EARTH SURFACE PROCESSES, VOL.

1, 63-70 (1976)

STABILITY OF MINERALS IN SURFACE

WEATHERING REACTIONS: A GENERAL
THERMOCHEMICAL APPROACH
C . D. CURTIS
Sorby Laboratory, Department of Geology, The University, Shefleld, England

SUMMARY
A list of formational free energy data is provided for a range of common rock forming minerals. Free energy changes
accompanying their decomposition to realistic stable products are calculated and shown to be in agreement with
persistence observations. Weathering mechanisms and environmental controls are discussed.

INTRODUCTION
Loughnan (1969) gives a recent and comprehensive summary of the present state of opinion concerning
the breakdown of minerals in surface weathering reactions. Whereas the mobility (that is the potential
ease of removal of elements in true solution) of different chemical species is fairly well established
and can be linked with fundamental atomic, ionic or molecular properties, the influence of crystal
structure on resistance to weathering is much less clearly understood. It is the purpose of this paper
to consider this particular aspect of Weathering.
One of the most important early contributions to the field was made by Goldich (1938). He docu-
mented the persistence of common rock forming silicates in soils. This turned out to be the reverse
of the sequence of minerals which normally separates during fractional crystallization of silicate melts.
Those minerals stable at the highest temperatures and pressures apparently were least stable in low
temperature and pressure weathering environments.
The minerals associated with Bowen’s reaction series are equilibrium assemblages: successive crystal
crops represent solid-melt equilibria in a decreasing pressure and temperature (also with changing compo-
sition) environmental series. It is not unreasonable to infer from Goldich’s work, therefore, that equilib-
rium (thermochemical) factors play an important role in weathering reactions.
Other workers have attempted to link resistance to weathering with crystal structural properties such
as degree of silicon-oxygen tetrahedral polymerization, bond type and lattice energy as estimated from
element-oxygen bond energies. Notable in this field is the paper by Keller (1954).
A different approach has been to define ‘weathering potential indices’ based simply on the chemical
composition of the mineral in question. Apart from demonstrating that stability depends on chemical
composition (which presumably follows from the general statement of the 2nd law) it is difficult to
see how such an empirical approach can contribute materially to our understanding of the weathering
process.
It seems to me that the conclusions drawn by Loughnan (1969 pp 60-61) from an entensive literature
survey are still valid:
‘From the above discussion it is apparent that the precise role played by the crystal structure in
determining the rate of decay of a specific mineral is still far from understood. Whereas the Goldich
stability sequence for the common rock-forming minerals seems well-established, an entirely satisfactory
explanation has not yet been advanced to account for the similarities between this sequence and Bowen’s
reaction series. Moreover, the apparent correlation between the weathering stability of a silicate material
Received 23 May 1975
@ 1976 by John Wilcy & Song Ltd.

63
64 C. D. CURTIS

and the degree of polymerization of the silica tetrahedra within the mineral structure cannot be extended
beyond the few rock-forming minerals listed by Goldich. Zircon and andalusite, for example, are resistant
to chemical breakdown, yet both are members of the nesosilicates, which, like olivine, do not contain
polymerized tetrahedra. Again, the marked disparity in weathering stability between albite and anorthite,
both of which are tectosilicates, and between the phyllosilicates, muscovite and biotite, suggests this
apparent correlation is somewhat fortuitous’.

ENERGY CHANGES IN WEATHERING REACTIONS

Certain developments have occurred in other fields of geochemical research since the writing of Lough-
nan’s text that suggest another approach to structural stability in weathering. In particular much accurate

