American Mineralogist, Volume 58, pages 291-301, 1973
The FluorineContentand Some PhysicalProperties
of the Amblygonite-Montebrasite
Minerals
Ivn ennNn, psrR irnNi,
eNo Ronnnr B. FencusoN
Department ol Earth Sciences, Uniuersity ol Manitoba,
IAinnipeg, Manitoba, Canada, R3T 2N2
Abstract
Specificgravities,refractiveindices,differential thermal behavior, unit cell dimensions,and sensitive
X-ray powder diffraction reflectionswere studied on a seriesof analyzed amblygonites-montebrasites,
to find a reliable method for indirect determination ofthe F-content, and to check the existing determinative graphs reported as doubtful by some authors. Specificgravities were found to comply with
the relations establishedearlier, but they proved to be very sensitiveto alteration and inclusions,
The variation of the refractiveindex 7' differs from that given by Winchell and Winchell (1951), and a
thorough study of the optics is desirable.Differential thermal analysesshow gradual changesin the
number and temperature of the reaction peaks, and in the composition and proportions of the
metastableheating products; however, the thermal behavior seemsto be strongly dependenton the
experimental setup. Unit cell dimensions,refined with high accuracy and precision, show small but
regular changeswith the F-content, and they appear to be not affectedby NazO contents up to about
1.5 wt percent.Two determinativegraphs were constructed, one based on separationsofthe uncalibrated X-ray powder reflections iOt, itO, i2t, Ott, O2l, and l2O, and the other on the 131 peak
(Moss el al., 1969)calibrated against CuKal quartz reflections.These two methods seemto be the
most practical for determining indirectly the F-content, provided the examined samplescontain less
than about 1.5 wt percent NazO, and are optically homogeneous.The chemistry of the amblygonitesmontebrasitesused for structure determinations (Simonov and Belov, 1958; Baur, 1959) and the
possibleexistenceof high- and low-temperature forms are discussed.
Introduction
In the present study, the specific gravities, refractive
indicesy', differentialthermal records,X-ray
In 1969-1970,the first author investigated the
powder
diffraction data, and unit cell dimensions
distribution and composition of the amblygonitehave
been
determinedfor 15 chemicallyanalyzed
montebrasiteminerals,(Li, Na)AlPOn(F,OH), in the
We describeour results
amblygonites-montebrasites.
Tanco pegmatitedepositat Bernic Lake, S. E. Manimethods,
we
estimatethe practical
obtained
by
these
toba (iern5,1970; iern| et al., 1972).The studywas
methods
for
indirect
determination
value
different
of
prompted by the wide range of amblygonite-monteof pegF
and
the
attention
of
contents,
we
draw
brasite compositions in this Li-Ta-Cs deposit and
inthe
widespread
chemical
investigators
to
matite
their potential petrological significance,but it also
primary
both
and
secondary
origin
homogeneities
of
helped to establish methods for separatingamblygonites-montebrasitesfrom refractory-grade spod- in amblygonites-montebrasites.
umene in which they are a highly undesirableconSelectedSpecimensand Their
taminant.
ChemicalCompositions
During the previous study, the large number of
specimensto be examinedunderscoredthe need to
About half of the specimensexaminedcamefrorn
estimatetheir (F,OH) content indirectly instead of the Tanco pegmatite, a convenientsource of large
by lengthy direct chemicaldeterminations.However, quantitiesof fresh material which givessomeindicathe then available physical methods were found to tion of the compositional changesby variations in
be either experimentallydifficult and time-consum- color. All these specimensare deposited in the
ing (specific gravity), or unreliable becauseof the mineral collectionsof the Department of Earth Sciwide scatter of the known data (optical properties, ences,IJniversity of Manitoba. Other sampleshave
Winchell and Winchell, 1951; see objections by been obtained from various collections.The list of
Ginzburg,1950;StanEk,1960).
is givenin Table 1.
specimens
291
292
vrvt
CERNA, CERNY, AND
FERGUSON
TABLE 1. List of Specimens
for chemical analysis,this report is concernedonly
with Na-poor ( < 1 wt percentNa2O) amblygonitesSpecimensfrom the Tanco pegmatite,Bernic Lake, Manitoba:
montebrasites;
the Hebron sample(AF-55 with 2.55
A-1:
Pale pink core of a zoned crystal (see A-2), slightly
wt percentNa2O) is shownto give someindication
contaminated by secondary montebrasite (-5%).
of the influenceof higherNa contents.
A-2l
Yellow outer rim of a zoned crystal (see A-l).
In the samplesanalyzedby K. Ramlal and R. M.
A-3:
Waterclear colorless crystal; analyzedby P. PovonHill, Al2Os and P2O5 werc analyzed by X-ray
dra, F-content by J. L. Dalton.
A-4:
White material surrounding and veining A-5;
fluorescencespectrometry,alkali metals and alkaline
chemical analysis by R. M. Iljll et al. (A-4B) and
earthsby atomic absorptionspectrography,and H2O
P. Povondra et al. (A-4A).
gravimetrically.
Fluorine contentswere determined
A-5:
Yellow translucent, veined by A-4.
(Mines Branch, Ottawa) using
Dalton
by
J.
L.
A-13: White to pinkish cleavablemass.
A-22: Brownishtranslucent,massive.
neutron activation. In the specimen analyzed by
A-29'. Milky white cleavablemass.
P. Povondra,PzOuand H2O were determinedgraviA-60: Milky white cleavagefragment.
metrically, alkali metals and alkaline earthsby flame
A-98: Pale yellow translucentmass.
photometry,Al2O3 by potentiometrictitration, and
F by pyrolytic decompositionand complexometric
Other Specimens:
titration.
AF-l:
Montebras, France, Dept. Earth Sciences,Univ. of
Manitoba No. M-,1421; milky white opaque mass
In the followingsections,the changesin individual
stuffed with dusty anisotropic inclusions.
physical characteristicsare related directly to wt
AF-43: Nesbitt spodumenemine, Gunnison County, ColopercentF as found by the chemicalanalysis,since
rado ; U.S.N.M. No., I 17,775; grayish-whitecleavage
the recalculation to atomic content is slightly infragment.
