American Mineralogist, Volume69, pages307-318,1984

Heat capacity and thermodynamicfunctions for gehleniteand staurolite: with comments on the Schoitky anomaly in the heat capacity of staurolite
Bnucr S. HeutNcwAY AND Rlcneno A' RostB

U.S. Geological Survey Reston, Virginia 22092 Abstract The heat capacitiesof a synthetic gehleniteand a natural staurolite have been measured of calorimetry,and the heatcapacities to from 12and 5 K, respectively, 370K by adiabatic staurolite have been measured to 900 K by differential scanning calorimetry. Staurolite exhibits a Schottky thermal anomaly having a maximum near 21 K. Smoothed values of the thermodynamic properties of heat capacity, entropy, and enthalpy function and Gibbs energy function are given for integral temperatures. The K at CazAlzSiOr, 298.15 and l bar is 210-1t0.6J/(mol'K), which entropyof gehlenite, includes a configurational contribution of 11.506J/(mol ' K). The entropy of staurolite at J/(mol ' K) for H2Al2 as 298.15K and I bar is reportedfor two compositions 1019.6112.0 ' K) for (H3Alr.rsF4.to)C'e3.tzf4iaTi6.66Mn6.62 FeaAll5,SiaOas ll0l.0-r12.0 J/(mol and entropy contributionsare 34.6 and where Alt.rs)(Mgo.aaAlrs.zdSieO+e the configurational (mot . K), respectively.In addition, the entropy value reported for the second 121.0J staurolite composition contains an additional l0 J representingthe estimated contribution of the magnetic entropy below about 5 K.

lntroduction The heat capacities of gehlenite were measured between 12 and 380 K in this study in order to reduce the uncertainty associatedwith the rather large extrapolation necessaryto obtain the entropy contribution below 50 K from the older heat capacity data that were measured between50 and 298 K by Weller and Kelley (1963).Hightemperatureheat-contentdata for gehlenite were reported by Pankratz and Kelley (1964). The heat capacitiesof staurolite were measuredin this study between 5 and 368 K by low-temperatureadiabatic calorimetry and between 340 and 900 K by differential scanningcalorimetry. We are not aware of previous heatcapacity measurementsfor staurolite. The single set of heat-capacityresults has been corrected to two staurolite compositions, and estimatesof the Schottky heat capacity have been made upon the basis of several simple models for determining the lattice heat capacity for staurolite. Similarly, simple models have been used to estimateconfigurationalentropy terms for each staurolite formulation. Materials The gehlenite sample was a portion of the sampleused by Woodhead (1977). A complete description of the sample preparation and of the physical and chemical properties of the sample was given by Woodhead. The sample was designated75001H by Woodhead. The geh0003-004)V84l030,{-0307$02.00

for lenite was annealedat 1525"C 20 hours from a glassof gehlenitecomposition. The annealedsamplewas crushed and material less than 150 mesh was removed' The g. were a = samplemasswas 28.0825 The cell parameters cell and 7.68658(23)A c : 5.06747(lDland the calculated J/bar (Woodhead,1977 Table 3' volumewas 9.01559(51) , 3). Woodhead also reported that l0% of the glass remained after the 20 hours of annealing. The chemical and physical properties of the staurQlite samplewere describedby Zen (1981,Table 4, page 124, sample 355-1). The sample representeda separateof natural staurolite from the Everett Formation collected near Lions Head, Conn. The crushed sample was dry sieved to remove material smaller than 150 mesh. The sample mass was 31.9650g for the low-temperature calorimetric measurementsand 39.900mg for the differential scanningcalorimetric measurements' Experimental results The low-temperature adiabatic calorimeter and the methods and procedures followed in this study are deRobieet (Robieand Hemingway,1972; scribedelsewhere of The heat capacities staurolitewere al.,1976and 1978). measured from 340 to 900 K by using a differential scanning calorimeter and following the procedures outlined by Krupka et al. (1979) and Hemingway et al. ofthis naturalstauro(1981). The onsetof decomposition lite samplewas 910t10 K, at a heating rate of l0 trUmin.

