Differential spectra of samples containing up to 10% N-ethyl-m-toluidine gave results similar to those obtained nith the aniline derivatives. This was also true for the studies on stability, slit width, and interferences. Low concentrations of n'-methylaniline in N-methyl-N-2-hydroxyethylaniline have been determined by a near-
infrared method essentially the same as that described for the AT-ethyl derivative. ACKNOWLEDGMENT
The authors wish to thank J. P. Reeves for purifying the samples which were used in this investigation.
Quantitative Ana lyses by X-Ray Spectrography
LITERATURE CITED
(1) Haye, w., Spectrochinz. A C ~ U 6, 257-87 (1964). (2) Siggia, S., Hanna, J. G., Kervenski, I. R., ASAL. CHEM. 22, 1295-7 (1950). (3) Siggia, S., Kervenski, I. R.,Ibid., 23, 117-8 (1951).
R~~~~~~~ for review january 7 , 1957 Accepted March 20, 1957.
FIuorescent
Determination of Germanium in Coal a n d Coal Ash WILLIAM J. CAMPBELL and HOWARD F. CARL' Eastern Experiment Station, Bureau of Mines, U. S. Department of the Interior, College Park, Md. CHARLES E. WHITE University of Maryland, College Park, Md.
,The determination of germanium in coal and coal ash by fluorescent x-ray spectrography was investigated in order to develop a rapid quantitative method for a large number of samples. Four general methods of correlating line intensity or line intensity ratios to concentration were developed: ( 1 ) direct comparison with standard samples, (2) internal standardization, (3) addition techniques, and (4) preliminary chemical separation with subsequent excitation of the concentrated germanium fraction. Studies on the loss of germanium upon thermal ashing showed that no detectable amounts were lost below 600" C.; losses did occur upon rapid oxidation a t higher temperatures. Method 1 can be used to survey 50 to 100 samples per d a y for germanium. Methods 2 and 3 are accurate to =+=lo%of the amount present for concentrations exceeding 0.1 % germanium with decreasing accuracy for lower values. Method 4 i s believed to b e accurate to + 10% for concentrations exceeding 0.00 1 The lower limit of detection b y Method 1 , 2, or 3 is approximately O.OI%, b y method 4, 0.0002 to 0.000470.
yo.
T
occurrence of germanium in coals and coal ashes has been the subject of rather extensive study (23), particularly in the United Kingdom ( 2 ) . Because evaluation of the germanium content of coals may require many HE
Present address, Davison Chemical
Co., Division of W. R. Grace and Co., Baltimore 26, Md.
analyses, a rapid quantitative analytical method was desired. Therefore, the determination of germanium by fluorescent x-ray spectrographic techniques was investigated. As determination of germanium by these techniques can be considered fairly typical of many determinations for elements in minor or trace amounts and in variable matrices, some detailconcerning theory and instrumentation has been included. INSTRUMENTATION
The apparatus used in this study was a modified Norelco fluorescent x-ray spectrograph, as shown in Figure 1. An open tube, 0.75 by 0.75 inch and 4 inches long, A , was placed between the sample and the analyzing crystal. As this tube allowed considerable scattered radiation, a cylindrical shield, B, xas placed around the analyzing crystal. The collimation was provided by Soller slits 4 inches long with 0.005inch spacing, C, placed between the analyzing crystal and detector. A scintillation counter, D,consisting of a sodium iodide thallium-activated phosphor with an E. hI. I. photomultiplier tube was used in certain phases of this investigation. The original Korelco timer was replaced by an electric stop watch, E , which provided increased timing accuracy. The ratio of line intensities for various elements excited by the molybdenum vs. the tungsten target tube is compared in Figure 2. Results indicate t h a t the tungsten tube is the most gen-
