Anal. Chem. 1980, 52, 833-837 (11) Uhl, F . A. Fresenius' 2.Anal. Chem. 1937, 770, 102-117. (12) David, D. J. Analyst(London) 1961, 8 6 , 730-740. (13) Abu-Samra, A,; Morris, F. S.; Koirtyohan, S. R.; Vogt, J. R. "Proceedings of the 2nd International Conference on Nuclear Methods in Environmental Research", Vogt, J. R., Meyer, W., Eds.; University of Missouri-. Columbia, Mo., Conf-740707, 1974, 206-212. (14) Fishman, M. J.; Mallow, E. D. J . Water Polut. Control Fed. 1968, 40(2) (pt.2), R67-R71. (15) Chan, K. M.; Riley, J. P. Anal. Chim. Acta 1966, 3 6 , 220-229. (16) Kim, Y. S.;Zeitlin. H. Anal. Chim. Acta 1969, 46, 1-8. (17) Head, P. C.; Burton, J. D. J . Mar. &0/. Assoc. U.K. 1970, 50, 439-448. (18) Riley, J. P.; Taylor, D. Anal. Chirn. Acta 1968, 40, 479-485. (19) Riley, J. P.; Taylor, D. Anal. Chirn. Acta 1968, 47, 175-178. (20) Kawabuchi, K.; Kuroda, R. Anal. Chim. Acta. 1969, 46, 23-30. (21) Chau, Y. K.; Lum-Shue-Chan, K. Anal. Chirn. Acta 1969, 48, 205-212. (22) Fujinaga, T.; Kusaka, Y.; Koyama, M.; Tsuji, H.; Mitsuji, T.; Imai, S.; Okuda, J.; Takamatsu. T.; Ozaki, T. J . Radioanal. Chem. 1973, 73, 301-311. (23) Riley, J. P.; Taylor, D. Deep-sea Res. 1972, 19, 307-317. (24) Muzzarelli, R. A. A. Anal. Chim. Acta 1971, 54, 133-142. (25) Muzzarelli, R. A. A.; Rocchetti, R. Anal. Chirn. Acta 1973, 64, 371-379. (26) Van der Sloot, H. A,; Wals, G. D.; Das, H. A. Anal. Chim. Acta 1977, 90, 193-200. (27) Vanderborght. 8. M.; Van Grieken, R. E. Anal. Chem. 1977, 49, 3 1 1-3 16. (28) Knowles, H. B. J . Res. Nati. Bur. Stand. 1932, 9 , 1-7. (29) Weiss, H. V.; Lai, M. G. Talanta 1961, 8, 72-76. (30) Bowen, J. H. M. Int. J . Appi. Radiat. Isot. 1959, 5 , 227-232. (31) Nadkarani, R. A.; Halder, 8. C. Talanta 1969, 76, 116-118. (32) Savariar, C. P.; Arunachalam, M. K.; Hariharan. T. R. Anal. Chim. Acta 1974, 69,305-311. (33) Fulton, J. D.; Robinson, R. J . Chern. SOC.1939, Part 7 , 200. (34) Welcher, F. J. Org. Anal. Reagents 1947, 3 , 240. (35) Ryan. D. E.; Stuart, D. C.; Chattopadhyay, A. Anal. Chim. Acta 1978, 100, 87-93. (36) Brooks, R. R. Geochim. Cosmochim. Acta 1965, 29, 1369-1370. (37) Van der Sloot, H. A,; Luten, J. B. "Proceedings of the International
(38) (39) (40) (41)
833
Symposium on Measurement, Detection and Control of Environmental Pollutants", IAEA, Vienna, 1976, 435-448. Darras, R.; May, S.; Engelmann, Ch. Reference 37, pp 339-355. Guegneniat, P.; Gandon, R.; Hemon, G.; Philippot, J. CI. Reference 37, pp 369-382. Young, E. G.; Smith, D. G.; Langille, W. M. J . Fish. Res. Board Can. 1959, 76, 7-12. Sugawara, K.; Okabe, S. J . Tokyo Univ. Fish., Spec. Ed. 1966, 8(2),
.-- . . .