307

308

HEMINGWAY AND ROBIE: GEHLENITE AND STAUROLITE Table 2. Experimental specificheatsfor staurolite. Series8 and 9 were results obtained by differential scanningcalorimetry r!rp. to;:lit"
J/(r. r)

The experimental specific heats for gehlenite and staurolite are given in Tables I and 2, respectively, in the chronological order of the measurements.The data have been corrected for curvature (Robie and Hemingway, 1972)but are uncorrected for chemical impurities. Thermodynamic properties of gehlenite and staurolite

r.rp.
I' S.r1.r 1 0 4 .r 3 r09.73 I I 5.37 t2l. t3 t26.92 t32.67 l3E.3E lra.0r t49. t5 1 5 5 .t 0 161.03 t66.66

to;:::t"
J/(r. f,) 5 0. 1902 0.2096 0.2297 0.2507 0 . 2 7t 5 0.2924 0.lttl 0.3335 0. t537 0. Itt5 0.t932 0.4125

r.rp.
K

tt;:lit"
J/(s. x)

3C l ar I r 29E.2 t 3 0 4 .E 5 3lt. t5 t19.09 325.35 0.7615 0.7152 0.7E77 0.8007 0 .E l 3 t

3.rla.9 {50.0 460.0 {69.9 a79.9 {09.9 r99.9 169.9 a t 9 .r 4 8 9 .t { 9 9 .E 5 0 9 .t 5 1 9 .r 529.0 tl9. E tt9. E t 5 9 .E 5 6 9 .E 5 7 9 .E 5E9.7 t99,1 609.7 619.7 629.1 639.7 649.7 619.0 629.t 639.7 649.7 659.7 669.7 579.7 6A9.7 699.7 709.1 719.7 729.7 739.7 719.7 759.7 ,89. I 779.7 7A9.1 799.1 8 4 9 .{ 89E.9 0.96t{ 0.979r 0.9853 0.995r r.00tl r.0r37 0.96{3 0.9901 0.9160 r.0027 1.0076 t.0r2t r.0l18 t.0262 t.0321 1.0393 r.047r 1.0518 t.0596 t.06a9 r.0713 1.079t r.0E75 t. O99t l. Il25 1.0920 r.0990 l. lol3 l. t066 l. lt20 l. ll74 t. l22l r. t262 r. r27E l. r33E r. l39E l.1425 l. 1477 l. l5l0 l.1493 r. 1655 l.1640 l.1705 t. lEO2 l. l67E r. l96r

The measured specific heat data were graphically extrapolated to 0 K from a plot of CIT vs. l. A more completedescription of the treatment of the low-temperature data for staurolite is given in a subsequentsection. Smoothed values of the thermodynamic functions of heat capacity, C!; entropy, Si, or entropy increment, Sfr - S$; enthalpy function, (Ift - Iil)lT; and Gibbs energy function, (GI - Iif;)lT; where r is the reference temperature, are given in Tables 3 through 7 for gehleniteand for two compositions of staurolite as (H3Al1. l5Feot.toxf'er'.t, Fefi*roTio.orMn6.62Al1 and 1e)(Mg6.aaAl15.26)SiEO4s as H2Al2FeaAll5SisOas. The reference temperature for the low-temperatureheat capacity data is 0 K whereas298.15 K is used in the high-temperaturetabulation. Tablel Experimental specific heats synthetic for gehlenite
renp. to;::lt" K J/(s.K) T e m p .t o ; : : I t . K ,r/(e.K)

8.81a.2 333.4t 340.62 311.72 354.75 3 6 1 .t t 36t.66 0.E253 0.t366 0.4473 0 .E 5 t 0 0.t679 0 .t 7 t 9

8.r1.r 6 0. t3l7 0.{509 0.4695 0 .{ E 7 E 0.5056 0.5212 0.539t 0. t5?5 0 . 5 7 4r 0. 5909 0. 6076 0.6249 0 . 6 4l 7
7

8.r1.. I 4. t0 5.t6 5 .6 t 6.5l 7.57 8.58 9.47 10.3E I l. 35 t2.49 13.83 15.31 r6.97 I8.82 20.86 23.06 25.46 27.43 30.19 32.96 36.41 {o.50 44.97 50.07 55.82

renp. toi::lt' K ,r/(g.K)

S e r ie s 1 12.16 13.41 14.68 16.07 17.68 19.63 21.60 23.64 26.10 29.03 3?.32 35.85 39.74 44.16 49.05 54.23 S e r le s 53.07 57.59 62.54 6 7. 7 5 72.73 77.25 81.85 86.91 92.30

