

"'-ENHANCED LORD 01t U!t-1


100 I I I I I i I I I I I
1::] 2b 21 22 23 24 25 26 27 28 29 30 31
l - e n e s 0 6 38, da t 2 Theta

Fig. 1. Diffractometer trace of original PMlo filter and trace of VM-1 filter loaded with concentrated
sample from ultrasonic extraction. Copper Ka radiation scanned at 0.02°C 20 step, and 4 s dwell.
Technical Note 2443

Table 1. Filter load data for three PM10 samples

Ambient conc. Tare wt Load wt Specific mass


Filter ID (#g m - 3) (g) (g) (/tg c m - 2)

SP1 13.9 4.2798 4.3008 52.3


SP2 18.9 4.3107 4.3405 73.0
SM 1 27.6 4.3318 4.3734 102.9

Table 2. Measured and calculated parameters fr three PMlo samples

Step, parameter SP SP2 SP3

(1),/a C (cm 2 g- 1) 36.26 36.26 36.26


(2), A T (cm 2) 176.7 220.9 220.9
M 1, M 2. . . . M n (g) 0.3616,0.3608 0.3733,0.3706 0.3774,0.3756
0.3703,0.3689 0.3731,0.3732 0.3747,0.3741
0.3728 0.3738
(3,4), Mf~g), ~/f(g cm -2 ) 0.00680,0.001385 0.00685,0.001395 0.00707,0.001440
MB(g), MBG(gcm -z) 0.01845,0.003363 0.02520,0.005297 0.03244,0.007324
(5), M 01, M o2. . . . M°(g) 0.3574,0.3566 0.3668,0.3654 0.3701,0.3676
0.3656, 0.3651 0.3676, 0.3651 0.3677, 0.3666
0.3666 0.3660
(6),/~B~ (cm 2 g- 1) 38.25 37.72 41.81
(7, 8), W~(Wt fraction)
Quartz 0.278 0.214 0.270
Calcite 0.418 0.254 0.384
Oligoclase 0.140 0.153 0.139
M uscovite/illite 0.164 0.200 0.146
Biotite -- 0.054 0.008
Kaolinite -- 0.125 0.052
(9),/~(cm 2 g- 1) 39.9 39. I 45.5
(10), Wbc, #b~(cm 2 g- 1) 0.675, 55.35 0.715, 52.31 0.821, 54.06
w',
Quartz 0.188(0.035) 0.152(0.027) 0.222(0.046)
Calcite 0.282 (0.058) 0.181 (0.047) 0.316(0.075)
Oligoclase 0.095 (0.021 ) 0.109 (0.020) 0.114 (0.024)
M uscovite/illite 0.110(0.028) 0.143 (0.029) 0.120(0.028)
Biotite -- 0.039 (0.016) 0.007 (0.004)
Kaolinite -- 0.090(0.027) 0.042(0.015)
Wa~ = Organic matter 0.325(0.101) 0.285(0.123) 0.179(0.149)

