American Mineralogist, Volume 70, pages 961-968, 1985

A single-dissolution technique for determining FeO and Fe2O3

in rock and mineral samples
SrrvsNF. Fnnz.lNo RonenrK. Popp
Department of Geology
Texas A&M Uniuersity
College Station, Texas 77843
Abstract
A single-dissolutiontechniquehas beendevisedto measureo/oFea and EFe as 7o FerO.
on small (<10 mg) samples.Colorimetric determinationof FeO is patternedafter Shapiro's
(1960)method with time-savingmodifications due to gravimetric dilutions. After the colorimetric measurements,
the solutionsare analyzedfor EFe as % FerO, by atomic absorption
spectrophotometry.The resultsfor both analysesgive valuesthat are in reasonableagreement
with other wet chemicalmethodsand with Mcissbauerspectraltechniques.
BecauseFeO and total iron are both measuredin a solution produced by a single digestion, weighingerrors of the dry sampledo not affect the ferric-ferrousratio obtained in the
analyses.This ratio should thus be determinableon vanishinglysmall sizes,provided that the
resultingsolutionslie within the detectionlimits of the analyticalinstruments.
Introduction
The iron content of silicatescan be determinedby several different analytical techniques.Electron microprobe
spectroscopymeasuresa sample'stotal iron-usually expressedas % FeO. M<issbauerspectroscopyis capableof
directly measuringthe Fe3+-Fe2+ratio, as well as determining site occupancyof iron within the crystal structure
of a mineral; however it does not yield quantitative
measurementsof a sample'siron content. Ironically, it is
the lessglamorouswet chemicalmethodsthat provide analytical versatility that can be used to obtain analysesof
both FeO and FerO.. From thesevalues,the Fe3+-Fe2+
ratio is easilycalculated.
Most wet chemicalmeasurements
of Fe2+-Fe3+ contents of rocks and minerals involve a 2-step dissolution
processin which a sampleis first analyzedfor total iron
expressedas FerO.. This is normally followed by a separate dissolution to directly determine the sample's FeO
content.
The manner in which a sampleis dissolveddictates the
valence of iron in the final solution used for total iron
analysis.The two most common techniquesemployed in
the dissolution of silicatesfor whole rock analysesare the
LiBO, fusion procedure (Medlin et al., 1969) and the
H3BO3/HF attack describedby Bernas(1968).Both techniques oxidize the iron so that only EFe as Fe2O3 can
be analyzed-typically by atomic absorbtion spectrophotometry. Alternatively,a samplecan be treated with
a strong reducing agent.In this case,the iron in the final
solution is ferrousso that EFe as FeO is determinedcolorimetrically (Shapiroand Brannock, 1962).
The next stepin the wet chemicalanalysisof iron usually
necessitatesa seconddissolution of the powdered sample
000H04x/85/0910-096I $02.00

for direct determination of FeO. This FeO determination

can be performedby titration or colorimetry.
Severaldifferenttitration techniqueshavebeendescribed
in the literature. In the technique of Schaefer(1966),the
dissolution and titration proceduresare carried out in a
nitrogen atmosphereto prevent oxidation by atmospheric
oxygen during digestionand analysisprocedures.The disadvantageof this techniqueis the tedious requirementof
working in an oxygen-freeenvironment.Inattentivenessto
this requirementmay lead to a low value for a sample's%
FeO (Fritz, 1976).The techniquesdescribedby Reichen
and Fahey (1962) and by Wilson (1955) as modified by
Whipple (1974)do not require an oxygen-freeatmosphere;
the sample is dissolvedand oxidized in the presenceof a
known amount of oxidizing agent. The excessoxidizing
agent is then titrated with a ferrous-bearingsolution, and
the Fe2+ content of the samplecalculated.Aside from the
disadvantageof relatively long dissolution times (overnight), titration techniquesgenerallyrequire more elaborate proceduresand well-honedanalyticalproficiency.
In the colorimetric determination of FeO, a sample is
dissolvedby a mixture of HF and H2SO4in the presence
of o-phenanthroline.This organic reagentservestwo purposes:(1) it preventsoxidation of the sample'sferrous iron
during dissolution by selectivelycomplexing with the releasedferrous iron during acid attack; and (2) it forms an
orange-coloredcomplex with Fe2+ whose intensity is directly proportional to the ferrous content of the digested
sample.This procedure,outlined by Shapiro(1960),is relatively tolerant ofatmospheric oxygenand thus obviatesthe
need for dissolution under oxygen-freeconditions: Morover, the l0 mg sample size prescribedby Shapiro (1960)
lends this technique to being modified to accommodate
dissolutionof samplesevensmallerthan 10 mg.

