Alkali feldspars: ordering rates, phase

transformations and behaviour diagrams for

igneous rocks*

WILLIAM L. BROWN
Centre de Recherches Petrographiques et G eochimiques, BP 20, 54501 Vandoeuvre-h!s-Nancy Cedex, France

AND

IAN PARSONSt
Department of G eology and Mineralogy, Marischal College, University of Aberdeen, Aberdeen AB9 l AS, Scotland

Abstract

Homogeneous and heterogeneous phase relationships in the alkali feldspars are reviewed, and behaviour
diagrams developed. Al,Si ordering is almost certainly continuous and higher order in both albite and
potassium feldspar and has been established reversibly or nearly so down to below 500 oc in albite
and possibly to � 200 oc in potassium feldspar. The degree of order in intermediate albite changes
strongly over a range of � 75-150°C depending on pressure, low albite being stable up to about 620-
650 oc and high albite above about 725 oc at low pressure. Symmetry is broken at � 980 oc mainly
by a cooperative shearing of the whole framework and not by Al,Si ordering alone; there is a thermal
crossover near 700°C, shearing being dominant above (high albite) and ordering dominant below
(intermediate albite).
In potassium feldspar symmetry is broken by Al,Si ordering at a temperature of about 500 oc. The
change in degree of order with respect to temperature has been followed easily and reversibly in
sanidine from � 1075 to � 550 oc and to a lesser extent in microcline from 450 to 200 oc. Ordering
rates in sanidine down to 500 oc and ordering rates in microcline between 450 and 200 oc are almost
as fast as in albite. Ordering in sanidine at 500 oc and below slows and then stops with the development
of the tweed orthoclase domain texture. The tweed texture acts as a barrier to further order because
the strain energy associated with the (incipient) twin domain texture balances or nearly balances the
free energy decrease resulting from ordering. Ordering stops not because of the kinetics of Al,Si
diffusion, but because the total driving force is very small or nil. Ordering can readily proceed to
completion, with the formation of low microcline, only if the domain-texture barrier is overcome by
processes involving fluids or strong external stresses. There is no barrier in albite.
The symmetry-breaking process in alkali feldspar changes with composition from mainly shearing
in albite to ordering in potassium feldspar. Symmetry is broken equally at a compositional crossover
(metastable with respect to exsolution) near Ab80_75 at low pressure and progressively displaced
towards Or at higher pressures. Ordering in pure albite occurs by a (nearly) one-step path which
progressively becomes two-step with substitution of Or. Diagrams showing the near-equilibrium
variation of the order parameters at low pressure with composition and T are given, as well as two
extreme phase and behaviour diagrams for complete coherent and complete incoherent (strain-free)
relationships. These diagrams can be used to understand feldspar relationships and microtextures in
hypersolvus and subsolvus rocks, the occurrence of orthoclase, and of intermediate and low microcline.

KEYWORDS: alkali feldspars, ordering rates, phase transformations, coherency.

* CRPG contribution 739.

t Present address: G rant Institute of Geology, University of Edinburgh, West Mains Road, Edinburgh EH9 3JW.

Mineralogical Magazine, March 1989, Vol. 53, pp. 25-42

© Copyright the Mineralogical Society
26 W . L . B R O W N A N D I. P AR S O N S

Introduction synthesized high albite, because almost all low­

pressure hydrothermal experiments were �ot car­
ALTHOUGH the polymorphic and subsolidus . .
ried out so as to obtain bracketed eqmhbnum.
phase relationships of the alkali feldspars have Direct hydrothermal synthesis leads initially to
been intensively studied, especially over the last growth of high albite at all temperatures, which
35 years, controversy still exists over many funda­ .
orders on annealing to an extent wh1ch depends
mental problems; nevertheless, important ad­ on T and time. By 1979 certain data (e.g. Raase,
vances in understanding have been made in the 1971; Mason, 1979) could be interpreted as point­
last few years. Albite occurs in rocks almost ing to an upper stability limit for low albite near
exclusively as low albite; this is because Ab-rich 680-700 oc at low pressures ( 0.1 GPa). The
�

feldspars are rare in rapidly cooled rocks, and in equilibrium relationships between an order-sensi­
more slowly cooled rocks, ordering rates in such .
tive parameter, �131, and T were finally uneqmvo­
feldspars are fast compared to cooling rates. Pot­ cally established by Goldsmith and Jenkins (1985)
assium feldspar, in contrast, occurs in nature at high pressures near 1.8 GPa using an NaCl
in several polymorphic forms known since the pressure medium in a piston-cylinder ap�aratus
nineteenth century. We discuss below the reasons although Salje et al. (1985) had shown contmuous
for this apparently different behaviour. A dis­
variation of the order parameters as a function of
ordered form of albite was first unequivocally T from thermodynamic reasoning. Goldsmith and
shown to exist when Tuttle and Bowen (1950) Jenkins measured �131 after quench as a function
described the form, now called high albite, synthes­ of run temperature and showed the transformation
ized hydrothermally in the classic study of the to be continuous (Fig. 1). Low albite is stable at
system NaA1Si308-KA1Si308-H20 by Bowen � 1.8 GPa up to about 660 oc and then changes
and Tuttle (1950). This discovery led to three lines rapidly and reversibly over a T range of about
of research: (a) the characterization of high albite 1 10 °C (intermediate albite) to high albite above
and the high-low albite relationships, (b) the study 780 °C. High albite changes slowly from 780 oc to
of solid-solution series between triclinic high albite reach a steady value above 975-980 oc. They
(analbite) and monoclinic high sanidine and the estimated that the curve should be 45-50 oc lower
monoclinic-triclinic phase transformation as a at low pressure, on the basis of data at intermediate
function of composition, temperature, pressure pressure (Fig. 1) and from the Clausius-Clapeyron
and order, and (c) the determination of the solvus relationship.
relationships. In this paper we discuss recent Also plotted on Fig. 1 are those data from
results for alkali feldspars and attempt to show existing hydrothermal experiments at low P
the supposed equilibrium variation of the order (mainly 0.1 GPa) which show the large�t differ­
parameters. We propose simple subsolidus phase ences in order parameter from start to fimsh. New
diagrams for alkali feldspars and show how they data in the critical region at 680 and 700 oc, 0.1
can be used to understand alkali feldspar in GPa, using albites synthesized by Mason (1�79)
hypersolvus and subsolvus rocks. Readers may as starting material, run in 2M NaOH solutiOn,
refer to Carpenter (1988) for a review of the are given in Table 1 and plotted in Fig. 1. The
thermodynamics of alkali feldspars, in which inten­ new data are consistent with all prevwus . data
sive use is made of Landau theory. except for one run at 700 oc. This difference is
We use the following terminology; T10, T1m, outside experimental error for �131, but could be
T20 or T 2m are the four types of tetrahedral site
due to slight differences in T, as the precision and
in triclinic alkali feldspars. T1 and T 2 sites are not accuracy in the latter in long runs is not better
respectively subdivided in monoclinic feldspars, than + 10 oc. A smooth curve has been drawn
when average equal AI occupancy occurs in the whichis consistent with all the data. It is much
two sets of subsites. t10 . . . etc. , refers to the steeper than the curve at � 1.8 GPa a�d lies at a
fraction of each site occupied by AI. t10 is one in T 30-70 oc lower. There is no compelling reason
fully ordered An-free feldspars, 0.25 when fully for suggesting a discontinuous transformation
disordered. The relevant ordering parameters of (Tuttle and Bowen, 1950; Raase, 1971; �enderov
Thompson (1969) are defined as: Y t10-t1m,
=
and Shchekina, 1976; Senderov, 1980; Smtth, 1983;
Z =(t10+t1m)-(t20+t2m). Salje (1985) intro­ Kroll and Voll, in Ribbe, 1983). If the transform­
duced Qod (t10-t1m)/(t10+tlm).
=
ation were discontinuous it would be hard to
understand why two distinct albites have never
been found in ordering experiments (particularly
The high-low albite transformation
below 700 oq and how albite in contact with a
This transformation has been the subject of hydrothermal fluid can order continuously to
controversy since Tuttle and Bowen (1950) first reach values of �131 as low as 1.5-1.3. Until
 OR D ER I NG R A T E S I N A L K A L I F E L D S P AR S 27

T(K)
800 900 1000 1100 1200 1300
1.0

Eberhard 1967
E 1.0
--
1.1 LA ...... d h e n 1985
' G ��b ���� s�it� ��k i
K MacKenzie 1957
....... M
1.2 M Mason 1979
n+ "-r R Raase 1971
2.]\ S Sen derov & Shchekina 1976
en 1.3 tM
_jT T This work, Table 1 0.9
Q)
Q)
....
____J.\
M I
Hydroth ermal

runs
)"'!
Finish
Start
V'L':.l
Sol1"d-me d1um
runs
"
Ol *
• No Change o ( P in GPa)
Q) I
:g Partial (complete) conversion
tM s IIA of natural low albite
�
tj

::::l
I
I
�s I
0.8
0
1M
..- I
C')
..-

CD
C\1 E I 0.7
I
<:I
T

s 0.6
0.1 GPa

2.1�..:------::-:1-:,.-------=!-::-----�=-----�-:!-o------:-='::-::----l
500 600 700 800 900 1000
T (OC)
FIG. 1 . Variation at room temperature of !!.28( 1 3 1 ) (Cu-Ktx) as a function of run temperature for albite, mainly at
P of 0. 1 GPa for hydrothermal runs and at pressures between 0.4 and 1 .0 GPa for solid-medium runs. The light
curve at 1 .8 GPa is from G oldsmith and Jenkins ( 1985). The heavy, partly broken curve is the proposed
�

equilibrium curve for albite at 0 . 1 GPa. The asterisk shows the behaviour after annealing in air at 900 °C, of an
albite sample synthesized in a short run at 350 oc by Senderov and Shchekina ( 1 976). Compare with Senderov and
Shchekina ( 1 976, Fig. 3) and Senderov (1980) who gave a discontinuous interpretation of the ordering transformation.
LA low, lA intermediate and HA high albites, AA analbite. On right approximate values of (t10-t1m) and of
(t10+t1m) assuming ordering in albite deviates slightly from the one-step path (Kroll andRibbe, 1 983).