166-1 7 I,
(42) Sugawara. K.; Okabe, S. J . Earth Sci. 1960, 8 , 93-107. (43) Black, W. A. P.; Mitchell, R. L. J . Mar. Biol. Assoc. U . K . 1952, 30, 575-584. (44) Zhavoronkina, T. K. Sov. Ocean. 1963, No. 3 , 27-29. (45) Morris, A. W. Deep-sea Res. 1975, 22,49-54. (46) British Chemical Standards, Bureau of Analysed Samples, Ltd., Newham Hall, Middlesbrough. U.K., 1966. (47) Dybczynski, R.; Tugsavul, A.; Suschny, O., IAEA, Vienna, IAEAIRL140, 1976. (48) Van Grieken, R. E. Universiteit Antwerpen, private communication, 1979. (49) Certificate of Analvsis, NBS. Washinaton. D.C., Standard Reference Material 1577, 1977. (50) Van der Sloot, H. A.; Wals, G. D.; Weers, C. A.; Das, H. A. Netherlands Energy Research Foundation, Petten, Rep. ECN- 79-096,1979. (51) Nadkarni, R. A.; Morrison, G. H. Anal. Chern. 1973, 45, 1957-1960. (52) Lievens, P.; Verzieck, J.; Cornelis, R.; Hoste, J. J . Radioanal. Chern. 1977, 3 7 , 483-496. (53) Currie, L. A. Anal. Chem. 1966, 40, 586-593
RECEIVED for review October 24, 1979. Accepted January 28, 1980. This research was supported by a research operating grant (=A-9977)from the Natural Sciences and Engineering Research Council Canada. The award of a study leave to one of the authors (A.I.K.) by the University of Peradeniya, Sri Lanka, is gratefully acknowledged.
Quantitative Analysis of Small Samples with a Guinier-Hagg Type Focusing Camera in Powder X-ray Diffraction Per-Erik Werner" and Thommy Ekstrom' University of Stockholm, Department of Structural Chemistry, Arrhenius Laboratory, S- 106 9 1 Stockholm, Sweden
An X-ray method is described for quantitative analysis of small sample amounts using a Guinier-Hagg focusing camera. No internal standard need be added to the samples as the method uses a direct analysis of the absorption conditions for each sample. The method has been tested on a sereis of cristobalite spiked dusts from a quarrying plant and from an iron foundry. Cristobalite concentrations of a few percent were analyzed In total sample amounts of 100-300 pg.
The powder X-ray diffraction methods used for quantitative analysis of crystalline components are generally based upon data obtained from diffractometers. Although powder diffractometry and photographic X-ray camera techniques have often been used in parallel, it is commonly experienced that high quality intensities can more easily be measured on a diffractometer. The use of a focusing film camera technique introduces, however, several physical advantages, which can be summarized as follows: (a) The photographic film is a sensitive multidetector, collecting intensities simultaneously from all diffraction angles in a wide angle interval. One Cu Kcu photon per pm2 of an Ilford G-emulsion corresponds to an optical density = 1 ( I ) . The complete record can be collected in less than a tenth of P r e s e n t address: N a t i o n a l B o a r d of O c c u p a t i o n a l Safety a n d H e a l t h , S-171 84 Solna, Sweden. 0003-2700/80/0352-0833$01 .OO/O
the time needed for the same quality of diffractometer data. (b) With a bent quartz monochromator, i t is possible to eliminate all K a 2 radiation from the X-ray source. For instance, in this study all the Guinier-Hagg diffractograms used have been recorded by strictly monochromatized Cu Kcul radiation. A corresponding possibility is usually not available for a powder diffractometer, because of the necessary sizes of primary radiation beam, of sample and of detector slit. (c) The half-widths of the lines on a Guinier-Hagg photograph are normally less than 2 / 3 of the corresponding diffractometer peak half-widths. We have used a Philips diffractometer system (APD-10) for the comparison: we must, however, stress the point that no exact comparison between the two techniques is feasible. The half-widths in a Guinier-Hagg diffractogram are somewhat dependent on the diffraction angle, and in the case of a diffractometer recording, upon chosen detector and divergence slits. The resolution will therefore in the latter case depend upon the time available for the recording. (d) The use of a thin rotating sample in transmission geometry reduces the preferred orientation problems in a Guinier-Hagg camera. However, serious problems for fibrous samples such as chrysotile arise for all X-ray powder diffraction methods. (e) One densitometer on-line to a small computer can be used for measurements of line positions and integrated intensities from photographs recorded on several powder cameras. This may reduce the cost per diffractogram and make C' 1980 American Chemical Society
834
ANALYTICAL CHEMISTRY, VOL. 52, NO. 6 , M A Y 1980
the system economically favorable. One of the most important advantages with the GuinierHagg method is that it requires only very small sample quantities. Also the detection limit is lower than that obtainable by standard powder X-ray diffractometry. I t is therefore of interest to study the possibilities inherent in the focusing camera technique. An important example would be analysis of work environmental dust samples, of which very small amounts normally are available. We have developed a new and quite simple technique for quantitative analysis by use of Guinier-Hagg camera data ( 2 ) . The results of this study are presented below.