962

FRI'|Z AND POPP: DETERMINING FeO AND FerO,

Apart from the tedium of dissolvingtwo dry aliquots of

the same sample,the disadvantageof the standard twostage dissolution processto determine Fe2* and Fe3* is
that inhomogeneityof the powderedsamplemay result in
selectionof a non-representative
dry aliquot for anlaysesof
FeO and EFe as FerO3. This problem is minimized if both
iron analysesare performedon a solution obtained from a
singledissolutionof one dry aliquot.
Roth et al. (1968)deviseda schemeby which FeO and
FerO. could be obtained from a single dissolution.They
dissolved their samples according to Shapiro's (1960)
method to obtain %oFeO. Thereafter,they addedNH2OH
'Cl (hydroxylamine hydrochloride) to reduce all the soluble iron to the ferrous state. After a 24 hour wait. these
solutionswereanalyzedfor EFe as FeO by colorimetry.
This study describesa simpler and quicker wet chemical
procedurefor analysisof FeO and FerO. in silicate samples by a singledissolutiontechnique.Shapiro's(1960)dissolution procedure is adopted with time-saving modificationsbrought about by gravimetricdilutions using a toploading balance.Following the colorimetric determination
of FeO, the solutions are analyzedby atomic absorption
spectrophotometryfor determination of EFe as FerO..
The techniquerequiresrelativelyfew stepsrequiring quantitative measurements,and is carried out on instruments
availablein most modestlyequippedlaboratories.
To gauge the accuracyof this technique,10 rock standards and 10 previously-analyzedamphiboleswere treated
as unknowns,and the observedvaluesare comparedto the
publisheddata. Moreover, the Fe3+-Fe2+ ratios obtained
for the amphibolesby this wet chemicalmethod are compared to ratios obtained from publishedMiissbaueranalyses.
Analytical procdure
The list of equipment and reagentsneededto perform
the FeO and FerO. analysesis given in the appendix.
Weighing proceilure
Specialcare must be exercisedin precision weighing of
extremelysmall samples.One source of error that results
from transferralof samplefrom weighing paper to the digestionvesselcan be circumventedif the sampleis weighed
directly into the vessel,in this casea preJabelled,4 ounce
polypropylenebottle. The most troublesomeweighingdifliculty is weight gain or loss of the sampleduring the weighing procedure.To insure the samplesare free of absorbed
water prior to weighing,the rock standardsand unknowns
should be dried at 110"C for at least 12 hours, removed
and allowedto cool in a desicator.
The uncappeddigestionbottle and its cap are placedon
the pan ofan analyticalbalanceand weighedto the nearest
0.01 mg. This requires patiencebecausea newly opened,
clean polypropylene bottle sorbs water such that three
minutes sometimeselapse before the bottle's weight has
stabilized.With the balance still in full pan position, the
balancewindow is openedand a small amount of sampleis

dropped into the bottle. The balance window is quickly

closed and the weight is recordedto the nearest0.01 mg.
Becausethe accuracyof this weighingstepis crucial to the
accuracyof both the FeO and FerO3 analyses,the sample
weight should be measuredin a minimal amount of time in
order to minimize adsorption of water by the powdered
sample.With practice,this weighingstepcan be performed
in less than 20 seconds.All standards and unknowns
should be accuratelyweighed with the same dispatch so
that exposureof all samplesto the laboratory'shumidity is
of the sametime interval.
Next, a small amount of crystalline o-phenanthrolineis
placedinto the bottle containingthe weighedsampled.The
exact amount of this reagentis not critical. The only requirement is that enough o-phenanthrolinebe presentto
complex all of the FeO in the sample. One mole of ophenanthroline(180grams)can complex2 molesof Fe2*.
Thus, the addition of 20 mg of o-phenanthrolinewill complex the ferrous iron content of 16 mg of pure FeO. Because ex@ss o-phenanthroline is not deleterious to the
colorimetric technique,any small amount over 20 mg will
suffrce.Weighing of the standardsand unknowns can be
done days in advanceof the actual digestionsand spectrophotometric measurements.
Shapiro (1960)recommendsweighing 10.0 mg of rock
powder for the FeO determination.To achievethis exact
weight necessitatesexposing the rock powder to the humidity of the laboratory for times greatly exceeding20
seconds.A singledropping of the powderedsampledirectly
into the 4 oz. polypropylene bottle minimizes the error
causedby undue sorption of water by the sample.Moreover, the variation of samplesizeresulting from this rapid
weighing method does not affect the overall analytical results.
Figure I is a plot of absorbance of the ferrousphenanthrolinecomplex versusmilligrams of sample dissolved for two rock standards.Samplesizesrange from 4
to 20 mg. The figure demonstratesthat for the different
rock standards,absorbanceis strictly a function of the absolute amounts of FeO in the dissolvedsampleand is thus
independentof the varying solution matrix causedby dissolution of different weights of the same sample.The extremesimilarityof the L% FeOIL abs.relationsof the two
lines of Figure 1 means that if absorbancewere plotted
versus absolute %oFeO, all 9 samplesof these two rock
standards would plot on a line having a slope of about
29.18% FeO/1.@0 absorbanceunit, and an intercept of
0.01% FeO. Thus, it is possiblethat only one rock standard can be used for the FeO calibration curve, provided
that a wide rangeofweights ofthat accuraterock standard
is employed.This would be advantageousfor those analysts whosestock of rock standardsis extremelylimited.
Becausethe absorbanceof all unknowns must be bracketed by absorbanceof rock standards,the concentrationof
FeO in an unknown solution must be lessthan that of the
highest rock standard used in calibration. In this study,
the rock standard, "Fe-Mica", a high-iron biotite