Table 1 Effect of Stbsequent hydrothennal treatment at a higher taoperature

than initial synthesis on i\131 of synthetic albites, at 0 .. 1 Gl:a.

.'.!:initial /!; 131 Irun time Ll131 change in

(oC) (029 CUKa) (°C) (h) /!;131

350 1.196 680 1701 1.204 +0.008

500 1.307 680 1701 1.263 -0.044

500 1.394 680 1701 1.361 -0.033

525 1.613 700 1682 1.620 +0 .007

625 l. 799 700 1682 1.865 +0. 066

Albites initially synthesized by Mason ( 1979)

28 W . L . B R O W N A N D I . P AR S O N S

direct proof t o the contrary, the most reasonable derov and Shchekina, 1976) probably occurs by a
conclusion is that the transformation is continuous heterogeneous solution/precipitation mechanism.
both at low and high pressures. This is in accord­
ance with the analysis by Salje et a!. (1985). The
Interaction between the high-low and the
ordering path is 'one-step' in albite, with AI
monoclinic-triclinic phase transformations in
migrating to T10 sites from all other sites equally,
NaA1Si308
or nearly so. Data on ordering states in Or-bearing
compositions (Parsons, 1968) are inadequate and The high-low albite transformation is slow,
unbracketed. They suggest however, that Or-bear­ diffusive and continuous, and occurs between
ing high albites have very similar degrees of order about 620 and 725 oc at low pressure (Fig. 1).
to Or-free high a!bites down to at least 700 oc, Fully disordered albite transforms rapidly and
but that they are less ordered below 700 oc and displacively from triclinic analbite to monoclinic
especially below 600 °C. The ordering path in such mona!bite at a temperature near 980 oc (Laves,
feldspars deviates more and more, with increasing 1960; Grundy et al ., 1967; Winter et al., 1979;
Or content, from a one-step path, because signifi­ Kroll et a!., 1980). The transformation is second­
cant transfer of AI from T2 to undifferentiated T1 order (Salje et a!., 1985) and the driving force is
sites occurs while the feldspar has monoclinic spontaneous shearing of the framework as a whole
symmetry. and not Na vibration (Brown et a!., 1984; Salje
and Kuscholke, 1984).
The kinetics of the high-low albite transformation
Salje (1985), Salje et a!. (1985), Salje (1986), and
Carpenter (1988) analysed the transformations in
In all hydrothermal synthesis experiments, the albite using a Landau-type free-energy expression
ordering rates depend on T and the composition with two parameters which are related by strain­
of the fluid. Maximum ordering rates in water or induced coupling. One parameter (Q) describes
in the presence of sodium salts occur at T in the structural distortion during the displacive
the range 550-350 ac. The absence of a marked transformation between monalbite and analbite
increase in ordering rates below 700 oc (cf. Fig. 1 ) (high albite) and the other (Qod) Al,Si ordering,
i s surprising, a s the driving force /1 G for ordering where Qod is (t10-t1m)/(t10+t1m) and t10 and
should increase rapidly just below 700 ac. Unlike t1m are the atomic fractions of AI in the T10 and
the situation at high pressure in a solid-medium T1m sites. Both Qod and Q vary from zero in
apparatus (Goldsmith and Jenkins, 1 985), it is monalbite to one in low albite at low temperature.
probable that ordering is not homogeneous and The symmetry-breaking process, shearing of the
that the rate-determining step involves a hetero­ framework as a whole, allows, under equilibrium
geneous reaction with the fluid (Mason, 1980a, b). conditions, a slight increase in order in high albite
The marked increase in hydrothermal ordering as temperatures are lowered, but only because of
rates below 600 oc, especially in the presence of the existence and strain coupling of the second
sodium salts, is best explained as due to changes parameter. The order-disorder transformation
in the physical properties of the solution, in sensu stricto alone would start at a lower tempera­
particular to an increase in ionic dissociation ture; a crossover (Salje et a!., 1985) with respect
(Franck, 1981). to temperature (thermal crossover) between the
Low albite can be converted into high albite processes occurs. The effect on 11131 measured at
either by dry heating or hydrothermally. Disorder­ annealing temperature can be estimated from the
ing rates on dry heating at atmospheric pressure data of Grundy and Brown (1969). The determined
vary from specimen to specimen (Tuttle and cell parameters at temperature for albites with a
Bowen, 1950; Schneider, 1957; McKie and range of Al,Si order from high to low, and /1131
McConnell, 1963), but are of the order of 5-10 values which would be observed at T for an
days at l 050°C and 150-200 days at 950°C. equilibrium Al,Si distribution can be calculated
Ordering has never been achieved on dry annealing (Fig. 2). T.hear is the temperature of the symmetry
at atmospheric pressure in the laboratory (com­ break for fully disordered albite ( � 980 oc at 0.1
pare ordering in sanidine). Goldsmith (1987) deter­ MPa). By extrapolating the room-temperature
mined disordering times as a function of pressure curve of 11131 against T, an apparent temperature
between 0.6 and 2.4 GPa and showed that times for the symmetry break due to ordering alone
decreased greatly as P increased, the effect levelling (I;,,d) of
� 700 oc is obtained, about 280 oc lower
off above � 2 GPa. He suggested that hydrogen (see, however, Carpenter, 1 988). The effects of the
was responsible for promoting Al,Si interdiffusion. thermal crossover are clearly seen in both curves
Hydrothermal conversion of low to high albite in Fig. 2 and account for the existence of a range
(Tuttle and Bowen, 1950; MacKenzie, 1957; Sen- of stable high albites and metastable analbites.
 O R D E RI N G R A T E S I N AL K AL I F EL D S P A R S 29

ordering t fI the�al
sheanng

annealing-temperature curve
(calculated)

1.0

room-temperature curve 0.8 .......

<D
{observed) E
T""
C\1
lA 0.6
-
<I
0
T""
�
resid �al
ordering
tt sheanng 0.2

M
0.0

300
K

FIG. 2. Interpretation of the room-temperature (lower curve, from Fig. 1) and annealing-temperature (upper curve,
calculated from Grundy and Brown, 1969) variation of the values of 8211( 1 3 1 ) as a function of annealing temperature
for equilibrium albites at low pressure (from Smith and Brown, 1 988). LA low, IA intermediate and HA high
albites, AA analbite, MA monalbite. The effect of the ordering transformation is best seen in the room-temperature
curve and that of the thermal shearing in the annealing-temperature curve. T,h,., I;,,d and the thermal crossover
are explained in the text. An increase of pressure displaces To,d (cf. Fig. I) and T,h.., to higher temperatures (Hazen,
1976; see however Angel et a!., 1 988). The value of (t10-t1m) at room temperature is given on right.

Ordering dominates below the crossover, shearing Moreover, the fact that the macroscopic and
above. Comparison of the curves of Lll31 mea­ microscopic order parameters are identical (from
sured at room temperature and estimated at the Raman spectra) shows that the phase trans­
annealing temperature shows that the thermal­ formation is homogeneous, so that domain tex­
shearing effect is important in both curves but tures do not develop, in contrast with orthoclase.
is more clearly seen in the latter curve. Similar
effects are seen in the familiar plots of rx and y, The phase transformation in KAISi308 and Or­
rx* and y* or the obliquity (rp b 1\ b*) as a
=
rich feldspars
function of annealing temperature and at room
temperature. Salje et al. (1985) showed that the Polymorphism in KAISi308 has been known
true parameters of spontaneous strain are x4 and to exist for over 100 years but has been little
x6, which are approximately cosa* and cosy. understood until comparatively recently. This was
30 W. L. B RO W N A N D I . P A R S O N S