Table I. Estimates of mineral standard free energy of formation values

Mineral Formula AG;* Reference
1. Forsterite Mg,SiO, -491.9 1
2. Clinoenstatite MgSi0, - 394.4 1
3. Diopside CaMg(SiO,), - 752.8 1
4. Anthophyllite Mg,Si,Ozz(OH)2 -2716.3 3
5. Tremolite a 2 M g 5Si802 - 2779.1 1
6. Anorthite CaAl,Si,O, - 955.6 1
7. Low Albite NaAISi,O, - 884.0 1
8. Microcline KAlSi308 - 892.8 1
9. Muscovite KA1,Si301,(OH), - 1330.0 2
10. Fayalite FezSiO - 329.7 1
11. ‘Ferrous Silicate’ FeSi0, - 257.0 4
12. Laumontite CaA1,Si40,, .4H20 - 1598.0 2
13. Prehnite Ca2A12Si3010(0H)2 - 1389-8 2
14. Wairakite CaAlzSi40,,.2H,0 - 1477.3 2
15. Zoisite Ca2AI3Si,O zOH - 1552.7 2
16. Wollastonite CaSiO, - 370.3 1
17. Grossular Ca3A12Si30I , - 1501.0 1
18. Spinel Ma12°4 - 523.0 1
19. Jadeite NaAI(SiO,), - 677.2 1
20. Lawsonite CaA1zSiz07(0H)2.H,O - 1076.9 1
21. Kyanite AI,SiO, - 577-0 2
22. Clinochiore Mg5A1$i3010(OH)8 - 1974.0 2
23. Pyrite FeS, - 38.3 1
24. Methane CH; - 121 1
* units Kcal mol-

Table I and Table I1 References

1. Robie R. A. and Waldbaum D. R. (1968). ‘Thermodynamic properties of minerals and related substances at 298.15”K (250°C)
and one atmosphere (1.013 bars) pressure and at higher temperature U. S.’, Geol. Suru. Bull. 1259, 256p.
2. Zen E-An (1972). ‘Gibbs free energy, enthalpy, and entropy of ten rock-forming minerals: calculations, discrepancies, implications’,
Amer. Mineral, 57, 524-553.
3. Zen E-An (1971). ‘Comments on the thermodynamic constants and hydrothermal stability relations of anthophyllite’, Amer.
J . Sci. 270, ‘ 13&’150.
4. Latimer W. M. (1952). Oxidation mtentials. 2nd Edn. New York. Prentice-Hall.
5. Rossini F. D., Wagman D. D., Evans W. H.,’Levine S.’and Jaffe 1.’(1952). ‘Selected values of chemical thermodynamic properties’,
Natl. Bur. Standards Circ. 500, U.S. Dept Commerce.
6. Reesman A. L. and Keller W. D. (1968). ‘Aqueous solubility studies of high-alumina and clay minerals’, Amer. Mineral, 53,
929-942.
 STABILITY OF MINERALS 65
Table 11. Formational free energy values for potential reactants 'and products stable
in the weathering ('average') environment

Phase Formula AGP (Kcal mol-') Reference

1. Hydrogen Ion Zero 0

2. Sodium Ion - 62.5 1
3. Potassium Ion - 67.7 1
4. Magnesium Ion - 108.9 1
5. Calcium Ion - 132.2 1
6. Bicarbonate Ion - 140.3 1
7. Sulphate Ion - 177.3 5
8. Oxygen Gas Zero 0
9. Water Liquid - 56.7 1
10. Hematite - 177.7 1
11. Quartz - 204.7 1
12. Gibbsite - 550.3 6
13. Kaolinite - 904.0 6

For References, see Table I.

information concerning the chemical composition of precipitate, soil, surface and groundwaters has
been gathered together. This has been utilized to give a clearer picture of migration of chemical elements
in the surface geochemical cycle (Garrels and Mackenzie (1971)). Consequently it is possible to write
realistic overall equations for weathering reactions including reactant and product aqueous solutions.
Experimental work at high temperatures and pressures in different laboratories has enabled compila-
tion of free energy of formation values for a considerable list of rock-forming and accessory minerals.
Simply by combining these two sources of information it clearly should be possible to calculate
reasonably accurately the net change in free energy associated with weathering reactions. It is particularly
fortunate that the standard state for expressing formational free energy values (25.0"Cand 1 atmosphere
total pressure) approximates so closely to 'average' weathering conditions.
Table I lists free energy of formation values readily available in the literature for a number of minerals.
Accuracy of the data is difficult to assess and depends to some extent on the actual material studied
and the experimental procedure utilized. The reader is referred to an excellent discussion of these matters
in E-An Zen (1972). While absolute accuracy must remain in doubt it is possible to select sets of
data that are internally consistent. Where this is done (a somewhat subjective procedure) the resulting
estimates of reaction energies will also be internally consistent and such a condition is all that is required
for the present purpose.
Table I1 lists free energy of formation values for reactants and products (solid, liquid and gaseous)
thought to be stable in a range of weathering environments. The fact that amorphous silica rather
than quartz may form in soil reactions does not affect the calculations which are only intended to
demonstrate trends. Quartz is the ultimately stable phase in many situations. For ions or neutral solvated
'
species in aqueous solution the standard state concentrations is 1 mol kg- (ideal solution).
It is appreciated that solution concentration and gas partial pressures vary in the natural environment
but adoption of these standard states facilitates and is sufficient for the purpose of comparison of
different mineral stabilities under a given set of environmental conditions. Testing the response of a
given mineral (or assemblage) to different weathering environments is possible and will be the subject
of a separate report.