FRITZ AND POPP: DETERMINING FeO AND Fe,O,

@
o

.9
p
=

16

g,

i4

E
a
E

(t,

L
(,P

o
o
L
dt
cl

c,
Cla

E
(t
U)
o6

((t
!

.90

C o r r e c t e dA b s o r b a n c e
Fig. 1. Variation of absorbancewith sample size of two rock
standards,G-2 and W-1. Reported % FeO for G-2 and W-l are
1.44V.a:nd8.73%,respectively(Abbey, 1980).Becausethe absorbance of different weights of samplesof the samestandard fall on a
line, the linearity of the absorbance-% FeO relationship is not
affectedby matrix effectscausedby dissolvingdifferentamountsof
the samesample.The number adjacentto eachdatum point representsthe total gramsof solution to which the digestedsamplewas
diluted with deionizedwater.

(FeO : 18.99%, total Fe as Fe2O3:25.76% (Abbey,

1980)),servesto anchor the highestposition of the calibration curve for both FeO as well as for IFe analyses.As
long as a weighedunknown has an absolute FeO content
lessthan that of Fe-Mica, the absorbanceof that unknown
will be bracketed by rock standards.Unknowns having
very high ferrous contents would, of course,have to be
weighedin amountssubstantiallylessthan that of the most
FeO-rich standard.
Dissolution procedure
Standards and unknowns are routinely dissolved in
batchesof sevenbecauseseven4 oz. polypropylenebottles
form a tightly-packedhexagon that fits neatly into the 4
liter glassbeakercontainingboiling water.The useof automatic pipets in volumetric dispensinggreatly reducesthe
time required for analyses.Three ml of loo/oHrSOn are

963

dispensedinto each of the bottles containing the weighed

sample and the o-phenanthroline.This is followed by an
input of 0.5 ml of 48% HF deliveredfrom a syringeinert to
HF. The cap is replacedand tightened until resistanceis
first felt. Back off one quarter turn and tape the cap to the
bottle's side with electrical tape. This small opening prevents gas build-up during digestion.The securedcap also
insures no contamination of the digesting sample by
splashingfrom the water bath.
The 7 bottles are arrangedin a tightly-packedhexagon
to obtain a circular pattern.The bottles are bound together
with two rubber bands and this assemblyis placedinto a 4
liter glass beaker which is about half filled with boiling
water. Samplesare digested for exactly 30 minutes and
added 20 ml of
removed.Into each bottle are successively
10% Na-Citrate and 5 ml of 5% boric acid solution. The
order in which these two solutions are delivered to the
samplesshould be the same as that in which the HrSO*
and HF were dispensedprior to thermal digestion. The
orange-color denoting the presence of the ferrousphenanthrolinecomplex usually appearsafter the addition
of the Na-Citrate. However,for highly ferrous samples,the
orangecolor may have alreadybeen developedduring the
digestion.
After the addition of the boric acid, the solutions are
immediately diluted with deionizedwater. Shapiro (1960)
recommendeddiluting each solution with deionizedwater
in a 100 ml volumetric flask. This step is done faster and
more accuratelyby directly adding deionizedwater to the
solution in the digestionbottle, which is placedon the pan
of a top-loading balance.Deionized water is added until
the total weight of the solution * bottle + cap is about lfi)
gm greater than the original weight of the dry bottle and
cap. The total solution weight should be recorded to the
nearest 0.01 gm. The total solution weight need not be
exactly 1@.00gm. The only critical considerationin diluting the orange-coloredsolution is that the pH of the final
solution to be measuredcolorimetrically falls within the
rangeof 2.8to 3.5(Roth et al., 1968).
In Figure 1, the number adjacent to each datum point
representsthe total gramsof solution to which the digested
sample was diluted with deionizedwater. The wide range
of total solution weights in Figure L (97.6 to 129.7 gm)
indicatesa moderatetoleration of dilution with respectto
maintaining the solution'spH within the rangewherecolor
intensity is simply a function of the concentrationof the
complex.This toleration of dilution
ferrous-phenanthroline
of highly
can be beneficialin the absorbancemeasurements
ferrous unknowns that must be diluted in order to be
bracketed by the most FeO-rich standard. Occasionally,
the intensity of the orange-coloredsolution of an unknown
after the additions of the citrate and boric acid solutionsis
visually observedto be darker than that ofthe highestrock
standard.This ariseswhen too much powder of the highly
ferrous unknown was weighted into the 4 oz. polypropylene bottle. To insure that the absorbanceof this
unknown is lessthan that of the highestrock standard,the