S e r ie s 3

S e r ie s 7 247.63 247.tt 252.65 258.32 264.09 269.81 275.47 287.20 256.91 ?92.82 298.73 0.6583 0.6706 0.6803 0.6903 0.7001 0.7092 0.7186 0 . 72 76 0.7356 0.7452 0.7538

0 . 0 0I 5 79 98.01 o.2823 0.002245 1 0 3 . 1 1 0 . 3 0 0 9 0.003088 108.10 0.3180 0.004?t7 113.26 0.3357 0.005851 I 18.59 0.3541 0.008314 0 . 0 1 1 2 1 S e r ie s 4 0.01477 0.01 43 110.43 0.3244 9 0.02s69 I 15.64 0.3422 0.03408 120.65 0.3591 0.04418 I25.58 0.3749 0.05638 0.07727 Series 5 0.08865 0.1088 135.90 0.4075 141.13 0.4230 2 146.32 0.4382 151.43 0.4528 0.1044 156.46 0.4665 0.1222 161.49 0.4801 0.1418 166.67 0.4937 0.1625 t 72 . 1 3 0 . 5 0 7 8 0.1828 t77.72 0.5220 0.2012 183.20 0.5352 0.2197 0.2396 S e r ie s 6 0.2607 180.54 0.5312 185.99 0.5441 19t.27 0.5559 19 6. 5 5 0 . 5 6 8 1 2 0 1. 9 0 0 . 5 8 0 2 2 0 7. 3 3 0 . 5 9 2 1 2r2.84 0.6039 2t8.26 0.6154 223.61 0.6264 228.94 0.6357 234.20 0.6469 239.4t 0.5573

172.37 r78.03 0.003254 Itt. 76 0.003375 It9.46 0 . 0 0 3 9 3 1 1 9 5 .2 l 0 . 0 0 4 5 6 E 2 0 0 .9 5 0.005343 206.tr 0 . 0 0 6 1 7 72r2.54 0.005905 zr0. 42 o . 0 0 7 7 0 E 224.17 0. 008444 2to.61 0.009165 237.09 0 . 0 0 9 9 2 7 2 4 3 .5 6 0.0lo7l 0.0t145 Setl! 0.0t220 0.0r29E 249.t4 0.01375 255.51 0.01468 26r.88 0.0r577 264.32 0.01717 274.a8 0.01922 28t.56 0.02236 288.29 0.02675 295.2t 0.0325r 302.06 0.0405r 0 . 0 5 1 4 1 Serte!
3{0. lo 350. l0 360.l0 370. l0 3 E O .l 0 390.l0 400.00 4lo. 00 420.00 430. O0 4{0. o0

0.655r O.6709 0.6857 0.7001 0.7146 0. 7290 0.7433 0.7570 0,7699 E 0. E397 0. E550 0.8690 0. E794 0.8921 0.9061 0.9154 0. 9289 0. 9394 0.9417 0.957t

Srlea 4 55.82 61.41 56.8E 73.02 79. ll 84.95 90.75 96.35 101.90 0. O5170 0.06333 0.07595 0.09191 0. 1093 0. r2t0 0. 1455 0. r638 0. IE26

S e r le s 8 296.20 302.13 308.00 313.75 0 . 75 0 3 0.7587 0 . 7 6 7| 0.7741

Serles 9 310.52 3t7.32 323.97 330.57 337.13 343.65 350.12 356.57 363.26 3 7 0 .l 8 377.07 383.93 0.77t6 0.7801 0.7893 0.7977 0.8062 0.8136 0.82r0 0.8?82 0.8356 0.8435 0.8503 0.8556

Following Ulbrich and Waldbaum (1976),a zero point entropy, S$,contribution of I 1.506J/(mol ' K) is included in the Gibbs energy function for gehlenitein Table 3. The entropy ofgehlenite at 298.15K and I bar, is, therefore, 210.110.6J/(mol . K). The values for the Gibbs energy function tabulated for the two staurolite compositions do not include a zero point entropy contribution. This contribution is discussed,at length, below. It should be noted that Woodhead (1977)has considered the question of how disorder in Si/Al on the Tz site would be reflected in the crystal structure of gehlenite. Woodhead concluded that gehlenite forming at low temperatures would obey the aluminum avoidance rule for

