American Mineralogist, Volume 90, pages 10331041, 2005

A synthesis of systematic mineralogy

L. BRUCE RAILSBACK
Department of Geology, University of Georgia, Athens, Georgia 30602-2501, U.S.A.

ABSTRACT
This paper presents several new diagrams that document previously unappreciated patterns and predictability in systematic mineralogy. They illustrate, for example, that sulde minerals of single cations contain only intermediate to soft cations (cations with at least some outer-shell electrons). On the other hand, hard cations (those with no outer-shell electrons) only enter into a few sulde minerals, all of which also contain intermediate cations. Among the simple chloride, uoride, oxide, and carbide minerals of hard cations, ionic potential (charge/radius) of cations is proportional to that of anions (e.g., from KCl to SiC). Among the simple nitrate, sulfate, carbonate, phosphate, and silicate minerals of hard cations, ionic potential of inter-radical cations is inversely proportional to the ionic potential of the cations in the oxysalt radical (e.g., from KNO3 to ZrSiO4). Across both of these sets of minerals, non-simple minerals (H2O- and OH--bearing minerals or minerals with extra O2) in each group typically involve cations of higher ionic potential than those of the simple minerals. The result is predictability and pattern in the compositions of minerals.

INTRODUCTION
Systematic mineralogy is conventionally treated as a listing of minerals in chemical groups dened by anions or anionic groups (e.g., suldes, oxides, carbonates, etc.). Both university textbooks (e.g., Klein 2002; Nesse 2000) and advanced reference works (e.g., Clark 1993; Gaines et al. 1997) provide little explanation of variation within these groups, except among the silicates, and little attempt is made to show relationships between these different groups. This paper attempts to rationalize systematic mineralogy by showing patterns and relationships in mineral compositions across the entire range of systematic mineralogy. For example, it answers such questions as Why are there CaF2, CaO, CaSO4, and CaCO3 minerals, but no CaCl2, Ca2C, or Ca(NO3)2 minerals? The results are a series of diagrams that contextualize the chemical formulae of minerals.

FUNDAMENTAL PRINCIPLES
Three fundamental concepts are essential to understanding this paper. First, cations and anions can be divided into categories ranging from hard to soft (Pearson 1968; Stumm and Morgan 1996) (Fig. 1). Hard cations have no outer-shell electrons, and so their outer shells are like those of a noble gas. Na1+, Mg2+, and Al3+ are good examples of hard cations. Intermediate cations have some outer-shell electrons, and Fe2+ and Fe3+ are good examples. Soft cations have many outer-shell electrons and are typically large; Au1+ and Hg1+ are good examples. Hard anions are those in the second row of the periodic table with full outer shells (F1, O2, N3, and C4), whereas anions become progressively softer or more deformable downward through the table from S2 to Bi2 (Fig. 1). The signicance of this scheme is that hard * E-mail: rlsbk@gly.uga.edu
0003-004X/05/00071033$05.00/DOI: 10.2138/am.2005.1842

cations coordinate well with hard anions, whereas soft cations coordinate more extensively with soft anions (Pearson 1968; Stumm and Morgan 1996). Second, both cations and anions can be characterized by their ionic potential. Ionic potential is dened as ionic charge divided by ionic radius, or z/r, and it is a measure of the density of charge of an ion and thus of the strength of the ionic bonds made by that ion. On the periodic table, ionic potential increases across the hard cations from less than 1 for K1+ to more than 40 for N5+ (Fig. 1). Across the intermediate to soft cations, it increases from less than 1 for Au1+ to more than 10 for Se6+ (Fig. 1). In contrast, ionic potential varies across the anions only from 0.5 for I1 to 1.5 for C4 (Fig. 1). Ionic radius for any one ion varies with coordination number [e.g., r = 1.00 for Ca2+ in sixfold coordination vs. 1.06 in eightfold coordination, in the data from Shannon (1972) used here]. However, the effect on ionic potential (2.00 vs. 1.89 in this example) is small relative to the large range of ionic potential across the various ions. Ionic potential has long been used to characterize cations (e.g., Cartledge 1928a, 1928b; Goldschmidt 1937), but this paper will show that ionic potential of both cations and anions is also useful in understanding trends in mineralogy. Both of these concepts are incorporated in the periodic table used in this paper, which is a simplied version of the Earth Scientists Periodic Table of the Elements and Their Ions (Railsback 2003, 2004). This new table arranges chemical entities by charge, rather than by elemental (uncharged) condition, and as a corollary it shows some elements multiple times (e.g., sulfur as S6+, S4+, S0, and S2 from left to right across the table). This allows the table to separate the hard and intermediate-to-soft cations graphically and to use contours across the table showing variation in ionic potential of both cations and anions (Fig. 1). Third, all denitions of a mineral include the stipulation that minerals be naturally occurring (Klein 2002; Nesse 2000) or have formed as the result of geological processes (Nickel

1033

1034

RAILSBACK: A SYNTHESIS OF SYSTEMATIC MINERALOGY

1995). Occurrence in nature requires that, for a substance to be a mineral, it must have a melting temperature above at least the coldest earth-surface conditions, it must be sufciently insoluble to not dissolve after condensation of atmospheric water vapor, and it must be sufciently non-reactive that it survives long enough to be sampled and identied. Minerals are thus, among all the inorganic chemicals that can be synthesized, the subset that have sufcient chemical stability to survive in nature. The goal of this paper is to show that existence within this subset can be predicted in terms of hardness and softness of ions and in terms of ionic potential.

THE -IDE MINERALS

One of the most fundamental observations about systematic mineralogy is that cations forming suldes, bromides, tellurides, iodides, and bismuthides are almost entirely intermediate to soft cations, rather than hard cations (Fig. 1). Of the 56 sulde minerals of single cations, all contain intermediate or soft cations rather than hard cations (Fig. 1). Of all 136 sulde minerals listed by Clark (1993) and Mandarino and Black (2004), only 13

contain any hard cations at all, and then typically only as minor constituents of rare minerals, as in rasvumite (KFe2S3). Among the intermediate to soft cations, those of lesser ionic potential, such as Ag1+, Au1+, and Hg1+, coordinate with softer anions like Br1, I1, and Bi2. Thus the softness of the intermediate to soft cations is generally correlative with the softness of the anions with which they combine to form minerals (Fig. 1). Another reection of the propensity of the cations that bond with soft anions to not bond to hard anions such as 02 is that only 4 of the 350 suldes, selenides, tellurides, arsenides, and bismuthides listed by Clark (1993) and Mandarino and Black (2004) are hydrous, and none contain F1. In contrast to the behavior of intermediate to soft cations, hard cations form minerals almost exclusively with the hard anions high in the periodic table (Cl1, F1, O2, and C4, or with the O2 of oxysalt radicals) (Fig. 1). Cl1, intermediate between the hard and soft anions, combines to make minerals with cations of all sorts, from hard cations like Na1+ and K1+ in halite and sylvite to intermediate cations like Mn2+ in scacchite to soft cations like Hg1+ in calomel (Fig. 1).