These load levels are generally below even that suitable for mass absorption balance of equation (A-6). The mass ab-
analysis on P T F E membrane filters (Davis and Johnson, sorption balance results may be verified by applying the raw
1982). Therefore the stripping/reconcentration procedure data of Tables 1 and 2 to the equations in Appendix A. The
outlined above was used for these analyses. Figure 1 presents advantage gained by ultrasonic concentration is evident
a diffractometer trace of the original unconcentrated filter from Fig. 1 and in the results of Table 2; only the weight
sample (cut directly from the filter) and the reloaded VM-1 fractions for calcite and quartz could have been determined
filter. Only the two strongest component peaks for quartz from the original filter diffraction trace, and for SPI the
and calcite are recognizable for the unconcentrated sample. amount of sample was too small with just four stripped
After completing the analysis of SP1 the decision was made circles for the biotite and kaolinite peaks to be observed.
to increase the number of 7.5-cm diameter circles stripped With five circles and a larger ~/B for SP2 and SM1 a satisfac-
fom 4 to 5. Otherwise the procedure for the analyses is tory quantitative evaluation of all significant components
identical. Table 2 presents the results of the procedure listed was achieved. There is no reason to suspect that the various
according to the 10 steps enumerated above. The mass components are being ultrasonically stripped at different
absorption coefficient of the composite sample (/~BG) ob- efficiencies.
tained in step (6) was used along with specific mass data to It should also be pointed out that X-ray mass absorption
correct the integrated intensities needed in steps (7 and 8) analysis cannot distinguish between "organic" and graphite
according to our earlier published procedures (Davis and carbon; the terms "organic matter" or "carbonaceous mat-
Johnson, 1982, equation (3)). Steps (7 and 8) provide a "pro- ter" are used here and in other papers to describe any
visional" analysis, using equation (A-4), resulting in the de- combination of organically derived or combustion derived
termination of weight fractions of all crystalline components carbon-bearing material; the mass absorption coefficient for
that could be identified in the concentrated sample. The final this material is set of 6 cm 2 g-1 for CuK, radiation (the
set of weight fractions, which includes values for both cry- average value of coal - - pure carbon is 4.2) to reflect a var-
stalline and amorphous components, were obtained from iety of oxygenated and nitrogen-bearing compounds in-
a second application of equation (A-4) combined with the cluded in most forms of carbonaceous matter.
2444 Technical Note

