- Email: [email protected]
Oxygen stoichiometry in the geochemical standards of the U.S. Geological Survey: A blind study
Chemical C,eolo~,. 20 (1977) 8~, 31 © Elsevier Scientific Publkhing Company, Amsterdam -- Printed in The Netherlands
86
Short Notes OXYGEN STOICHIOMETRY IN THE GEOCHEMICAL STANDARDS OF THE U.S. GEOLOGICAL SURVEY: A BLIND STUDY*'
ALEXIS VOLBORTH', GEORGE E. MILLERI and CLAUDIA K. GARNERI ~A'orth Dakota State UnJvemffy. Fargo. N.D., 68102 (U.S.A.),2 z Department o f Chemistry. Unfversify o f Calffornia, lrvJne, Calif.. 92717 (U.S.A.)
(Raceived July 12, 1976; accepted for publication September 20, 1976)
ABSTRACT Volborth. A.. Miller. G.E. and Garner. C.K., 1977. Oxygen stoichiometry in the geochemical standards of the U.S. Geological Survey: a blind study. Chem. Geol.. 20: 96--91. A blind study of total O content in nine USGS standard rocks G-2, AGV-1, GSP-1, DT8-1, BCR-I, SCo-I, MAG-1, RGM-I and QLO-1, Reagent Grade oxalic acid and Spacpure SiOswss performed usins the activation method by 14.3-Me V neutrons and a dedicated dual sample transfer and counting system. Over a composition range varying from 43 to 76% O absolute deviation from stoichiometric 0 varied from 0.07 to 0.88%. A simple computer program w~ used to calculate the stoinhiometrie O from chemical mnmlysesof the rocks and to show summations including the FNAA oxygeL Accurate O data permit the evaluation of oveildl accuracy of rock sn,dysis and help to detect deficiencies, errors, and incomplete work by eneblinl the chemist to cainulate the analysis unambiguously.
INTRODUCTION Accurate and rapid analysis o f O b y fast.neutron activation is becoming increasingly accessible. (Volborth et al., 1975). The chemical analysis of complex systems for major and trace elements presents great difficulties when one attempts to balance the results stoichiometrically. The large n u m b e r o f determinations to be s u m m e d up causes accumulative errors. One generally is satisfied with summations o f I00 ± 1.0%; nevertheless, in a carefully perf o r m e d analysis one strives for at least a 100 +_ 0.5% summation. This necessitates the taking into a c c o u n t of O in the s - m of major and s o . a i l e d trace and minor constituents, because together, the minor and the trace elements often form a total of 2 wt.% in an analysis o f silicate rocks (Volborth et al., 1966). It would be helpful if all these constituents were determined independently and s u m m e d as to reflect the stoichiometry o f these m u l t i c o m p o n e n t *' This work was supported by NSF Grants GA-40511 and DES 74-01945 AOl. .2 Visiting Professor, University of California, Irvine, Calif., 92717 (U.S.A.)
86 a~
.:
<<
•~
o
~
~
O
O
O
O
0
~
O
O
~
O
~
. O
, O
O
O
O
O
as
G ¢J
.
.
. . . O O
• O
O
,
. O O
.
G .
D~
rE_
.
J
_
m
~
.
.
.
.