erally useful, but molybdenum is more effective for elements of atomic numbers 32 t o 39. The Compton scattering is very strong from samples such as coals and coal ashes. If a tungsten target x-ray tube is used, some of the modified TIr& lines overlap the weak germanium spectrum, making the analysis difficult. Although the second-order N o K p lines are close t o the GeK, lines, it is possible to resolve their spectra lvith the modified spectrograph shown in Figure 1. Table I lists important characteristics of the analyzing crystals evaluated in this study. Quartz and topaz crystals, 1.818 and 1.398 il., respectively, resolved the germanium Kal line from possible interfering spectra with a reduced but not restrictive line intensity. Four detectors were evaluated for the determination of germanium: an Anton krypton-filled Geiger tube, Norelco 62019 argon-filled and 62030 neon-filled Geiger tubes, and a scintillation counter consisting of a sodium iodide thallium-activated phosphor with an E. 31. I. 6260 photomultiplier tube. A comparison of detector efficicncies for various wave lengths relative to the argon Geiger tube is shoivn in Figure 3. The Norelco 62019 argonfilled Geiger tube was used for the germanium determination. This detector is very efficient in the germanium K a wave length region, is somewhat insensitive t o possible overlapping higher order radiation, and has a low noise level compared to the scintillation counter. The high dead time of the Geiger tubes was not a problem in trace VOL. 29, NO. 7 , JULY 1957
1009
analvses because of the low line inten-
,s,.*9*',,au(.. iii,i.E
t
WNTIUATION
X
coal ashes. The GeK,, line intensity for the 5% iron oxide was approximatc>ly 1.5 times that for 25% iron oxide These results indicated that accurate analyses cannot be obtainrd by a single calibration curve because of t,he relatively high absorption of GeK,, by iron as compared to the absorption by silicon or aluniinuin. Howcver, if standards of 15% iron oxide are used, tlic analytical error generally does not exceed i 2 5 7 , of the germanium prwent. Although this is rather large th(% method is very useful for det’ermining whether or not germanium is prcsrnt, and it’s approximate concentration. Method 2. Internal Standards. Internal standardization consists of addition to the sample of a referencP el+ ment whose characteristic radiation will be excited and abs0rbt.d in the saiiir manner a s the cliaractrristic radiation of the element under det,ermination. The effects of the matrix on the conip:nison line intensities can he grouprd into three general rlas.
COUNTER
KRYPTON GEIGER TUBE NEON GEIGER TUBE
0
\
01
1 01
’
I
I
03
1 1 , , I , 06
I I
x
I
I I I I I
3
6
Figure 3. Comparison of detector efficiencies for various wave lengths relative to argon Geiger tube as unity
I
2 3 4 WEIGHT PERCENT GERMANIUM
5
Figure 4. Experimentally observed relationships of germanium Kcvl line intensities to weight per cent germanium in various matrices
Le
GeKa R K a
A. The matrix has a higher ahsorption for the comparison line of longer w v c length. €3. An element has a n absorption edge 1)etween the cwnparison linrs. C. Emission from the matrix preferentially excites the romprison clcment of lower atomic nuni1)cr. For this study the matrix was (wisidered as a third "element" addrd to :I mixture of the rr>ferencc c$l(,mcnt and r>lrmentsubject to analysis. The lines and ahsorption edges that result in these three classes of matriws are shown in Figure 5 . If the niatris “element” emits lines 1,1 or Lar or has absorption edges El or ESsthesample will lie in Class A, i l n example of Class R is the presence of a third elrnirnt n i t h an alisorption rdge of E*. If a spectral line I,2 is present, refrrrnce c.lcmrnt R will be preferentially rxritrd and it is c:onsiderc.d as a Class c‘ matrix r+fcrt. The variation of tlir c.onil):irison linc intensities with a class A matrix is considered in dctail, as this is tlir theoret,ical basis for intrrnal standardization. The relationship of tlic ratio of line intrnsitirs to the niatris is given in the following rquation, wherr R is t’hc rtfrrencp clcnwnt. ZG,K,+
IRK,,
-
Pio
K,T17C, f I.lt;eKa,
/ YL, -.
pir
(:%)
If K 1 = K 2 and W,, = TI-R, I,dividing through by p i e the following cxprcssion can be derivrd ( 4 ) :
’A Figure 5.