Cations and anion that form simple* carbides

Chemistry of Simple Minerals

* "simple" denotes minerals of one cation without H2O, OH-, or other anions.

Cations and anion that form simple* oxides Cations and anion that form simple* fluorides Cations and anion that form simple* chlorides Cations and anion that form simple* sulfides Cations and anions that form simple* bromides or iodides Cations and anion that form simple* bismuthides Cations that form oxysalt anions (e.g. S6+ as sulfate, SO42-) Contours of equal ionic potential (charge radius, or z/r)

Hard Anions
z / r = -1

C z / r = -2 Si

6 N

7 O

8 F

Trajectory of Figs. 4 & 5

14 P 3 15 S 2 16 Cl

17

As

33 Se

34 Br

35 z/ r = 1/ 2

Sb

51 Te

52 I

53

Hard Cations
(Cations with no outer-shell electrons)
z / r = -1
ory of Fig. 6

Bi

83

At

85

z/ r = 4 z/ r = 2

Li

3 Be2+

4 B 3+

5 C 4+

6 N 5+

ect Tr a j

z / r =16

z / r = 1/ 2

z / r = 32 Na z/ r = 1
3+ +

Intermediate Cations
(Cations with some remaining outer-shell electrons)
V 22 22 V
3+ 4+

11 Mg 2+ 12 Al 3+

4+ 14 P 13 Si

5+

15 S 6+

16

4+

16 As 50

5+

33 Se6+ 34

Soft Anions

23 23

Cr

3+

24 Mn3,4+ 25 Fe3+ 26 Co
2+ 2+

3+

27 Ni

3+

28 Cu2+ 29 Zn
2+

Sn4+ 30 Ga
3+

Sb

5+

51 Te

6+

52

19 Ca 2+ 20 Sc3+ 21 Ti 4+ 22 V 5+ 23 Cr 6+ 24

Ti Ti

z/ r = 4 Mn Cr2+ 24 Mo
4+

31

2+

25 Fe

26 Co

2+

27

Ni 2+

28 Cu

29

Ge

4+

32 As

3+

33 Se4+ 34 z / r =16

Rb+

37 Sr

2+

38

Y 3+

39

Zr 4+

40 Nb

5+

6+ 41 Mo 42

z/ r = 8

42

Ru3,4+ 44 Rh

2+

45 Pd

2+

46 Ag

47 Cd

2+

48 In

1,3+

49 Sn

2+

50 Sb

3+

51 Te

4+

52 I

5+

53

z/ r = 2 Cs
+

55 Ba 2+ 56 La3+ & 57- Hf REEs 71

4+

72 Ta5+ 73 W 6+ 74 Re7+ 75

4+

74 Re

4+

75 Os

4+

76 Ir

4+

77 Pt

2+

78 Au

79 Hg

80 Tl

81 Pb

2+

82 Bi

3+

83 z/ r = 4

z/ r = 8

Fr

87 Ra2+ 88 Ac 3+ 89 Th

4+

90 Pa

5+

91 U 6+ 92

U4+

92

z/ r = 1

z/ r = 8 z/ r = 2 z/ r = 4

(Large cations with many remaining outer-shell electrons)

z/ r = 2 z/ r = 1

Soft Cations

Hg2+ 80 Tl 3+

81 Pb

4+

82 Bi

5+

83

z/ r = 4

FIGURE 1. Existence of simple carbide, oxide, uoride, chloride, sulde, bromide, iodide, and bismuthide minerals across the periodic table. Note the presence of symbols for suldes, bromides, iodides, and bismuthides among the intermediate to soft cations in the middle of the table, and the absence of such symbols from the hard cations at the left of the table. Minerals are from Clark (1993), Gaines et al. (1997), and Mandarino and Black (2004). Format of the periodic table used here is an abridged and simplied form of the Earth Scientists Periodic Table of the Elements and Their Ions (Railsback 2003, 2004). The left side of this diagram, which shows hard cations, is the eld shown in Figures 2 to 5 and 8.

RAILSBACK: A SYNTHESIS OF SYSTEMATIC MINERALOGY

1035

CHLORIDES, FLUORIDES, OXIDES, AND CARBIDES OF

HARD CATIONS

Both the properties of minerals and the existence of minerals vary regularly across the portion of the periodic table containing hard cations (left side of Fig. 1). For example, melting temperature, hardness, and Be C N Li+ bulk modulus of oxides are B3+ -56 2577 450 1570 2500 greatest, and solubility is 500 2000 2+ Na+ Mg Al3+ Si4+ P5+ 1500 least, in a swath from Be2+ 1000 2053 z/ r = 1 1132 1722 562 2825 4+ to Zr (Fig. 2). Simple 1500 2000 K+ 1000 2500 Ca2+ oxide minerals form with V Sc3+ 4+ Ti 2485 197 Cr 2698 1843 670 350-861 cations in the same swath 2+ 2438 Y3+ 500 Sr Rb+ (Fig. 2). This variability can Zr4+ 801 Mo 1512 Nb5+ 2709 2531 400 be rationalized in terms of La3+ & REEs 2774 Cs+ Ba2+ 1784 ionic potential. Cations of W Hf 490 1472 1972 2304 Ta5+ 3390 3000 low ionic potential, such z/ r = 8 z/ r = 4 Th4+ U6+ Melting Temperature z/ r = 2 as K 1+ , Na 1+ , and Ba 2+ , Be C N Li+ B3+ -7.4 2.77 have such diffuse posi4.4 0.0 Na+ tive charge that they only Mg2+ Al3+ -5.0 Si4+ P5+ -5.0 -3.9 -1.37 make weak bonds with O2, -8.1 z/ r = 1 9.9 0.0 -2.4 5.0 K+ and they fail to shield the Ca2+ 10.0 V Sc3+ 2 Ti4+ 15.0 charge of the anions (O ) 1.4 Cr -7.6 14.0 Sr2+ from each other. Cations Y3+ Rb+ 4.3 4+ Zr 28.9 Mo Nb5+ of intermediate ionic poLa3+ & REEs Cs+ Ba2+ 3+ 4+ tential, such as Al to Ti , W Hf 6.7 Ta5+ have sufciently dense posiz/ r = 2 4+ z/ r = 8 6+ z/ r = 4 Th U Insolubility tive charge to make strong Be C N Li+ B3+ bonds to O2 and to shield 9 the change on the anions Na+ Mg2+ Al3+ Si4+ P5+ 4 from each other, so that 6 z/ r = 1 8 9 7 5.5 stable three-dimensional 8 K+ Ca2+ V Sc3+ cation-oxygen frameworks 6 4+ >9 Ti
2+ 4+ 5+ 6+ 4+ 6+ 2+ 4+ 5+ 6+ 4+ 6+ 2+ 4+ 5+