EXPERIMENTAL The synthetic samples were prepared from chemicals of analytical purity, fractionated in alcohol to yield particles 5 5 pm; and from Scandinavian quartz standard, Fyle quartz ( 5 5 pm). The cristobalite used was prepared from quartz standard heated at 1500 "C for a couple of days and then quenched t o room temperature. The dust samples were collected on membrane filters (Millipore AAWP, 0.8-pm pore size) in various industrial plants. They were fractionated in alcohol to yield particles 5 5 Irm. For all samples the powder X-ray patterns were registered in a Guinier-Hagg camera with strictly monochromatic Cu Kcv, radiation, using an exposure time of 1 h. The X-ray sources was a Philips 2113/00 fine focus type (0.4 X 8 mm focus spot), run a t 40.0 f 0.2 kV and 20.0 f 0.1 mA. The film used was single coated, Test-X H, manufactured by CEA, Sweden. The exposed photographs were developed during 240 f 5 s, fixed for 10 min, and washed in water for 10 min. To avoid losses of material during the Guinier-Hagg recording, during which the rotating sample is vertical, all samples were covered by a thin zapon lacquer coating. The zapon lacquer was diluted with amylacetate, and a drop of solution was allowed to dry on the sample after the tape was prepared. Our tests have shown the X-ray absorption in the zapon lacquer film to be negligible. The powder film intensity data were obtained from measurements made with a SAAB model 2 drum film scanner (3). The scanner is coupled on-line to an IBM 1800 computer and the data reduction programs are described in references 4 and 5 . THEORY With identical procedures for exposure and development, the relative intensities I , and I z of any one powder diffraction line from two samples of the same material, with mass ml and m2, respectively, will have the relationship:
Figure 1.
Determination of A,,
pd,, and A, from pd,
In this equation, 8 is the diffraction angle and )I is the angle between the incident primary X-ray beam and the normal to the specimen surface. The factor S in Equation 3 will here be treated as a scale factor although it is dependent on 8 and $ (6) which are, however, constant in the experiments. The absorption factor A ( p d ) is normalized so that lim A = 1 as pd 0. In this way S will be obtained for a certain 8 and $.
-
From Equation 1, i t follows that the ratio between the absorption factors can be directly determined from measurements as
(4) Moreover, the absorption factor A as a function of p d can easily be obtained by computer evaluation of Equation 3. The additional conditions for determining A , and A z can be obtained from Equations 2 and 4. A simple way t o find A , and A 2 is illustrated in Figure 1. Let ~d~vary from 0 to a value exceeding any possible pd for the specimens, (for example pd = 3). Let the computer calculate, a t every step, pd,, the values of A, (from Equation 3), pd, = (rn2/m,).pd,(from Equation 2) and A, (from Equation 3), as indicated in Figure 1. Find the minimum of the difference: E = IA,/A, - (ml12)/(m211)l (5) The values of A, and A, obtained at the minimum (ideally = 0) represent the absorption factors A , and A2. One unique solution is obtained because of the fixed relation between pd, and pd, and the continuous decrease in A with increasing F d t
A , and Az are relative absorption factors for the two samples. They need not include absorption in the film and film emulsion, in any metal foil that may be used for reduction of fluorescence radiation, in the air, or in the film protection paper; these factors are equal for the two samples. A , and A2 represent strictly the absorption within the different amount of samples. If the samples are uniformly spread over equally large irradiated areas, we have the relation
where p represents the linear absorption factor of the material and d l and d2 are the average thicknesses of the two mounted samples. For absorption in the two samples we have, according to Sas and De Wolff (6),the following expression:
S
-
A = - [exp(-pd/cos
&
$1
-
exp(-pd/cosl28
-
rl.l)]
(3)
(cf. Equation 3). By use of the known A , and A2,an absorption corrected intensity per mass unit can be calculated from Equation 1. The quantities p , d l , and d 2 need not be used explicitly in the evaluation. They are generally unknown and the basic idea with the proposed procedure is that the product pd is treated as one variable. Thus, in principle, the absorption corrected intensity per mg of sample can be determined from measurements on two different quantities of sample. As the I,,,, values for the X-ray reflections of any substance to be determined are directly proportional to its concentration in the sample, I,,, can be used to determine unknown concentrations.