964

FRITZ AND POPP: DETERMINING FeO AND Fe"O,

highly ferrous unknown can be diluted with as much

deionizedwater allowableby the volume of the 4 oz. bottle.
The rock standard having the highestferrous content can
then be diluted with substantiallyless water such that its
dissolvedferrous concentrationbecomesgreater than that
of the highly-dilutedunknown.
Spectr ophotometr ic measurements
In the colorimetric determination of FeO. the color intensity of the ferrous phenanthroline rcmplex is timedependent(Shapiro, 1960; Roth et al., 1968).As such, the
absorbanceof the colored solutions must be performedin
chronologicalorder at preciseintervals measuredfrom the
time of digestion. A two-person operation facilitates this
requirement as well as expediting the runs. One person's
responsibilitiesare to handlethe addition of the reagentsto
the samplesas well as to managethe digestionprocedure.
The other analyst'schore is to dilute the colored solutions
with deionized water and to record the colorimetric
measurements.
Absorbancereadings are also performed in batches of
seven,commencingwhen the seventhsamplein that batch
has been diluted with deionizedwater. Absorbanceof the
diluted oiange-coloredsolution was performed on a dual
beam spectrophotometer,(a Bauschand Lomb Spectronic
2000),using matchedglasscuvetswith a path length of 10
mm. The referencesolution was a blank, againstwhich the
absorbanceof all standardsand unknowns was compared.
The blank consistedof rock crystal quartz digestedin the
same fashion as all the other samples.Absorbanceof all
sampleswas measuredat 555 nm, with background readings taken at 640 nm. The absorbancereading at 640 nm
correctsfor turbidity of the solution arising from possible
incompletedigestionof the sample(Shapiro,1960).
Figure 2 is a typical calibration curve of standardsfor a
set of FeO analyses.The linearity of the concentrationabsorbancerelationshiptestifiesto the obedianceof Beer's
Law up to highly ferrous rock standardssuch as Fe-Mica.
The high value for the linear correlation coefficientfor this
calibration curve was typical of all FeO runs. The different
positioning of the samerock standardsalong the line is due
to differing initial weightsof their dry powders,as well as
to their varying total solution weights.
Becauseboth the dry sample weight and final solution
weight vary from sample to sample, all blank- and
background-correctedabsorbancedata must be normalized against some arbitrary referencevalues in order to
obtain absolute FeO contents of the unknowns. In the
analysesreportedhere,the referencevalueswereselectedto
be 10.0 mg dry sample weight, and 100.0gm of solution.
The "normalized o/oFeO" valuesplotted on the ordinate of
Figure 2 have been corrected in this manner. Thus, the
"normalized % FeO" of a rock standard is equal to its
reported % FeO multiplied by Dilution Fraction One (Fr),
where
(100grams)
weight of solution in grn

(l)

o/oFeO = 31 327 Abs + 0 007

r:09998

o'.
II

E'"
N
(t

E^
o

0100

0200

0300

0400

0500

0600

C o r r e c t e dA b s o r b a n c e
Fig. 2. TypicalFeO calibrationcurveshowingnomrc;lized
%"
FeO of rock standards vs. their corrected absorbance readings.
These rock standards were used in this study for FeO and lFe
analyses for the amphiboles. Normalized % FeO is equal to the
reported % FeO of a rock standard multiplied by Dilution Factor
One (see text).

The corrected(blank and background)absorbanceof an

unknown measuredagainstthe "normalized % FeO" standard curve translatesto a "raw" % FeO. This FeO value
must be corrected for the unknown's sample weight and
the total weight of the diluted solution by dividing the
unknown sample'sraw o/oFeO by its Fl. An example of
this calculationis provided below.
For a singleanalysisof grunerite CK-L, 3.21mg of amphibole were dissolved and diluted to a final solution
weight of 158.88gm. This gives an F, for this sample of
0.2O2O.
A background-and blank-correctedabsorbanceof
0.290 was me.rsuredagainst the absorbanceof the rock
standards shown in Figure 2. Thus, the raw % FeO
(9.09%)of this unknown is obtained by substituting O.29O
absorbanceinto the equation given in Figure 2. The absolute oh FeO (45.01%)is obtained by dividing the ravt %o
FeO by Fr.
Total iron expressedas oh Fe2O3was determinedon the
orange-coloredsolution of the FeO analysesby atomic
absorption spectrophotometry,using a dual beam unit
(Perkin-Elmer Model 603). Absorbancewas measuredat
248.3nm and the instrument was zeroed with the quartz
blank. Absorbance is measured against normalized %o
FerO. of the rock standardwherethe normalizedvalue for

965

FRITZ AND POPP: DETERMINING FeO AND Fe,O.