WEATHERING EQUATIONS
The standard free energy change of a reaction is the sum of free energies of formation of all the
reaction products (in their standard states) minus the sum of the free energies of the reactants (also
standard states).
AG: = AG: (products) - AG;! (reactants).

E.S.P. 1/1-E
66 C. D. CURTIS

Consider the reaction of fayalite with oxygen to give hematite plus quartz:
Fe,SiO,(s) + f02(g) = Fe203(s) + Si02(s)
(equation 10, Table 111: (s) and (g) refer to the physical state of the substance in question-solid and
gas respectively)
+
AG: = AGg (Fe203) AG; (SiO,) - AGF (Fe2Si04)- AG? (40,)
This is the amount of energy released on weathering of fayalite to quartz plus hematite. For anorthite
weathering it would be reasonable to write

CaA12Si20&) + 2H+(aq)+ H20(1) = A12Si20,(OH),(s) + Ca"(aq)

where (1) refers to the liquid state and (aq) to hydrated species in aqueous solution--equation 6, Table
111. This reaction includes hydration and exchange of hydrogen ions for metal cations. The most common
source of hydrogen ions is dissolution of carbon dioxide;
H20(1)+ C02(g)$H2C03(aq)+H+(aq)+ HCO;(aq)
This can be taken account of by adding bicarbonate ions to both sides of most of the equations in
Table 111. This, of course, in no way affects the energy calculations!
The first nine minerals in Table IV were placed together to facilitate comparison with Goldich's
stability series on the one hand and Bowen's series (both continuous and discontinuous) on the other.
The absence of thermochemical data for realistic amphiboles or any biotite is regretted. Overall, musco-
vite and then potassium feldspar are seen to be least unstable (in energy terms) followed by sodic
plagioclase. Calcic plagioclase is seen to be less stable still. The overall ranking is reasonable since
muscovite can persist almost indefinitely with but slight modification whereas olivines, pyroxenes and
amphiboles rarely survive for long in the weathering environment. The calculations take account of
structural and compositional factors that separately, were the basis of earlier approaches.