Possible interfering lines,
I, and absorption edges, E
whrre
u = 1
X3RKn, - ___h3GeKal
VOL. 2 9 , NO. 7,JULY 1957
0
101 1
Gallium and arsenic are the two reference elements usually used in the determination of germanium. When the reference element has a loner atomic number than germanium,-that is, gallium-u has a negative value and I G ~ K ~ / I exceeds R K ~ ~ unity. The opposite is true for arsenic, as u is positive. The calculated u values for gallium and arsenic are -0.22 and +0.21, respectively. Calculated intensity ratios of I G ~ aK1 IRK,^ agreed with the experimental values within &5%. A Type B matrix contains a n element that has an absorption edge between the comparison lines, and Equation 4 is no longer applicable. The relationship between the absorption coefficients for two lines across a n absorption edge (11) is shown as: pl/p2
= 7.21 Xo.zs4
ANALYTICAL CHEMISTRY
edges studied in this series of samples are shown in Figure 7 . The small line preceding the Kg, line for each element corresponds to the absorption edge for that element. The first mixtures studied were formed by dilution of germanium-
I.o
0.9
(5)
where p 1 and p z are the absorption coefficients a t the absorption edge and immediately beyond. Because the magnitude of the discontinuous absorption is proportional to the concentration of the absorbing element, i t is necessary t o know this concentration in order to apply suitable corrections. However, the presence of such elements can be readily detected and by proper choice of standard a Type B matrix is avoided. The preferential excitation of one of the comparison elements, a Type C matrix, is also proportional to the concentration of the exciting element. The relationship in Equation 4 requires addition of a correction term which is proportional to the concentration of the exciting element and to the magnitude of the preferential excitation yield. As for Type B matrices, reliable analyses require knowledge of the sample composition. Use of proper comparison lines will avoid this preferential excitation. To evaluate the reliability of internal standardization, a number of synthetic samples were prepared, covering all types of matrices. These samples are placed in two groups: germanium-gallium, where the analytical line is of shorter wave length than the reference line, and the opposite example, germanium-arsenic. More general studies were made by Adler and Axelrod (1) and von Hevesy (19). A 1 to 1 atomic ratio of germaniumgallium or germanium-arsenic was diluted 9 to 1 with a mixture of silicon dioxide and diluent. Diluents were sodium bromide, selenium, zirconium dioxide, cupric oxide, cobaltic oxide, stannic oxide, zinc oxide, and arsenious oxide or gallium oxide. These samples were excited a t 45 kv., while the current was varied to keep the intensity of GeKal nearly constant for all samples. The measured 1012
GeKal line intensities were normalized to a current of 50 ma. by multiplying the GeKa, intensity by 50 ma. per ma. used. The decrease in the absolute line intensity is shown in Figure 6. The positions of the spectral lines and
0.2
0
20
40 60 WEIGHT PERCENT DILUENT
00
100
Figure 6. Variation in intensity ratio germanium Kal (diluent)/germanium Kal (SiOz) with increasing percentage of various diluents
Table II.
Evaluation of Arsenic as an Internal Standard for Germanium Deviation from Unity,
Matrix, G. I. Sn02 0.2 0.4 0.6 0.8 11. Zr02 0.2 0.4 0.6 0.8 111. NaBr 0.2 0.4 0.6 0.8
IV.
Se
0.2 0.4 0.6 0.8
V. Ga203 0.2 0.5 VI. ZnO 0.2 n.4
0.6 0.8 VII. co203 0.2 0.4 0.6 0.8
0 /O
39.8 30.2 23.3
95.5 95.5 97.3 94.8
-4.5 -4.5 -2.7 -5.2
79.7 67.0 55.6 48.8
92.6 100,7 94.5 96.0
-7.4 0.7 -5.5 -4.0
94.8 85.9 76.7 72.1
93.0 94.2 95.2 95.2
-7.0 -5.8 -4.8 -4.8
97 - . . .0
84.3 74.5 67.7
119.2 i26.5 132.1 136.5
19.2 26.5 32.1 36.5
77.5 61.6
167.0 206.4
67.0 106.4
45.3 27.8 18.0 15.0
97.5 96.8 95.8 99.5
-2.5 -3.2 -4.2 -0.5
51.5 34.0 27.0 25.5
96.3 95.8 103.3 95.9
-3.7 -4.2 3.3 -4.1
gallium and germanium-arsenic samples with stannic oxide. Tin strongly absorbs spectral lines in the region of GeK,, as shown in Figure 6, b u t the comparison-line ratios remained relatively constant, as shown in Tables I1 and 111, with a slight shift in favor of the shorter wave lengths. The characteristic lines and absorption edge of tin fall on the short wave length side of the comparison elements and result in a n El and L, Class A matrix. Zirconium dioxide also resulted in a n El and L1 Class A matrix, and the line ratios were similar to those obtained for stannic oside. Addition of sodium bromide gave a special case of an L1 and El Class A matrix, as the BrK,,,, lines are very close to the arsenic K absorption edge. However, Tables I1 and I11 show negligible difference in line ratios compared to tin or zirconium additions. As illustrated in Figure 7 , the SeKcrl,* lines fall between the absorption edges of germanium and arsenic and preferentially excite germanium. This places the sample in a n L2 Class C group, in which the intensity of the line of
Table 111.