"a
El
m'-
i ~
~
~
°
87 mixtures. Because most of the trace elements are heavier than the majors their contribution to stoichiometric O is disproportionate. This further complicates the estimates of stoichiometric O. We a t t e m p t here to give actual examples of recalculation of a silicate analysis and provide accurate additional data on O content in some widely used rock standards. RE8ULTS
In January, 1975, Dr. Hugh T. Millard of the USGS Reactor Facility in Denver, Colorado, kindly submitted eleven rock powder samples u n k n o w n to us and labelled USGS-A to USGS-K. We analyzed these using our dual fast-neutron activation m e t h o d and reported the data, after which the identity of samples was revealed to us. We had analyzed standard rocks RGM-1, BCI~I, DTS-1, SCo-1, MAG-1, G-2, AGV-1, GSP-1, and QLO-1, and a Spacpure quartz, SiO2, by Matthey and a Reagent Grade oxalic acid, (COOH)2. 2H20. We also had analyzed previously (Volborth and Vincent, 1967), and in this laboratory (1974) from other batches the USGS rocks andesite-AGV-1, granodiorite-GSP-1, granite-G-2, dunite-DTS-1, basalt-BCR-1, and peridotitePCC-1. Table I compares our results with data calculated from the literature (Flanagan, 1969) and with analyses of these substances submitted by the USGS (Hugh T. Millard, personal communication, 1975). Recalculated and critically reviewed data by Abbey (1970, 1973) are also used. There are two columns giving stoichiomeuric O based on USGS data and data by Abbey, as well as three columns for oxygen by FNAA for the different splits analyzed by us. Absolute and relative differences of FA'AA data from calculated are given, also the differences of summation from 100% calculated from classical analysis. The calculations were performed using a single computer program written in BASIC language for the PDP-11 computer (Digital Equipment Corporation). TABLE II The effect of moisture on total oxygen in rock powders Sample
% Oxygen in dry sample
Relative standard deviation
% HzO (1100C)
% Oxygen "as is"
0.16 0.22 0.34 0.22 0.24 0.28 0.36 0.28 0.21
0.24 0.47 0.07 1.77 2.82 0.08 0.73 0.07 0.27
49.46 45.16 43.67 50.87 46.66 48.26 47.02 46.03 47.99
(%) RGM-1 BCR-1 DTS-I SCo-1 MAG-I G-2 AGV-I GSP-I QLO-I
49.87 44.95 43.64 50.18 45.66 48.23 46.71 48.00 47.88
88 TABLE IX[ Calculate % o z y p n in compovnd, UBGS-F, MAG-1 Total determined O by Na ? 46.66 % oxygen .
.
.
.
.
.
.
.
.
.
?
% A I , O= determined = Multiply by .47085 M u l t i p l y by .52915
? 16.44
% Fe=O= determined : Multiply by .80087 Multiply by .89948
? 2.64
%FeO=
? 3.66
.
.
.
metals .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
: 28.2505
% TiOs Multiply by .40049 Multiply by .59951 % MIIO = Multiply b y . 3969 Multiply by .6031 % CaO = Multiply by .2868 Multiply by .7147
P=O.
.
: 26.4895
= 7.74077 :
8.69923
:.798605 : 1.8465
Multiply by .22268 Multiply by .77752
%
.
49.74
% 8i0 s determined : Multiply by .53266 Multiply by .46744
% Na=O = Multiply by Multiply by %K,O: Multiply by Multiply by % MnO = Multiply by Multiply by
.
=.812782 = 2.88722
? .70 =.280848 =.419657 ? 2.98 : 1.18276 = 1.79724 ? 1.50 =.42795
: 1.07205 " 3.9
.25814 .74186
: 1.00675 -
2.89325
:
2.98858
9 3.6 .16924 .88016
:.611424 .10
.22654 .77446
=
:.022554 :.077446
.32
Multiply by .56358 Multiply by .48642
% CO= = Multiply by .72708 Multiply by .27292 %H=O* = Multiply by .8881 Multiply by .1119 %I4,0" =
MulUply by .8881 Multiply by .1119 Oxygen throuJhHzO= 46.8467
=.180546 =.139654 9.51 =.370811 -
.139189
5.22 = 4.68888
=.584118 o 2.58 -
2.2913
:.288702
89
TABL~- I I I
(continued) % oxygen
C o n t i n u e traces, yea = 1 n o : 0 ? 1 % ZrOz : ?0 % I ~ O , ffi ?0 %BaO= ?0 % SOs : ?.15 M u l t i p l y b y .59951 Multiply by .40049 %SrO=
?0
%L~O:
?0
:.899265E-1 :.600735E-1
% Rb=O = ? 0 % Cr10 s = ? 0 % NiO ? 0 O x y g e n in c o m p o u n d n o t corr. for F, S, Cl, etc.