EXPERIMENTAL TESTS Recording of the Calibration Curve of Cristobalite. As stated above, it is necessary with this method that a number of factors can be reproduced within sufficiently narrow limits from one sample to another. Examples of such factors are
ANALYTICAL CHEMISTRY, VOL. 52, NO. 6, MAY 1980
835
Table I. Results of the Analysis of a Series of Cristobalite-Spiked Iron Foundry Dust, Using the Method of Absorption Corrected Intensity weighed in cristobalit e
analyzed cr ist ob alit e
sample amount,pg
%
ug
I,,,,,
1000
20
200
lOOlb
20 15
54 152 39
1000
15 10
255 1015
0.763
3 32 746b
0.070 0.350
0.937
0.090 0.175
0.917
100
253 518
10 5
26 51
152 241
0.045 0.145
0.963
270 990
5
72 138
0.039 0.005
0.969
2
14 20
275
2
6
38
0.002
1.0
270 1015
Icolxlmg
A
pd
0.260
} } } }
0*692 2 60
0’836 o’864
1312 1061 619 275
}
138
A is the difference between weighed in and calculated amount of cristobalite. with equal amounts. (I
exposure time, development procedure of the X-ray films, and a uniform distribution of the specimen. To test for this, and to construct a calibration curve for the used measuring system, equal amounts of dust from a quarrying plant, with known added amounts of cristobalite have been recorded. Analysis of the cristobalite-free dust from the quarrying plant showed X-ray diffraction from three crystalline components only. The main crystalline phase was a-quartz. The two others were NaAlSi30s and KA1Si3O8. The absorption factors are of the same order of magnitude for all components, so one can expect pd to be almost independent of the amount of added cristobalite and approximately constant for the same total amount. The integrated intensity of the strongest cristobalite reflection, a t d = 4.057 A, as a function of the cristobalite concentration would therefore yield a linear plot. Least-squares fitting of data from two series of measurements, on different total sample amounts of cristobalitecontaining dust yielded the equations
I = 48.2(1.4)X
+ 153(15) for 1443(13) pg sample
(6)
and
I = 42.3(1.9)X + 145(22) for 1029(9) p g sample (7) In these equations X is the cristobalite concentration (wt 70)and the values within the parentheses are the standard deviations. T h e lines do not pass through zero at X = 0 because of overlap with a line a t d = 4.024 A, arising from crystalline NaA1Si308. The occurrence of such an overlap can cause a small deviation from linearity at high cristobalite concentrations, but in the chosen range X 5 20, this effect may be neglected. T h e linear relations, Equations 6 and 7, imply that absorption is dependent only on the total amount of sample, as was assumed. They also show that reproducible recordings are possible. We can therefore choose an arbitrary cristobalite concentration, e.g., X = 20%, and calculate the absorption for the two sample amounts. From Equations 6 and 7. we obtain, after subtraction of the contribution of NaAlSi&, I = 964 and 845 for the sample amounts of 1443 and 1029 pg, respectively. With these values and the known diffraction angle 8 for the cristobalite reflection, the absorption is evaluated from the minimum of expression 5 . m, u g Iobs Icorr ctd A 1443 1029 1000
Y 64
845
1969 1403 1365
0.679 0.478
0.490 0.603
%
19.2
ug
A;pg
192
8
52 158
-6
40 91
-1 9
23 41
11
2
15.6 9.1
3
4.0 11
3
20
0
6
0
2.0
Mean value of two different samples