longer wave lengths is preferentially increased. As noted in Table 11,the increase in the measured lines ratio is proportional t o the concentration of selenium. With gallium as the reference element an L1 and El Class A matrix is obtained upon addition of selenium. If gallium is the third component in the germanium-arsenic system, preferential absorption of the shorter wave length ASK,, characterizes this as belonging to a n E2 Class B matrix. Table I1 shows the variation in the comparison lines ratio with gallium concentration. Addition of arsenic to a mixture of germanium and gallium results in preferential excitation of the longer wave length GaK,, which places this sample in L1 Class C. Zinc oxide has an absorption edge between the germanium and gallium K,, lines, and thus the sample containing this oxide belongs to the Ez Class B group. Addition of zinc oxide to germanium-arsenic gives a n E3 Class A matrix. Even though the zinc K edge is very close to the GeK,, line, no preferential absorption was noted. Therefore, the nearness of the
Evaluation of Gallium as an Internal Standard for Germanium
Matrix, G.
IG~K,(~)' x 100 IG~K,(~)
I x b x 100 IR
Deviation from Unity,
%
I. SnO, 0.2 0.4 0.6 0.8 11. ZrOz 0.2 0.4 0.6 0.8
59.5 39.4 30.4 25.0
99.1 98.9 01.6 97.3
-0.9 -1.1 1.6 -2.7
83.0 66.8 58.0 51.3
03.1 01.1 99.4 01.3
3.1 1.1 -0.6 1.3
0.2 0 4 0 6 0 8
100.0 89.3 81.5 77.6
02.4 104.3 103.0 101.3
2.4 4.3 3.0 1.3
0.2 0 4 0 6 0 8
100.0 88.0 76.5 69.4
103.3 102.5 106.0 101.7
3 3 2.5 6.0 1.7
0.2 0 4 0 6 0.8
87.2 75.7 68 2 64.3
87.1 78.8 62.8 60.8
-12.9 -21 2 -37.2 -39.2
0.2 0 4 0.6 0.8
47.9 28.9 21.9 16.2
62.0
-38 -54 -60 -67
0.2 0.4 0.6 0.8
69.1 55.3 44.3 41.5
00.6 02.1 07.3 97.7
0 6 2 1 -2.7 -2 3
52.1 36.2 27.8 20.8
97.8
-2.2 0.7 2.8 0.0
111. NaBr
Se
IV.
V.
AS203
VI. ZnO
VII. CUO
VIII.
0.2 0.4 0.6 0.8
cozo3
45.6 40.0 32.9
.oo . i
02.8 -00.0
0 4 0 1
absorption edge is not important as long as the comparison lines occur on the same side of the edge. Cobaltic oxide and cupric oxide resulted in E3 and L3 Class A matrices; the results given in Tables I1 and I11 are in close agreement with those found for El and LI matrices. Theoretical calculations and the above experimental results showed t h a t internal standards are reliable for Class A matrices only. The presence of interfering elements can be detected by a preliminary scan of the sample, which permits the proper choice of reference element. Possible overlapping lines can be determined from tables similar to those of Campbell and Parker ( 7 ) . Spectral lines and absorption edges that result in Class B and Class C matrices are determined from tables by Fine and Hendee (1.2). K i t h the internal standard technique the measured comparison-line ratio varied at high counting rates, owing to counting losses in the Geiger tube. Csing a scintillation counter or operating a Geiger counter at lo^ counting rates effectively eliminated this counting-loss problem. Another solution was the adjustment of the amount of reference element so that the lines to be compared are equal in intensity. Comparison of line ratios from calibration standard and the unknown at equivalent counting rates also solved this nonlinearity problem. As the comparison line ratio varies with the applied voltage, the calibration standard and the unknown sample should be excited at the same voltage, preferably three t o four times the critical excitation value of either element. K i t h a n excitation voltage of 45 kv., a variation +1 kv. has no detectable effect on the ratio of germanium K,, to gallium K,,. The necessary calculations for analysis n i t h the internal standard method were made with the following equation: Wt. yo Ge = 0.694 X wt. % reference X dilution factor X intensity calibration X I G ~ K ~ J I(unknon-n) RK~~ (6) where 0.694 = weight fraction Ge/GeOa Wt. sc reference = weight reference ouide/(n-eight of reference material weight of sample) Dilution factor = (wt. of sample wt. of reference material)/mt. of sample
+
+
Intensity calibration = I K K , , / I G ~measK~~ ured at 45 kv. on 1 to 1 ratio reference oxide/GeO* diluted in coal or coal ash. I G ~ K , ~ / I R(unknomi) K~~ = intensity ratio determined on unknonm sample at 45 kv Calibration standards were prepared by dilution of a mixture of 1 to 1 weight ratio of germanium oxide and reference oxide in a germanium-free coal or coal ash, followed by wet grinding with acetone in a hand mortar. The reference element must be VOL. 2 9 , NO.