= 46.9366
Continue anions ? 1
% oxygen
%F:
% metals
% non metals
?.12 = .12
:-.505248E-I
Multiply by-.42104 O ffi - . 5 0 8 2 4 8 E - I O x y g e n corr. f o r F = 4 6 . 8 8 6 1
F : (I/2)
% S ffi
?.62 = .52
=-.25947
Multiply by -.49898 S : O : -.25947 O x y g e n corr. f o r F a n d S : 4 6 . 6 2 6 7 % CI :
? 2.86 : 2.86
"-.64533
Multiply by -.22564 C1=(1/2) 0 :-.64633 O x y g e n corr. f o r F , 8, CI = Metal + non metals :
45.9813
50.60
Total oxygen determined f r o m line 2 : 4 6 . 6 6 S u b t r a c t .12 t i m e s . 6 1 4 NA oxygen determined, corr. for F: Rtoiehiometrie oxyjen in compound repeat: % compound • : C a r b o n , etc.
4~5983 45.9813
? 2.37
N A o x y g e n + m e t a b + F, S, CI + • c a l c u l a t e d o x y g e n + m e t a l s + F , 8, CI + •
total - 99.6617 t~tal - 98.9447
90
This program permits a rapid recalculation of classical chemical analyses reported in oxide percentages and the separate addition of total O and total metals with stoichiometric corrections for F, CI, S, etc., and nuclear correction for F interference performed by the computer. DISCUSSION There appears to be good agreement between stoichiometric and FNAA oxygen in these standard rocks, which is expected assuming that most of these standards have been analyzed by numerous chemists around the world and the data critically assembled (Flanagan, 1969; Abbey, 1970, 1973). In order to evaluate the effect of moisture (H20-) in these samples on the total O value the H20- was determined separately on each while the sample was packed. Table II gives O de~ermined on dry vs. "as is" samples. It is apparent that in order to analyze for and report O accurately in powders one does have to take the hygroscopic water into account. Note that the small differences in O percentages in this table when compared to Table I are due to corrections due to F present, not yet performed at this stage. We suggest that the accurate 0 data can be used as an additional check of accuracy and stoichiometric balance of rock analyses. As an example of our ability to indicate that something is missing from an analysis we show a provisional analysis of marine mud MAG-1 by USGS recalculated by our computer (Table III). This material contains 0.12% F, 0.52% S, 2.86% CI, and 1.43% C (Hugh T. Millard, personal communication, 1975). The elements S. Cl and C are not regularly determined in a rock analysis. With an independent determination of O by FNAA the sum (not counting the above elements) is about 94%, indicating that further work is necessary. Suppose that one did not know of the large quantity of CI and C in this rock, then assuming the accuracy of the method, one could indicate that the analysis was not yet complete and that some major constituents remained undetected or some significant errors were made in the determination of the major elements. CONCLUSIONS This study shows the application of accurate O deterrnination to evaluation of completeness of a complex analysis of a multicomponent system, a rock in this case. This approach is applicable to a multitude of geochemical and industrial analytical problems when a total analysis is desired. Based on the close agreement of the stoichiometric O with the determined O in all of the USGS rock standards, one may suggest that the quality of the analytical work on these substances is high. Considering the difficulties in summing up major and trace elements one may also suggest that in terms of O stoichiometry little could be improved by additional work in this case on these standards except the marine mud, MAG-1.
91
NOTE A c o m p u t e r program used t o recalculate the data can be o b t a i n e d f r o m Prof. A. Vo lb o r th , either at N o r t h D a k o t a State University, Fargo, N o r t h Dakota, 58102, or at University o f California, Irvine, California, 92717.
REFERENCES Abbey, S., 1970. U.S. Geological Survey Standards. A critical study of published analytical data. Can. Speetrcee., 15: 10--16. Abbey, S., 1973. Studies in "standard samples" of silicate rocks and minerals, 8. 1973 Extension and revision of "usable" values. Geol. Surv. Can., Pap. 78--36: 1--25. Flanagan, FJ., 1969. U.S. Geological Survey Standards, II. First compilation of data for the new U.S.G.S. rocks. Geochim. Ccamochim. Acta, 88: 81--120. Volborth, A., Miller, G.E. and Garner, C.K., 1976. Oxygen: accurate determination of a neglected element and its potential in chemistry. Am. Lab. 7(10): 87---98. Volborth, A. and Vincent, H.A., 1967. Determination of oxygen in USGS rock standards by fsat-neutron activation. Nuclear Appl., 3: 701--707. Volborth, A., Fabbi, B.P. and Vincent, H.A., 1968. Total nondestructive analyak of CAAS syenite. Adv. X-Ray Anal., 11: 158--168.