HEMINGWAY AND ROBIE: GEHLENITE AND STAUROLITE Table 3. Molar thermodynamic properties of gehlenite, Ca2Al2SiO7.
TEI.IP. HEAT ENTROPY ENTHALPY GIBBS NERGY E CAPACITY FUI{CTIOI,I FUNCTION

309

T
KELVI N

c;

(s;-s;)

(H;-H;)/T

-(ci-H;)/T

J/(mol.K)

10
15

20 25 30 35 40 45 50 60 70 80 90 100 ll0 120 130 140 150 160 170 lB0

I On

.0306 .239 . 9I 4 2.42t 4 . 74 2 7.690 11.44 15.69 20.36 25.37 36.04 47.15 58.25 69.01 79.25 8 8 .9 1 98.04 106.7 1I5.0 1230 . 130.6 . 1379 144.9 151.6 157.9 164.0 169.7
t/f.r

.0104 .0785 .279 .725 1.501 2.6t3 4.075 5.875 7.989 10.39 15.95 22.34 29.36 36.85 44.65 52.66 6 0 .8 0 68.99 7 7. 2 0 85.41
o1 RO

.0075 .0583 .2tL .555 1.147 1.980 3.059 4.367 5.881 7.577 Lt.42 15.73 20.35 25.17 30.07 34.98 39.86 q4.67 4 9. 4 0 5 4. 0 4 58.59 63.04 67.40 71.65 75.81 79.86 8 3. 8 2 87.67 9t.4? 9 5. 0 8 98.64 102.1 105.5 r08.8 tI2.0 115.1 118.2 t?t.z 1 2 4. 1 t26.9 t29.7 132.4 135.0

11.506 11.526 I 1.566 11.676 11 , 8 5 6 1 2. 1 3 6 t2.522 1 3. 0 1 4 13.614 14.320 16.03 18.11 20.51 23.t9 ?6.09 29.19 32.44 35.82 39.31 42.87 46.51 50,19 53.92 57.68 61.46 65.26 6 9. 0 6 72.87 76.69 80.49 84.29 88.08 91.85 95.61 9 9 .3 6 103.1 106.8 110.5 It4.t 117,8 121.4 L?5.0 128.5 89.27 9 8 .6 6

- So and (Hr - Ho)lT should be 0.2% or less, which is within the experimental uncertainty of the data. The measuredheat capacitiesof natural staurolite were corected to the two compositionslisted above by assuming the principle of additivity and representingthe sample 140 as, fust, for Tables 4 and6, 1665. g of staurolite, 4.282 g g of corundum 1.514 of lime, 4.394g of zincite, 0.503g , of FeO, 0.403g of periclase,0.479g of rutile, 0.426g of P2O5, 0.468 g of ice, and a deficiency of 0.142 e of g manganosite,and second, for Tables 5 and 7, 1703.737 of staurolite, 356.263 g of pyrophyllite, 17.029 g of brucite, 73.399e of magnesioferrite,5.430g of hematite, g 328.725 of corundum,2.243 g of lime, 10.227 of rutile, e g and 6.510g of 0.710g of P2O5,1.9E6 of manganosite, zincite. The corrections representa changein the specific properties of Table4. Low-temperature molarthermodynamic (Hdlr.rsFe6.i,o) g2All staurolite, Ge35zFe31+Tio.GMn6 rs)(Mgo 44 for chemicalsiteAl15 The 26)SisOas. data af,e uncorrected contributions the entropy. to configurational
TEIIP.

200 210 2?O 230 240 250 260 270 280 290 300 310 320 330 340 350 360 370 3B0 2 73 . t 5 298.r5

101.7 109.8 117.8 r25.8 133.6 141.4 149.0 156.6 164.1 l7t.4 t 78 . 7 185.8 t92.9 199.8 206.7 213.5 220.1 226.7 233.2 239.6 245.9 252.1 180.9 198.6