(i.e., oxide minerals) can form. Cations of high ionic potential, such as C4+, N5+, and S6+, have such focused charge that they make very strong cation-oxygen bonds, but their charge is so dense that the anions cannot shield one positive charge from another, so that the resulting cation-cation repulsion precludes
5+

1 z/ r = 32 S6+

6+

z/ r = 16

5+

1 z/ r = 32 S6+

6+

z/ r = 16

Properties of oxides
1 z/ r = 32 S6+

5+

3.5

3-3.5

Cr6+

z/ r = 16

Rb+ Cs+ Ba2+

Sr2+

Y3+ La3+ & REEs

4
Zr4+

6.5 7
Hf4+

Nb5+

3-4

Mo6+

F IGURE 2. Variation in properties of mineral and nonmineral oxides of hard cations, shown across the periodic table (cf. Fig. 1). Melting temperatures (C) are from pages 438 to 496 of Lide (2003). Solubilities, expressed as the log of the activity of dissolved species, were calculated from thermodynamic data from Robie et al. (1979), Robie and Hemingway (1995), Drever (1988), Langmuir (1997), and Stumm and Morgan (1996). Hardness data are from Nickel and Nichols (1991) with data on TiO2 from Dubrovinskaia et al. (2001). Data on bulk moduli (Ks in GPa) are from Bass (1995), Knittle (1995) and Leger et al. (1993). Circles show the relative size of hard cations across the periodic table.

Hardness

6
z/ r = 2 Li+ z/ r = 4

Ta5+ z/ r = 8 B3+

W6+ U6+ C4+ N5+

Th4+
Be2+

251
Na+ z/ r = 1 K+ Mg2+ Al3+

2.93
Si4+

1 z/ r = 32

P5+

S6+

160
Ca2+

253.5

200

37.8

200 150
Y3+ Sc3+

150

100 50
V5+ Cr6+ z/ r = 16

114.7
Rb+ Cs+ Ba2+ Sr2+

215.5 152 145

Hf4+ Zr4+

Ti4+

87.3

100
La3+ & REEs

Nb5+

Mo6+

70.7

Ta5+ z/ r = 8 B3+

W6+ U6+ C4+ N5+ 1 z/ r = 32 Si4+ P5+ S6+

Bulk Modulus

z/ r = 2 Li+

z/ r = 4

Th4+
Be2+

Bromellite
Mg2+ Al3+

Na+ z/ r = 1 K+

Periclase
Ca2+

Corundum
Sc3+

Quartz
V5+ z/ r = 16

Rutile
Ti4+

Lime
Rb+ Cs+ Ba2+ Sr2+ Y3+ La3+ & REEs Hf4+

Baddeleyite
Zr4+

Schcherbinaite
Nb5+

Cr6+

Molybdite

Mo6+

Tantite
Ta5+ z/ r = 8

W6+ U6+

Simple XOn Minerals

z/ r = 2

z/ r = 4

Thorianite 4+
Th

1036

RAILSBACK: A SYNTHESIS OF SYSTEMATIC MINERALOGY

general pattern is thus a progression from cations of low ionic formation of oxide minerals. Analogous but shifted patterns can be seen in the melting potential combining with anions of less-negative ionic potential temperatures of chlorides, uorides, nitrides, and carbides (Fig. to form minerals, to cations of high ionic potential combining 3). Chlorides of Ra2+, Ba2+, and La3+ have the highest melting with anions of more-negative ionic potential to form minerals. temperatures of chlorides, whereas uorides of La3+ and Sc3+ have The result is predictability of the chemistry of the chloride, the highest melting temperatures of uorides, oxides of Th4+ and uoride, oxide, and carbide minerals (Fig. 4). Mg2+ have the highest melting temperatures among Be C N Li+ B3+ 1 2350? >2100 oxides, nitrides of Hf4+ and z/ r = 32 Na+ Mg2+ Al3+ Al3+ have the highest meltSi4+ P5+ S6+ 2830 z/ r = 1 2100 ing temperatures among K+ 2000 Ca2+ nitrides, and carbides of V Sc3+ z/ r = 16 Ti4+ 2500 4+ 4+ Cr 3067 Zr and Ti have the high2300 3000 3500 Sr2+ Y3+ est melting temperatures Rb+ Zr4+ Mo 3532 Nb5+ >1700 among carbides. The highs La3+ & REEs Cs+ Ba2+ ~3000 in melting temperature thus 2360 W Hf Ta5+ 2500 ~1700 progress through ions of z/ r = 2 z/ r = 8 z/ r = 4 Th4+ U6+ Carbides increasing ionic potential, Be C N 3+ Li+ 1 2966 B 2200 from Ba2+ to Ti4+, as one 813 z/ r = 32 progresses through anions 2+ 3+ Na+ Al 1000 Mg Si4+ P5+ S6+ 1500 of more negative ionic po1900 3000 z/ r = 1 800 ~1500 2000 25003000 tential, from Cl1 to C4 (Fig. K+ Ca2+ V Sc3+ z/ r = 16 Ti4+ 3). 3000 Cr 1195 2500 2000 1500 The progression in meltSr2+ Y3+ Rb+ 1000 Zr4+ Mo Nb5+ ing temperatures observed 1200 2960? La3+ & REEs Cs+ Ba2+ from chlorides to carbides W Hf Melting 600 3305? Ta5+ is mirrored by the existence >500 2820 z/ r = 2 z/ r = 8 z/ r = 4 Th4+ U6+ Nitrides Temperatures of minerals in each group Be C N Li+ B3+ 1 -56 (Fig. 4). Simple chlorides 2577 450 1570 2500 z/ r = 32 500 form only with cations of 2000 2+ 3+ Na+ Mg Al Si4+ P5+ S6+ 1500 lowest ionic potential (K1+ 1000 2053 z/ r = 1 1132 1722 562 2825 1+ 1500 and Na ), fluorides form 2000 K+ 1000 2500 Ca2+ V Sc3+ z/ r = 16 with cations of low ionic Ti4+ 2485 197Cr 2698 1843 670 350-861 1+ 2+ potential (from K to Mg 2438 Y3+ 500 Sr2+ Rb+ Zr4+ and La3+), oxides form with 801 Mo 1512 Nb5+ 2709 2531 400 + La3+ & REEs 2+ 2774 Cs Ba cations of intermediate ionic 1784 W Hf 490 1472 1972 2304 potential (e.g., Al3+ and Ti4+), Ta5+ 3390 3000 z/ r = 2 z/ r = 8 z/ r = 4 Th4+ U6+ Oxides and the lone carbide mineral Be C N Li+ B3+ 1 of a hard cation forms with -183.6 552 -126 848.2 z/ r = 32 a cation of relatively high Na+ Mg2+ Al3+ Si4+ P5+ S6+ 0 ionic potential, Si 4+. The -93.8 -90 z
2+ 4+ 5+ 5+ 6+ 6+ 4+ 6+ 2+ 4+ 5+ 5+ 6+ 6+ 4+ 6+ 2+ 4+ 5+ 5+ 6+ 6+ 4+ 6+ 2+ 4+ 5+

/r = 1 K+

996

1263

Ca2+

858
Rb+ Sr2+

Sc3+

500
284
Zr4+ Ti4+

V5+

1500 1500 3+
Y

1418

1515

1000

19.5 80
Nb5+

17.5 Cr6+

z/ r = 16

F IGURE 3. Variation in melting temperatures of chlorides, uorides, oxides, nitrides, and carbides of hard cations. The ve panels are arranged in order of ionic potential of the anions (see Fig. 1); note the overall progression to the right of the shaded elds of highest melting temperatures. Data are from pages 438 to 496 of Lide (2003).