due in part t o difficulties i n characterizing the a!., 1981; Kroll and Knitter, 1985), 2t1 being
various forms until the advent of X-ray precession estimated from cell dimensions. Brackets have also
methods for the study of the lattice geometry of been obtained by the dry heating at atmospheric
microcline (Laves, 1 950), and of transmission pressure of a high sanidine from the Eifel (Hertel­
electron microscopy for the study of the microtex­ mann e t a!. , 1985; Gering, 1985), state of order
tures of orthoclase and microcline (McConnell, being estimated from optics and neutron diffrac­
1965; Nissen, 1967; Eggleton and Buseck, 1 980; tion respectively. Times to reach the latter brack­
McLaren, 1978, 1984; Fitz Gerald and McLaren, eted values during dry heating were very short, a
1982). These studies have shown that the difference few hours at 1050 oc to a few weeks at 750 °C,
between orthoclase and microcline is fundament­ and they are very similar to the times required in
ally one of scale in the microtextures, as postulated hydrothermal runs both for sanidine (Senderov e t
originally by Mallard (1876), coupled, however, a!., 1981; Kroll and Knitter, 1985) and albite
with differences in Al,Si order (Laves, 1 950, 1960; (MacKenzie, 1 957). Ordering rates are thus very
Goldsmith and Laves, 1954). Orthoclase has a similar in both sanidine and albite. Moreover,
'tweed' texture consisting of a very fine-scale evidence that the presence of Ab may affect the
alternation of partially Y-ordered and anti-ordered equilibrium state of order in sanidine (Senderov
domains only a few unit cells thick, which retain et a!., 1981) is inconclusive. Kroll and Knitter
overall monoclinic symmetry to X-rays; these found similar rates for an anorthoclase (Or28)
correspond with the 'left' and 'right' Albite and from the Canary Islands.
Pericline twin domains in the coarser 'tartan' The extremely rapid variation in order in sani­
intergrowth of microcline. Our lack of understand­ dine from the Eifel is possibly exceptional, other
ing was and still is also due in part to the common sanidines showing much less rapid variations
belief (Smith, 1 974; Yund, 1974) that Al,Si order­ (Priess, 1 981; Bertelmann et a!., 1985). Further­
ing in KA1Si308 is very much more sluggish than more, the rates of change were strongly reduced
in albite, whatever the structural state of the former. by pre-annealing at 650 oc for long periods (which
The available experimental data, which we review produced, however, no change in 2Y,;) or for
below, show, in fact, that this is not the case, and shorter times at higher temperatures. Pre-anneal­
that the rates of ordering in both sanidine and ing for even longer times caused loss of the ability
microcline are similar to those in albite. On the to order and caused disordering rates to approach
other hand ordering rates in orthoclase are many those of other sanidines (Bertelmann e t a!., 1985).
orders of magnitude slower, because of the devel­ The Eifel sanidine crystals are structurally very
opment of a domain texture. Unlike albite, how­ nearly perfect, with extremely low dislocation
ever, ordering in KA1Si308 follows a two-step densities (Gering, 1985). The cause of the high
path (Ribbe, 1983), so that there are two (at reactivity is not known at present, but could be
least) independent Al,Si order parameters. We deal related to the high H20 contents of 0.013-0.036
below first with sanidine, then with microcline and wt. % occurring as molecular water in the M site
orthoclase. (Beran, 1 986). Bertelmann e t a!. (1987) showed
that the loss of reactivity starts from the surface
of crystal slabs and proceeds inwards with time.
Ordering states and rates in potassian sanidine They observed the rapid development of large,
irregularly shaped 'inclusions' or pores, mainly in
For K-feldspars, similarly to albite, direct hy­ the (010) plane and associated with dislocations.
drothermal synthesis from gels or glasses leads to At low temperatures condensation of a fluid or
the disordered form, high sanidine, which orders solid (ice) phase could be observed in the pores,
on further annealing, especially in the presence of but most were more or less empty. It is possible
potassium salts (Martin, 1974b; Goldsmith and that these large, optically visible pores result from
Newton, 1974; Senderov et al., 1975, 1 981). As in the exsolution of water from the feldspar, most of
the case of albite, ordering rates decrease as T which is lost to the outside by diffusion along
decreases, but unlike albite, annealing after direct dislocation cores. It is also possible that small
synthesis leads only to a maximum degree of order 'bubbles' of water might be seen by TEM as in
of 2t1 """0.71 at all T below 500 oc (Fig. 3). This quartz (McLaren e t a!., 1983). Ordering rates are
limiting value in hydrothermal experiments is possibly considerably lower for many natural
reached both at high and low pressure (Goldsmith sanidines (see for example Scott e t a!., 1971). This
and Newton, 1 974; Senderov e t a!., 1975). may be because they never were as reactive as the
Above 500 oc the state of order has been roughly Eifel sanidine or more likely that they lost this
bracketed by hydrothermal crystallization or treat­ property on slow cooling, like the pre-annealed
ment of ordered starting materials (Senderov et Eifel sanidine.
 O R D E RI N G R A T E S I N AL K ALI F EL D S P A R S 31

Ordering i n microcline and the low-sanidine­ whereas their main effect is to destroy, at least
microcline transformation partially, the twin-domain texture and hence to
increase the driving force. Tweed orthoclase is in
Direct synthesis of rum-scale single crystals of a state of constrained equilibrium, where the
microcline was achieved by Flehmig (1977) at constraint is the preservation of the domain tex­
room temperature in alkaline aqueous solutions ture. It is frequently described as being stranded
in an attenuated silica-gel matrix in 180 days, and or in metastable equilibrium.
authigenic low microcline occurs in sediments Euler and Hellner (1961 ) claimed to have prod­
(Kastner and Siever, 1979). The presumption is uced triclinic potassium feldspar by hydrothermal
that these microclines do not have monoclinic synthesis, on the basis of optical and powder
precursors, although Flehmig's products exhibited diffraction data, but attempts to confirm this were
twinning believed to be different in type to that unsuccessful (Martin, 1974a). Preliminary TEM
characteristic of cross-hatched microcline. It is studies on low sanidine produced at high pressure
generally agreed that low microcline is stable at by J. R. Goldsmith did not show the presence of
low temperatures, but there is no concensus on tweed orthoclase (unpublished data). It does not,
the stability or otherwise of high and intermediate therefore, seem to be possible to proceed further
microclines. Although the most Or-rich feldspars in ordering experiments using truly monoclinic
which have been synthesized by hydrothermal precursors. Senderov and Yas'kin (1975, 1976)
methods have 2t1 � 0.71, more ordered feldspars studied the disordering of natural low microcline,
are common in igneous and metamorphic rocks. hydrothermally at T between 200 and 450 ac,
They generally have complex twin-domain tex­ together with the ordering of two 'sanidine' samples
tures, which were formed during the symmetry (Fig. 3). Both 'sanidine' samples were obtained
transformation from monoclinic low sanidine to hydrothermally by annealing microcline at 0.05
pseudo-monoclinic orthoclase or triclinic micro­ GPa, one for 64 h at 950 oc and the other for 53 h
cline. Unlike albite, the symmetry-breaking process at 900 oc; corresponding values of 2tb read from
in Or-rich feldspars is ordering, except possibly at the grid of Kroll and Ribbe (1983), are 0.56 and
very high pressures (Hazen, 1976; see however 0.61, respectively. The two samples behaved quite
Angel et al., 1988). It starts by a continuous differently on subsequent hydrothermal treatment.
spinodal-like process and coarsening probably The 2t1 value of the more disordered 950 oc sample
proceeds eventually by a nucleation and (partial) increased only slightly to � 0.66 even after 120
growth mechanism with the production of ordered days at 350 oc. The behaviour of the less-dis­
and anti-ordered domains which are geometrically ordered 900 oc sample was quite different. Run
related to each other by complex twin-like laws products were either microcline (plus kalsilite with
(McConnell, 1965, 1971; McLaren, 1978, 1984). or without residual 'sanidine') or, in one case
Orthoclase has a very fine tweed texture with 2t1 with less concentrated KOH, sanidine alone. The
in the range 0.7 to 0.8 and microcline has coarser highest values reached were 0.89 at 450 oc and
tartan twins with (t10+t1m) in the range 0.75-1. 0.9 15 at 200 oc, and are shown on Fig. 3, along
Once formed, tweed orthoclase, which is a com­ with the results for the two low microcline starting
mon mineral even in slowly cooled rocks, persists materials. The unannealed, less ordered microcline
because of the difficulty of coarsening which re­ samples showed virtually no change at 350 and
quires reversal of the ordering sense in adjacent 450 oc, whereas at 550 and 650 oc, they showed
domains. Eggleton and Buseck (1980) roughly very large changes to values which fall within the
calculated the strain energy involved in the tweed band defined by ordering in sanidine. The more
texture and showed that it would balance or ordered microcline samples showed slight increases
nearly balance the decrease in volume free energy in disorder at 350 and 450 °C. The approximate
obtained by complete ordering. See also Carpenter values of (t10-t1m) can also be deduced from the
(1988). This means that the total free energy change lattice parameters of Senderov and Yas'kin (1975,
for the formation of tweed orthoclase from low 1976) and are given in Fig. 3. Moderately to
sanidine is very small. Thus, the to tal driv ing force strongly triclinic microclines were obtained at T
for ordering is very small in such a domain texture, of 450 oc and below, whereas at 550 and 650 oc
and consequently the rate of formation of ordered the run products were monoclinic. Unfortunately,
domains is very low. The effect of this very low the precision in the lattice parameters is not high.
driving force has led to the widely-held view that Senderov and Yas'kin (1 975) suggested that the
ordering rates in Or-rich feldspars are intrinsically different behaviour of the two 'sanidine' samples
very low, whatever the domain texture. Extrinsic was due to the preservation of triclinic domains
factors such as fluids or deformation are often in the 900 oc sample, but not in the 950 oc sample.
invoked as a means of increasing ordering rates, This seems to be the most probable explanation
32 W. L. B R O W N A ND I . P A R S O N S

600 8oo 1000 T (K) 1200

1.0 ror---��-----.----��-----.----��----.-----��----.----,

0. 5 I
1 -0 95 I Gering 1985
_ __

-- --...LM-...... --.:::_- ....