HEAT T C A P AIC Y

E N T R O P Y E i l T H A L P Y G I B E SE I I E R G Y FUI{CTI0N FUilCTI0ll

ci
5.37 t2,32 t7.55 2t.04 24.03 28.19 3 4 .5 4 43.05 53.98 67.15 99.59 139.5 1 95 . 8 23t ,3 292.6 350.5 410.0 470.2 5 3 0 .3 5 9 0 .0 648.5 705.8 751.1 814.5 866.0 915.6 9 6 3. 4 1009.2 1053.0 1094.8 I134. 7 tt7z.7 1208.9 t243.3 1276.0 1306.9 1336.5 1365.0 t392.2 t4t7 .7 1442.0 1454.0

(si-s;) 1xi-Hitrr -(c;-H;)/T

. .t/(mol K) 2,99 8 . 87 14.95 2 0 .5 0 25.51 30.23 35.04 4 0 .l 8 45.85 52.2t 5 7. 2 1 85.48 I 107. 131.9 159.7 190.3 2 2 3. 4 258.5 295.6 334.2 374.2 {15.2 4 5 7 .r 499.7 542.8 s 8 6 .3 630.0 673.8 7t7.7 7616 . 805.3 848.8 8 9 2 .I 935.2 9t7.9 t020.2 t062.2 I 1 0 3 .7 1144.9 I 185.6 t225.9 124 .9 5 8 6 2 .5 970.0 2.09 5.41 8 .7 l 11.38 13.51 15.56 17.89 2 0. 4 7 23,57 2 7. 2 5 36.50 48.29 62,53 7 9. 0 4 9 7. 6 0 118.0 139.8 162.9 187.0 2119 . 2 3 7. 4 263.2 2 8 9. 4 315.6 341.9 3 6 8. 0 3 9 4. 0 4t9.7 445.2 . 4704 495.2 519.6 543.5 567.1 590.2 612.8 635.0 6 5 6. 6 677.9 698.7 719.0 729.0 5 2 7. Z 585.9 0.91 3. 4 0 6,24 9.t2 11.90 14.57 I7.15 19.70 22,29 24.96 30.70 3 7 .t 8 44.54 5 2. 8 3 52.11 72.35 83.54 95.64 108.6 t22.3 1368 . 152.0 167.8 184.1 201.0 218.3 236.0 254.1 ? 72 . 5 29r.2 310.1 329.3 348.6 3 6 8 .I 387.7 4 0 7- 4 421 2 . 4 4 7. l 467.0 4 8 7. 0 506.9 516.9 335.3 384.0

KELVI T{
a

180.3 185.2 190.0 194.6 199.0 203.2 2 0 7. 2 211.1 ?14.9 218.6 ?z?.0
tza.L

2?8.1 23t.2 233.8 1960 . 206.s

l0 l5 20 2S 30 35 40 45 50 50 70 80 90 100 ll0 t20 130 140 150 160 170 180 190 200 2t0 220 230 240 250 260 270 ?80 290 300 310 320 330 340 350 360 365

r03.2 111.4

the T2 sites. Following the interpretation given by Woodhead, the configurational entropy for gehlenite would be 5.75J(mol . K). Gehlenites synthesized at higher temperaturesappear to have greater disorder leading Woodhead to conclude that disorder of Si/Al on the T2 site would be temperature dependent. The heat capacitiesfor gehlenitewere not corrected for the l0Voof uncrystallized glass (Woodhead, 1977).Robie et al. (1978)have shown that the heat capacities of the feldsparsanalbite, high sanidine,and anorthite differ little from the heat capacities of glassesof the same composition. These observed difierences yielded calculated differencesof 0.8 to 2.3% in Sr - So. We can assume(as a first approximation) that the difference in the gehlenite system should be no larger than that in the feldspar system.Consequently,the uncertainty in the functions 51

273.t5 1184.3 2 9 8 .l 5 1 2 7 0 l.