833
Cs+

1000
1477
Ba2+ La3+ & REEs

~1150 1400s

2.3 Mo6+

Fluorides

703

1368
z/ r = 2 Li+

>970
z/ r = 4

Hf4+

95.1
Ta5+ z/ r = 8

W6+

1100

64
U6+ C4+ N5+

Th4+
Be2+

610
Na+ z/ r = 1 K+ Mg2+

415
Al3+

107

B3+

-22.6

1 z/ r = 32

Si4+

P5+

S6+

800

714
Ca2+

192.6
Sc3+

-68.7

sp = 160
V5+ Cr6+

0
-24.12
Ti4+ z/ r = 16

771
Rb+ Cs+ Sr2+

775 721
La3+ & REEs Y3+

874 715
Ba2+

500

Zr4+

sp = 331
Hf4+

204.7 Nb5+ 275

Mo6+

1000
859
z/ r = 2 z/ r = 4

962

Chlorides

645 1000

770 Th4+

216 Ta5+
z/ r = 8

W6+

Ra2+

177
U6+

RAILSBACK: A SYNTHESIS OF SYSTEMATIC MINERALOGY

1037

NITRATES, SULFATES, CARBONATES, PHOSPHATES,

AND SILICATES OF HARD CATIONS

Increasingly negative ionic potential of anion

The overall pattern seen in Figure 4 can also be seen in the compositions of minerals formed by oxysalt anions, the -ate minerals (Fig. 5). Simple nitrates (nitrates with no H1+, no H2O, no OH1, and no extra O2) form only with cations of lowest ionic potential (K1+, Na1+, and Ba2+, as in NaNO3). Simple sulfates form only with cations of slightly higher ionic potential, from K1+ to Ca2+. Simple carbonates form with cations of still higher ionic potential, from Na1+ to Mg2+. Simple phosphates form with cations of yet higher ionic potential, from Na1+ to Al3+. Simple silicates form with cations of even higher ionic potenBe C N Li+ B3+ 1 tial, from Na1+ to Zr4+ (as in z/ r = 32 2+ 3+ Na+ Mg Al Si4+ P5+ S6+ ZrSiO4) (Fig. 5). Thus, as SiC Moissanite z/ r = 1 one progresses from the catK+ Ca2+ V Sc3+ z/ r = 16 ion of highest ionic potential Ti4+ Cr to form an oxysalt radical, Sr2+ 3+ + Y Rb Zr4+ Mo Nb5+ N5+ in nitrate, through this La3+ & REEs Cs+ Ba2+ sequence to the cation of W Hf Ta5+ lowest ionic potential to z/ r = 2 z/ r = 8 z/ r = 4 U6+ Th4+ Carbides form an oxysalt radical, Si4+ Be C N BeO Li+ B3+ 1 Bromellite in silicate, one progresses Diaspore z/ r = 32 AlO(OH)Akdalaite + Na Mg2+ Al3+ from inter-radical cations 4Al2O3.H2O Si4+ P5+ S6+ Al2O3 MgO SiO2 3SiO2.H2O z/ r = 1 Corundum Periclase of low ionic potential (e.g., Quartz Silhydrite K+ CaO Ca2+ Navajoite K1+) to high ionic potential Sc3+ TiO z/ r = 16 V2O5 V 2 Ti4+ V O .3H2O Cr Lime Schcher- 2 5 Rutile binaite (e.g., Zr4+). This progresSr2+ MoO3.2H2O Y3+ Rb+ ZrO2 MoO3 Zr4+ Sidwillite sion can be rationalized as Nb5+Molybdite Mo Baddeleyite La3+ & REEs Tantite WO3.H2O Tungstite Cs+ Ba2+ the result of repulsion of all WO3.2H2O Hydrotungstite Ta2O5 W Hf Ta5+ but the weakest cations by Thorianite UO3.2H2O Schoepite z/ r = 2 z/ r = 8 z/ r = 4 ThO2 Th4+ U6+ Oxides the dense positive charge Be C N LiF Li+ B3+ 1 Griceite of N5+, and by the progresz/ r = 32 + Na Mg2+ Al3+ sive acceptance of more Legend: Si4+ P5+ S6+ NaF AlF(OH)2 MgF2 Zharchikhite z/ r = 1 Villiaumite Sellaite densely charged inter-radiOH-bearing K+ Ca2+ cal cations (i.e., from K1+ minerals V CaF2 Sc3+ z/ r = 16 KF Ti4+ Cr Fluorite 2+ 4+ 4+ Carobbite H2O-bearing to Mg to Al to Zr ) by (otherwise Sr2+ Y3+ Rb+ Zr4+ simple) radical-forming cations of Mo Nb5+ minerals La3+ & REEs Simple Cs+ Ba2+ lesser ionic potential (i.e., minerals (Ce,La)F2 BaF2 W Hf (no H2O Ta5+ from N5+ to S6+ to C4+ to P5+ Frankdicksonite Fluocerite and no OH-) z/ r = 2 z/ r = 8 z/ r = 4 Fluorides Th4+ U6+ 4+ to Si ) (Fig. 5). Be C N Li+ B3+ 1 Figure 5 shows that there Korshunovskite Hydrohalite
2+ 4+ 5+ 5+ 6+ 6+ 4+ 6+ 2+ 4+ 5+ 5+ 6+ 6+ 4+ 6+ 2+ 4+ 5+ 5+ 6+ 6+ 4+ 6+ 2+ 4+ 5+

is also predictability in the compositions of non-simple minerals of oxysalt anions. Fields for H2O-bearing minerals, OH1-bearing minerals, and minerals with extra O2 progress to the right from the elds for simple minerals (Fig. 5). This trend can be rationalized in terms of ionic potential and cation-cation repulsion, in that incorporation of cations of progressively higher ionic potential to the right in these diagrams is allowed rst by neutral groups (H2O) that simply shield cations from each other and then by anionic groups (OH1 and extra O2) that help cancel positive charge. An alternate rationalization in terms of aqueous speciation would explain this trend as resulting from failure of

Cadwaladerite NaCl.2H2O Na+ Mg2+ Al3+ AlCl(OH)2.4H2O 4+ Si NaCl MgCl3.6H2O AlCl .6H O z/ r = 1 Halite 2 2 Bischofite Chloraluminite
K+ Ca2+

Mg2Cl(OH)3.4H2O

z/ r = 32

P5+

S6+

F IGURE 4. Chloride, fluoride, oxide, nitride, and carbide minerals of hard cations (there is no nitride mineral of a hard cation). Note the shift of the blue eld for simple minerals (minerals of one cation with no H 2O or OH1) to the right as one progresses upward through the diagram. The ve panels are arranged in order of ionic potential of the anions (see Fig. 1).