Equation 1 given by Sas and De Wolff (6), has not been previously applied in any case known to us. It was therefore of interest to test the plausibility of the results obtained above. We may use a sample density = 2.5 g/cm3 and a linear absorption factor = 86 cm-l. These values are strictly correct for the main component, a-quartz, but can be assumed to be approximately valid for the total sample. Moreover, the irradiated surface of the sample is a circle of approximately 0.3-cm diameter. For a 1.0-mg sample, d = 0.001/(~-0.15~~2.5) = 0.0057 ern and pi = 0.49. This is very close to the value pd = 0.478 calculated above for 1.029 mg. Thus, the measurements made have resulted in a “calibration curve”, i.e., a relation between the absorption corrected intensity per mg sample and the cristobalite concentration X in wt %: Icorr(d = 4.057
A, 1-mg sample) = 68.25*x
It should be stressed however, that although Equation 8 is independent of the absorption in a particular sample, it is valid only for the measuring system used, i.e., a given camera with constant primary beam, a given film type, constant exposure time, identical sample holders, etc. In order to check the stability of the system, we have found it useful to save a sample holder with an arbitrary reference sample. Twice a week the reference sample has been exposed and the obtained photographs measured. Quantitative Analysis of an Iron Foundry Dust. The validity of the described method was checked on a series of iron foundry dust samples. The samples were spiked with known amounts of cristobalite. For all specimens, two different sample amounts were weighed in and the Guinier-Hagg photographs were recorded in the same manner as the cristobalite standards described in the previous section. Thus, Equation 8 could be assumed to be valid for all samples. Absorption and absorption corrected intensity per mg sample were derived for each pair of samples, by means of a computer program, AESPU, which calculates the minimum difference (5) using the method described above. The results of the quantitative analyses of cristobalite are summarized in Table I. As expected, the relative errors in calculated amounts of cristobalite are larger for the lowest cristobalite concentrations, whereas the absolute errors, A in Table I, show no tendency to increase. As can be seen, l c l shows a tendency to decrease with the cristobalite content. There is no obvious interpretation of this since the separate variations in p- and d-values are not known. Although it is expected that iron minerals of high absorption are present
ANALYTICAL CHEMISTRY, VOL. 52, NO. 6, MAY 1980
836
Table 11. Results of the Cristobalite Analysis of Iron Foundry Dust Samples Found by Setting A = 0.9 for All Samples weighed in cristople a- balite mount, __ 14. os 7 PCrg 7% Pg I,.,,, mg 270 20 54 332 1230 260 15 39 253 973 255 10 26 1 5 2 596 270 5 14 72 267 275 2 6 38 138
analyzed cristobalite Icorr/W
%
1367 1081 662 297 153
20.0 15.8 9.7 4.4 2.3
Pig A ,
pg
54 41 25 12
0
-2 1
6
0
Pg
Iobs
Iobd2O0 p g sample
15 20
21 0 215 205 220 21 5
86 198 399 593 770
82 184 38 9 539 716
cristobalite,
sample,
5%
Pg
2 5
295 300 300
%
2 5 10
10
15 310 20 315 a These results are used ality constant.
I = 35.25X + 15 for 200-yg sample
2
Table 111. Results of a Series of Analyses on Cristobalite-Spiked Iron Foundry Dust Using Sample Holders Adapted for Small Sample Amounts‘ cristobalite, sample,
holders, we w ill obtain 4 diffraction lines from metallic copper, which is a component in the brass holder, because of the 3-mm height of the primary X-ray beam. These diffraction lines do not interfere with the quantitative analysis of such materials as quartz or cristobalite, however, since the first copper line appears a t a fairly high diffraction angle. Table I11 shows two measurement series, with the new sample holder, on iron foundry dust with known amounts of cristobalite. From these results least-squares refinements produced the following linear relationships for the strong cristobalite peak a t d = 4.057 A.