AF-44: Moguk, Upper Burma, U.S.N.M. No. 1O,7341, fluenced by errors in the determination of other
colorless to yellowish, waterclear fragments; Fcomponentsand by assumptionsinvolved in calculacontent not determined.
tion of crystallochemicalformulae. The theoretically
AF-46: Karibib, South Africa, U.S.N.M. No. 105914;pale
most useful atomic ratio F/(F + OH) is considbluish cleavagefragments,
AF-47: Plumbago, Newry, Maine, U.S.N.M. No. 5841; erably affected by errors in the determination of
IlzO (cl. Ribbe and Rosenberg,1971) and the plots
bluish waterclear crystals from pegmatite vugs, with
quartz and cleavelandite;F-content not determined, of physical propertiesagainstthis ratio show poorer
AF-48: Newry, Maine, U.S.N.M. No. 5906; waterclear correlation than
when F content alone is used.
colorlesscrystals from a pegmatitevug; no chemical
The
linear
regression
equationsand the correladata available.
tion
for
coefficients
most
of
the plots of physicalconAF-50: Coolgardie,West Australia, U.S.N.M. No. 10432;
pale brownish transparent mass.
stants u.r wt percent F are given in Table 3. They
AF-55: Hebron, Maine, U.S.N.M. No. 62,576; colorless were calculatedon the basisof only those specimens
cleavable mass; the only sodium-rich specimen for
which the F content was directly determined
examined.
chemically,
exceptfor the Na-rich specimenAF-55.
AF-65: Chursdorf, Saxony, B.M.N.H. 14, 274; translucent
fragments with faint bluish tint; partial chemical
analysisby Moss et al. (1969).
AF-66: Newry, Maine, Dept. of Geological Sciences,
Harvard University; waterclear colorless crystals
from a vug; no chemical data available.
As shownin Table 2, full chemicalanalyseswere
obtainedfor 15 specimens;two other sampleswere
analyzed.for all major componentsexcept F (AF44, AF-47), and the partially analyzedspecimenZ
of Moss et al. (1969) was included in the examined series(AF-65). Two other specimens(AF-48
and AF-66) were checked only for physical properties becauselack of material precluded chemical
analyses.Sincemost of the Na-rich varietiesexamined in this study did not yield material sufficient
SpecificGravity
The specificgravities were measuredwith a Berman torsion balance,using toluene checkedfor temperature as a suspensionmedium. Three to five
grainswere usedfor eachsample,the specificgravity
being determinedtwo to four times per grain. The
purest grains availablewere selectedfor each specimen for this determination,and their refractive indices were checked.However, in some casesonly
milky, clouded grains were available. Microscopic
examination showed that liquid inclusionswere the
main impurity in all cases, with clayJike andfor
mica-like particles observedonly occasionally.
The averagedspecificgravities plotted againstwt
percent F in Figure 1 fall close to the widely used
293
AM BLY G ON IT E-MON TEB RA SI T E M I N E RA LS
Tlsre 2. Chemical Composition and PhysicalProperties
Al2o3
A- 22
AF-44
F-43
A-9E
3 5 .r 0
35.90
3 4 .6 0
34,46
.085
M8o
. 003
9.47
Lt
20
Na2O
,08
. 004
K2o
49.38
Pzo5
.10
*2-
9.52
. 005
49.TT
.07
5.98
H2O+
.30
F
t 0 0 . 2 74
-O-2F
1.40
100.682
.005
9.98
100.092
.92r
. 004
.II
.17
. 002
. 002
.014
, 002
. 004
gr.
49.21
49.48
48.98
4 9. 2 6
.04
.08
.07
.08
.10
5.45
5 .1 l
4.43
4. 09
4.34
3.30
3.39
l- 4s
100.81
1 0 0 .7 2
3.44
101.591
- r.45
-1.536
1 0 0 .3 5 9
1 0 0 .r 4 l
I00.151
3.024
+ .010
2.992
+.005
3.033
1.638
5. 1896
+08+08+09
7.1685
+10+92+I0
3.015
+.008
3.007
+.010
3.022
r.636
s.r924
1.635
5,1812
.924
.969
.001
(.970)
,9A9
(. 989)
r. 015
4.033
,265
.702
3,022
+ .003
3.037
r,626
5.1801
+07+08
7.1155
+
11
5.0424
5.0422
5,0448
+06+07
5. 0438
tt2ot L.g'
-+9 E o o1. 4. '0 1
s.o3s.
+
tu
5.tJ407
+07+07+08
4
t12019.4'
-+g t " s t0. t' '9 '
ttz"zl.t,
-+9 7 o 5 0
5 .. g7' '
rr2o28.i'
-+g 7 o 5 0t r,. 7z ''
tr2oz6.7l
-+g 7 o 5 07 ,. 37' '
-0t",,4. '
o
+
0.9'
160,76
+ .03
sz.tt,
52.t46
+
0.7'
6104a.5t
+
0.1'
160.46
+ .03
-+0 7 0 5 0
0 .. 8 '
t'
0.8'
+
150.48
+ .03
-+6 7 0 4 1
7 .. 30' t
tt2ot 3.l'
-+9 8 o o 0. . 7 '
t'
0 .7 '
+
-6ioL3.8'
+
0.71
1 6 0 .5 9
+ ,03
+
0.6'
160.04
+ .03
A
v(I-)
o2or,, r".". +
o 2 e 1 3 rc a 1 c ,+
52.24
52.22
52.22
52.35
52.229
52,234
52.217
52.361
4.05
4.51
5,56
r0r.882
-1.71
ro2,44a
-1.898
103.099
-2.34r
r00.159
I00.172
.995
. (t02
( . 9 9 7'
.992
(.992)
1.023
4,O92
.242
.626
7.1685
+
13
0.91
+
-67"45.r'
+
0.7'
159.9r
+ .03
52.386
radia
line which connectsthe specificgravities of montebrasite and amblygoniteend members (2.98 and
3.11respectively)
asgivenby Palacheet al. (1951).
However, different grains from a single specimen
show considerablespread,and the measurementof
three to five grains seemsto be the only way of
obtaining reasonable averages. (The differences
among individual grains of a singlespecimenappear
to be real, since repeatedmeasurementsof a single
grain fall within :t.0015 and variationsamongdifferentgrainsreachup to .03). Most of the specimens
that plot below the 2.98 - 3.11 line contain abundant inclusions,particularlythe samplesAF-l and
AF-46.