.9
'f:t Goldsmith and Newton 1974 (1.1-1.9 GPa)

...
0. --... 0 Kroll and Knitter 1985
0. -- 9 D ......_
"""'0.6 \
9 NS::I: Nanev and Swanson 1980 (0.7GPal

-0.7
I -0.7 '-' \ 11-tl Senderov and Vas· kin 1975, 1976

'( Senderov et al. 1975

\\ ']:� Senderov et al. 1981

\1M
I

o
o.6 -

D McDowell 1986 authigenic

0.5�----�-- --�� ----�-- --��----L---��,---�----�� ----�

200 400 600 (DC) 800 1000
T
FIG. 3. Variation of 2t1 as a function of annealing temperature for Or-rich feldspars. Heads of arrows show final
values in all cases, tails show starting values for experiments of Senderov and Yas'kin (1975, 1976) and Senderov
et al. (1981). Double-headed arrows indicate Or-rich feldspars coexisting with Ab-rich ones, and which are therefore
Na-bearing. The three broad arrows indicate changes in 'sanidines' produced initially by hydrothermal treatment
of natural microcline. Solid rectangles indicate samples which did not change with time. The open rectangle is for
a sample recovered from a borehole at 360 oc in the Salton Sea geothermal field. Numbers show approximate values
of (t10-t1m) for triclinic feldspars. LM low, IM intermediate microcline (including high microcline); LS low, HS
high sanidine. Curved bands (widths indicating uncertainty) show supposed equilibrium variation of 2t1 for
KA1Si308 and Or-rich feldspars. Solid circle is approximate position of monoclinic-triclinic transformation.
Horizontal dashed line gives approximate limit of order reached in synthetic Or-rich feldspars because of development
of a domain texture (orthoclase). Compare with Senderov and Y as'kin (1976) who showed a first-order transformation.

and it is supported by a dry-annealing study of The above experiments may be considered to

the disordering of low microcline by Blasi e t al . give reasonable brackets on the value of 2t1 =

(1984); they showed that it occurred by a one-step (t10+t1m) at temperatures of 350, 450 and 550
path and that triclinic symmetry was still detected oc (Fig. 3). When combined with data for ordering
in specimens whose b-e values would imply a 2t1 in sanidine between 650 and 1075 oc, it is possible
value of 0.62. Such triclinic starting materials
� to draw a curve or band taking account of all data
are particularly suitable for approaching equilib­ on Or-rich feldspars, which is sigmoidal in shape
rium in microcline from either side, because there is and very similar to that for albite (Fig. 1 ), but
no domain-texture barrier to overcome. Although with the steep part displaced to lower T. Also
they involve metastable states of order, this is shown is the (t10+t1m) value deduced from 2Vx
unlikely to affect significantly the final states of from an authigenic microcline occurring at 360 oc
order in experiments of long duration. A system­ in a borehole in the Salton Sea geothermal field
atic study of ordering in microcline using such (McDowell, 1986) which is consistent with the
starting materials is highly desirable. experimental data. We conclude that low micro-
 O RD E R I N G R A T E S I N A L K A L I F E LD S P A R S 33

cline i s stable u p to 450 o c and low sanidine above sanidine and highly ordered microcline at � 450
about 550 °C. Only unbracketed experimental data and 500 °C. This difference of interpretation arises,
exist between these two T. Goldsmith and Laves we believe, because of the formation and stranding
(1954) obtained partial conversion of a microcline of tweed orthoclase and confusion between stab­
from Madagascar by hydrothermal treatment at il ity, attainabil ity and preservation of intermediate
� 0.06 GPa in 24 h at 525 oc and no conversion microclines. In the absence of conclusive experi­
at 500 oc in 500 h. They showed that conversion ments between 450 and 550 ac in which the
occurred by solution and reprecipitation. Tomi­ transformation has been accomplished reversibly,
saka (1962) hydrothermally annealed an impure we can draw, like Kroll and Voll, only on the
perthitic low microcline (AnL6Ab29.90r68.5) at evidence of feldspars in rocks. Many studies of
various pressures and showed that �131 was subsolvus igneous rocks, reviewed by Smith (1974),
effectively unchanged at Tup to 450 oc and that
� Parsons (1978a), and Kroll and Ribbe (1983) have
it decreased at T 500 oc. Steady-state values were shown that the Or-rich feldspar is almost always
not reached in times up to 1000 h. either orthoclase or nearly low microcline with a
Goldsmith (1988, and pers. comm. 1987) investi­ high degree of order, or m ixture s of the two, often
gated the 'dry' disordering rates of an authigenic demonstrably within the confines of a single crystal
microcline and a K-exchanged low albite at grain. Intimate mixtures of tweed orthoclase and
750900 °C. As in albite, disordering rates are gre­ tartan microcline have been imaged using TEM
atly enhanced at high pressures and were similar by Fitz Gerald and McLaren (1982), though the
for the two microclines. Goldsmith reported the exact geometrical relationships may differ from
length of time to convert the microclines into straightforward M twinning. Studies of 'trans­
sanidine with a sharp monoclinic pattern, but the formation isograds' in metamorphic regions (e.g.
exact structural state of the product was not Bambauer and Bernotat, 1982; Bernotat and Bam­
known. Below 1.3 GPa at 800 oc, 1.5 GPa at bauer, 1982) also show the common coexistence of
850 oc and up to at least 2.5 GPa at 900 oc, the optically monoclinic feldspar and low microcline.
potassium feldspar disordered considerably more Krause et al. (1986) determined the AI occupancy
slowly than Clear Creek albite and the effect of the T10 site in Peri cline-twinned microcline
is particularly marked below � 1 GPa when lamellae coexisting with orthoclase using electron
disordering rates of the microclines decreased channelling enhanced microanalysis in an electron
sharply. At higher pressures the microcline con­ microscope (ALCHEMI, Tafto and Buseck, 1983).
verted to sanidine slightly more rapidly than low They obtained a value for t10 of approximately
albite disordered. Goldsmith suggested that a 0.56; the corresponding values of (t10+t1m) and
change of diffusion mechanism occurs in K-feld­ (t10-t1m) were not determined but could be in
spar in the region of 1 GPa, perhaps because the ranges 0.8-0.9 and 0.3-0.2 respectively, hence
transient re-coordination of AI in the framework, showing that the lamellae consist of high to
or formation of OH groups, may be inoperative intermediate microcline. The coexisting orthoclase
below that pressure. More data are required on has a value for 2t1 of � 0.88 derived from X-ray
the nature of the transformation, and the textural powder diffraction. This value is high compared
changes, in these experiments. with a value for 2t1 of 0.71 for a Madagascar gem
The monoclinictriclinic phase transformation, orthoclase (Tafto and Buseck, 1983), which was
whatever its nature, must lie between 550 and presumably free of microcline lamellae.
450 oc at low P. There is a large change in 2t1 McLaren and Fitz Gerald (1987) combined
from � 0.65 at 550 oc to � 0.9 at 450 oc and it another new electron-beam technique, conver­
is possible to draw either a smooth curve (Fig. 3) gent beam electron diffraction (CBED), with
or a sharp break (Senderov and Yas'kin, 1976). As ALCHEMI. They showed that for a high sanidine
is the case for albite, there is no compelling reason devoid of microstructure and a low microcline with
to suggest the presence of a sharp break with a coarse, cross-hatched twin microstructure, there was
first-order transformation. By analogy with albite, excellent agreement between 2t1 values obtained
the transformation may be close to tricritical from X-ray diffraction and from ALCHEMI.
(Carpenter, 1988). We thus propose that the degree CBED can be used to obtain the symmetry of very
of order in potassium feldspar varies strongly and small volumes of crystals; in this case, McLaren
continuously with T at temperatures just below and Fitz Gerald obtained patterns from regions
� 500-525 oc near the start of triclinic ordering with a diameter of c. 50 nm in a 'tweed ortho­
at 2t, � 0. 7-0.71 (Fig. 3). The continuous nature of clase' (previously illustrated by Fitz Gerald and
this transformation has also been called in doubt McLaren (1982) and with monoclinic symmetry
by Kroll and Voll (in Ribbe, 1983, pp. 29-30) who from normal selected-area diffraction), corre­
proposed a first-order transformation between low sponding to an average of about 10 modulations
34 W . L. B R O W N A ND I. P A R S O N S