7,JULY 1957
1013
thoroughly incorporated unto the unknonn sample because of tlie small depth of the sample analyzed. The difficulty in obtaining the necessary homogeneity has been stressed by Adler and L4velrod (1) and 3Iortimore, Romans, and Tews (2.2). Initially the use of very fine powders was tried for the reference element and unknown. in an effort to obtain equivalent paths for the comparison lines. However, the finest screens pass particles of 30- t o 40-micron diameter, which are too large to mix adequately. The reference material and unknown may be heterogriic'ous because of moisture or static charges. Therefore, shaking or tumhling devices in themselves may not inrorporate the reference satisfactorily into the sample. To obtain a uniform sample. it was necessary to grind the materials together after the mixture had been shaken thoroughly and passed through a 200-mesh screen. Previous n ork on tlie preparation of mixed oxide itandards of tantalum and niobium by Campbell aiid Carl (6) demonstrated the advantages of n e t grinding; therefore, the coal ash plus the reference was I\ c.t ground n it11 act.tone to ensurr roiiiplete mixing. Adlcr and Axelrod [ I ) suggested the rdtlitioii of a highly abrasive material such as silicon carbide, so that both sample and reference \rould be ground unifornily-, but it is not necessary for the softer coal ashes. Another mixing technique was use of a flus to fuse the components into a solid solution; sodium carbonate TI as used satisfactorily with arescnious oxide and roal ash. However. thc use of a fluxing agent is undesirable in the very loir eoncentration rangc. as the original saniple is diluted by a factor of 5 to 10. T h r results obtained by various w n p l e preparation trchniqueb are suniniarizd in Table IV.
Table IV. Comparison of Intensity Ratios of Internal Standard Calibration for Arsenic-Germanium System Obtained b y Different Methods of Sample Preparation
Preparation of Standards' Giounddryinmortar
Xnalysis SO. 1 2
I k s ~ e i
I(,c~ that nic,asured within 10% on triplicate smiples. If it is necessary to extend tlie liiiiit of detection, a larger starting ~ a i ~ i p lniay e be usrtl. For exaniplr. 2 to 4 1i.p.m. can lie detected in a 1gram co:il samlilc.. hut approximately 0.2 to 0.4 p.p.rii. could lie detectcd if tlie original coal saniple n-eiglic.tl 10 grams or mow. It \voultl lx, possiIjlr: to deterniiiie microgr:ini amounts to an accurarj- oi ivitliin i IT,, if the collection of the (,lenient \\-:is qumititntirc, anti a uniforni layer was fornietl. A s cliemical pre-
06
A C C U R A C Y OF FLUORESCENT X-RAY SPECTROGRAPHIC M E T H O D S
Three kinds of line nieasurements are of interest for the analytical methods described : measurement of a line intensity n.herc the Iiackground is negligihlc. a veak line abore background, and line intensity ratios for the internal standard and addition methods (20). U-htln the hackground is ncgligitjle-
SeKa GeKa ZnKa BrKa A5Ka GaKa CuKa
1
04
cipitations in the microgram range gave considerable difficulty, application of ion exchange membranes was considered (16). illthough a suitable ion exchange technique for germanium was not developed, in favorable cases. such as with zinc or cobalt ions in dilute solutions, the exchange was 98 to 99% complete and the distribution of the ion on the niembrane was uniform. so that the precision of analysis n a s nithin i 1.5% in the microgram range.
12
IO
08
1 x 1
14
WAVELENGTH
Figure 7.