86
Short Notes OXYGEN STOICHIOMETRY IN THE GEOCHEMICAL STANDARDS OF THE U.S. GEOLOGICAL SURVEY: A BLIND STUDY*'
ALEXIS VOLBORTH', GEORGE E. MILLERI and CLAUDIA K. GARNERI ~A'orth Dakota State UnJvemffy. Fargo. N.D., 68102 (U.S.A.),2 z Department o f Chemistry. Unfversify o f Calffornia, lrvJne, Calif.. 92717 (U.S.A.)
(Raceived July 12, 1976; accepted for publication September 20, 1976)
ABSTRACT Volborth. A.. Miller. G.E. and Garner. C.K., 1977. Oxygen stoichiometry in the geochemical standards of the U.S. Geological Survey: a blind study. Chem. Geol.. 20: 96--91. A blind study of total O content in nine USGS standard rocks G-2, AGV-1, GSP-1, DT8-1, BCR-I, SCo-I, MAG-1, RGM-I and QLO-1, Reagent Grade oxalic acid and Spacpure SiOswss performed usins the activation method by 14.3-Me V neutrons and a dedicated dual sample transfer and counting system. Over a composition range varying from 43 to 76% O absolute deviation from stoichiometric 0 varied from 0.07 to 0.88%. A simple computer program w~ used to calculate the stoinhiometrie O from chemical mnmlysesof the rocks and to show summations including the FNAA oxygeL Accurate O data permit the evaluation of oveildl accuracy of rock sn,dysis and help to detect deficiencies, errors, and incomplete work by eneblinl the chemist to cainulate the analysis unambiguously.
INTRODUCTION Accurate and rapid analysis o f O b y fast.neutron activation is becoming increasingly accessible. (Volborth et al., 1975). The chemical analysis of complex systems for major and trace elements presents great difficulties when one attempts to balance the results stoichiometrically. The large n u m b e r o f determinations to be s u m m e d up causes accumulative errors. One generally is satisfied with summations o f I00 ± 1.0%; nevertheless, in a carefully perf o r m e d analysis one strives for at least a 100 +_ 0.5% summation. This necessitates the taking into a c c o u n t of O in the s - m of major and s o . a i l e d trace and minor constituents, because together, the minor and the trace elements often form a total of 2 wt.% in an analysis o f silicate rocks (Volborth et al., 1966). It would be helpful if all these constituents were determined independently and s u m m e d as to reflect the stoichiometry o f these m u l t i c o m p o n e n t *' This work was supported by NSF Grants GA-40511 and DES 74-01945 AOl. .2 Visiting Professor, University of California, Irvine, Calif., 92717 (U.S.A.)
86 a~
.:
<<
•~
o
~
~
O
O
O
O
0
~
O
O
~
O
~
. O
, O
O
O
O
O
as
G ¢J
.
.
. . . O O
• O
O
,
. O O
.
G .
D~
rE_
.
J
_
m
~
.
.
.
.