each rock standard is obtained by multiplying that standard's F, by its reportedvaluefor EFe as % FerO..
Figure 3 showsa typical calibration curve for total iron
analysis.The reportedlinear working rangefor iron on the
P-E 603 unit is 5 ppm. This translates to a total iron
content ol 7.14o/oFerO, of the dry sampleif the sample's
F, were unity. For this reason,the calibration curve begins
to depart from linearity at higher concentrations.Despite
the nonlinearity of the absorbance-EFerelationship,there
is a high degreeof confidencein fitting the data points to a
paraboliccurve(Fig. 3).
The total iron of the unknownscan be measuredwith or
without further dilution. If an unknown solution is directly
aspirated into the atomic absorption unit, its computed
raw EFe value is then divided by its F, to obtain the
absolute value of XFe expressedas Vo FerO3. The disadvantageof this methodlies in noisy signalsfor high absorbance valuesthat translateto large relative errors due to a
decreasedsensitivityon the non-linear segmentof the analytical curve. The analyzed amphiboles discussedbelow
were all iron-rich such that noisy signals were especially
troublesomeduring absorbancereadingsof thesesamples.
Since the stability of absorbancereadingsis greater for
samples containing lower iron concentrations,the unknowns werediluted with the quartz blank to yield absorbancesfor thesesamplesthat lay within the linear working
range of the instrument.Quartz blank is usedas a dilutant
rather than deionized water to minimize matrix eflects.
This dilution is performedgravimetricallyon a topJoading
balance.A small amount of the orange-coloredsolution is
placedin a tared, 4 oz. polypropylenebottle and the solution weight is recorded. Quartz blank is added to an
amount necessaryto achievethe desireddilution, normally
about two to threetimes that of the unknown solution.The
total solution weight is recorded.This dilution givesrise to
Dilution FactorTwo (Fr) which is:

> Fe as Fe?Or = -0 010 + 70 316 Abs + 27 m6

r=09998

s
@
C'
c

o
-1

:1
(g

o
E
o
N
C,

E
L

o1oo

o2oo

oaoo

C o r r e c t e dA b s o r b a n c e
Fig. 3. Typical EFe as %oFerO. calibrationcurve showing

normalized oh FerO'' of rock standards vs. their blank-corrected

absorbance readings. The dashed line shows the extension of the
linear segment of the calibration curve whose slope is 73.48. For
analyses of high iron unknowns, samples were diluted with a
quartz blank such that their absorbances usually ranged from 0.50
to 0.100 absorbanceunits.

bring the total solutionweightto 15.35grams.Thus,F, for

(2) the sampleis 0.5036.The absorbance
of this solution,0.069,
was measuredagainst the zero absorbancevalue of the
Before this method is adopted,the analyst must be con- quartz blank and those rock standardsdefining a straight
vinced that the absorbanceof the lower-iron rock stan- line passing through the origin for which the slope is
unit (Fig. 3).The raw (i.e.,
dards definesa straightline passingthrough the origin. The 73.48%FerO./l.0@ absorbance
oh
dashed line of Figure 3 shows that such a trend exists. uncorrected) FerO'' is equalto:
Having establishedwhich standardsplot on a line passing
73.48V"FerO, x 0.069abs.unit
:5.07'/' Fe2O3.
through the origin, the analystis free to chooseany one of
1.00abs.unit
theserock standardsas the basisby which to comparethe
absorbanceof the unknowns to that of the standard for If this value is divided by Ft(0.2020)and Fr(0.5036),the
purpose of calculating the raw EFe as % FerO, of the result,49.86%,is the absolute2Fe of the sample,expressed
unknowns. Alternatively, all rock standardsfalling on the as o/oFerO..
linear segmentof the calibration curve could be used to
The main sourceof error in the atomic absorptiondeterdefine the slope. In either case,the absolute EFe as % mination of EFe residesin fluctuation of absorbancereadFerO. of an unknown is equal to its raw XFe value divid- ings about the observedvalue.Although thesefluctuations
ed by both F, and F2.
are generallyon the order of only 0.001-0.002absorbance
Grunerite CK-l is again usedas an exampleof the cal- units, this uncertainty is divided by Ft and F2 to yield
culation procedure.Into a clean4 oz. polypropylenebottle potentially large errors about the observedvalue for XFe.
was poured 7.73 gramsof the orange-coloredsolution left Severalsteps can be taken to minimize this uncertainty.
over from the FeO analysis.Quartz blank was added to Sincethe percentrelative error is inverselyproportional to
Fz:

(gm unknown)

(gramsunknown + gm quartz blank)

966

FRITZ AND POPP: DETERMINING FeO AND Fe,O"

Table 1. Observed versus reported values of FeO and EFe as

FerO. of 10 common rock standards.Reported values obtained
from Abbey (1980).
Rock
Standard

Analyzed
total Fe
as tFezo3

Reported
total Fe
as tFe203

Anal
Feo
w t . g"vzed

Reported
wt.t Feo

DT-N

0.12 (0.06)

0.10

0.20 (0.30)

GH

0.6e (0.02)

0.84

1.29 (0.08)

1.34

1.48(0.05)

t.62

2.22 (0.10)

2.77

0.66

GA

1.29(0.06)

1.32

2.73 (0.18)

2.83

G5-N

1.s4 (0.13)

I .66

3.43 (0.39)

3.77

AGV-1

?.21 (0.02)

2.03

6.84 (0.28)

6.82

NR-N

5 . 1 2( 0 . 1 3 )

5.3?