Characteristic lines and absorption edges
3OC CtkCHONIhlE
;EF"~A\~NCLYB~ATE
MOLYBERldY
iT
TARGE'
45 k v ,
45 ma
TOPAZ
CRYSTAL
X-RAY X.RAY
TUBE
dz280A
m
a $00 > i= v r
1
-
z **wz
I
I
0
~
I00
K
!
i I ZOO ! GEQMANILJM, MICROGRAMS i
i 300
e.g., line-background ratios exceeding 10 to 1-the standard deviation approximates the square root of the number of counts collected. However, in trace analysis, the ratio of line height to background is a very important factor ( 3 . 1 4 ) . K i t h a lycstandard deviation on both measurements, the analytical line can be tlrterniineti when the ratio of analytical within =t5yc line to background is unity. Thc lon-er limit of detection with a 1% standard deriation is an analytical line approximately 57c abow background. R l t h rither the internal standard or addition method it is necessary to knonthe standard deviation from the nieasurcmriit of tlie intensity ratio. With a standard dwiation of 1% for cach of the comparison lines ahove a ni&iblo l,acskground, the stnndarcl dcriatioii for thc ratio is 1.417,. Count,ing statisticss usually arc not the limiting factor in the determination of line intwsity ahovil hackgrountl. bec a u s ~the cffwt of impurities must also he considcwtl. It is ncwssary to d(,t,iJrminr the liackgrountl a t :I position free of interfering lines, as the barkground should the intensity of th(1 scattrrctl radiation only. For thc topaz crystal (2d = 2.T9G A.) a value of 0.80" 28 bc,lon. tlie GeKal line location on the goniomcter was used, as it m s found to lie frer of interfering spwtral lines. ( K i t h the molylitir~nuni tub(, operatid a t 50 k r . , 45 ma. the line intensity is approximately 8 counts per srcond for O . O 1 ~ o germanium ahove a background of 18 count's per second.) Recausc of these limiting factors. the l o w r limit of tlctcc%ion for 1lethods 1. 2 . and 3 was conclutled to lie ap. O . O 1 O ~ o gcrmnniuni. h l is c.onsider:tbly higher tlian reported by optiml spectrographic t w h niquw, an econoniic amount of gernianium in any ( a i l ash woultl he determined. Headlcc ( 1 7 ) rci)orts that only ( a i l ashcs containing more than 0 . 0 5 ~ ,gtJrninniuni niay Iir of cwxioniic value. By tlic. UT of 1lethod 1. .50 to 100 samplrs of coal ashes ran bc analyzcti in an S-hour day, x i t h minimum snmplc iireparation. The coal ashcs n-ith n germanium content of >0.05% can be more ncwiratdy analyzed Iiy 1Irthods 2 or 3. The suggcsted procedure for samples containing < 1 % gcrmanium is 1Irthotl 3 with a gc.rIii:iiiium-ricli lignitr ash as a sourcC of the elcment untlrxr dctcrniination. 1Icthotl 2 iq recommended for higher conrcntration (>1%,) or samples n-here the quantity of coal ash is limited. 1Icthod 4 is the most acc.uratc procedurc, for trac.c,s ticrc~loprtl in this l ~ study and thc lon-cr limit of detection is npprosimatrly 2 to 4 y or 0.0002 to 0.00047, in a 1-grain f;amplc. Conccn-
I
400
I
Figure 8. Calibration curve for cinchonine germanomolybdate filter paper
i
VOL.
29, NO. 7,JULY 1957
1015
Table
VI.
Determination of Germanium by Various Analytical Methods"
Germanium, 1Vt. yo Sample No.
Emission spectrographic
M-6-16 hI-6-2 H-5-4 17-6-18 Ma-6-1 Ma-6-18 PH-2-3 BB-5-1 17-6-19 V-6-20 OB-6-2 V-6-1 L'I-6-1 Ha-7-4 OB-6-1 T-1
0.003' 0 .O O i b 0.011h 0.02 0.02 0,02 0,008 0 . 032b 0.026 0.037 0.046 0.046 0.07 0.07 0.12
T-9 Ash G Lignite A Lignite G
Phenylfluorone colorimetric
A
Fluorescent X-Ray Spectrographic B C D
... ... 0.010
0: 025
... ...
O:i)i)5 0.012 0.022 0,020 0,023 0.046 0.039 0,058 0,040 0.049 0.046 0.044 0.076 0.12
...