KCl Sylvite
Rb+ Cs+ Ba2+ Sr2+

CaCl2.2H2O Sinjarite
Y3+ La3+ & REEs

Sc3+

Ti4+

V5+ Cr6+

z/ r = 16

Zr4+

Nb5+

Mo6+

Hf4+

Ta5+ z/ r = 8 B3+

W6+ U6+ C4+ N5+ 1 z/ r = 32 Si4+ P5+ S6+

Chlorides

z/ r = 2 Li+

z/ r = 4

Th4+
Be2+

Na+ z/ r = 1 K+

Mg2+

Al3+

Ca2+

Sc3+

Ti4+

V5+ Cr6+

z/ r = 16

Rb+ Cs+ Ba2+

Sr2+

Y3+ La3+ & REEs Hf4+

Zr4+

Nb5+

Mo6+

Ta5+ z/ r = 8

W6+ U6+

Iodides & Bromides

z/ r = 2

z/ r = 4

Th4+

1038

RAILSBACK: A SYNTHESIS OF SYSTEMATIC MINERALOGY

Liberite Li+ Li2BeSiO4
Be2+ 3+ C4+ N5+

Be2SiO4 Be4Si2O7(OH)B 2 Phenakite Bertrandite

z/ r = 32 Natrosilite Na Si O .5H O 3+ 2 2 5 2 Na2Si2O5 Chrysotile Al Si4+ P5+ S6+ Mg2+ Mg Si O (OH) Al2SiO5 + 4 Al Si O (OH) Na8Al6SO4Si6O24 Mg2SiO4 3 2 5 NaHSi2O5.H2O Na 2 2 5 4 Na3CaPO4SiO3 Nosean Clinophosinaite Forsterite Grumantite Kaolinite Sc3+ CaSiO3 Ca2+ Wollastonite 4+ 5+ z/ r = 16 K+ Ti V Cr6+ CaSi2O5.2H2O CaTiSiO5 (Zn,Ni,Cu) Al V Si O .27H O Ca3Si3O8(OH)2 8 8 2 5 35 2 Titanite ZrSiO4 (Sphene) Kurumsakite Rosenhahnite Y3+ Zircon Rb+ Mo6+ 5+ Sr2+ Zr4+ Nb Cs+ Ba2+ La3+ & REEs (Y,Yb)2Si2O7

Ca5(CO3)2Si2O7 Tilleyite

Hafnon HfSiO4
Hf4+

Ca4Nb6Si5O24(OH)1o.nH2O Mogolite
W6+ Ta5+

Silicates

Kievyite ThSiO4Huttonite BaSi2O5 BaSi2O5.3H2O Sanbornite z/ r = 2 z/ r = 4 Th4+

2+

UO3.7SiO2 z/ r = 8 Uranosilite U6+

4+

Be C 3+ Li3PO4 Li+ Be2PO4(OH).4H2O B Na3MgCO3PO4 Lithiophosphate Moraesite Bradleyite Olympite Na+ Na3PO4 Al Mg2+ Al3+ 2PO4(OH)3 Si4+ P5+ Na2HPO4 Mg3(PO4)2 Mg2PO4OH Augelite Na3CaSiO3PO4 Farringonite AlPO4 z/ r = 1 Nahpoite (Mg)3(PO4)2.8H2O AlPO4.2H2O K+ Ca2+ Sc3+ Ti4+ V5+

N5+

1 z/ r = 32 S6+

CaAl3SO4PO4.(OH)6 Woodhouseite
Cr6+ z/ r = 16

Rb+ Cs+ Ba2+

Sr2+

Ca5(PO4)3(OH) ScPO4.2H2O Kolbeckite Apatite YPO4 Xenotime

La3+ & REEs Y3+

YPO4.2H2O Churchite

Zr4+

Mo6+ Nb5+

Decreasing ionic potential of cation in oxysalt radical

Phosphates

Monazite (La,Ce)PO4.H2O Hf4+ (La,Ce,Nd)PO4 Rhabdophane z/ r = 2 z/ r = 4 Th4+

W6+ Ta5+ z/ r = 8 U6+

Rb+ Cs+

Be2+ C4+ N5+ Li2CO3 Li+ B3+ 1 Zabuyelite Ca4Mg(B4O6)(CO3)2(OH)6 Na2CO3 Borcarite Bradleyite z/ r = 32 Natrite Na2CO3.10H2O Na3MgPO4CO3 MgCO3.5H2O Al3+ Na+ MgCO NaHCO3 Si4+ P5+ S6+ 3 Nahcolite Magnesite Mg2+ NaAlCO3(OH)2 Ca Si O (CO3)2 Na6(SO4)2CO3 Mg5(CO3)4(OH)2.4H2O Dawsonite 5 2 7 z/ r = 1 Tilleyite Burkeite Hydromagnesite K+ CaCO3 KHCO3 V5+ Sc3+ z/ r = 16 Ca2+ Calcite Ti4+ Kalcinite Cr6+ CaCO3.6H2O Na4Zr2TiO4(CO3)4 Sabinaite SrCO3 Ikaite Sr2+ Y3+ Zr4+

Legend:
Minerals with extra O2-

Strontianite
Ba2+

Y2(CO3)3.2-3H2O La3+ & REEs Nd(CO3)(OH)

Hf4+

Nb5+ W6+

Mo6+

(Ta,Nb)(OH)3(O,CO3)
Ta5+ z/ r = 8 U6+ B3+

(La,Ce)2(CO3)3.8H2O

Carbonates

BaCO3 Witherite

z/ r = 2 Li+

Na6BaTh(CO3)6.6H2O Tuliokite Th4+

Be2+

UO2CO3 Rutherfordine UO2CO3nH20 Joliotite

C4+

OH-bearing minerals H2O-bearing (otherwise simple) Simple minerals minerals H2O Minerals (no and with H+ no OH-) (bicarbonates,
1 z/ r = 32

bisulfates, etc.)