These results appear to confirm the views of
Nicolas( 1968) who concluded,aftera detailedstudy
3.65
3.008
+ .008
3,042
r,626
5. 1798
7.1689
49.32
L o r .i 2 5
-1.536
.959
. 00r
.004
(.e64)
.002
.985
( . 9 8 7)
1.008
4,062
.279
,659
7,1699
.^(8)
t
10 1 . 6 8 7
-0.70
,989
(0.989)
I.019
4.022
.r44
.834
.964
J. d)
3.65
I. 8E
101.059
(0.e24)
7. 168r
+
l0
.054
.05
(.964)
b^(8)
.044
.003
4 9. 2 6
1-016
(r.016)
I.003
4.000
,130
.870
1.6367
5. 1957
. 039
. 005
9,86
.06
I.008
(1.008)
1.0r3
4.039
. 108
.853
r'^
a^ (I)
-+09
.055
.025
L0,29
49.66
1.005
(I. Oo5)
l. OII
4. 00
.04
.90)
celc.
9.47
.09
49.38
^.
(Ferr+AI)
Sp. gr. meas.
Sp.
r0.14
9.86
9.90
. 015
(.925)
(oH)
,02
A-t
34.53
. 0 74
. t24
(Ll+Na.rce)
P
. 11
A-5
34.86
. o2a
.020
F€-
o
3 4 . 75
.166
.001
0.933
. o07
,004
<.e44)
Ne
Ce
A-4 B**
34.45
.t7
.0r2
-0.59
r00. r74
9.47
A-4 A**
.030
F"203
Cao
A-2
35.53
100.550
1 0 0 .7 5 E
.928
. 002
.004
(.934)
r.010
,002
. oo2
(r.014)
.964
.004
. 001r
.998
(.998)
1.0r0
3. 984
.3r2
. 70 4
1 .0 0 3
(r. 003)
1.018
.348
.536
3. O22
+ ,002
3. 036
t.624
5,1783
+07
7.1770
f12
3.037
+.008
3.059
1.626
5.L192
09
+
7.1736
5.0468
+09
'l,tz"
49.3'
+
0.7'
-9804.
E'
0.91
+
67042,5'
0.7'
r
r59.95
+ .03
5. 0451
+I0+09
<.s72)
. 002
.989
( . 9 e1 )
1.0I5
4.025
.427
.548
3.037
+.010
3.067
1.61s
5.1581
+
l0
7.1816
t14J10
5.0501
rI2o45.s'
-+9 8 0 0 00 .. 59' '
tt3o5.6'
0 I. .72I'
-+9 8 0 1
-+6 7 0 4 1
3 .. 60' '
-r 6 7 " 3 50,.69' r
+
1.0'
159.94
+.04
0.7'
+
159.39
+.03
52.42
52,42
52.56
52.4L2
52.391
52.558
tion
of actinolite, that inclusions and other crystal imperfections may play a much more important role
than doesthe chemicalcompositionin causingslight
variationsin specificgtavities.
these problems
For amblygonites-montebrasites,
can be demonstratedby comparingthe measured
specific gravities to the values calculated from unit
cell volumesand contents.Using the valuesof 161
A3 and 158 A' for volumes of montebrasiteand
amblygonite,respectively(cf . the X-ray powder diffraction sectionbelow), the calculateddensitiesare
3.01for LiAlPOl(OH) and 3.11for LiAlPOnF.The
calculated density of montebrasiteis higher by .03
than that estimatedby Palacheet al. (1951). As
shown in Figure 1, all the calculated densitiesare
distinctly higher than the measureddata, and the
294
itnv.q', innwv, AND FERGUSzN
Tlnre 2. Continued
oI2o3
A-60
^-29
33.53
32.86
.146
.r32
. 002
A_3
34.63
F"203
A_I3
AF-50
AF_46
AF_l
34. 18
34.26
35.16
3r.62
.08
.013
.025
,015
.0r0
Li20
. 003
9 .9 0
N.20
. r28
.o47
Kzo
. 005
. 004
!,c0
, ro,
.22
9 .9 0
48,85
.015
.008
1o.22
9.30
.039
. 005
, 003
. 004
.004
4 9- 2 2
49,rr
.08
.06
Hzs
3.32
3.2L
2 . 70
6.30
6.3{J
3 .3 7
6,43
6,I7
L O I . 7 75
- 2.65
99.512
Tocal
.997
.003
.004
(Ll+Na+Ca) (.984)
o40
A1
fr.3**Arl f.6is)
I.014
3 . 9 77
.479
.544
P
o
F'
(0H)
Sp.
gr.
Sp,
gr.
r'^
ao (I)
r r8r
co {8)
p
r
v(83)
o2er'
zul
3I
neas.
3.031
+ .012
3.052
L.6T4
5.1694
+L9
7. r815
+12
5,0507
+
10
lr:og. t ,
+
0.8'
9 8 0r t , 2 '
+
1,5,
67032.6l
+
1,.2'
159 -32
+ .05
m e a s .+ 52.61.
caIc.
calc. T
103.424
-2.65
1 0 2 , 7 75
100.774
100.075
48.85
3.29
7, 0 7
3.094
+ .012
3,030
+ .012
3.025
+ .010
1.5923
s.1483
+
17
7.207r
20
!