l o b s / 300
sample 118 120 248 248 50 6 506 620 600 829 790 f o r evaluation of the proportion-
I = 36.56X
where X is the cristobalite weight percent. The small difference that can be seen between the slopes of the two lines indicates that the absorption in this case is very high. In spite of this, and because of the linearity, these values can be used to determine the proportionality constant of the measuring system. As previously we chose the data for the 20% cristobalite sample; the line equations yield I = 705 and 731 for the 200-1g and 300-1g samples, respectively, disregarding the constant terms. Using these figures as input to the computer program ABSPU (see above), the absorption corrected intensity per milligram sample was calculated
Pg
Iobs
in the dust, nothing is known about the presence of amorphous or crystalline components with low absorption. The cristobalite concentration may also affect the density and, hence, the thickness of a sample. It can be observed, however, that the absorption corrections for small total sample amounts need not be very accurate in order to obtain useful analytical results. This is illustrated in Table 11,where the absorption factor for all samples of the 1/4-mgseries of iron foundry dust has been set to 0.9. Quantitative Analysis of S m a l l S a m p l e Amounts. In order to optimize the method for the analysis of very small sample amounts, we modified the sample holders so that the diameter of the circular area of the sample is reduced from the normal value 5.0 mm to 1.5 mm. With such sample
+ 73 for 300-yg sample
m, pg
lobs
Icorr
20 0 300 1000
705 731
1435 2133 7141
Pd
A
0.667
0.491 0.343
1.000
The results obtained, I,,, = 7141/mg sample, must be regarded as somewhat uncertain because of the strong absorption. However, this value can be used to demonstrate the principle of the method; the absorption corrected intensity per milligram sample as a function of the cristobalite concentration can be written as I,,, = 357.X. This relation has been tested on the spiked quarrying plant dust samples used earlier in this study. By the reduction of the diameter of the circular sample area from 5 to 1.5 mm, the area decreases to about 1/11 of its original value, making the sample more “concentrated”. This will give a corresponding increase in the sample absorption; in an attempt to reduce the absorption, sample amounts of 100 and 200 fig were used in each pair of samples. A normal weight error on the microbalance used is about f 5 hg. We found it impractical to reduce the sample amounts further, since the weight error would then become significant. Each pair of samples was evaluated using the program ABSPU, and the wt 7’ cristobalite was calculated from the proportionality con-
Table IV. Results of the Analysis of a Series of Cristobalite-Spiked Quarrying Plant Dust, Using Sample Holders Made for Small Total Sample Amounts weighed in sample amount, p g 210
analyzed cristobalite
cr ist obalit e %
MLg
I4.US7
18
38
67 7
Pd
0.535
100 195
18
18
25
436 461
0.255 0.705
0.767
13
115 195
13
15
8
16
37 2 327
0.416 0.695
0.645 0.477
110 195
8
9 10
25 6 19 3
0.392 0.610
0.661 0.522
0’471
’A
I,lmg
A
0.567
)
}
5686
Pg
5016
5
16
2 -2
15.9 14.0 16
19
1895
105 5 5 141 0.328 0.708 is the difference between weighed in and calculated amount of cristobalite.
-3 -3
9.9
1
j’
AfMg
33
27
3519 5
%
’L
5.3
11 10
-2
6
-1
0
Anal. Chem. 1980, 52, 837-842
837
samples of respirable foundry dust (11,12). With the physical advantages of a focusing camera, as described above, similar or better results might be expected. With improved accuracy of weighing, sample amounts less than 0.1 mg may be used.
stant derived above. A simplified treatment assumes the contribution of an overlapping peak from NaAISiaOBin the matrix material to correspond to 3% cristobalite. The results are shown in Table IV. T h e results indicate that the weight errors are reasonably small also for 100-pg sample amounts. With the measuring system employed, a suitable compromise would be to use sample holders with 2.5-mm diameter to reduce sample absorption. The optimal conditions to be used have to be tested out for each measuring system. The error in the weighing procedure sets the lower limit for the sample amounts.