"a
El
m'-
i ~
~
~
°
87 mixtures. Because most of the trace elements are heavier than the majors their contribution to stoichiometric O is disproportionate. This further complicates the estimates of stoichiometric O. We a t t e m p t here to give actual examples of recalculation of a silicate analysis and provide accurate additional data on O content in some widely used rock standards. RE8ULTS
In January, 1975, Dr. Hugh T. Millard of the USGS Reactor Facility in Denver, Colorado, kindly submitted eleven rock powder samples u n k n o w n to us and labelled USGS-A to USGS-K. We analyzed these using our dual fast-neutron activation m e t h o d and reported the data, after which the identity of samples was revealed to us. We had analyzed standard rocks RGM-1, BCI~I, DTS-1, SCo-1, MAG-1, G-2, AGV-1, GSP-1, and QLO-1, and a Spacpure quartz, SiO2, by Matthey and a Reagent Grade oxalic acid, (COOH)2. 2H20. We also had analyzed previously (Volborth and Vincent, 1967), and in this laboratory (1974) from other batches the USGS rocks andesite-AGV-1, granodiorite-GSP-1, granite-G-2, dunite-DTS-1, basalt-BCR-1, and peridotitePCC-1. Table I compares our results with data calculated from the literature (Flanagan, 1969) and with analyses of these substances submitted by the USGS (Hugh T. Millard, personal communication, 1975). Recalculated and critically reviewed data by Abbey (1970, 1973) are also used. There are two columns giving stoichiomeuric O based on USGS data and data by Abbey, as well as three columns for oxygen by FNAA for the different splits analyzed by us. Absolute and relative differences of FA'AA data from calculated are given, also the differences of summation from 100% calculated from classical analysis. The calculations were performed using a single computer program written in BASIC language for the PDP-11 computer (Digital Equipment Corporation). TABLE II The effect of moisture on total oxygen in rock powders Sample
% Oxygen in dry sample
Relative standard deviation
% HzO (1100C)
% Oxygen "as is"
0.16 0.22 0.34 0.22 0.24 0.28 0.36 0.28 0.21
0.24 0.47 0.07 1.77 2.82 0.08 0.73 0.07 0.27
49.46 45.16 43.67 50.87 46.66 48.26 47.02 46.03 47.99
(%) RGM-1 BCR-1 DTS-I SCo-1 MAG-I G-2 AGV-I GSP-I QLO-I
49.87 44.95 43.64 50.18 45.66 48.23 46.71 48.00 47.88
88 TABLE IX[ Calculate % o z y p n in compovnd, UBGS-F, MAG-1 Total determined O by Na ? 46.66 % oxygen .
.
.
.
.
.
.
.
.
.
?
% A I , O= determined = Multiply by .47085 M u l t i p l y by .52915
? 16.44
% Fe=O= determined : Multiply by .80087 Multiply by .89948
? 2.64
%FeO=
? 3.66
.
.
.
metals .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
: 28.2505
% TiOs Multiply by .40049 Multiply by .59951 % MIIO = Multiply b y . 3969 Multiply by .6031 % CaO = Multiply by .2868 Multiply by .7147
P=O.
.
: 26.4895
= 7.74077 :
8.69923
:.798605 : 1.8465
Multiply by .22268 Multiply by .77752
%
.
49.74
% 8i0 s determined : Multiply by .53266 Multiply by .46744
% Na=O = Multiply by Multiply by %K,O: Multiply by Multiply by % MnO = Multiply by Multiply by
.
=.812782 = 2.88722
? .70 =.280848 =.419657 ? 2.98 : 1.18276 = 1.79724 ? 1.50 =.42795
: 1.07205 " 3.9
.25814 .74186
: 1.00675 -
2.89325
:
2.98858
9 3.6 .16924 .88016
:.611424 .10
.22654 .77446
=
:.022554 :.077446
.32
Multiply by .56358 Multiply by .48642
% CO= = Multiply by .72708 Multiply by .27292 %H=O* = Multiply by .8881 Multiply by .1119 %I4,0" =
MulUply by .8881 Multiply by .1119 Oxygen throuJhHzO= 46.8467
=.180546 =.139654 9.51 =.370811 -
.139189
5.22 = 4.68888
=.584118 o 2.58 -
2.2913
:.288702
89
TABL~- I I I
(continued) % oxygen
C o n t i n u e traces, yea = 1 n o : 0 ? 1 % ZrOz : ?0 % I ~ O , ffi ?0 %BaO= ?0 % SOs : ?.15 M u l t i p l y b y .59951 Multiply by .40049 %SrO=
?0
%L~O:
?0
:.899265E-1 :.600735E-1
% Rb=O = ? 0 % Cr10 s = ? 0 % NiO ? 0 O x y g e n in c o m p o u n d n o t corr. for F, S, Cl, etc.
= 46.9366
Continue anions ? 1
% oxygen
%F:
% metals
% non metals
?.12 = .12
:-.505248E-I
Multiply by-.42104 O ffi - . 5 0 8 2 4 8 E - I O x y g e n corr. f o r F = 4 6 . 8 8 6 1
F : (I/2)
% S ffi
?.62 = .52
=-.25947
Multiply by -.49898 S : O : -.25947 O x y g e n corr. f o r F a n d S : 4 6 . 6 2 6 7 % CI :
? 2.86 : 2.86
"-.64533
Multiply by -.22564 C1=(1/2) 0 :-.64633 O x y g e n corr. f o r F , 8, CI = Metal + non metals :
45.9813
50.60
Total oxygen determined f r o m line 2 : 4 6 . 6 6 S u b t r a c t .12 t i m e s . 6 1 4 NA oxygen determined, corr. for F: Rtoiehiometrie oxyjen in compound repeat: % compound • : C a r b o n , etc.