310

HEMINGWAY AND ROBIE: GEHLENITE AND STAUROLITE

Table 5. Low-temperature molar thermodynamic properties of staurolite, H2Al2FeaAll6SisOas. The data are uncorrected for chemical site-confi gurational contributions to the entropy.
TEi'lP. T r(ELVIr,{ 5 l0 20 ?5 30 35 40 45 50 60 70 BO 90 1C0 110 t20 130 140 150 160 170 lB0 I90 200 210 220 230 ?41) 20c ?70 280 29t) 300 310 3?0 330 340 350 360 365 3.38 18.28 25.38 29.03 31.47 34.71 40,08 57.73 70.12 101.1 . 1395 194.6 ?354 . 2 9 4. 4 348.2 407.5 4677 . 528:0 qc7 0 6 4 7. 0 704,6 760.5 sl4.4 366,5 916.9 965.4 I012.n I056.6 1099.0 1139.5 I178.1 rzt4.9 r249.9 1233.1 1314.5 I 344.6 t?73.7 1401.6 t427.7 rq52.3 t464.7
IiTAT EIITiI(]PY CAPACITY II]TI]ALPY FIIIICTI()II

0t;LlsE,,Eil0Y FUI]CTIO]I

ci

1sl-sit lrri-rrilrr -1t;i-rrillr

'l J/ ( r , r or ) 4.07 1 2. 9 7 21.3C 2 9 . 75 4 ?. 4 9 43.23 54.05 66.9? 82.33 100.7 12?.? 146.8 174.4 B 204. 2376 . ?72.6 309.5 340.0 347B . 428.8 470,6 513.2 556.3 599.9 643.6 6i17.6 731.6 775,6 819.5 En3.2 906.7 950.0
oo? a

2.97 3.19 1 2- 3 7 16.51 19.z',d 21.54 2 3 . 7I ?6,?8 29.19 32.65 41.36 5 2. 5 5 66.10 82.11 100.2 120.1 141.5 164.3 198.1 212,n 239.1 263.8 2',d9.9 316.1 342.3 363.5 39+.5 4204 . 4,i5,9 4 71 , ? 496,? 5 2 t )7 , 544.3 563.6 614.6 637.0 659.9 580.3 701.3 731.9 52ti.4 587.6

1.10 9.01 L 7. 2 3 20.95 2 4. 4 4 27.77 31.03 40.96 4S.15 56.03 64,72 74.29 84.76 96.11 103.3 1?1.4 135.? 149.7 164.9 180.3 I97.1 214.C 231.3 249.1 267.2 285.5
JUq. J

assuming that the impurities can be represented by phasesfor which specific heat data are available. Most geochemistsbelieve that ideal additivity does not truly prevail, but where the required corrections are small (i.e., where the impurities representonly a small percentageof the sample) the error associated with the correction is often less than the uncertainty in the measured specific heat. The potential errors associatedwith this type of correction may be minimized by combining the impurity components into phasesthat are structurally similar to the phase under study (Robie et al., 1976and Klotz, 1950),or by usingacorresponding statesargument(Robie etal.,1982, or Stout and Catalano,1955).In either case,the colTections may be applied directly to the measured specific heats (Robie et al., 1976)in order to provide corrected thermodynamicparametersas a function of temperature, or the integratedproperties may be corrected at a specific temperature (Westrum et a1., 1979). Both procedures should yield the same results at the same temperatureif the same componentsare chosen. Two models are presentedin Table 8 in which the heat capacities of staurolite as given in Tables 4 and 6 are approximated by the summation approach. In model l, the heat capacity of staurolite is approximated by the summation of the heat capacities of equivalent oxide

323.3 342.5 351.9 401.1 440.7 460.7 440,6 500.7 5?0.7 530.7 34U.6 3 9 7. 4

1035.5 7 107 .7 1119,5 1161.0 t?02.0 L?42.5 t 2 6 2. 1 876.9 935.0

properties for Table6. High-temperature molarthermodynamic is naturalstaurolite. formulafor staurolite givenin Table4. The J(mol ' K) and A chemical site-configurational entropyof 121.0 entropyhave been an additional10.0J/(mol' K) of magnetic added the entropyat298.15 givenin Table4. Theequation to K data deviation 0,5Vo. of fit the experimental with an average
TEMP.
T'

HEAT CAPACT! I

ENTRUPY ENTHALPY F Ur , l c T l o N sr (nT-n296)/T

GIBBS BNERCT I ' UN C I O N T -(cT-Ha9E)/T

273.15 11Ct.9 298.15 r277,1

K E L V IN 298.15 325 350 315 400 425 450 475 500 525 550 575 600 625 o50 675 700 725 750 775 800 825 850 875 900 Il70.l | J 4 9. 2 14t3.0 1469.4 1519,7 r 564.9 1505.8 t643.t 1677,4 I 708.9 r738.I 1755.3 1790.6 1814.3 1836.5 1857.5 t877,2 1895.9 19t3,6 1930.3 1946,2 1961.3 i975.5 1989.3 2002.4