o f the tweed pattern. The symmetry o f these small may coarsen to tartan in response to interactions
volumes was triclinic. ALCHEMI, on regions of with deuteric fluids (e.g. Parsons, 1978a) or to
c. 100 nm diameter, gave 2t1 of 0.67±0.02 (S.D.), deformation (see Smith 1974, vol. 2, p. 384). If
and, at this scale the sample is an intermediate intermediate microcline had no stability field,
microcline. Further work of this type on carefully coarsening of tweed orthoclase under these cir­
selected suites of samples is desirable in order to cumstances would lead directly to low microcline.
understand details of the orthoclase-microcline If on the other hand intermediate microcline has
transformation. a stability field, coarsening should occur by way
of intermediate microcline as in the specimen
studied by Krause et a!. (1986). If ordering rates
0.5
are, as we suggest, rapid, it would not be pre­
served except under special cooling conditions.
In coherent (or semicoherent) intergrowths (see
"E below) the nature of the Or-rich phase depends
:- 0.7 on the bulk composition of the intergrowth (Brown
0
£0.8 and Parsons, 1984b). Brown and Parsons (1984a)
0 discussed the development of different perthite
� 0.9
"'
morphologies and the presence of intermediate
microcline with respect to cooling in strain-con­
trolled coherent cryptoperthites in a compositional
time
range around Ab600r40; their conclusions are
FIG. 4. Schematic variation of the degree of order in Ab summarized in Fig. 5. For intermediate microcline
and Or-rich feldspars as a function of time at a T near to develop, rather rapid cooling rates in a limited
the upper limit for low microcline or during slow cooling range between C and C' are required. Intermediate
in rocks. In albite ordering proceeds to completion, microcline occurs in these coherent intergrowths,
whereas in Or-rich feldspars it is blocked at orthoclase,
rather than tweed orthoclase, because of coherency
unless some 'unzipping' event (such as interaction with
strain imposed by the intergrown Ab-rich phase;
fluids or external stresses) intervenes (circles).
there is no barrier to the formation of microcline.
Orthoclase on the other hand forms only in Or­
rich or in Ab-rich compositions where coherency
On the face of it, the common natural occurrence strain is much less; dislocations develop to relieve
of two coexisting K-feldspars supports the hypo­ coherency stress. For more Or-rich bulk compo­
thesis of a discontinuous transformation. How­ sitions we have at present not even a semi-quanti­
ever, intermediate microcline was first described in tative estimate of cooling rate. Parsons and Brown
Or-rich crystals by MacKenzie (in Adams, 1952) (1984) reviewed the rather fragile petrological
and there have been many subsequent descriptions evidence for the upper temperature limit of the
(e.g. Mergoil-Daniel and Chevalier, 1984; Stewart microcline field; it is in rough agreement with but
and Wright, 1974). It occurs in coherent cryptoper­ at slightly lower T than in Fig. 3.
thites (Brown and Parsons, 1984a, b) and inter­ In conclusion, we believe that the ordering
grown with orthoclase (Krause et a!., 1986). These transformation in KAISi308 is very probably
observations are not easily reconciled with a dis­ continuous in nature and similar in part to that
continuous transformation (cf. discussion on al­ in albite. Carpenter (1988) has suggested that it
bite). We suggest that the comparative rarity of may be tricritical. High-to-low sanidine and high­
intermediate microcline is a result of the relative to-low microcline are stable phases characterized
rapidity of Al,Si ordering, once the barrier of by different degrees of AI,Si order, whereas ortho­
tweed is overcome. This is shown schematically in clase is a thermodynamically metastable (or con­
Fig. 4 in which ordering is compared in albite and strained stable) configuration. It is common in
potassium feldspar during slow cooling. At the nature because it is a kinetically stranded domain
beginning of cooling, potassium feldspar is as­ texture which requires the intervention of appro­
sumed to order only slightly more slowly than priate geological events to overcome the energy
albite, but ordering slows down dramatically near barrier to domain coarsening. Intermediate micro­
2t1 �0.7, because of the reduction in driving clines are rare (as are intermediate albites!) because
force resulting from the build-up of strain energy their temperature range of stability is small and
associated with the twin-domain texture. If, how­ ordering rates very fast geologically; their preser­
ever, this texture can be by-passed, by some process vation may require a particular limited range of
that allows nucleation and growth of microcline, cooling rates in the absence of interactions with
then ordering can again proceed rapidly. Tweed deuteric fluids.
 O RD E RI N G R A T E S I N A L K A L I F E LD S P A R S 35

Solidus
COHERENT INTERGROWTHS

800

Volcanics 4 km diameter
stock, 1 km in
700

600

Spinodal C cl
decomposition \ :
__ M�T Or-rich [Jt!�se
500

L \ �I���\\�����
�����-�--
:
���0�
��
�����t
SANI NE INTERM EDIATE MICROCLINE
I
I
\ I
I
,\=21 45 140nm
400 l

Homogeneous
I
I
I
:
\ \ \ 1 � �!:_'{ LOW MICROCLINE
Straight lamellar IWavy Zig-zag
I
I
!(diagonal association
I 1 - M- twinning )
t yrs
FIG. 5. Continuous cooling-transformation diagram for coherent perthitic intergrowths with bulk compositions
around Ab600r40 showing the limited range of cooling rates suitable for the development of intermediate microcline.
A is the periodicity of regular lamellar cryptoperthitic intergrowths. (Modified from Brown and Parsons, 1 984a).

Order parameters and phase transformations in from Kroll et al. (1980), the symmetry is broken
alkali feldspars by shearing, whereas on BC it is broken by Al,Si
ordering. Neither the position of the singular point
Alkali feldspars at equilibrium are monoclinic at B, where symmetry is broken equally by shearing
high temperatures and triclinic at low tempera­ and ordering nor that of the line BC is known
tures. The symmetry-breaking process is shearing with certainty. The position of BC depends on
in albite (Fig. 2) and Al,Si ordering in potassium the way the Na/K ratio is supposed to affect
feldspar. If the phase transformation is continu­ equilibrium Al,Si ordering, and this is also not
ous, there can be only one line which crosses the well known. Most authors consider that Na,K
Ab-Or-T diagram separating monoclinic from mixing favours Al,Si disorder, so that lines of
triclinic feldspars. As the symmetry can be broken equal order are shown sloping inwards from both
only once, the symmetry-breaking process must sides (MacKenzie and Smith, 1961; Wright, 1964;
change from shearing to ordering at a point on Parsons, 1968; Senderov, e tal., 1981) or an inward­
this line. Fig. 6 shows the proposed constrained sloping binary loop is drawn which implies the
equilibrium relationships in the alkali feldspars, same thing (Wright, 1967; Luth et al., 1974;
where the feldspars are considered to be homo­ Martin, 1974b; Smith, 1974). Laves (1952; 1961,
geneous (cf. Carpenter, 1988). The symmetry­ in discussion to MacKenzie and Smith, 1961)
breaking line is ABC, A being at 980 oc and C suggested that the ratio had little effect on Si,Al
taken at � 500 oc. On the segment AB, taken order; only Zyrianov (1977), Thompson and Hovis
36 W. L . B R O W N A ND I. P A R S O N S

1100 the smaller sodium ion in Or-rich feldspar will

have less effect on framework order than the
{oC) substitution of potassium in Ab-rich feldspars.
HS
Feldspars below the dotted solvus (see next sec­
tion) are metastable relative to exsolution.
.!'n��oclinic The stippled band in Fig. 6 corresponds with
rapid Y ordering with decreasing T. It crosses the
lS line AB and its dashed continuation (Brown et al.,
1 984) at the point B and below; see, however,
Carpenter (1988). Compared to Fig. 2, there is
700 both a thermal and a compositional crossover at
B. From A to B, shearing is dominant, whereas
triclinic from B to C, Y ordering is dominant. The stippled
'
500 '
band in Fig. 6 corresponds to values of Y of
zero (or nearly zero) and �0.9. Finally, possible
framework '....._
shearing variation of Z is shown in Fig. 7, lines of equal Z
'-
300�---=�--��--��--J_--� being horizontal at higher T but sloping inwards
0 20 40 80 100 from the Ab side below 700 oc.
�

mol.
FIG. 6. Diagram showing supposed equilibrium vari­
ation of symmetry-breaking line (ABC) as a function of
T and composition (modified from Smith and Brown, r-------, 1100

1988). The stippled band shows region of rapid Y (oC)

ordering. Along AB symmetry is broken by shearing and
along BC by ordering. At B a thermal and compositional
cross-over occurs at which symmetry is broken equally
by shearing and ordering. MA monalbite, HA high, l A T
intermediate, L A low albite; HS high, L S low sanidine; (K)
IM intermediate, LM low microcline. Dotted curve is
solvus from Fig. 8.

(1979) and Carpenter (1 988) proposed diagrams

in which Ab favoured order in KA!Si308.
Above about 700 oc the Na/K ratio appears to
have little effect on non-convergent Al,Si ordering
(Figs. I and 3; see Parsons, 1968, and Senderov 300 L-------�

et al., 1981), whereas this is not so below 700°C. FIG. 7. Supposed equilibrium variation of Z in alkali
The major departure from horizontal is near feldspars as a function of T. Above line ABC (from Fig.
the Ab end member because significant ordering 6) feldspars are monoclinic. Ordering follows a one-step
occurs at a higher T in albite (due to the displacive ordering path (or a near one-step path) in Ab100 whereas
transformation) than in potassium feldspar. Thus it is two-step for other alkali feldspars.
a stippled band is shown in Fig. 6 which occurs
at about 650-720 oc on the Ab side (intermediate
albite) and slopes down rather steeply towards B Solvus relationships and possible partial and
to become nearly horizontal towards Or (inter­ complete stable equilibrium alkali feldspar phase
mediate microcline) . This downward slope is based diagrams
on a comparison of data which is partly non­
equilibrium for albite at low pressure. At high In this section we discuss briefly the various
pressure it is deduced from data in Goldsmith and types of solvi and the relationship between the
Jenkins (1985) for pure albite and Goldsmith and monoclinic-triclinic transformation line and the
Newton (1974) for potassian albite coexisting solvi. The solvi vary in position with respect to
with a potassium feldspar. The band is shown degree of Al,Si order, occurring at higher T in
horizontal on the Or-rich side of the diagram for more ordered feldspars (Muller, 1 971). They also
want of any convincing evidence to the contrary depend on the nature of the interface between the
(see e.g. Thompson and Hovis, 1979; Carpenter, phases. In coherent intergrowths the framework
1988). It is, moreover, likely that substitution of is continuous and strained by coherency stresses;
 ORDERING RATES I N ALKALI FELDSPARS 37