9 . 1 3( 0 . 3 8 )

9.69

NII.I-N

7.48 (0.33)

7 ,3 0

8.91 10.22)

8.91

u-I

8.76 (0.05)

8,73

1 1 . 5 3( 0 . 7 0 )

l 1. 1 0

1 4 . 5 9( 0 . 3 0 )

14.63

1 6 . 7 7( 0 . 4 8 )

16 -97

ilIl'l-D

the magnitude of the true absorbancereading,it is advisable to base the calibration on the highest iron-bearing
standardsthat still lie on the linear segmentof the calibration curve. Furthermore, the unknown's dilution with the
quartz blank should be enough to render an absorbance
reading below that of the standard; yet, this dilution
should not be so large as to render a very low absorbance
of this mixture. Third, all EFe analysesshould be performed in triplicate for eachdiluted mixture.
Results
In order to test the accuracyand precisionof the technique, well documentedrock standardswere employed.A
seriesof ten rock standardswere analyzedas unknowns.

using sevenother rock standardsto obtain the calibration

curvesfor both FeO and total iron as FerO.. The resultsof
three replicateanalysesare given in Table 1; the numbers
in parenthesesrepresentl-standard-error values.In general, the agreement between the observed and reported
values is reasonable.Even though the relative error for
total iron is large for severalof the Fe-poor samples(e.g.,
DT-N), the absolutedifferencesbetweenthe observedand
reported values is never more than 0.2 wt.% when the
l-standard-errorvaluesare considered.The relativelylarge
absoluteerrors for total iron analysesmay result from the
fluctuationsof the absorbancereadingsas discussedabove,
or from possibleinhomogeneityof the small samplesizesof
the rock powders.
The technique was tested further for ten natural amphibole mineral separatesobtained from a number of different sources.The resultsare shown in Table 2. The onestandard-errorvalues given in parentheseswere obtained
from three replicate analysesof all samplesexcept CK-l,
CA and SC, for which the number of replicate analyses
were six, four and two, respectively.As might be expected,
the agreementbetweenthe observedand reported valuesis
poorer than for the well-calibratedrock standards.Compared to the reported wet chemical analyses,four of the
amphiboles(TP-l, SLO-5, C-4980,and Y42-BX) show differencesthat are consideredto be excessive.However.for
three of those samples(TP-l, SLO-5, and C-4980)the observed molar Fe3*/(Fe3*+ Fe2+) values show better
agreementwith the valuesreported from Miissbauerspectral analysesthan do the valuesobtainedfrom the reported
wet chemicalanalyses.Miissbauer analysesare not available for sample Y42-BX. Overall, the observed ferricferrous ratios are in reasonableagreementwith those ob-

Table 2. Observedversusreportedvaluesof FeO, EFe as FerO., and molar ferric-ferrousratio of 10 natural amphiboles.

A n p hbi o l e

Average Number
Sanple
of
Height Samples
in mg
Analyzed

0bseryed
% FeO
(t 1o)

Reported
% FeO

F e 3 + 7 ( F e*3 F " z * ) r-ltolar

ll
observed
Reported !,et
Reported
(t 1o)
Chemical
ltossbauer

';T:'::
(t to)
"liol'

F"?03

TP-1 (glaucophane)a

5.02

5.63(0.06)

6.12

8.10(0.05)

0.298

o . z 3 oo, . z u "

SC(magnesioriebeckite)a

6.2?

7.1I(o.08)

7.O7

11.62(0.26].

1 7. 7 4

0.5s5(0.007)

0.557

0.542,0.526e

S L 0 - 5( g l a u c o p h a n e ) a

6.35

16.22(0.06)

r( 02

n 1 ? q f nn l q \

0. 104

0. 165

1 7 5 - A( g l a u c o p h a n e ) a

6.94

e . 2 1( 0 . 2 0 )

9.2r

14.57(0.73)

14.87

0.302(0.022)

0.312

0,293

409 (crossite)a

7.45

1 0 . 7 90( . 0 4 )

r0.31

18.82(0.18)

19.40

0.353(0.008)

0.409

0.397

423 (glaucophane)a

7.48

1 2. 2 s ( 0 . 2 6 )

12.O2

17.93(0.37)

17,71

0.241(o.oo7)

0.246

o.273

6 . 6 3 (0 . 4 6 )