J/(oor.K) tI0I.0 12I4.0 1316,3 14r5.8 1512.3 1 6 0 5 ,E I 696.4 L784.2 1E69.4 1952.0 2 0 3 2. 2 2IlU.l 2r85.7 2259.3 2330.9 2 4 0 0 .b 2468.5 2 5 3 4. 7 2 5 9 9. 3 2 6 6 2. 3 2 72 3 . 9 2 78 4 . O 2A42.7 2 9 0 0. 2 2956.4 0.0 l0E.3 I99.2 2 8 2 .r 357.9 491.9 5 5 1. 5 607.0 658.7 707.1 152.6 79 5 . 3 E35,5 873.7 909.7 943,9 976.4 1007.4 1036.9 I 065.0 1092.0 iIri.8 rt 42.5 rr66.2 r I0l .0 1t05.7 rtl7.t

heat (J/g . K, the quantity actually measured)of less than O.lVo for the first case given above for heat-capacity valuesabove 125K and ofless than lVofor valuesfrom 8 to 125K, and for the secondcomposition,of lessthan 1% above 300 K, less thar, 2Vofrom 150 to 300 K, and less than 3Vofrom 50 to 150 K. Below 50 K, the sum of the specific heats of the impurity phasesbecomesnegligible compared to the measuredspecific heat of staurolite. It is both instructive and important to examine the assumptionsand procedures underlying the corrections applied to the specific-heatdata, particularly in light of the rather large mass correction necessaryto obtain the heat capacities of staurolite as listed in Tables 5 and 7. Samples are initially chosen on the basis of chemical purity and/or approximately correct stoichiometry. Corrections for small deviations from ideality are made

rr33.7

1t54.4 1t7E.2 I 204.5 t232.7 t2b2 .4 1293.3 t 3 2 5 .I I357.5 I 3 9 0 .4 t423.7 L457.3 1490.9 1524.6 1558.3 r 5 9 r. 9 1 6 2 5. 5 I 65E.8 L59Z.O 1725.O 1757.8 1790.3

HEMINGWAY AND ROBIE: GEHLENITE AND STAUROLITE Table 7. High-temperaturemolax thermodynamic properties for ideal staurolite, H2Al2Fe4Alr6Si8O4s. chemical siteA configurationalentropy of 34.6J/(mol . K) has beenaddedto the entropy at 298.15 K given in Table 5. The equation fit the experimentaldata with an averagedeviation of 0.EVo.
TEIiP. T K E L V IN 2 9 E .t 5 t25 350 375 400 425 450 475 500 525 550 515 500 625 650 675 700 125 75 0 775 800 825 850 8 75 900 r27l .l 1357.4 L422.3 1479.6 r 530, 7 1576.8 1618.5 1556.6 1691.6 L 72 l . 9 1753.9 1761.9 1806.0 1E32.6 1855.7 LA71.6 1898.3 t9t 7.9 I936.6 1954.4 1971.4 1987.7 2003,2 2 0 18 . 2 2 0 3 2 .5 HEAT C A P A CT Y I C^
E

3ll

Table E. Model I and 2 compositional approximations to natural staurolite.

llodeI
Phase

ilode I

t2 liloles

t2 iloles
Tl 02 7.47 itr0 Hzo 0.268 0.27 0.17 Pyroplrylllte Fe51 03 0.08 0.02 1.50
i ??

ENTROPY

ENTIIALPY FIJITCTION

CIEBS UN};RGY IUNCTION

sr

(Hr-rt298)/T -(cT-d298)/T

Sl02 Al 203

8.00 8.80 ?.67

0.08 0.02

J/ (Do1. K)

l0l9 .6 1133.2 L 2 3 6. 2 I 3 3 6 .4 1433.5 t527.7 L 70 1 . 6 1793.5 r876.8 t957.7 2036.3 2rt2.7 2r87.0 2 2 5 9. 3 2329.8 239A.4 2465.4 2530.7 2594.5 2656.8 2 7r 7 . t 2777.3 2835.6 2 8 9 2. 6

r5r9.o

0.0 108.9 200.5 283.9 3 6 0. 2 430.5 495.3 555.4 511.4 66J.6 7t2.5 75E.4 80r.6 a42.3 880.9 911.4 952.0 985.0 I0t6.4 I046.4 1075.0 1r02.5 1128.7 rr53.9 ll78.l