Na+ Na2SO4 z/ r = 1 Thenardite

Mirabilite Na2SO4.10H2O

Na6(SO4)2CO3 Burkeite
Si4+

N5+ Na3NO3SO4.H2O

Darapskite

Mg7(SO4)5(OH)4.H2O
Mg2+

KHSO4 Mercallite
Rb+ Cs+

K SO4 K+ 2Arcanite

MgSO4.1-7H2O Al2SO4(OH)4.7H2O Na8Al6(Si6O24)SO4 Al2PO4SO4(OH).9H2O Al2(SO4)3.17H2O Ca4Al6O12SO4 Nosean Alunogen Ye'elimite Ca2+
Sc3+ Ti4+ V5+ Cr6+ z/ r = 16

Al3+

P5+ Sanjuanite S6+

CaSO4 CaSO4.2H O Anhydrite Gypsum 2

Sr2+

SrSO4 Celestite
Ba2+

Y3+ La3+ & REEs Hf4+

Zr4+

Nb5+

Mo6+

Sulfates

BaSO4 Barite
z/ r = 2 Li+ Na+ z/ r = 4

Ta5+ z/ r = 8 B3+

W6+ U6+ C4+ N5+ 1 z/ r = 32 Si4+ P5+ S6+

Th4+
Be2+

NaNO3 Nitratine
z/ r = 1 K+

Mg2+

Al3+

Mg(NO3)2 .6H2O Nitromagnesite Ca(NO3)2 .4H2O Nitrocalcite

Y3+ Ca2+

KAl7(NO3)4Cl2(OH)16.8H2O Sveite
Sc3+ Ti4+ V5+

Na3SO4NO3.H2O Darapskite
z/ r = 16

KNO3 Niter
Rb+ Cs+

Cr6+

Zr4+

Sr2+ Ba2+

Nb5+

Mo6+

La3+ & REEs Hf4+ W6+ Ta5+ z/ r = 8 U6+

Nitrates

Ba(NO3)2 Nitrobarite

z/ r = 2

z/ r = 4

Th4+

FIGURE 5. Nitrate, sulfate, carbonate, phosphate, and silicate minerals of hard cations. Note the shift of the blue eld for simple minerals (minerals of one cation with no H2O or OH1) to the right as one progresses upward through the diagram. The ve panels are arranged in order of ionic potential of the cation in the oxysalt radical (see Fig. 1). Not all possible minerals are shown for lack of space.

RAILSBACK: A SYNTHESIS OF SYSTEMATIC MINERALOGY

1039

Log Ksp (Greater solubility

cations of higher ionic potential to dehydrate or dissociate from ligands before entering mineral structures. The result is the same: increasing incorporation of H2O or anions with incorporation of cations of higher ionic potential (Fig. 5). Minerals far to the right in the diagrams (i.e., minerals incorporating cations of high ionic potential) also commonly incorporate cations of low ionic potential (e.g., Na1+) that can be viewed as diluting the ionic potential of the entire population of cations in a mineral. For example, mongolite [Ca4Nb6Si5O24(OH)10nH2O] is a silicate in which the high ionic potential of the principal cation, Nb5+, is diluted by that of a cation of lower ionic potential (Ca2+), is shielded by H2O molecules, and is countered by the negative charge of both ten OH1 groups and four O2 in excess of those dictated by the stoichiometry of SiO4 tetrahedra. 4 In contrast to the elds of non-simple minerals to the right of the elds for simple minerals in Figure 5, elds for H1+-bearing minerals consistently fall to the left. In this respect, the bicarbonates, biphosphates, and bisulfates can be viewed as minerals in which cations of low ionic potential provide sufciently little density of positive charge that H1+ ions of undissociated HCO1, 3 HPO2, and HSO1 are tolerated in minerals, opposite to the con4 4 siderations described in the above paragraph.

orthorhombic carbonates with ninefold coordination (Fig. 7). Coordination of three O2 anions around C4+ in these carbonate minerals is likewise dictated by radius ratio. At the ends of the ranges of cation radius for each group, solubility is high (Fig. 7), illustrating the importance of bond length in determining whether a cation forms a mineral within a particular isostructural group. However, the existence of synthetic solid BeCO3 (Wiberg 1971, p. 614) again disproves the conventional wisdom that the nonexistence of BeCO3 as a mineral is because Be2+ is too small, relative to the cations of rhombohedral and orthorhombic carbonates, to form stable structures. The trends shown in this paper (Figs. 4 and 5) would instead suggest that Be2+-C4+ repulsion is the more likely reason that BeCO3 is sufciently unstable that it is not known as a mineral.
Ionic potential
1.6 1.4 1.2 1.0 0.8

A
2.5 Synthetic crystalline LiCl Synthetic crystalline 2.0 LiNO3 1.5 1.0 0.5 Synthetic crystalline LiSO4 NaNO3: Nitratine Nitrates NaCl: Halite

IONIC RADIUS AND RADIUS RATIO ARE NOT SUFFICIENT TO EXPLAIN EXISTENCE AND NON-EXISTENCE OF
MINERALS

Chlorides

KCl: Sylvite

KNO3: Niter

A skeptic might suggest that the patterns shown in Figures 4 and 5 need not be explained in terms of ionic potential but might be explained simply as a matter of radius ratio and coordination. For example, the conventional wisdom regarding the non-existence of LiCl, LiNO3, and Li2SO4 as minerals is that Li1+ is too small relative to the cations in the stable comparison group to form stable structures (in this example, the comparison group would be Na and K chlorides, nitrates, and sulfates). However, there are in fact synthetic crystalline LiCl, LiNO3, and Li2SO4 phases (Lide 2003), so the non-existence of these compounds as minerals is not for lack of a possible structure. Likewise, the conventional wisdom regarding the non-existence of B2O3 as a mineral is that B3+ is too small compared to Al3+ to form a stable structure like that in corundum. However, there is in fact synthetic crystalline hexagonal B2O3 solid (Lide 2003), again demonstrating that non-existence as a mineral is not for lack of a possible structure. For the example of LiCl, LiNO3, and Li2SO4, Figure 6 shows that these non-mineral crystalline compounds are more soluble than the corresponding mineral-forming compounds, and Figure 2 likewise shows that non-mineral crystalline B2O3 is much more soluble than Al2O3. This greater solubility is clearly not for lack of a possible crystalline structure. It instead probably results from a characteristic of that structure, such as greater cation-cation repulsion (e.g., Li1+-Li1+, Li1+-N5+, Li1+-S6+, and B3+-B3+) between cations of higher ionic potential than that found in the mineral-forming compounds of cations of lower ionic potential. This is of course not to say that cation size, radius ratios, bond lengths, and coordination are unimportant to mineralogy. For example, simple carbonates of divalent cations occur in two structural groups, with smaller cations forming rhombohedral carbonates with sixfold coordination, and larger cations forming