5.23

23.49(0.06)

22.58

0.687(0.016)

0.743

0.6M

4 4 . 3 6 r( . 1 3 )

44,99

49.08(1.05)

50.00

C - 4 9 8 0( m a g n e s i o r i e b e c k l t e ) a 4 . 4 3

C K - 1( g r u n e r i t e ) D

CA-( ri ebecki te) c

Y 4 2 - B X( c u m i n g t o n i t e ) o

3.48

12.54(0.24)

r?.tl

o 60

,?eln

n1t\

-0.004(o,oo8)

0,000

3.94

20.80(0.31
)

20.67

34.73(0.11)

34.60

0.334(0.011)

0.336

4 .1 9

21.86(0.48)

23.t?

2 4 . 7 7( O . 4 8 )

26.8t

0.019(0.006)

0.042

a, SamPleobtained from ll.G. Ernst. l.let chemlca1analyses perfomed by several different analysts as reported by Ernst and ilai (1970). ilossbauer
analysis also from Ernst and Uai (1970).
b . S a m p l eo b t a i n e d f r m C . K l e i n . A n a l y s i s r e p o r t e d i n K l e i n ( 1 9 6 4 ) .
c . S a m p l eo b t a i n e d f r o m l r , C . G i l b e r t . F e Oa n a l y z e db y U h i p p l e u s J n g t e c h n i q u eo f t l h i p p l e ( 1 9 7 4 ) . T o t a l F e f r o m e l e c t r o n n i c r o p r o b e a n a ] y s l s b y
l , l , C .G i ' l b e r t .
d . S a n p l eo b t a i n e d f r o m P . R o b i n s o n , A n a l y s i s r e p o r t e d i n R o b i n s o na n d J a f f e e ( 1 9 6 9 ) .
e . S e c o n dm o s s b a u evr a l u e f r o m B a n c r a f t a n d B u r n s ( 1 9 6 9 ) , a s r e p o r t e d b J E r n s t a n d l l a i ( 1 9 7 0 ) .

FRITZ AND POPP: DETERMINING FeO AND Fe"O,

tained from Miissbaueranalysis:the largest differencebetweenthesetwo setsof measurements

is 3.4 absolute% for
sample409. It should be noted, however,that for sample
409 the ferric-ferrousratio from the reported wet chemical
analysisagreeswell with the Mdssbauervalue.For sample
423, the observedand reported ratio are in good agreement, but differ from the Mcissbauervalue by approximately 3 absoute %. Consideringthe possiblesourcesof
inhomogeneity and impurities in micro-sized samples of
natural mineralseparates,
the resultsare encouraging.
The most iron-rich sample,gruneriteCK-l, requiresadditional discussion.Becauseof its high iron content, a
small averagesamplesize,3.48mg (Table 2), was necessary
in order to bracket the solution by the most iron-rich rock
standard.Shown in Table 3 are the six individual analyses
obtained for the sample.For three of the analyses(replicates # 1, 5, and 6) both FeO and total iron are significantly lower than the reportedvalues.Thesedifferencesare
interpreted to result from weighing errors, possibly from
sorbtion of water onto the sample or digestion bottle
during weighing. Despite the differencesin absolute FeO
and total iron, the ferric-ferrous ratio of all six analyses
agreeswell with the reportedvalue.
BecauseFeO and total iron are both measuredin a
solution producedby a singledigestion,weighingerrors of
the dry sample do not affect the ferric-ferrous ratio. If a
weighingerror of, say,l0%oexists,the absoluteamounts of
FeO and total iron will be in error by lO%, but this error
will cancel when the ratio is calculated.Thus, determination of a ferric-ferrousratio can, in point of fact, be
carried out independent of any knowledge of the dry
sampleweight. If only the ratio and not absoluteamounts
are desired,the analysescould be carried out on unweighed
samples,provided, of course,that the dilutions are carried
out accurately,as discussedabove.Even though weighing
errors for samplessmaller than -2 mg may preclude the
determination of accurate absolute FeO and total iron
values,the lower limit for samplesizeon which the ferricferrous ratio can be determinedshould be limited only by
the detectionlimits of the analyticalinstruments.
Conclusions
l. The method describedhere can be used to determine
absoluteamountsof FeO and total iron as FerO, on small
samplesizes( < 10mg).
2.'lhe technique gives results that, in general, are in
reasonableagreementwith other wet chemicaltechniques
and with M<issbauerspectraltechniques.
3. The more cumbersomeproceduresof volumetric dilution are replaced by faster and inherently more accurate
gravimetric dilution techniques.Those steps that do require volumetric measurementsrequire only the precision
of automatic pipetsor of syringes.
4. The method is quite versatilein terms of dry sample
weights and dilutions. High accuracy in dilution is required, but preciseduplication of weightsbetweensamples
is not required.