4~5983 45.9813
? 2.37
N A o x y g e n + m e t a b + F, S, CI + • c a l c u l a t e d o x y g e n + m e t a l s + F , 8, CI + •
total - 99.6617 t~tal - 98.9447
90
This program permits a rapid recalculation of classical chemical analyses reported in oxide percentages and the separate addition of total O and total metals with stoichiometric corrections for F, CI, S, etc., and nuclear correction for F interference performed by the computer. DISCUSSION There appears to be good agreement between stoichiometric and FNAA oxygen in these standard rocks, which is expected assuming that most of these standards have been analyzed by numerous chemists around the world and the data critically assembled (Flanagan, 1969; Abbey, 1970, 1973). In order to evaluate the effect of moisture (H20-) in these samples on the total O value the H20- was determined separately on each while the sample was packed. Table II gives O de~ermined on dry vs. "as is" samples. It is apparent that in order to analyze for and report O accurately in powders one does have to take the hygroscopic water into account. Note that the small differences in O percentages in this table when compared to Table I are due to corrections due to F present, not yet performed at this stage. We suggest that the accurate 0 data can be used as an additional check of accuracy and stoichiometric balance of rock analyses. As an example of our ability to indicate that something is missing from an analysis we show a provisional analysis of marine mud MAG-1 by USGS recalculated by our computer (Table III). This material contains 0.12% F, 0.52% S, 2.86% CI, and 1.43% C (Hugh T. Millard, personal communication, 1975). The elements S. Cl and C are not regularly determined in a rock analysis. With an independent determination of O by FNAA the sum (not counting the above elements) is about 94%, indicating that further work is necessary. Suppose that one did not know of the large quantity of CI and C in this rock, then assuming the accuracy of the method, one could indicate that the analysis was not yet complete and that some major constituents remained undetected or some significant errors were made in the determination of the major elements. CONCLUSIONS This study shows the application of accurate O deterrnination to evaluation of completeness of a complex analysis of a multicomponent system, a rock in this case. This approach is applicable to a multitude of geochemical and industrial analytical problems when a total analysis is desired. Based on the close agreement of the stoichiometric O with the determined O in all of the USGS rock standards, one may suggest that the quality of the analytical work on these substances is high. Considering the difficulties in summing up major and trace elements one may also suggest that in terms of O stoichiometry little could be improved by additional work in this case on these standards except the marine mud, MAG-1.
91
NOTE A c o m p u t e r program used t o recalculate the data can be o b t a i n e d f r o m Prof. A. Vo lb o r th , either at N o r t h D a k o t a State University, Fargo, N o r t h Dakota, 58102, or at University o f California, Irvine, California, 92717.
REFERENCES Abbey, S., 1970. U.S. Geological Survey Standards. A critical study of published analytical data. Can. Speetrcee., 15: 10--16. Abbey, S., 1973. Studies in "standard samples" of silicate rocks and minerals, 8. 1973 Extension and revision of "usable" values. Geol. Surv. Can., Pap. 78--36: 1--25. Flanagan, FJ., 1969. U.S. Geological Survey Standards, II. First compilation of data for the new U.S.G.S. rocks. Geochim. Ccamochim. Acta, 88: 81--120. Volborth, A., Miller, G.E. and Garner, C.K., 1976. Oxygen: accurate determination of a neglected element and its potential in chemistry. Am. Lab. 7(10): 87---98. Volborth, A. and Vincent, H.A., 1967. Determination of oxygen in USGS rock standards by fsat-neutron activation. Nuclear Appl., 3: 701--707. Volborth, A., Fabbi, B.P. and Vincent, H.A., 1968. Total nondestructive analyak of CAAS syenite. Adv. X-Ray Anal., 11: 158--168.