Table 111. Weathering equations written with 'primary' minerals as reactants

1. +
Mg2Si04(s) 4H+(aq)= 2Mg2+(aq) 2H20(1) Si02(s)+ +
2. + +
MgSiO,(s) 2H+(aq) = Mg2+(aq) H20(1) SiO,(s) +
3. + + +
CaMg(Si03)2(s) 4H+(aq) = Mg2+(aq) Ca2+(aq) 2H20(1) 2Si02(s) +
4. Mg,Si,O,,(OH),(s) + +
14H+(aq)= 7Mg2+(aq) 8H20(l) 8Si02(s) +
5. Ca2Mg,Si8022(OH)2(s)+ 14H+(aq)= 5Mg2+(aq)+ 2Ca2+(aq)+ 8H,0(1) + 8Si02(s)
+ +
6. CaA1,Si208(s) 2H+(aq) H20(1) = A1,Si20,(OH)4(s) Ca2+(aq) +
+ + + +
7. 2NaA1Si,08(s) 2H+(aq) H,0(1) = Al2SiZO,(OH),(s) 4 SiO,(s) 2Na+(aq)
+ + + +
8. 2KA1Si308(s) 2H+(aq) H20(1) = AI2Si,O5(OH),(s) 4SiO,(s) 2Ka+(aq)
9. 2KA13Si,0,,(OH),(s) + + +
2H+(aq) 3H20(1) = 2K+(aq) 3AI,Si20,(OH),(s)
+
10. Fe2Si04(s) f02(g) = Fe203(s) Si02(s) +
+
11. 2FeSi03(s) jO,(g) = Fe203(s) 2SiO,(s)+
+ +
12. CaA1,Si4Ol2.4H20(s) 2H+(aq) = CaZ+(aq) A12Si,05(OH)4(s) 2SiOz(s) 3H20(1)+ +
13. Ca2AlzSi3010(OH)2(s) + 4H+(aq) = 2Ca2+(aq)+ A12Si205(OH)4(s)+ Si02(s) H20(1) +
+ + +
14. CaAl2Si4OI2.2Hz(s) 2H+(aq)= Ca2+(aq) A12Si20,(OH)4(s) 2Si02(s) H20(1) +
+ + +
15. 2CazA13Si30120H(s) 8H+(aq) H20(1) = 4Ca2+(aq) 3Al2Si2O5(OH),(s)
+ +
16. CaSi03(s) 2H+(aq) = Ca2+(aq) H20(1) Si02(s) +
+ +
17. Ca3A12Si,0,z(s) 6H+(aq) = 3Ca2+(aq)+ A12Si20,(OM),(s) H20(1) Si02(s) +
18. MgAl,O,(s) +
2H+(aq) + 2H20(1) = AI20,.3H2O(s) + Mg2+(aq)
+
19. 2NaA1Si20,(s) 2H+(aq)+ H20(1) = A12Si,0,(OH)4(s) + Si02(s) 2Na+(aq) +
+ +
20. CaA12Si,0,(OH)2 .H20(s) 2H+(aq) = Caz+(aq) A12Si205(OH),(s) H20(1) +
+ +
21. 2A1,Si05(s) 5H20(1) = A12Si205(OH),(s) A1203.3HzO(s)
22. Mg,A12Si3010(OH)8(s)+ 10H+(aq)= 5MgZ+(aq)+ A1,Si20,(OH)4(s) + SiOz(s) 7H20(1) +
+
23. 2FeS2(s) + 4H20(l) 7+o,(g) = Fe20&) = 4SO:-(aq) 8H+(aq) +
+
24. CH4(g) 202(g) = H,0(1) +
H+(aq) HCO;(aq) +
 STABILITY OF MINERALS 67

Energy values calculated for breakdown of iron-rich silicates, are particularly interesting (equations
10 and 11, Table 111). The free energy liberated by oxidation of ferrous to ferric iron swamps that
derived from molecular rearrangements. Such a result could not have been anticipated from an analysis
of structure alone but only by considering reactions. Rapid weathering of almost any mineral rich
in ferrous iron is a familar phenomenon. The behavioural discrepancy between muscovite and biotite
is readily explained on this basis.
The 'special position' of midation reactions in weathering may be tested rather easily by considering
the weathering of pyrite; a mineral stable at low pressures and temperatures and in the presence of
water. The AG; associated with equation 23 (Table 111) is - 17.68 kcal g atom- (Table IV). This is
of the same order as hydrocarbon combustion (equation 24), nor is this unrealistic; pyritic shales weather
readily in oxygen-rich environments. Spontaneous combustion of shale tips associated with coal mining
is not uncommon. Pyrite-rich shales frequently occur together with coal seams.

Table IV. Gibbs free energy values for weathering reactions

Mineral G:, Kcal mol- I GP, Kcal g atom-

1. Forsterite - 44.0 Magnesian
2. Clinoenstatite - 20.9 - 2.98
3. Diopside - 38.1 - 2.72 silicates with
4. Anthophyllite - 137.2 'magmatic
- 2.49
5. Tremolite - 123.2 - 2.24 affinity'
6. Anorthite - 23.9 - 1.32
7. Low Albite - 23.1 -0.75 Feldspars
8. Microcline - 15.9 -0.51
9. Muscovite - 17.3 -0.32 Muscovite
Ferroan silicates
10. Fayalite - 52.7 -6.58
with 'magmatic
11. 'Ferrous Silicate' - 73.1 -6.64
affinity'
12. Laumontite - 17.7 -053
13. Prehnite -40.0
14. Wairakite - 25.0 -0 9 2 Zeolites
15. Zoisite - 78.7 - 1.43
16. Wollastonite - 23.3 - 3.32
17. Grossular -61.0 -2'34
18. Spinel - 22.8 - 1.52 Various
19. Jadeite - 27.3 'metamorphic'
20. Lawsonite - 16.0 -0.76 minerals
21. Kyanite - 16.8 -0.54
22. Clinochlore - 76.1 - 1.65
23. Pyrite - 583.5 Highly reduced
24. Methane - 184.9 - 20.54 phases