5. 060r
+
12
1.638
5.1960
+09
7 . L694
+20
5. 0378
+o7
1 1 2 o 1 8 r. '
+
.8'
9 7 0 5 3 .t ,
f
.9'
6 7 0 4 9. 5 1
+
.9'
1 6 0 . 79
+ .03
52.13
I 0 I . 30 4
100.01
.994
<.994)
1.015
4 076
.486
.438
.982
(.982)
1.0r3
3.956
.496
.548
.985
(.985)
1.008
3.920
,545
0 .s 3 5
3.048
+ .020
3.072
I.6T4
5.1668
3.032
+ .009
3.050
1,616
5. I705
+
15
7 . 1799
+t4
s. 0509
+
11
Ittoz,. g,
5 2 . 5 75
09
5. 0482
tl:oa- z,
0.6'
9808.8'
|
-+6 7 0 30 7. .93 '
t
0.9'
1
159.32
+.04
52.56
52.558
7.24
I00.302
.954
(.954)
1.026
3.903
.490
.527
+
4 . 93
t03.32
-3.31
, 008
. 002
<.e23)
rr
5,0513
+09+07
1 1 3 " 5 .8 '
1 .1 '
t
98010,7,
r.3'
J
67035,4t
1.0'
1
159.38
+ .04
52.59
.IL
2.92
-2.70
. 004
(. e38)
I
+
1,0'
9809.6,
67036.3'
+
I.2l
159.48
+.05
52.5J
48.85
.05
1O4.364
1. 001
.002
. 006
(1._009)
7.r782
49.27
1O3.272
. 980
. 002
.003
(.985)
3.039
+ .008
3-O52
1 .6 1 6
5.1691
L.2
.97
.06
1 0 2 .10 2
-2.59
8.8
.o97
49,26
F
8.93
.174
.05
3+
AF_48
,005
9.60
.082
Hzo-
Fe
AF_66
.007
Ca0
Li
Na
Ca
AF_65
.933
. 004
< . e 3 7)
. 906
.o47
(0.953)
1.001
(1.00r)
1.007
3,977
.553
. 4 70
,941
(. e41)
l. 04
4.003
.627
. 2 70
3.000
3.035
+ .010
+ .010
3.057
3.065
r.612
1. 609
5.L670
5.1643
+
l0
+08
7. 1825
7 , 1863
+07
+09
5. 0497
5.0533
+07
+09
irlorr. s'
t r:"t2. : '
+
0.6'
+
O.1'
980r0.81
98015.4,
+
1.0,
+
0.8'
67034.4|
6 7 0 3 L . 9r
+
o.7
+
0.71
r 5 9. 2 2
t59.18
+ .03
+.03
52,62
52.63
52.6L9
trend of their change with F-content follows rea_
sonably well the 3.01-3.11 line connectine the the_
oretical end-membervalues.
The fairly large difference between the measured
and calculated densities of F-poor montebrasites
seems to be due to the secondary nature of most
of these minerals, which usually contain a high con_
tent of impurities. The gem-quality montebrasites
from drusy vugs that should be very poor in F (ac_
cording to 7' and X-ray diffraction data) show
specific gravities fitting the OH-rich end of the re_
gression line for calculated densities (Table 2).
Refractive fndex y,
The refractive index 7, was measured in white
light by the Becke method. The refractive index
of the immersion liquids was measured bv means of
a Zeiss refractometer. The highest index of cleavage
3.007
.01.0
3.043
1.605
5.1606
+
11
7.t879
13
t
5,0499
+
U
rr3o23-i'
I.3'
1
98017,3'
+
1.7'
67032,Ot
r.4'
t
15E.86
+ .05
J
tt+oo.:'
1.3,
98038.3,
+
1-6'
67014.51
1.6'
J
1 5 8 .1 5
+ .05
53.00
J
52.13I
t.637
5.1.946
+09
1. r 7 1 2
+
t3
5,0430
+12
rr2o2r.5'
+
r.0,
97054.81
+
1.3,
670sr.61
I60.92
+ .O4
5 2 .1 5
52. t46
fragments showing the highest birefringence was
measured and designated y,. The true y index is
distinctly higher in most cases,since the optical direction Z deviates for most amblygonite-montebrasite
compositions from the {100} cleavage.
Our data plotted in Figure 2 show very good alignment which considerably deviates from the y/F rclation derived from widely scattered data by Winchell
and Winchell (1951). Our 7, values are significantly
higher than the 7 values given by them for about
two thirds of the range of F-contents. Since the
errors inherent in our technique could only decrease
the y' values, the difference between our and the
Winchells' best fit lines for true y should actually
be still larger, and thus the plot of F os refractive
indices given by the quoted authors is undoubtedly
erroneous (as suggestedearlier by, e.g., Ginzburg,
1950, and Standk, 1960).
295
AM BLY GO N IT E-MO N TEBRASITE M I N ERALS
TrsLB 3. Regression Equations and Correlation Coefficients
'0087
x
ul.% F
+
2.9849
('989)
-'ooars
x
ut.%F
+
r.64223
('989)
=
.0750
x
wr.%F +
52.1060
(.997)
crK.r
ca1c. -
.0774
x
wr.%F
+
52.O9L9
(.993)
o2€
c.Koj (ior
- i1o) =
.0058
x
wt.% F
+
.4699
(.810)
o2e
c'K" (iot - i.zr) =
-.0593
x
w E . %F
+
1.1099
(.985)
o29
cuK. (iol
- o2r) =
-.0511
x
vr.% F
+
1.6302
(,987)
o29
c.K.1 (ior - 011) =
.0170
x
ut.% F
+
1.0077
(.95,e)
cuKq (ior - 120) =
-.0140
x
et.7. F +
1.4781
(.942,
=
-.0043
x
wr,%F
+
5.1960
(.983)
bo =
'0029
x
ut,% F
+
7.1631
(.927)
-
.0018
x
ut,% F +
5.0384
(.966)
x
wt.% F
+
112,1735
(.991)
x
ut.
F
+
97.7833
(.965)
x
er.% F
+
67.9190
<.974)
er.% F
+
150.8696
(.986)
=
speciflc gravitv
f,'=
o2er'
o2er31
o2e
l,r,l,t,t,t
2
3
4
cuK" I eeas.
rtrt,t,t,trl
5
6
7
I
9
l0
ll
t2
13
I
wtXF
Frc. 1. Specific gravities (measured-dots; calculated-x;
theoretical for ideal end members-triangles) plotted against
F contents. Points in parentheseswere not used in calculations of regression equations. Solid line-regression line for
the measured data; heavy broken line-connects the 2.98
and 3.11 values given by Palache et al. (1943) for the end
members; light broken line-regression line for the calculated data. Empty circle is the plot of the Na-rich Hebron
sample (AF-55).
c
o
o
P = .o6sr
f, = -.0:rr
''1 =
'.2374
x
to lower temperatureswith decreasingF-content.