incoherent phases have separate frameworks and the solvus is truncated by melting if small amounts
are unstrained ('strain-free') i.e. there are no coher­ of water are present. On Fig. 8a we therefore show
ency stresses. The coherent solvus lies at lower T a composite, probable stable-equilibrium solvus
than the corresponding strain-free solvus (see at low P from Brown and Parsons (1984a) con­
Yund and Tullis, 1983, for a review). structed from the solvi obtained in homogeniza­
In principle, if the solvi intersect a phase trans­ tion experiments by Muller (1971), using feldspar
formation line or even a temperature zone rep­ starting materials with known degrees of order.
resenting the rapid onset of Al,Si ordering (Fig. We also show (Fig. Sa, b) the phase transformation
6), there is likely to be some change of slope (e.g. line and a band of rapidly changing Y order from
Laves, 1960; Bambauer et al., 1974). However, Fig. 6. A coherent solvus representing equilibrium
there are serious practical problems in interpreting with respect to Al,Si order has also not been
solvus data. Most studies of the strain-free solvus obtained. We show a coherent solvus on Fig.
have involved data which were not bracketed (see 8b constructed on the basis of the temperature
Parsons, 1978b); refined treatment of such largely difference between the disordered strain-free solvus
invalid data (as carried out by Martin, 1974a, and of Smith and Parsons (1974) and the disordered
Merkel and Blencoe, 1982) cannot locate subtle coherent solvus of Sipling and Yund (1976) near
changes of slope, and only closely reversed data the solvus critical temperature, and the ordered
should be used for this purpose. Even then we strain-free solvus of Bachinski and Muller (1971)
come up against a severe observational problem: and ordered coherent solvus of Yund (1974) on
the Y-ordering band (Fig. 6) intersects the strain­ the limbs. Although not fully satisfactory, we
free solvus near its crest on the Ab-rich limb at consider that these two curves are the best pre­
temperatures around 600 oc (at 0. 1 GPa), and sently available estimates for complete 'strain-free'
above this temperature it is extremely difficult to and coherent thermodynamic equilibrium, which
measure phase compositions because the X-ray take into account both the most reliable experi­
diffractions are poorly resolved (see Smith and mental data and theoretical considerations. A
Parsons, 1974). Special high-resolution techniques direct determination of these curves using reversed
(e.g. synchrotron radiation) might solve this prob­ data points would be a formidable task.
lem, but at present this has not been done. Reversed
experiments at high P (Goldsmith and Newton, Alkali feldspars in magmatic rocks: phase and
1974) do not extend to high enough temperatures behaviour diagrams
to locate any likely intersection (see Parsons,
1978b, Fig. 3). We conclude that no data -set exists Fig. Sa, and b represent two extremes which are
at present which can be used to demonstrate the rarely found in feldspars in rocks, except over
existence or non-existence of changes in slope on restricted T ranges. Fig. 8a is a normal phase
any solvus at the monoclinic-triclinic transform­ diagram which shows supposed-equilibrium
ation, and although we expect that high quality relationships of separate but coexisting feldspars
work could demonstrate such changes, we have whose compositions lie on the strain-free solvus;
drawn smooth curves on the diagrams in the each feldspar is a single crystal which is homo­
following section. Merkel and Blencoe (1982) geneous at every T and acts as a perfect open
gave a complete review of possible relationships system. Alkali exchange occurs only between
between the strain-free solvus and spinodal, and crystals and there are no fine-scale intergrowths.
the symmetry transformation, but the data are not Furthermore, the homogeneous feldspars are sup­
of sufficiently high precision to enable a choice to posed to contain no microstructures resulting from
be made . the monoclinic-triclinic phase transformation, i.e.
they change symmetry essentially as single crystals
The strain-free and coherent solvi for equilibrium
without developing complex twin microtextures.
ordered feldspars
Fig. 8b represents the other extreme. Crystals
act essentially as closed systems, alkali exchange
The many attempted determinations of the occurring only within and not between crystals.
'strain-free' alkali feldspar solvus at low P have Moreover, all feldspars below the heavy line con­
been reviewed by Parsons (1978b). However, all sist of complex twin and/or intergrowth microtex­
these curves are likely to be metastable to a large tures which are fully coherent. Labelled fields
extent with respect to Al,Si order on the Ab-rich within the solvus refer to bulk compositions and
limb, and are certainly so on the K-rich limb at not phase compositions. Fig. 8b is thus not a phase
low T. At high P, stable equilibrium may have but a behaviour diagram. It is possible to combine
been nearly reached on the Ab-rich limb (see parts of these diagrams to show behaviour in
Goldsmith and Newton, 1974), but the crest of common rocks which may be either essentially
38 W . L . B R O W N AND I. P A R S O N S

1100
Fully in coherent

One sanidine

,
�
orthoclas..e - - - ....'
. '
' , ,o t h o c l a s e
low albite + low micro cline anti- / T OT
perthi- ,' E T DT
,
, , p erthltes 3 00
tes 1 ' '
I
Two sepa rate feldspars \ '
( low mlcrocline \r M'
1 mesoperthites • \
1 ) p e r hites
30 0o
��2
��--4L---6
��--8 �---1�
1 00 0 0 0 0 00
mol. % Or
FIG. 8. (a) Diagram showing phase relationships for An-free alkali feldspars under complete (incoherent) equilibrium.
Outside the strain-free solvus (SFS) all feldspars are homogeneous and devoid of any exsolution or twin domain
textures. For bulk compositions inside the strain-free solvus, two phases coexist whose compositions are given by
the solvus at any T and are independent of bulk composition; each feldspar is devoid of domain textures and the
interfaces between them are stress-free (no strain energy). (b) Behaviour diagram showing different constrained
equilibrium states as a function of bulk composition and T for completely coherent intergrowths (below heavy line).
TOT twin-domain microtexture; ETDT exsolution and twin-domain microtexture. Inside the coherent solvus (CS)
the microtextures depend on the proportions of the phases and thus on the bulk compositions (dashed lines are
purely schematic). Low microcline occurs only in mesoperthites whose bulk composition is in the region Ab600r40,
whereas orthoclase occurs in more Ab or more Or-rich bulk compositions. MA, HA, IA, LA, HS, LS, IM and LM
as in Fig. 6.

hypersolvus or subsolvus, depending on rock com­ whose bulk composition lies close to the centre
position and crystal growth conditions. of Fig. 9a, whereas intermediate microcline and
Hypersol vus rocks contain only one alkali feld­ orthoclase occur in more Or-rich bulk compo­
spar which crystallized from a magma with low sitions; this is because of the effect of the coherency
water content (Tuttle, 1 952; Tuttle and Bowen, stresses, which vary with the relative amount of
1958). The compositions of the alkali feldspar in the Ab-rich phase (Brown and Parsons, 1984b).
such rocks are limited to the centre of the diagram Such coherent perthites and mesoperthites are
(e.g. Brown et al., 1983), generally in the range common even in large, slowly cooled hypersolvus
Or20-0r80 (Fig. 9a). Once the alkali feldspar has igneous intrusions.
crystallized from the magma, there is usually little They may be considered to be in constrained
exchange between crystals through diffusion in equilibrium, where total coherency is the imposed
such dry rocks. On further cooling exsolution will constraint. Such microtextures may change only
occur within each crystal below the coherent solvus, if some way is found to overcome coherency.
usually by spinodal decomposition, which gives Parsons (1978a) discussed various ways in which
rise to extremely regular, fully coherent cryptoper­ this may occur and showed that water was the
thitic or microperthitic exsolution microtextures main agent in igneous rocks, though external
(Brown et al., 1983; Brown and Parsons, 1984a, stresses may act in the same way. The coherent
b). Such microtextures occur homogeneously interfaces became 'unzipped' and 'catastrophic'
throughout all the crystals of a given rock sample coarsening may occur irregularly within a rock or
and are independent of defects in the crystal. Each within alkali feldspar crystals during the deuteric
crystal acts as a perfect closed system at this stage, or hydrothermal stage. The very regular coherent
and should continue to do so down to lower and cryptoperthitic microtextures may be locally pre­
lower T, as diffusion slows. The microtextures served (e.g. Moreau et al., 1987) and should be
which develop depend on the bulk composition of carefully looked for even in the most altered rocks.
the crystal, as does the structural state of the Or­ The catastrophic coarsening consists of an abrupt,
rich phase. Low microcline occurs only in crystals often more than thousandfold increase in the scale
 O R D E R I N G R A T E S I N A L K A LI F E L D S P A R S 39

T @ @ Subsolvus rocks
T
Direct crystallization
of two separate
._-,--- f eldspars -�__.

cs sanldlne

JM mesoperthlte
orthoclase
LM mesoperth lte perthlte SFS

Partial deuteric u n m l x l ng Partial deuteric

unmixing crypto- to
LA-LM patch perthite with
relics of coherent mesoperthite macro perthlte