DISCUSSION
The first essential observation is that, without exception, the calculated free energy changes are negative.
Thus far, at least, the analysis is sensible for this result implies that all the reactions listed would
proceed spontaneously as written in the natural environment. A zero value is the condition for equilib-
rium of reactants with products whilst a positive result would imply 'reverse weathering'. Intuitive
reasoning suggests further that the greater the negative value of AG:, the greater should be the tendency
to react and this should be reflected in instability.
The free energy changes for overall reaction (Table IV,column 3) show a considerable range. Clearly,
however, these numbers are affected by the sizes of molecular formulae as written. Chemical formulae
68 C. D. CURTIS

and equations express only the relative numbers of atoms within them and may be multiplied or divided
by any factor. Comparisons between different chemical equations are therefore difficult. The simplest
(and perhaps only) way of correcting is to rephrase as gram atom values (i.e., dividing by the number
of product atoms). In this way the comparison is between different amounts of energy, liberated for
a given number of product atoms. If weathering is to be linked with energy liberation, low negative
values of AG: indicate relative stability whereas high negative values suggest rapid breakdown in the
wet, oxygenic and low temperature and pressure conditions inherent in the calculations. It would appear
from Table IV that this is so: indeed there is a very fair degree of correlation between energy and
persistence in weathering.
A variety of iron-poor metamorphic minerals is listed in Tables I, 111 and IV items 12 to 22. AGY
values lie within the same range as those for 'magmatic affinity' minerals. Less information is available
as to the relative stability of such minerals in any given soil regime. Predicted stabilities, however,
do seem to be reasonable.
Finally, it must be stressed that these reactions all involve complete oxidation such as would be
anticipated in well-drained yet wet soil situations. The persistence of minerals containing Fe2+, Mn2+,
Cu', S2-etc in waterlogged soils or in sedimentary environments would be expected to be much greater
since oxidation would not then occur. Specific environmental situations such as these may be investigated
by writing appropriate equations and then calculating energy changes.

KINETIC AND MECHANISTIC CONSIDERATIONS

The net energy change associated with a chemical reaction comes from molecular-scale rearrangement
to new structures (i.e., breaking of bonds followed by formation of new ones in different configurations)
which are more stable in the temperature, pressure and compositional conditions of the reaction environ-
ment. Although the ultimate criterion of stability of a chemical compound relative to possible decomposi-
tion products is the sign of the net energy change involved, the rate of decomposition of unstable
species is not generally linked quantitatively with the magnitude of the energy change. Such a relationship
exists between reaction rate and the Activation Energy of reaction. This Activation Energy relates to
a specific mechanism. Many chemical reactions can proceed via more than one mechanism. Catalysts
induce acceleration of reactions by providing an alternative reaction mechanism with lower Activation
Energy. The net energy change of a reaction however, remains independent of mechanism.
This paper has revealed a general relationship between the rate of different weathering reactions
and their free energy changes. The Appendix is devoted to further analysis of this slightly surprising
result.

CONCLUSION
The breakdown of rock-forming minerals in soil profiles liberates many chemical species into aqueous
solution. The nature and rate of evolution of those soluble species largely determine the actual and
potential fertility of soils. Whilst some minerals react to yield essential nutrients and buffer soil solutions
to narrow and acceptable pH ranges, others produce acids and heavy metals in abundance with disas-
trous consequences. A quantitative understanding of the interaction between minerals and their aqueous
environment obviously would be of benefit to many.
This paper has outlined one approach to the problem. It has demonstrated that persistence of a
particular mineral in weathering correlates well with the total energy released by its breakdown into
realistic weathering products. The importance of oxidation/reduction reactions is well documented in
these calculations.
Unlike most earlier approaches, the chemical environment is considered as well as the solid material
reacting. Although not demonstrated the same approach can be used to compare and contrast different
weathering environments in terms of their reactivity-with respect to mineral breakdown.
 STABILITY OF MINERALS 69