(3) A reactionat 795-8I1."C is the most prominent
for the F-richest specimens.Although rather broad,
it sharpensand increasesin intensity in the 7 to 4
percentF range,then fadesout rapidly. In the range
3.3 to 2.5 wt percent F it becomesperceptibly douDifferential Thermal Analysis
bled, and then is united again. (4) A closely-spaced
The thermal behavior of 11 specimenshas been doublet, absentin the two F-richest specimens,destudied using the automatic apparatus DrA-13M
(R. L. StoneCo.). About 0.1 g of a 0.104-0.177
1 6 5 0mm fraction ( 150-80 mesh) washeatedin a platinum
dish; AlzOe was used as a referencestandard.The
1.640rate of heating in air varied only negligibly about
-\
100"C/10 min. The records were corrected for
r630unbalance in the weights of sample and standard
which causesvertical shifts, and for changesin the
1620 conductivity of the sample during heating by deducting the zero curve of a secondrun from that of
r 6 1 0the previously analyzed specimen.
The d.t.a. records (Figure 3 and Table 4) show 1 6 0 0severalcharacteristicendothermicreactions: (1) A
1 5 9 0very weak reaction at 636 to 698'C seemsto be
more pronouncedin F-poor minerals.(2) The major
0123456789r0il1213
wt %F
reaction at 751 to 793"C forms only a broad shoul'y' plotted against the F
refractive
indices
The
Frc.
2.
der on a higher-temperatureendotherm in the two
contents. The solid line is the regression line calculated
F-richest specimens,then increasesin intensity up to
from these data; the broken line is that given by Winchell
around 3 wt percent F, and decreasesagain for the and Winchell (1951) for true'y. Empty circle is the plot
F-poorestsamples.This endothermshiftsfrom higher of the Na-rich Hebron sample (AF-55).
This conclusionis supportedalso by the y values
of 1.638 for a montebrasitewith 1.29 wt percentF
(Gallagher, 1967), and 1.645 for a montebrasite
with sp gr 2.98 (Haapala,1966).
0
296
inny/,
innNy, AND FERGUSzN
| | | | | | | | | | | xF
The heights of the blocks for any one specimenare
rc.v proportional to the amounts of the phasespresent;
no absolute amounts are implied, and they are inAF-I
7.85 tended to show only the changesin mutual proportions among the five phasesin different specimens.
A-29
The
tridymite- and cristobalite-typeforms of AlpOr
6.30
always occur together and in nearly equal amouns,
A-60
6 . 1 7 their most prominent powder diffraction peaksbeing
roughly of the sameintensity;thus they are shown
togetherin Figure 4.
A-4
4.05
For sample .{-60 with 6.17 wt percent F, the
heatingproductsafter eachendothermwere X-rayed
A-2
3.65 (at room temperature),and the phasesidentified
and semiquantitatively
estimated.The results illusA-98
3 . 4 4 trated in Figure 5 show the complicated reactions
that take place during the heatingof an intermediate
amblygonite-montebrasite.
A- 103
2.20
The phase assemblages
obtained at IOOO.Cafter
the differentialthermal treatmentare understandably
r.8E not equilibrated.Moreover,productsof d.t.a. runs
AF-43
repeated on previously run samplesdiffer consid1 . 4 0 erably from the assemblages
A-22
shownin Figure 4. This
suggeststhat the relative phase abundancescould
lttttl
ttltl
0
200 4oo 600 800 1000.c be expectedto be quite differentnot only after equilibrium heatihg, but also under difierent experimental
Frc. 3. Difterential thermal analysis records of l0 chemconditions of the differential thermal analysisitself.
ically analyzed specimens;weight percentagesof F are given
at the right-hand side, Temperatures of reaction peaks are
This is documentedby the d.t.a. resultspublished
listed in Table 4.
by Manly (1950) and Boruckii 0966). Manly reported fusion of his sample at 850oC, whereas
Boruckij
found much of his charge unchangedat
velopsgradually in the 6.5 to 3.4 wt percentF range
90OoC.
Their
d.t.a. patterns,as well as thosepubin the 850-879"C region. From 3.4 wt percentF
lished
(in Vlasovet a1.,1964),Ivanova
by
Ginzburg
downwards,the first reaction becomesreducedto a
(1961),
and
Correia
Neves and Lopes Nunes
mere shoulder and a broad slope on the higher(1968),
generally
either
deviate from our records
temperature reaction, which becomes much more
or
approach
those
with
different
F-percentages.
intensefor specimenswith the F-content lower than
3 wt percentF and occursat 896"C.1
The products of the d.t.a. treatment after heating Tlsrn 4. Temperatures of Differential Thermal Reactions*
to 1000oC have been identified by X-ray powder specl@o Fz wt. !550'c .zoo'c .7ffi
diffraction as LiaP2O7,the high-temperatureform AT-55
(557)
r0,17
(793) 811
of LLPO4, and the berlinite-, tridymite-, and cristoAT-I
(636)
7.85
(780)797
balite+ypeforms of A1PO4.Figure 4 showsthe rela780
810 (8s4) (870)
tive amounts of individual phasesproduced by the
A-60
6.l7
777
805
heating of a given specimenas estimatedfrom the
A-4
(698)
4,05
(8s0) 879
764
807
relative intensities of their most prominent peaks.
AF-55
1 The only exception
from these trends are the samples
A-60 (6.17 wtVo F) and,A-29 (6.30 wt% F) that would fit
the general scheme better if their positions in Figure 3 (and
also in Figure 4) are interchanged. perhaps their actual
F-contents are reversed, since the determined values lie
within the limits of analytical error.
a-2
3.65
(588)
761
808
A-98
3,44
(689)
760
8L2
A-103
2. 2 * *
(690)
760
Ar'-43
1.88
(688)
A-22
1.40
(684)
* ( ) - weEk reactloni
** F-content
eetlpated
_
bFx-ray
884 895
862
879
(795)(805) (850) 881
(800)
897
75r
Btrong reactLon.
pdder
dlffractloD
lethod.