ACKNOWLEDGMENT We express our sincere thanks to Peder Kierkegaard and Staffan Krantz for the research facilities placed at our disposal and their kind interest in this work. LITERATURE CITED Morimoto, H.; Uyeda, R . Acta Crystallogr. 1963, 16, 1107-1 119. Hagg. G.; Ersson, N. 0. Philips Bull. 1971, No. 7000.38.0050.11. Abrahamsson, S. J . Sci. Instrum. 1966, 43, 931-933. Werner, P.-E. Ark. Kemi 1969, 31, 505-51 1. Malmros, G.; Werner, P.-E. Acta Chem. Scand. 1973, 27, 493-502. Sas, W. H.; De Wolff, P. M. Acta Crystallogr. 1966, 27,826-827. Leroux, J.; Powers, C. A. Staub Reinhalt. Luft. 1969, 29, 197-200. Leroux, J.; Davey, A. 6.C.; Paillard, A. A m . Ind. Hyg. Assoc. J. 1973, 3 4 , 409-417. Altree-Williams, S.; Lee, J.; Mezin, N. V. Ann. Occup. Hyg. 1977. 20. 109-126. Altree-Williams, S. Anal. Chem. 1977, 49,429-432. Higgins, R . I.; Dewell, P. BCIRA J . 1961, 9 , 117-125. Higgins, R . I.; Dewell, P. BCIRA Rep. 1976, X . 4 4 , 1-10,
DISCUSSION X-ray diffractometers have long dominated the field of silica analysis from working-place environments. This is not surprising, since operation is easy and good results have been obtained by direct analysis on metal foil or PVC and Nucleopore filters (7-10). The detection limits for quartz reported in these studies are often of the order of a few micrograms, which in most instances is sufficiently good. Looking forward it is, however, desirable to develop X-ray methods capable of handling sample amounts of about 0.5 mg. Thus, the use of photographic X-ray techniques are of obvious interest. For instance, by the use of a DebyeScherrer camera, Higgins and Dewell have shown that it is possible to quantitatively analyze cristobalite and quartz in sub-milligram
RECEIVED for review February 7 , 1979. Accepted December 26,1979. Grants from the Swedish Natural Science Council and from the Swedish Work Environment Fund are thankfully acknowledged.
Simultaneous Multi-Mycotoxin Determination by High Performance Thin-Layer Chromatography K w a n Y. L e e , Colin F. Poole, and Albert Zlatkis" Department of Chemistty, University of Houston, Houston, Texas 77004
Zearalenone produces an estrogenic syndrome in swine ( I , 4-8). Screening of feedstuffs to control the level of mycotoxins in the diet is essential. Analytical methods for the determination of individual mycotoxins by thin-layer chromatography (TLC), gas chromatography, high performance liquid chromatography (HPLC), and polarography have been developed (for recent bibliographies of references, see 9, 10). However, it is not uncommon for more than one mycotoxin producing fungi to be isolated from the same contaminated food source and, indeed, some fungi possess the ability to produce more than one mycotoxin simultaneously (11-13). Consequently, there is no rationale for the selection of a particular mycotoxin contaminant in any particular foodstuff for analysis, and the individual analysis of each mycotoxin by the procedures described above is tedious and time consuming. A rapid screening method for their simultaneous analysis would be advantageous. Several multi-mycotoxin screening methods have been reported (7, 8, 14-24) of which TLC (16-24) owing to its simplicity is most frequently used. Conventional TLC systems are often slow, many require the use of several solvent systems and have been developed to the stage of providing mainly qualitative information useful for identification purposes or for semiquantitative analysis.
An analytical method is described for the simultaneous determlnatlon of 13 mycotoxins by high performance thin-layer chromatography (HPTLC). I n seven contlnuous multiple developments with two solvent systems of different polarity a base-line separation of sterigmatocystin, zearalenone, citrinin, ochratoxin A, patulin, peniclllic acid, leuleoskyrin, and the aflatoxins B1, B,, GI, Gz, M,, and M, was obtained. About 1 h was required for the separation and quantitation of all 13 mycotoxins from one spot. By using in-situ scanning of the HPTLC plate, detection llmlts in the low nanogram range were obtained by UV-vlslble absorption (reflectance mode) and in the low picogram range by fluorescence with a relative standard deviation of 0.7 to 2.2% in the nanogram range.
Animals, including man, are continually exposed to numerous substances known or suspected of being carcinogenic which are present as trace contaminants in the food supply. Among these toxicants there is increasing apprehension concerning the mycotoxins - the toxic metabolites produced by fungal contamination (1-4). The aflatoxins, patulin, and sterigmatocystin, were found to be carcinogenic in animal feeding studies and ochratoxin A, penicillic acid, citrinin, and luteoskyrin produced kidney and liver damage in rats. 0003-2700/80/0352-0837$0 1.0010
t=.
1980 American Chemical Society