0 20 40 60 80 100 20 40 60 80 100
mol. % Or mol . % Or
FIG. 9. Diagrams showing approximate ranges of feldspar compositions and microtextures in slowly cooled igneous
rocks. Temperatures depend directly on An-content, which raises strain-free and coherent solvi. (a) Hypersolvus
rocks. Only one bulk feldspar occurs. (b) Subsolvus rocks. Two bulk feldspars coexist due to intersection of solidus
and solvus. After crystallization feldspar crystals behave essentially as closed systems and develop complex
microtextures, except at low T where deuteric or hydrothermal alteration has occurred.

of the textures, with very marked phase separation. 0.85 from X-ray powder diffraction (e.g. Guidotti
This is shown schematically at the bottom of et al. 1973; Stewart and Wright, 1974). Ab-rich
,

Fig. 9a. Microprobe analyses of such coarsened plagioclase, developed following coherent exsol­
textures would give two extreme compositions, ution, is generally insufficient in amount to allow
one almost pure albite (Ab ;;:, 98) and the other coarser microcline to develop by coherency stresses
very Or-rich (Or95-98), whereas the clear crypto­ (Brown and Parsons, 1984b). Because orthoclase
perthitic areas, with exsolution textures con­ has a very fine coherent tweed texture, it has no
siderably finer than the probe beam, would give place on a phase diagram for total incoherency
bulk-compositions near the centre (Brown et a!., such as Fig. 8a. Again, as in hypersolvus rocks
1983; Moreau et a!. , 1987). (Fig. 9a), the coherent orthoclase microtexture
Subsolvus rocks contain two separate feldspars may be catastrophically unzipped at low T through
which crystallized directly from the magma, either the action of fluids or external (local) stresses
because of the presence of An which raises the giving rise to coarse, cross-hatched low microcline.
solvus or because of the presence of dissolved H20 Large volumes of intermediate microcline formed
which lowers the solidus (Tuttle, 1952; Tuttle by coarsening of orthoclase within crystals can
and Bowen, 1958). The compositions of the two arise only under unusual conditions where the
feldspars lie towards the Ab and Or ends with a unzipping occurs in the stability field of intermedi­
central portion of very variable width unoccupied ate microcline (Fig. 8a) under conditions in which
by bulk feldspar compositions. In some hypabyssal the unzipping agent (water, or stress) does not
rocks this gap is almost closed (Henderson and persist into the low microcline field. Most Or-rich
Gibb, 1983), but in most typical subsolidus granitic alkali feldspar crystals in subsolvus rocks consist of
suites, the gap is quite wide (Fig. 9b). The two relics of orthoclase with coarsened low microcline
feldspars are usually an Or-rich alkali feldspar and (e.g. McLaren, 1984) giving a typical bimodal
an An-poor plagioclase in the albite or oligoclase distribution in the obliquity index. Coarsening
range. In markedly subsolvus rocks the Or content therefore almost always occurs or continues in the
of the plagioclase may be < 5 mol % and it is low microcline stability field.
generally not perthitic.
The Or- rich feldspar is (perthitic) orthoclase Conclusions
with a very fine tweed texture. It arises during the
monoclinic-triclinic ordering transformation and Although much further research is required
may be preserved, even in slowly cooled igneous to establish fully equilibrium homogeneous and
and metamorphic rocks, unless the coherent micro­ heterogeneous phase relationships in the binary
texture can be bypassed in some way. Such ortho­ alkali feldspars, it is now possible to resolve many
clase may have a value of 2t1 (derived from a of the problems concerning the alkali feldspars.
b-e plot) which exceeds 0 . 7 1 and may reach 0.8- At high temperatures disordered alkali feldspars
40 W . L . B RO W N A N D I . P A R S O N S

are monoclinic. O n slow cooling, they order and to largely incoherent. Exsolution in alkali feldspars
transform to triclinic symmetry. The most signifi­ most frequently occurs by spinodal decomposition
cant difference between Ab and Or-rich feldspars and the resulting fully, or almost fully coherent
lies not in the kinetics of Al,Si ordering as such, but microtextures are independent of defects and very
in the nature of the symmetry-breaking process. regular; this is also the case for tweed orthoclase.
In albite symmetry is broken by a long-range If coherency is disrupted by external factors, the
cooperative shearing of the feldspar framework textures become strongly dependent on defects
(displacive transformation) which is 'instan­ (both original and induced) and nearly complete
taneous' and reversible. There is no barrier to the strain-free relationships may apply. Phase and
symmetry transformation and the relatively coarse behaviour diagrams have been presented which
twin-domain texture which may, but does not can be applied to alkali feldspars in hypersolvus
necessarily, develop, is also not a barrier to order­ and subsolvus rocks and which adequately explain
ing. The high-low order/disorder transformation the occurrence of the natural forms of Or-rich
can be established experimentally and to a large feldspars.
extent reversibly, the only problem being the time
required to reach equilibrium below � 700 oc at Acknowledgements
low to moderate pressures. Intermediate and low
albites are stable below � 700 oc, the former The authors thank Michael A. Carpenter, Julian R.
having a narrow stability range of � 75 ac at low Goldsmith and Ekhard Salje for stimulating discussions
pressure, and a wider one of � 150 oc at 1.8 GPa. and Roger A. Mason for supplying some of his synthetic
albite samples for use in our hydrothermal experiments;
In Or-rich feldspars, the symmetry-breaking
we thank the last three and A. Blasi for critically reading
process is Al,Si ordering, and the symmetry change the typescript. The literature survey forming much of
occurs at a temperature of about 500 oc. Above this work was greatly facilitated by access to J. V. Smith's
that T ordering in sanidine occurs homogeneously, feldspar database, and we are extremely grateful for his
does not break the symmetry and occurs reversibly hospitality and encouragement. We thank the CNRS
and about as rapidly as in albite. Further ordering and NERC for financial support and NATO Scientific
does not occur so readily once the symmetry has Affairs Division for generous travel money.
been broken, because of the very fine-scale domain
texture which develops, the so-called 'tweed' ortho­ References
clase texture. Ordering in tweed orthoclase pro­
ceeds only slowly and partially, not because Adams, L. H. ( 1 952) Annual report of the Director of
ordering rates are intrinsically slow in triclinic the Geophysical Laboratory, Carnegie Institution of

microcline, but because the driving force in tweed Washington, 49- 5 1 .

Angel, R . , Hazen, R . M . , McCormick, T . C., Prewitt,
orthoclase is very slight. This arises because the
C. T., and Smyth, J. R. ( 1 988) Phys. Chem. Minerals,
strain energy in the tweed texture, which is always 15, 3 1 3 - 1 8 .
positive, counter-balances the free-energy re­ Bachinski, S. W . , and Muller, G. ( 1 9 7 1 ) J . Petrol. 12,
duction due to ordering in the triclinic microcline 329-56.
domains. Ordering in microcline devoid of small Bambauer, H-U., and Bernotat, W. H. (1 982) Schweiz.
tweed domains is very rapid and can be achieved Mineral. Petrogr. Mitt. 62, 1 8 5-230.
hydrothermally and reversibly at temperatures �- Kroll, H., Nager, H. E., and Pentinghaus, H. (!974)
down to � 200°C. As in the case of albite, Al,Si Bull. Soc. fr. Mineral. Cristallogr. 97, 3 1 3-45.

ordering in KA1Si308 is continuous and occurs, Beran, A. ( 1 986) Phys. Chem. Minerals, 13, 306 - 1 0.
Bernotat, W. H., and Bambauer, H-U. (1 982) Schweiz.
in the absence of the tweed domain texture, mainly
Mineral. Petrogr. Mitt. 62, 23 1 -44.
over a small temperature range of � 100 oc below Bertelmann, D., Fortsch, E., and Wondratschek, H.
500-525°C. ( 1 9 85) Neues Jahrb. Mineral. Abh. 152, 123-4 1 .
The symmetry-breaking process in alkali feld­ �- Walther, J., and Wondratschek, H. (1 987) Terra
spars changes from shearing in albite to Al,Si Cognita, 7, 257-58.
ordering in Or-rich feldspars; a compositional Blasi, A., Brajkovic, A., and De Pol Blasi, C. (1 984)
crossover between the two (metastable with respect Bull. Mineral. 107, 423-35.
to exsolution) occurs at about Abso- 750rz0- 25. Bowen, N. L., and Tuttle, 0. F. ( 1950) J. Geol. 58, 489-
511.
Homogeneous alkali feldspars intermediate in
Brown, W . L., and Parsons, I . ( 1 984a) Contrib. Mineral.
composition exsolve readily on lowering T and
Petrol. 86, 3 - 1 8 .
more rapidly than they order, because Na,K -- ( 1 984b) Ibid. 86, 335-4 1 .
interdiffusion is much more rapid than Al,Si �- Becker, S . M . , and Parsons, I . (1983) Ibid. 82, 1 3 -
diffusion. As in the case of the tweed orthoclase 25.
texture, strain energy is involved in the exsolution �- Openshaw, R. E., McMillan, P. F., and Henderson,
microtextures, which may vary from fully coherent C. M. B. ( 1984) Am. Mineral. 69, 1058-7 1 .
 O R D E R I NG RATES I N A L K A LI F E L D S P A R S 41

Carpenter, M . A. ( 1988) I n Physical properties and - (1980a) Ibid. 72, 329-33.