297
AM BLY GO N IT E.MO N TEBRA SI TE M I N ERA LS
Li3 PO4
L\P2O7 Ar PO4
At Pq
"crisd'+"trid."berlinite
h.t form
AF-55
AF-I
A-29
----I
A-60
A-2
T-
A-9E
I
AF.43
I
A-22
Frc. 4. Heating products of 8 chemically analyzed samples, as determined at room temperature by X-ray powder
diffractometry in the products of d.t.a. treatment to 1000'C.
Heights of blocks indicate the relative abundancesof the
individual phases.The tridymite and cristobalite forms of
AIPO, are always present in about equal amounts.
X-Ray
Powder Diffraction
I
012345678910
wt XF
Frc. 6. Measured values of the 2e CuKat angles for the
131 reflection plotted against the F contents. The solid
line is the regression line calculated for the measured data,
the dashed line is that derived from values calculated from
unit cell dimensions (see Tables 2 and 3).
Methods
Characteristic group ol reflcctions
Reflection 131
Moss el al. (1969) have shown that a group of
Mosse/ aI. (1,969)givea graphfor the determina- X-ray powder reflections between 26 and 29" 20
tion of F-content based on the changesin 21coxo CuKo is sensitiveto slight changesin the F content.
angle of the 131 reflection. Their diffractograms, We have developeda method for using this group
however,were not calibratedand the data plotted in of reflectionsfor quantitativelyestimatingthe F contentsfrom uncalibratedX-ray powderdiffractograms.
their Figure 3 show considerablespread'
In the present study, the diffractograms were
At the low-angleside of this group, the 101refleccalibrated with quartz as an internal standard.They tion is alwayswell definedand not affectedby overlap
were recorded using a slow scanning speed and a with other peaks.Thus it servesas a zeropoint from
moderatechart driving speedwhich gave l" 20 per which the angular distancesto the other reflections
5 cm. The absolutevalue of 20n, for amblygonite- (i21, i10,021,011,and I 20)areread.Thesedistances,
montebrasitewas correctedusing Frondel's (1963) plotted on the appropriatescaleon a paper strip, are
calculated 20 CuKa, anglesfor the ll22 and 2022 then comparedwith our Figure 7 which relatesthe
reflectionsof quartz. The graph presentedin Figure 6
0showsa much smallerscatterof individual plots than
doesthat of Moss et al.,but the bestfit line matches
very well our lines drawn for both measuredand
0.5_
calculatedvalues(seeTable 2). The maximum errors
in readings of F should be lower than .5 percent. U .
Y
t-
Al PO4 Li3 PO4
'tri*obolird'
htfom Li4 P207
omblygoniteberlinite "tridymire'
Al Pq
AIPO4
7w"-
8ro.
-
I
-I
1000'c
Frc. 5. Heating products of sample ,4'-60 (6.17 wt percent F) as determined by.X-ray powder diffractometry at
room temperature after each endotherm in d't.a. treatment'
Heights of blocks indicate the relative abundances of the
individual phases.
H r.0N-
1 . 5r,r,l,r,l,l,l'l'l'l'l'l'l'l
012345678910n1213
wtXF
Frc. 7. The variations in angular distances of the X-ray
powder diffractionreflectionsi21, i10, 021, 011, and 120 from
itt. iOt reflection. Regressionequations are quoted in Table3'
vr!,
CERNA, CERNY, AND FERGUSON
20.u*o positions of the six reflectionsto the F contents. With the "zero point" i01 reflection of the
examinedrecord sliding on the iOl tine of the graph,
the wt percent F in the examined mineral can be
found in the position of best fit of the other five
reflections;we regardthe accuracyas +.5 wt percentF
or better.
The only disadvantageof this methodis the partial
or complete overlap of two reflections in several
compositionalranges.However, with some practice
it is possiblein most casesto estimatecorrectlythe
positions of individual reflectionsin partially overlaping pairs, and there is always at least one resolved reflection presentwhich aids in this estimate
(cl. Figure 8). X-ray powder diffractogramsof high
resolution are of courseneededto permit reading of
the 20 valuesto within i.o2".
were measuredas closeto the top as was reasonable,
and 20 valueswere then correctedusing the theoretical CuKal-values for quafiz calculated in Frondel
(1963). Partial indexingwas done using the data
published by Haapala (1966) and Moss et al.
(re69).
A least-squares
refinementof unit cell dimensions
was performed using the self-indexingprogram by
Evans et al. (1963), modifiedby D. E. Appleman,
with an input of 25 to 37 reflectionsfor each specimen. The computer-calculated
standarderrorsof the
unit cell dimensions(Table 1) are mostlyvery small
and even if doubled, as deemedmore reasonablefor
feldsparsby Wright and Stewart(1968), would still
be smaller than the precision attained in earlier
amblygonite-montebrasite
studies.The large number
of uniquely indexed and well shapedpeaks obtainablefrom amblygonites-montebrasites
likely accounts
Unit cell dimensions
for this high precision, and the calibration by 6 to
Glass-slide mounts of powdered amblygonites- 8 quartz peaksdistributedthroughout
the whole patmontebrasites,with admixturesof subordinatequartz tern makesus believethat the
accuracyis alsohigh.
as internal standard,were run on a philips Norelco
The plot of unit cell dimensionsus F contentsis
diffractometer(Cu/Ni radiation, scanningspeed,
/+"
shown in Figure 9. The calculatedlines of best fit
20/min. and chart driving speed 10 giving approxi- representin our view a very good
approximationin
mately 5cmfl"20, 400 counts/sec.,time const. 4 the rangeof the lower F contents,
but are of a rather
sec.). Each sample was X-rayed three times, the doubtful valuein the high F range
becauseof scarcity
powder being remounted for each run. All peaks of data. It is possible
that the plots of some unit
A-13
6.43wt.%F
E'8rla,9 ,0
29
8
27
26 29
28
27
)Jt I tl
Jt | il |
E'8;:o , 9 , 0
,o
a&'8,3|g
"llll
Iil|r
'38e:e,g,o
,oa- R -er -,,a- 9
26 n
28
27
26 29
28
27
?b 29
28
27
26'2e
CuKcr
Ftc' 8' Typicalexamplesof the groupof 6 reflectionsusedfor indirect
determinationof the F contentby aid of Figwe 7. Note
the verygooddefinitionofall the 101reflections,
usedas zeropointsin readingthe angulariirtuo".r.