Na2SO4: Thenardite

0.0

Sulfates

K2SO4: Arcanite

-0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6

Ionic radius of cation () Ionic potential

8.0 6.0 4.0 2.0 Synthetic Synthetic Synthetic Synthetic Synthetic 3 crystalline crystalline crystalline crystalline crystalline BeCl2 Ca(NO3)2 MgCl2 Mg(NO3)2 CaCl2 2 Synthetic 0 crystalline BeSO4 -1 -2 -3 -4 -5 -6 -7 -8 -9 -10 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 Legend: CaSO4: Anhydrite Sulfates SrSO4: Celestite BaSO4: Barite 1

B
Synthetic crystalline Sr(NO3)2

Log Ksp (Greater solubility

Synthetic crystalline MgSO4

Synthetic crystalline Synthetic SrCl2 crystalline BaCl2 Ba(NO ) : 3 2 Nitrobarite

ls ra ine ls - m n er a n i No M Chlorides Nitrates Sulfates

Ionic radius of cation ()

FIGURE 6. Solubility of chlorides, nitrates, and sulfates of alkali metal cations (A) and alkali earth cations (B) as a function of ionic potential (tops of diagrams) and ionic radius (bottoms of diagrams). Filled symbols indicate minerals. Open symbols indicate crystalline solids (Lide 2003) not known from nature as minerals. Note that, contrary to conventional wisdom, failure of smaller cations to form minerals is not for lack of a possible crystalline structure. Solubility data are from Lide (2003).

1040

RAILSBACK: A SYNTHESIS OF SYSTEMATIC MINERALOGY

POLYHEDRAL LINKAGE
The concept of cation-cation repulsion between oxysalt radicals used to explain trends in Figures 2 and 5 is also useful in understanding the extent of polyhedral linkage in different groups of oxysalt minerals (Fig. 8). For example, in minerals of oxysalt

5 4 )
-7 -8 -9

3 1.8 1.6 1.4

Ionic potential 1.0 1.2

0.8

Log Ksp (Greater solubility

-10 -11

-12 -13

-14

CaCO3 SrCO3 MgCO3 Calcite CaCO3 BeCO3 Strontianite Magnesite Aragonite (Synthetic) BaCO3 Witherite Four-fold ZnCO3 coordiCdCO3 nation Smithsonite Otavite FeCO3 MnCO3 Siderite Rhodochrosite PbCO3 Cerussite Six-fold coordination Nine-fold coordination 0.2 0.3 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6

Cationic radius ()

FIGURE 7. Solubility of carbonates of alkaline earth cations as a function of ionic potential (top of diagram) and ionic radius (bottom of diagram). Filled symbols indicate minerals, with triangles for rhombohedral minerals and squares for orthorhombic minerals. Open symbol is for synthetic BeCO3 (Wiberg 1971), which is not known from nature as a mineral. Fourfold-coordination for Be2+ in BeCO3 is inferred from fourfold coordination in BeNO3 (Wiberg 1971). Dashed line indicates that solubility of BeCO3 is not well constrained. Solubility data for minerals are from Robie and Hemingway (1995).

radicals of cations of highest ionic potential (N5+, S6+, C4+, and P3+), no mineral structures involve polyhedral linkage (there is no sharing of O atoms by NO1, SO2, CO2, and PO3 radicals). As 3 4 3 4 a result, all members of these groups of minerals have chemical formulae that preserve the identity of the radicalfor example, all nitrates have chemical formulae involving NO3, all sulfates have formulae involving SO4, and so on (Fig. 8). In minerals of oxysalt radicals having cations of slightly lower ionic potential (Cr6+, Mo6+, and W6+), most mineral structures involve no polyhedral linkage, but a small proportion of minerals exhibit limited sharing of O atoms between polyhedra. Chemical formulae reect this, in that most chromates have formulae in which CrO4 appears, but a few have formulae like K2C2O7 reecting the sharing of an oxygen between paired chromate groups, analogous to the tetrahedral linkage in sorosilicates (Fig. 8). Finally, in minerals of oxysalt radicals of cations of still lower ionic potential (B3+, Si4+, V5+, Nb5+, and Ta5+), most mineral structures involve polyhedral linkage (Fig. 8). The most familiar example is the silicates, where only the nesosilicates have no polyhedral linkage, and where polyhedral linkage progresses from sharing of just one O atom between pairs of tetrahedra (as in the chromates) to sharing of all O atoms in frameworks (in the tectosilicates). The resulting chemical formulae obscure the identity of SiO4 groups, in that they end with SiO3, Si2O5, 4 and even SiO2. This extent of polyhedral linkage in the silicates is of course well known, but it is also characteristic of the borates, vanadates, niobates, and tantalates because of the similar ionic potential of B3+, V5+, Nb5+, and Ta5+ (Fig. 8). In this context, the nesosilicates can be viewed as the only

CaCO3
Li+ Be2+ B3+ C4+ N5+

Na N O 3
Nitrates

Bo r ates
Na+

Carbonates

No polyhe d ral linka ge

Mg2+ Al3+ S i4 + S i li cate s
P 5+ S6+

CaSO4

Phosphates Vanadates
V5+

Sulfates

Mg3(PO4)2 CaCrO4 K2CrO4 K2Cr2O7

K+

Ca2+

Sc3+

Ti4+

Ch rom at es
Cr6+

Multiple links between polyhedra

Rb+ Sr2+ Y3+

Poly hedra l linka ge

in most mi n e rals
Nb5+

Zr4+

in a few minerals
Mo6+

Nio b ates
1 Cs+ Ba2+ La3+

Mo ly bda t es

Hf4+

T a 5+

W6+

Tantalates

Tungst at es

Mg2SiO4 Zn2Si2O7(OH)2H2O Mg2Si2O6 Mg3Si4O10(OH)2 SiO2

(Fe,Mn)WO4 CaWO4 YW2O6(OH)3

FIGURE 8. Extent of polyhedral linkage by oxysalt radicals of hard cations across the periodic table. Representative chemical formulae are shown to right and below.

RAILSBACK: A SYNTHESIS OF SYSTEMATIC MINERALOGY

1041

silicate analogs of the phosphates, carbonates, sulfates, and nitrates, in which no polyhedral linkage occurs. The position of a nesosilicate, olivine, at the high-temperature end of Bowens Reaction Series can thus be rationalized as the result of a silicate structure minimizing tetrahedral repulsion, like those of the phosphates et al. On the other hand, silicates lower in the discontinuous series involve progressively more tetrahedral linkage, suffer more tetrahedral repulsion, and have lower melting temperatures.

predictions about chemical formulae of minerals, if we appreciate the distinction between hard and soft anions and between hard and soft cations, and if we use the ionic potential of cations and anions to understand minerals.

ACKNOWLEDGMENTS
The authors colleagues at the University of Georgia, and especially Michael F. Roden, provided many comments helpful in developing the ideas in this manuscript. Mark Leach of Salford University also helped the author develop his understanding of mineral chemistry, especially with regard to the variation among anions. Reviews by James H. Stout of the University of Minnesota and Cornelis Klein of University of New Mexico further improved the manuscript.