thermodynamic behaviour of minerals (Salje, E., ed.). -- ( 1 980b) Mineral. Mag. 43, 905-8.
Reidel, Dordrecht. McConnell, J. D. C. ( 1 965) Phil. Mag. 11, 1289-30 1 .
Eberhard, E. ( 1 967) Schweiz. Mineral. Petrogr. Mitt. 47, -- (1971) Mineral. Mag. 38, 1 -20.
385-98. McDowell, S. D. ( 1986) Ibid. 50, 75-84.
Eggleton, R. A., and Buseck, P. R. ( 1980) Contrib. McKie, D., and McConnell, J. D. C. (1963) Ibid. 33,
Mineral. Petrol. 74, 123-33. 581 -8.
Euler, R., and Hellner, E. (1961) Z. Kristallogr. 1 15, McLaren, A. C. ( 1 978) Chemistry and physics of solids
433-8. and their interfaces. Chemical Soc. London Spec. Rep.
Fitz Gerald, J. D., and McLaren, A. C. ( 1 982) Contrib. 7, 1 -30.
Mineral. Petrol. 80, 2 1 9-29. -- ( 1 984) In Feldspars and feldspathoids (Brown,
Flehmig, W. ( 1 977) Ibid. 65, 1 - 19. W. L., ed.). Reidel, Dordrecht, 373-409.
Franck, E. U. (1981) In Chemistry and geochemistry of -- and Fitz Gerald, J. D. ( 1 987) Phys. Chern. Minerals
solutions at high temperatures and pressures (Rickard, 14, 281 -292.
D. T., and Wickman, F. E., eds.). Pergamon, Oxford, -- Cook, R. F., Hyde, S. T., and Tobin, R. C. ( 1 983)
65-82. 9, 79-94.
Gering, E. (1985) SiliziumfAluminium-Ordnung und Mergoil-Danie1, J., and Chevalier, R. ( 1 984) Bull. Min­
Kristallperfektion von Sanidinen. Unpubl. Doctoral eral. 107 401 - 10.
,

thesis, University of Karlsruhe. Merkel, G. A., and Blencoe, J. G. ( 1 982) In Adv. Phys.
Goldsmith, J. R. ( 1 987) Contrib. Mineral. Petrol. 95, Geochem. (Saxena, S. K., ed.), 234-84.
3 1 1 -2 1 . Moreau, C., Brown, W. L., and Karche, J. P. ( 1 987)
- ( 1 988) J. Geol. 96, 1 09-24. Contrib. Mineral. Petrol. 95, 32-43.
-- and Jenkins, D. M. ( 1 985) Am. Mineral. 70, 9 1 1 - Muller, G. (1971) Ibid. 34, 73-9.
23. Naney, M. T., and Swanson, S. E. ( 1 980) Am. Mineral.
-- and Laves, F . ( 1 954) Geochim. Cosmochim. Acta, 65, 639-53.
5, 1 - 1 9. Nissen, H-U. ( 1 967) Contrib. Mineral. Petrol. 16, 354-
-- and Newton, R. C. ( 1 974) In The feldspars 60.
(MacKenzie, W. S., and Zussman, J., eds.). Man­ Parsons, I. ( 1 968) Mineral. Mag. 36, 1 06 1 -77.
chester Univ. Press. 337-59. - ( 1 978a) Ibid. 42, 1 - 1 7.
Grundy, H. D., and Brown, W. L. ( 1 969) Mineral. Mag. -- ( 1 978b) Phys. Chern. Minerals, 2, 1 99-2 1 3 .
37, 1 56-72. -- and Brown, W . L. ( 1 984) In Feldspars and
-- -- and MacKenzie, W. S. ( 1 967) Ibid. 36, 83-8. feldspathoids (Brown, W. L., ed.). Reidel, Dordrecht,
Guidotti, C. V., Herd, H. H., and Tuttle, C. L. ( 1 973) 3 1 7-7 1 .
Am. Mineral. 58, 705-16. Priess, U. (1981) Neues Jahrb. Mineral. Abh. 141, 1 7-29.
Hazen, R. M . ( 1 976) Science 194, 105-7. Raase, P. (1971) Tschermaks Mineral. Petrogr. Mitt. 16,
Henderson, C. M. B., and Gibb, F. G. F . (1983) Contrib. 1 36-55.
Mineral. Petrol. 84, 355-64. Ribbe, P. H. (1983) Rev. Mineral. 2. Mineral. Soc. Am.,
Kastner, M ., and Siever, R. ( 1 979) Am. J. Sci. 279, 435- 2 1 -55.
79. Salje, E. ( 1 985) Phys. Chern. Minerals 12, 93-8.
Krause, C., Kroll, H., Breit, U., Schmiemann, I., and - ( 1 986) Ibid. 13, 340-6.
Bambauer, H-U. ( 1986) Z. Kristallogr. 174, 1 23-4. -- and Kuscholke, B. ( 1 984) Bull. Mineral. 107, 539.
Kroll, H., and Knitter, R. (1985) Fortschr. Mineral. 63, -- Kuscholke, B., Wruck, B., and Kroll, H. ( 1 985)
127-8. Phys. Chern. Minerals 12, 99-107.
-- and Ribbe, P. H. (1983) Rev. Mineral. 2. Mineral. Schneider, T. R. ( 1957) Z. Kristallogr. 109, 245-7 1 .
Soc. Am., 57-99. Scott, R . B., Bachinski, S . W., Nesbitt, R . W., and Scott,
-- Bambauer, H-U., and Schirmer, U. ( 1980) Am. M. R. (1971) Am. Mineral. 56, 1 208-21 .
Mineral. 65, 1 1 92-2 1 1. Senderov, E. E. ( 1 980) Phys. Chern. Minerals 6,
Laves, F. ( 1 950) J. Geol. 58, 548-71. 251 -68.
- ( 1 952) Ibid. 60, 436-50 and 549-74. -- and Shchekina, T. I . ( 1 976) Geochem. Intern. 13/1,
-- ( 1960) Z. Kristallogr. 113, 265-96. 99- 1 12.
Luth, W. C., Martin, R. F., and Fenn, P. M. ( 1 974) In - and Yas'kin, G. M. ( 1 975) Ibid. 12/3, 1 39-45.
The feldspars (MacKenzie, W. S., and Zussman, J., -- -- ( 1 976) Geokhim. 7, 1038-54. [In Russian]
eds.). Manchester Univ. Press, 279-312. -- -- and Bychkov, A. M. ( 1 975) Geochem. Intern.
MacKenzie, W. S. ( 1957) A m . J . Sci. 255, 48 1 -5 1 6. 12/6, 1 1 6-25.
-- and Smith, J. V. (1961) Inst. 'Lucas Mal/ada' -- Bychkov, A. M ., Lebedev, Y. B., and Dorfman,
Cursillos y Conferencias VIII, 53-69. A. M . (1981) Geochem. Intern. 18/1, 122-34.
Mallard, F. ( 1 876) Ann. Mines 10, 60-196. Sipling, P. J., and Yund, R. A. ( 1 976) Am. Mineral. 61,
Martin, R. F. ( 1974a) Bull. Soc. fr. Mineral. Cristallogr. 897-906.
97, 346-55. -- ( 1 974) Feldspar Minerals, 2 vol. Springer, Heidel­
-- ( 1 974b) In The feldspars (MacKenzie, W. S., and berg.
Zussman, J., eds.). Manchester Univ. Press, 3 1 3-36. - - ( 1 983) Rev. Mineral. 2. Mineral. Soc. Am., 223-39.
Mason, R. A. ( 1 979) Contrib. Mineral. Petrol. 68, -- and Brown, W. L. ( 1 988) Feldspar Minerals Vol. 1
269-73. (2nd ed.). Berlin, Springer Verlag.
42 W . L . B R O W N A N D I. P A R S O N S

Smith, P., and Parsons, I . ( 1 974) Mineral. Mag. 39, Wright, T. L. ( 1 964) Ibid. 49, 7 1 5-35.
747-67. - ( 1 967) Ibid. 52, 1 1 7-36.
Stewart, D. B., and Wright, T. L. (1 974) Bull. Soc. fr. Yund, R. A. ( 1 974) In Geochemical transport and kinetics
Mineral. Cristallogr. 97, 356-77. (Hofmann, A. W., Giletti, B. J., Yoder, H. S. Jr., and
Tafto, J., and Buseck, P. R. ( 1 983) Am. Mineral. 68, Yund, R. A., eds.). Carnegie Inst. Wash. and Academic
944-50. Press, 1 73-83.
Thompson, J. B. Jr. ( 1 969) Ibid. 54, 341 -375; also 55, -- and Tullis, J. ( 1 983) In Feldspar mineralogy (Ribbe,
528-32. P. H., ed.). Rev. Mineral. 2. Mineral. Soc. Am.,
-- and Hovis, G. L. ( 1 979) Trans. Am. Crystallogr. 1 4 1 -76.
Ass. 15, 1 -26. Zyrianov, V. N. ( 1 977) Dokl. Akad. Nauk. SSSR, 233,
Tomisaka, T. ( 1 962) Mineral. J. 3, 261 -8 1 . 1 1 92-95 [in Russian].
Tuttle, 0. F . ( 1 952) J. Geol. 60, 1 07- 1 24.
-- and Bowen, N. L . ( 1 950) Ibid. 58, 572-83.
-- -- ( 1 958) Geol. Soc. Am. Mem. 74, 1 - 1 53.
Winter, J. K., Okamura, F. P., and Ghose, S. ( 1 979) [Manuscript received 3 September 1987;
Am. Mineral. 64, 409-23. revised 29 April 1988]