299
AMBLYGON ITE-MON TEBRASITE MINERALS
Discussion
5.20-
___\._:.-\1-.-..
5lE -
oo
_
----:-?
r
s.r I
5.r4-
.- -..=.-"----o--..-
:
-
7n-
.-5.'
bo 7lE 7 1 6_
5.06co
5 04 --o-'
t-o
L^---L}1=--;P'-
_
lt4.d
fl3'-
-lJ--
,r-
_-_-:-.-t
tst-'
ltltrrllrlllll
012345676910111213
rt IF
Frc. 9. Unit cell dimensions plotted against the F contents. Regressionlines are calculated from equations given
in Table 3. Empty circles are the plots of the Na-rich
Hebron sample (AF-55).
cell dimensions follow very flat arcs (particularly
those of bo, B, and V),but this can be established
only by collecting more data on F-rich amblygonites'
For this reason it is not advisableto take the regression intercepts at F : .00 as accurate unit cell
dimensions of pure montebrasite.
The deviations of individual plots from the best
fit lines largely exceed the standard error values.
They are probably caused by analytical errors in the
determination of F, and by chemical substitutions
other than F/OH andf or deviations from stoichiometry. Neverthelessthe general agreementis distinctly
better than that obtained by Moss et aL (1969),
although most of their best fit lines are close to our
regressions.Thus the graph presented in Figure 9
should yield estimates of the F-content correct to
within .i0.5 wt percent F (see Table 2). Not to
obscure the possibly curved alignment of our plots,
the earlier data by Haapala (1966) and by Moss
et al. (1969) shown in Figure 5 of the latter authors,
are not included in our Figure 9.
(1) From the different physical properties and
determinative techniques used in this study, specific
gravity and d.t.a. seem least suitable for the indirect
determination of F contents in the examined mineral
group. Specific gravity is too sensitive to the presence
of impurities, and difierential thermal behavior is too
strongly dependent on the experimental setup'
(2) Optical properties are likely to be very useful once they are properly studied. The preliminary
results obtained during the present study are good
for only approximate estimates of the F contents.
Nevertheless,we feel sure that these estimateswould
be much better than those based on earlier graphs
of optical properties which have been shown to be
unreliable (Winchell and Winchell, 1951)' A detailed re-examination of the amblygonite-montebrasite
optics is highly desirable.
(3) X-ray powder diffraction is evidently the best
method available at present. The 20 values of the
131 reflection, calibrated against quattz peaks, seem
to be the most accurate. The "six-peak method"
shows lower correlation coefficients but is faster in
investigations where many dozens of samples are to
be examined. Accurate unit cell dimensions should
also yield quite reliable estimates, at least in the
montebrasite half of the series, but the data collecting and computing process is time-consuming.
(4) It must be stressed, however, that fundamental prerequisites for the reliable use of the relationships and graphs presentedin this study are Na2O
content lower than about 1.5 wt percent and optical
homogeneity of the specimens. The influence of
higher NazO contents is illustrated by the plots of
the specimen AF-55. Since the determination of Na
on minute samples is easily accomplished in most
chemical laboratories, a series of specimens representative of all paragenetic types of amblygonitesmontebrasitesin a studied locality should be analyzed
for Na before the indirect checks on F-contents are
performed.
(5) As shown by iern6 (1970) and eetnb et al.
(1972), grains and crystals of amblygonite-montebrasite minerals frequently show primary compositional zoning and/or secondary replacement by
members of the same series with different F contents.
Both zoning and replacement may betray themselves
in hand specimens by slight changes in color, but
usually they can 'be detected only microscopically.
Thus thin section study and checks in immersion
300
vtv,
CERNA, CERNY, AND FERGUSON
liquids of materialdestinedfor X-ray diffraction work about 1 wt percentF, and we supportthe conclusion
of Moss et al. that the differencebetweenthe two
are necessaryto ensurereliable results.
(6) In detailedinvestigations
of individualspeci- structuresis probably due to compositionaldiffermensof amblygonite-montebrasite,
greatcareshould encesbetweenthe two minerals.Crystal structure
be taken to use only homogeneousmater.ial. A refinementsof analyzedamblygonites-montebrasites
specimenbeing consideredfor chemical analysis with differenl F/OH ratios would help to elucidate
should be studiedin thin sectionsbefore crushing, the crystalchemistryof this series.
and the required amount of the cleanestmaterial
Acknowledgments
separated out. The separatedgrains should be
This study was carried out during postgraduate studies of
checkedin oil immersionbefore final grinding,and
materialfor the determinationof physicalproperties the first author as a part of her Master of Science thesis
(dern6, 1970)with support from a National ResearchCouncil
shouldbe sampledfrom them at random.
grant to the third author (R.B.F.), and during the tenure of a
Differencesbetweenthe analyzedand the optically Postdoctoral Fellowship of the second author (P.e.) funded
studiedmaterialin many earlierstudiesare ths most by research grants from the University of Manitoba and the
probable explanationfor the poor correlation of National Research Council of Canada to the third author
optics with F content in Winchell and Winchell's (R.B.F.) and to A. C. Turnock from our Department. Our
(1951) graph.Differencesbetweenthe analyzedand thanks are due to all colleagueswho assistedin our searchfor
suitable samples, particularly to C. T. Williams and R. A.
physicallyexaminedmaterial can be demonstrated Crouse (Tantalum Mining Corporation of Canada, Bernic
on Haapala's(1966) Hunnako montebrasite,
which Lake, Manitoba), J. S. White, Jr. (National Museum of Natural
yielded 2.06 wt percentF in chemicalanalysisbut History, Washington, D. C.), P. G. Embrey (British Museum,
its refraction index 7 Q.624), specific gravity Natural History, London), J. Mandarino and R. I. Gait
(Royal Ontario Museum, Toronto), and F. eech (Charles
(3.023), and unit cell dimensionscorrespondto 4University, Prague.) Special thanks goes to Dr. P. G. Embrey
5 wt percentF, when comparedwith our Figures1, and his colleaguesat the British Museum (Natirral History)
2 and 9 (cl. also Moss et aI., 7969). The decrease whose comments on the thesis, later discussions,and critical
in volume with increasingF content explains (at reading of the manuscript were invaluable. We are grateful
least partly) the discrepancybetween Haapala's for chemical analysesto J. L. Dalton (Mines Branch, Ottawa),
to K. Ramlal and R. M. Hill (Department of Earth Sciences,
measured(3.023) and calculated(3.065) specific University of Manitoba), and to P. Povondra (Geological
gravity.