967

Table 3. Replicate analysesof grunerite (CK-l). Each replicate

representsa separatedissolution and the observedFeO and EFe
values are an averageof triplicate analyses.Reported valuesare
from Klein (1964).

A n a l y s i s#
X Feo

44.0? 45.30

45.66

t Fe as
X FeZ03

48.47 50.06

49.92 49.81

44.92

42.71

43.54

44.99

47.3A 48.82

50.00

-0.009 -0.006 -0.017 -0.002 -0.002 +0.009

0.000

5. The procedureis relatively rapid. After the dry samples have been weighedinto the digestion bottles, a twoperson analysisteam can typically analyze40 samplesfor
both FeO and total iron, using a 7 standard calibration
curve,in an 8 hour day.
6. The ferric-ferrous ratio should be able to be determined on vanishingly small sizes,provided that the resulting solutions lie within the detectionlimits of the analytical instruments.
Acknowledgments
TheauthorsthankW. G. Ernst,M. C. Gilbert,C. Klein,andP.
Robinsonfor providingmineralseparates
of previouslyanalyzed
amphiboles,
and J. Ricefor a helpfulreviewof the manuscript.
Researchwas undertaken with support of the National Science
Foundation (EAR-8312878),
which is gratefullyacknowledged.

References
Abbey, S. (1980) Studies in "Standard Samples"for use in the
general analysis of silicate rocks and minerals. Part 6: 1979
edition of "usable" values.GeologicalSurveyof Canada,Paper
80-14.
Bancroft, G. M. and Burns, R. G. (1969)Mrissbauerand absorption spectralstudy of alkali amphiboles.Mineralogical Society
of AmericaSpecialPaper2, 137-148.
Bernas,S. (1968)A new method for decompositionand comprehensive analysis of silicate rocks by atomic absorption spectrometry. Analytical Chemistry,40, 1682-1686.
Ernst, W. G., and Wai, C. M. (1970)Mossbauer,infrared, x-ray
and optical study ofcation ordering and dehydrationin natural
and heat-treatedsodic amphiboles.American Mineralogist, 55,
1226-1258.
Fritz, S. J.(1976)Physicaland chemicalcharacreristics
ofweathering rinds from severalplutons in the North Carolina Piedmont.
Ph.D. dissertation,University of North Carolin4 ChapelHill.
Klein, C. (1964)Cummingtonite-gruneriteseries:a chemical,optical, and X-ray study.AmericanMineralogist,49,96T982.
Medlin, J. H., Suhr,N. H. and Bodkin, J. B. (1969)Atomic absorbtion analysisof silicatesemploying LiBO, fusion. Atomic Absorption Newsletter,8, 25-29.
Reichen,L. E. and Fahey,J. J.(1962)An improved method for the

968

FRITZ AND POPP: DETERMINING FeO AND FerO'

determinationof FeO in rocks and minerals including garnets.

nation of FeO in rocks. U.S. Geological Survey Professional
U.S. GeologicalSurveyBulletin, 1l,l4B.
Paper 4008, 496-4.97.
Robinson,P. and Jaffee,H. W. (1969)Chemographicexploration Shapiro,L. and Brannock,W. W. (1962)Rapid analysisof silicate,
of amphibole assemblagesfrom central Massachussettsand
carbonate,and phosphaterocks. U.S. GeologicalSurvey Bullesouthwestern New Hampshire. Mineralogical Society of
tin 11214A.
America SpecialPaper2,251-274.
Whipple, W. E. (1974)A study of Wilson's determination of ferRoth, C. B., Jackson,M. L., Lotse,E. G., and Syers,J. K. (1968)
Geology,14,223-238.
rousironinsilicates.Chemical
Ferrous-ferric ratio and CEC changeson deferration of wea- Wilson, A. D. (1955)Determination of ferrous iron in rocks and
thered micaceousvermiculite. Israel Journal of Chemistry, 6,
minerals.Bulletin of the GeologicalSurvey of Great Britain, 9,
261-273.
5G58.
Shaeffer,H. N. (1966) The determination of iron (II) oxide in
silicateand refractarymaterials.Analyst,91, 763-790.
Manuscriptreceiued,
J anuary4, 1985;
Shapiro, L. (1960)A spectrophotometricmethod for the determiacceptedforpublication,May22, 1985.

Appendix
Equipment
r double beam atomic absorption spectrophotometer
o double beam colorimetric spectrophotometer and
cuvets having a l0 mm path length.
r analytical balance with readability of 0.01 mg
r top loading balance with readability of 0.01 grams
o 4 oz. polypropylene bottles
o j ml plastic syringe
o 3 automatic pipet dispensers
o a 4Jiter glass beaker
. ring stand and bunsen burner
Reagents
.
t
o
o
r
o

at least 6 well-characterized rock standards

lO% (vlv) HrSOn solution
5% (wlw) H.BO. solution
l0% (w/w) Na Citrate solution
crystalline o-phenanthroline
48% HF

matched