4. MAJORSOURCES OF UNCERTAINTY Stevens R. K. and Pace T. G. (1984) Overview of the math-


ematical and empirical receptor models workshop (Quail
The Reference Intensity Method (RIM) of quantitative Roost II). Atmospheric Environment 18, 1499-1506.
analysis has a complete variance error propagation algo-
rithm incorporated into the processing program (Davis, APPENDIX A
1981a). These errors are presented for the final weight frac-
tions W~ in parentheses in Table 2. Aside from these para-
meters, additional uncertainty may result from the presence Analytical equations for mass absorption balance
of more than one amorphous component, or because the (From Davis, 1984b; Davis and Johnson, 1987)
mass absorption coefficient of the amorphous component is
not known accurately. Mixtures of hydrous iron or manga- 1 [- lf-]
(XRT). (A-I)
nese oxides can create this undesirable situation, but such
occurrences in ambient aerosols would be uncommon. How-
ever, carbonaceous matter and silicious slag glass, fly ash, or #t is the mass absorption coefficient of the blank filter (glass),
limonite may be present together in many industrial environ- ~/r is the specific mass of the filter, If is the transmitted, direct
ments (Davis, 1981b); in these cases additional optical ana- X-ray beam flux through the filter, and lo is the unattenuated
lysis by polarized light microscopy would be needed to direct beam flux.
resolve the amorphous component contributions. The
weight data of Table 2 also provides an indication of the N s = - 1 ~ Fsin0gn/~ ] (SIDF). (A-2)
lateral homogeniety of the PM~o load distribution. The 2MfL l~d
means and standard deviations of the PMIo specific masses #fs is the mass absorption coefficient of the blank filter
for the 7.5-cm stripped circles for SPI, SP2, and SM1 are determined by substrate diffraction (SDIF) where Mt is the
found to be 96.55 (7.04), 142.32 (25.84), and 170.59 (9.63) specific mass of the filter, and Ifs and 1~o are intensities
/.tg cm 2, respectively. The homogeniety of SP1 and SM1 are diffracted from the substrate material at angle 20 through the
quite good; the set of stripped circles represent about two- filter, and without the filter in place, respectively.
thirds of the 8 x 10 inch original PMlo filter area.
This complete procedure has been tested only on PMlo 1 r l.a , . 'l
"quartz" filters; however, we have extracted for routine #BG=-2---~-~lsinO¢n--+2Mfgfl (SDIF). (A-3)
quantitative analysis particulate matter imbedded in high- M.~ L l~o J
volume filters which were known to consist of borosilicate #Be is the mass absorption coefficient of the composite
glass having mass absorption of about 48-50 cm 2 g-t. The sample plus substrate fibres (glass), MBo is the specific mass
results of the extraction process, including glass fiber con- of the composite sample, and l,Gs and l,o are the intensities
tamination levels, appear to be no different than for "quartz" diffracted from the substrate material at angle 20 through
microfiber filters. Therefore we believe that this technique the loaded filter, and without the loaded filter in place,
should be applicable to the high-volume filter as well. respectively, The filter effect is taken into account in this
equation by the second term within the square brackets.
#t may be replaced by /zts since it is the same quantity
measured by different techniques.
REFERENCES Fk. l.~-]-1
W j - - / ~ x-' l . (A-4)
Davis B. L. (1981a) A study of the errors in X-ray quantitat- kl~ k,J
ive analysis procedures for aerosols collected on filter Wi is the weight fraction of the jth component having
media. Atmospheric Environment 15, 291-296. reference intensity ratio ki and diffracting intensity lT in
Davis B. L. (1981b) Quantitative analysis of crystalline and a sample of i components with reference intensity ratios
amorphous airborne particulates in the Provo-Orem vi- k~and intensities l~. This equation is self-normalizing so that
cinity, Utah. Atmospheric Environment 15, 613-618. all the W~ sum to unity.
Davis B. L. (1984a) X-ray diffraction analysis and source
/ZBo= W~#f + WB#~ . (A-5)
apportionment of Denver aerosol. Atmospheric Environ-
merit 18, 2197-2208. #Bo is the mass absorption coefficient of the composite
Davis B. L. (1984b) Reference intensity quantitative analysis sample and glass fibre contaminant, Wo is the weight frac-
using thin-.layer aerosol samples. Adv. X-ray Anal. 27, tion of the glass material comprising the filter stripped by
339-348. ultrasonic cavitation, and Wa is the weight fraction of pure
Davis B. L. (1992) Quantitative phase analysis with reference sample material in the composite. #B is the quantity to be
intensity ratios. Accuracy in Powder Diffraction II, NIST determined in this process.
Special Publication 846, pp. 7-16.
Davis B. L. and Johnson L. R. (1982) On the use of various
[~B=~,Wi]~i-~Wam#a m . (A-6)
filter substrates for quantitative particulate analysis by Wi is the weight fraction of crystalline phases in the initial
X-ray diffraction. Atmospheric Environment 16, 273-282. RIM analysis; the ratios of these weight fractions are correct
Davis B. L. and Johnson L. R. (1987) The use of mass regardless of the presence of amorphous or unidentified
absorption in quantitative X-ray diffraction analysis. Adv. other phases in the sample and hence can be used to compute
X-ray Anal. 30, 333-342. the true mass absorption coefficient of the crystalline fraction
Gorden G. E., Pierson W. R., Daisey J. M., Lioy P. J., of the sample,/zBo which is equivalent to the term ~Wdz t in
Cooper J. A., Watson J. G. Jr and Cass G. R. (1984) this equation./h are the mass absorption coefficients of the
Considerations for design of source apportionment stud- crystalline components of the sample which must be meas-
ies. Atmospheric Environment 18, 1567-1582. ured or calculated prior to use this equation. W,m is the
Johnson D. L., Davis B. L., Dzubay T. G., Hasan H., weight fraction of amorphous material in the sample (not
Crutcher E. R., Courtney W. J., Jaklevic J. M., Thompson including filter matrix material which at this step has been
A. C. and Hopke P. K. (1984) Chemical and physical eliminated from the computations) to be determined, and
analysis of Houston aerosol for interlaboratory compari- #am is its mass absorption coefficient which must be known
son of source apportionment procedures. Atmospheric En- beforehand. Wsc (Table 2) is the crystalline weight fraction of
vironment 18, 1539-1533. the original sample and is numerically equal t o 1 - W~m.

You might also like