Institute of the CSAV, Prague). D. E. Appleman and D. B.
(7) The structureof theamblygonite-montebrasiteStewart (U. S. Geological Survey, Washington, D. C.) kindly
mineralshas been studied by Simonov and Belov provided us with the computer program for the refinement of
(1958) and by Baur (1959). Baur (1959) sug- unit cell dimensions, and G. A. Russell of our department
kindly made the DTA apparatus available.
gested the possible existenceof high-temperature
References
and low-temperaturephasesin the series,since the
(1959) Die Kristallstruktur des Edelamblydistributions of Li found in the two structure de- Beun, W. H.
gonits LiAlPO. (OH,F). Acta C ry stallogr. 12, 988-994.
terminationswere distinctly difterent.However,the Bonucrrr, B. E. (1966) Hebronite from the sodium-lithium
chemicalcompositionsof the mineralsusedin both
pegmatites of Siberia. Trans. Mineral. Mus. Acad. Sci.
structuralstudieswere not known. and Baur conU.,S.S.R.
17, 183-189(in Russian).
cluded from indirect evidencethat his crystal had CEnN{, Ivt' (1970) Mineralogy and paragenesis of amblygonite-montebrasite with special references to the Tanco
a F/OH ratio 1:1. Moss et al. (1969) usedtheir
(Chemalloy) pegmatite, Bernic Lake, Manitoba. Master
graphof F contentsus.unit cell dimensionsto derive
of Science Thesis, Department of Earth Sciences, Uni
compositions
for the specimens
of SimonovandBelov
versity of Manitoba. lO2 pp.
and of Baur. They found that the Simonov and
P. dsnNf, lNo R. B. FrncusoN (1972) The Tanco
pegmatite at Bernic Lake, Manitoba. III. AmblygonitesBelov valuesfall into the high-F half of the diagram,
montebrasites.Can. Mineral. 11, in press.
and that Baur's data approachclosely the monteConnere Nrvrs, J. M., ero J. E. Lopps NuNrs (1968) Pegbrasitecomposition.
matitic phosphatesof Alto-Ligonha Region (MozambiqueWe determinedthe unit cell dimensionsand other
PortugueseEast Africa). Rev. Cienc. Geologicas,1, Serie
physicalcharacteristics
A, 1-48, Lourenco Marques.
of a crystal (AF-66) from
the samesuite of Newry specimens(Palrc,heet al., EvlNs, H. T. Jn., D. E. Apprnr"reN,lwo S. S. HlNowBnrnn
(1963) The least-squaresrefinement of crystal unit cells
1943, depositedin the mineral collectionsof the
with powder diffraction data by an automatic computer
Harvard University) from which Baur obtainedhis
indexing method. (abstr.) Amer. Crystallogr. Ass. Meetmaterial.All the propertiesfit a montebrasitewith
ins, 3.
AM BLYGO N ITE-MON TEBRA SI TE MI N ERALS
FnoNoer, C. (1963) Dana's System ol Mineralogy, Tth ed.
Yol.3, Silica Minerals. New York.
GelucnEn, M. J. (1967) Phosphatesand other minerals in
pegmatitesof Rhodesia and Uganda. Mineral. Mag. 36,
50-59.
Grwzrunc, A. I. (1950) Montebrasite and the processesof
its alteration. Trans.Mineral. Museum Acad. Sci. U.S.S.R.
2.72-85.
Hueetn, I. (1966) On the granitic pegmatitesin the Per:iseiniijdki-Alavus area, South Pohjanmaa, Finland. Bzl/.
Comm. G6ol. Finlande224,98 pp.
IveNove, V. P. (1961) Thermograms of minerals. Mem.
All-Union Mineral. Soc. 90, 50 (in Russian).
MlNr-v, R. L. (1950) The.difterential thermal analysis of
certain phosphates.Am-er. Minerol.35, 108-115.
Moss, A. A., E. E. Fn-ren,eNo P. G. Elrgnsy (1969) On
the X-ray identificationof amblygonite and montebrasite.
Mineral. Mag. 37, 414-422.
NIcorls, A. Dispersionde Ia densitedans un mineral homogene. Bull. Soc. fr. Mineral. Cristallogr. 91, 182-184.
Prrecnr, C., H. BrnueN, ANDC. FnoNopr (1951) Dana's
Systemof Mineralogy, VoI. 2,7th ed. John Wiley & Sons,
N.Y.
301
PerlcHr, C., W. E. RrcnvroNo, eNp C. W. Wolre (1943)
On amblygonite.Amer. Mineral. 2Er 39-53.
Rrnss, P. H. lNo P. E. RospNnenc (1971) Optical and
X-ray determinativemethods fer fluorine in topaz. Amer.
Mineral. 56, 1812-1821.
SrrraoNov.
V. I.. eNDN. V. Brlov (1958) The determination
of the structure of amblygonite by the minimum function
method. Kristallografiya, 3, 428437 (in Russian).
SreNEr, J. (1960) Montebrasite,augelite,and lazulite from
Otov, Acta Acad. Sci. aechosl. Basis Brunensis,32' Tom.
1, Fasc. 4OO,17-32 (in Czech).
Vusov, K. A., Ed. (1964) Geochemistry and mineralogy
ol rare elements and genetic types of their deposits. VoI.
II, Mineralogy of rare elements. Moscow. English translation, Jerusalem,1966.
WrxcHerr-. A. N.. eNo H. Wrwcnsrr. (1951) Elements ol
optical rmineralogy. 4tt' ed., Part II. New York.
Wnrcnr, T. L., eNo D. B. Srewe'nr (1968) X-ray and optical study of alkali feldspar: I. Determination of composition and structural state from refined unit cell parametersand 2V. Amer. Mineral.53, 38-87.
Manuscript receiued,May 15, 1972; accepted
lor publication, October 30, 1972.