Finally, attention has been drawn to the fact that in many chemical systems, rates of reaction are
not directly related to net free energy changes. The reason why this appears to be the case in weathering
systems may possibly be due to the fact that an essentially similar reaction mechanism is operative
in many different mineral breakdown reactions (see Appendix). A more general account of energy rela-
tionships is given in Curtis (1975).
APPENDIX
Possible limits on reaction rates
Chemical reactions usually proceed in at least two steps. The overall rate is decided by the slowest
of these (the ‘rate determining step’). It is reasonable to write a carbonation reaction as two steps:

M
M-%silicate + pf + \0’-silicate
/
- BE kcal .mol- I

H’
followed by
M
\
O+-silicate --+ H-0-silicate (clay) + M+(aq)+ (AGr + AE kcal .mol- l )
/
H
where M is a metal.
The first of these reactions demands in input of AEA (the activation energy) which may or may
not be rate determining. The second reaction results in stable products. It is the net energy change
over both stages AG, that was considered earlier.
It is possible that the AGr term could influence the physical state of the stable products. The free
energy of formation of any solid (i.e. all those listed in Tables I and 11) is a function of its state of
subdivision since the structure and hence free energy of the surface phase must be different from that
of the bulk phase. The effects on total AG, are slight, however, unless the phase is very finely subdivided
(large surface area). This is clearly illustrated by Schindler’s (1967) article on heterogeneous equilibria.
In Figures 7 and 8, solubility constants for Cu(OH), and CuO are plotted as a function of molar
+
surface, s. Straight line relationships of the type log *Kso = A B . s are found where *Kso is the
solubility constant in question, A and B being constants.
In a chemical reaction in which a great deal of energy is liberated therefore, the potential state
of subdivision of the products is much greater than for a reaction in which little energy is released.
Solubility constant functions such as those described by Schindler could be used to predict the maximum
state of subdivision.
Returning to the problem of weathering reaction rates, highly energetic reactions could yield solid
products in a highly subdivided or even amorphous form whereas low energy reaction products would
be physically dense, perhaps crystalline and perhaps ‘welded’ to the reactant solid. The former ‘material
would allow free diffusive passage of reactants and products in true solution whereas dense solid products
would act as a barrier to their immigration. Under these conditions the overall rate of reaction would
be diffusion rate controlled. Experimental work has shown that kinetics of this type are common for
heterogeneous reaction. Alternatively, the solid products may be regarded as inhibiting the carbonation
reaction. Clearly dense solid products would inhibit the hydrolysis reaction to a greater extent than
would open structured products. For a simple but clear account of these factors, see Stevens (1965).
In this way a plausible link can be drawn between reaction energetics and rates. One encouraging
aspect of treatment is that it requires heterogeneous reaction: weathering reactions obviously are hetero-
geneous.
70 C. D. CURTIS

ACKNOWLEDGEMENTS

Dr. D. A. Spears took part in general discussions of aqueous geochemistry at Sheffield over a number
of years that led, albeit indirectly, to this paper.* Professor W. S. Fyfe F.R.S. some time ago drew
my attention to a possible link between products physical state and energy transfer during reactions.
These contributions are gratefully acknowledged but no blame should be attached for unwarranted
extrapolation of ideas!

REFERENCES
Curtis, C. D. (1975). ‘Chemistry of rock weathering fundamental reactions and controls’, in Geomorphology and Climate, Ed.
E. Derbyshire, London, Wiley.
Garrels, R. M. and Mackenzie, F. T. (1971). Eoolution of sedimentary rocks, New York, W. W. Norton.
Goldich, S. S. (1938). ‘A study in rock weathering’, J. Geol. 46, 17-58.
Keller, W. D. (1954). ‘Bonding energies of some silicate minerals’, Amer. Mineral, 39. 783-793.
Loughnan, F. C. (1969). Chemical Weathering of Silicate Minerals, New York, American Elsevier.
Schindler, P. W. (1967). ‘Heterogeneous equilibria involving oxides, hydroxides, carbonates and hydroxide carbonates’, in Equilib-
rium concepts in natural water systems, American chemical society, advances in chemistry series No. 67, 196221.
Stevens, B. (1965). Chemical Kinetics, London, Chapman and Hall.
Zen, E-An (1972). ‘Gibbs free energy, enthalpy and entropy of ten rock-forming minerals; calculations, discrepancies, implications’,
Amer. Mineral, 57, 524-533.

* Miss S. Wilgose is thanked for her help in typing the manuscript.