CONCLUDING REMARKS
Hardness and ionic potential of both cations and anions are useful in explaining existence and non-existence of possible chemical substances as minerals, and in explaining variation in mineral properties. These concepts have long been used to characterize cations, but their application to anions is also useful in explaining trends and patterns in mineralogy. For example, sulde minerals of single cations only form with intermediate to soft cations (Fig. 1). Conversely, hard cations only enter into a few sulde minerals, all of which also contain intermediate cations. The dearth of hard cations in sulde minerals means that most sulde minerals are anhydrous. Chloride, an anion intermediate between hard and soft anions, is distinct in forming minerals with hard cations, intermediate cations, and soft cations from Na1+ to Ag1+ (Fig. 1). Among the simple chloride, uoride, oxide, and carbide minerals of hard cations, ionic potential of the cation is proportional to that of the anion. Non-simple (H2O- and OH1-bearing minerals) in each group typically involve cations of higher ionic potential than those of the simple minerals (Fig. 4). Among the simple nitrate, sulfate, carbonate, phosphate, and silicate minerals of hard cations, ionic potential of the cation (e.g., Ca2+ in calcite) is inversely proportional to the ionic potential of the cation in the oxysalt radical (e.g., C4+ in calcite). Non-simple minerals (H2O- and OH1-bearing minerals and minerals with extra O2) in each group typically involve cations of higher ionic potential than those of the simple minerals. Such minerals also commonly incorporate other cations of low ionic potential (Fig. 5). Polyhedral linkage is increasingly common in oxysalt mineral groups in which the ionic potential of the cation in the oxysalt radical (e.g., C4+ in calcite) is lower (Fig. 8). Figures 1, 4, and 5 show that, amidst the diverse array of chemical formulae of minerals, we can nd patterns and make predictions. For example, Figure 1 shows that we can expect suldes of intermediate-to-soft cations, and we can conversely expect oxides and uorides of hard cations, but we are unlikely to work with or discover a sulde mineral solely of Mg2+ or an oxide of Ag1+. As another example, Figure 4 shows that we can expect to encounter chlorides of cations of low ionic potential, and we might discover a BaCl2 mineral, but we need not worry ourselves about BeCl2 minerals. We likewise should know about the SiC mineral (moissanite) and might someday discover a TiC mineral, but we neednt worry about a Ca2C mineral. Similarly, Figure 5 leads us to expect simple nitrates of cations of low ionic potential, and hydrous phosphates of cations of intermediate ionic potential, but we are unlikely to learn about or discover an Al(NO3)3 mineral or a K5PO4(OH)2 mineral. We can thus make

REFERENCES CITED
Bass, J.D. (1995) Elasticity of minerals, glasses, and melts. In Ahrens, T.J., Ed., Mineral Physics and Crystallography: A Handbook of Physical Constants, American Geophysical Union Reference Shelf Series 2, 4563. Cartledge, G.H. (1928a) Studies on the periodic system. I. The ionic potential as a periodic function. Journal of the American Chemical Society, 50, 28552863. (1928b) Studies on the periodic system. II. The ionic potential and related properties. Journal of the American Chemical Society, 50, 28632872. Clark, A.M. (1993) Heys Mineral Index, 852 p. Chapman and Hall, London. Drever, J.I. (1988) The Geochemistry of Natural Waters (2nd ed.), 437 p. PrenticeHall, Englewood Cliffs, New Jersey. Dubrovinskaia, N.A., Swamy, V., Muscat, J., Harrison, N.M., Ahuja, R., Holm, B., and Johansson, B. (2001) The hardest known oxide. Nature, 410, 653654. Gaines, R.W., Skinner, H.C.W., Foord, E.E., Mason, R., and Rosenzweig, A. (1997) Danas New Mineralogy (8th ed.), 1819 p. Wiley, New York. Goldschmidt, V.M. (1937) The principles of distribution of chemical elements in minerals and rocks. Journal of the Chemical Society, 655673. Klein, C. (2002) Manual of Mineral Science (22nd ed.), 641 p. Wiley, New York. Knittle, E. (1995) Static compression measurements of equations of state. In Ahrens, T.J., ed., Mineral Physics and Crystallography: A Handbook of Physical Constants, American Geophysical Union Reference Shelf Series 2, 98142. Langmuir, D. (1997) Aqueous Environment Geochemistry, 600 p. Prentice-Hall, Upper Saddle River, New Jersey. Leger, J.M., Tomaszewski, P.E., Atouf, A., and Pereira, A.S. (1993) Pressure-induced structural phase transitions in zirconia under high pressure. Physical Review BCondensed Matter, 47, 14,07514,083. Lide, D.E., Editor (2003) CRC Handbook of Chemistry and Physics (84th ed.), 2522 p. CRC Press, Boca Raton, Florida. Mandarino, J.A. and Black, M.E. (2004) Fleischers Glossary of Mineral Species, 309 p. The Mineralogical Record Inc., Tucson, Arizona. Nesse, W.D. (2000) Introduction to Mineralogy, 442 p. Oxford, Oxford University Press. Nickel, E.H. (1995) The definition of a mineral. Canadian Mineralogist, 33, 689690. Nickel, E.H. and Nichols, M.C. (1991) Mineral Reference Manual, 250 p. Van Nostrand Reinhold, New York. Pearson, R.G. (1968) Hard and soft acids and bases, HSAB, Part I. Journal of Chemical Education, 45, 581587. Railsback, L.B. (2003) An earth scientists periodic table of the elements and their ions. Geology, 31, 737740. (2004) An earth scientists periodic table of the elements and their ions. Geologic Society of America Map and Chart Series item MCH 092. Robie, R.A. and Hemingway, B.S (1995) Thermodynamic Properties of Minerals and Related Substances at 298.15 K and 1 Bar (105 Pascals) Pressure and at Higher Temperatures. U.S. Geological Survey Bulletin 2131, 461 p. Robie, R.A., Hemingway, B.S., and Fisher, J.R. (1979) Thermodynamic Properties of Minerals and Related Substances at 298.15 K and 1 Bar (105 Pascals) Pressure and Higher Temperatures. U.S. Geological Survey Bulletin 1452, 456 p. Shannon, R.D. (1972) Revised effective ionic radii and systematic studies of interatomic distance in halides and chalcogenides. Acta Crystallographica, A32, 751767. Stumm, W. and Morgan, J.J. (1996) Aquatic Chemistry (3rd ed.), 1022 p. Wiley, New York. Wiberg, E. (1971) Lehrbuch der anorganischen Chemie, 1209 p. Walter de Gruyter, Berlin.

MANUSCRIPT RECEIVED NOVEMBER 1, 2004 MANUSCRIPT ACCEPTED FEBRUARY 28, 2005 MANUSCRIPT HANDLED BY ROBERT F. DYMEK