I0I9.6

r024.3 r035.I 1052.5 1 0 7 3J . 1097.3 r123.7

ll52.I lr82.l I2I3,2 t245,2 1 2 77 . 9 r3ll.l 1344.6 r376.4 l4l2 .4 t 4 4 6. 4 r480.4 1514.3 l54E.l I 5 8 1. 8 1615.3 I 648.6 1581.7 r714.5

Fe203 MgO M 90 H 2 ( )

0.268 0.44

2.67

56 of the model 2 data at 298.15K differs by only +0.2Vo, whereas that of model I is nearly 8Vogreatet. At 900 K, model 2 differs by -lVo whereasmodel I drfrersby +6%6. The values of 51 - 56 have been corrected for the transitions in FeO, SiOz, and H2O. Consequently, the best overall fit is obtained from model 2. The success of the model 2 components over those chosen for model I lies in the similarity of the magnetic

components.Two major changesare presentedin model 2. First, approximation of the contribution of OH to the specificheat of staurolite in model I is made by assuming that model I contains ice and water, whereasin model 2, the OH contribution is introduced through pyrophyllite and brucite. Second. the contribution of Fe2+ as FeO in model I was replaced through the introduction of the differencein the specific heat of a mixed Mg-Fe enstatite and the specific heat of pure Mg enstatite (Krupka et al., 1978,unpublisheddata), where this difference represented the composition FeSiO3. Differenceswere calculatedby subtractingthe summed values for each model from the smoothed heat capacities given in Tables 4 and 6. These differences are shown in Figure l. At temperaturesgreater than 3fi) K, relatively little advantageis obtained by adopting one specific heat model over the other, with the exception of the effect of the a - B transition in quartz. The specific heat of a transition in an oxide component is usually removed by a smoothing process. For the model I data presented in Figure 1, the effect of the c - B transition has been preserved as a visual reminder that when using the integrated properties (e.g., entropy) approach, an additional correction may be required in order to maintain smooth estimated values. Although local deviations of the model 2 specific heat data from the measured heat capacity of staurolite are relatively large at low temperatures,the entropy as 51 -

z40

620
(J e U 40

z u

EXPTANATION o a Model 2 Model I

r00

200

300

400

500

600

700

800

900

TEMPERATURE KELVINS IN Fig. l The percentage deviation ofthe summations from the measured heat Positive deviations indicate values of smaller than the observed heat capacity model I and 2 (see text) capacity of staurolite. the models which are of staurolite.

312

HEMINGWAY AND ROBIE: GEHLENITE AND STAUROLITE

contribution of Fe2+ in enstatite and staurolite. The cooperative ordering efect in FeO is of a different form than the apparent Schottky-type anomaly seen in the dilute paramagnetic salts, staurolite and enstatite; see Figure 2. In addition, pyrophyllite is a better approximation of the contribution of OH to the heat capacity of staurolite than the separateoxide components of corundum, quartz, and ice. The heat capacity data for the natural staurolite sample have been corrected to two arbitrarily chosen compositions as an example of the procedure. These corrections are only first approximations becausewe do not have sufficient data to estimate mixing properties. These results may be adjusted in a similar manner to estimate values for other staurolite compositions. Magnetic entropy of staurolite Schottky (1922)postulatedthat the electronic systemof some atoms in a crystal would undergo excitation between a ground state and higher energy states. Each energy state is also characterizedby a degree of degeneracy. For a simplified casein which there are two levels, each with equal degeneracy, it can be shown that the contribution to the heat capacity of a crystal arising from the presence of electrons in an excited energy level is

A4 A

oB

V3 E z
F. dt

o
Ag

Atr

25
oo o
A

ao
t5
A

a$
AE EXP[ANATION o o ^ Stourolite Model 2 Model I

50

r00

r50

200

250

300

TEMPERATURE, IN KELVINS Fig. 2. A comparisonof the heat capacity of staurolite and the values obtained from the model I and 2 summations(see text). The data are presented as CJT to emphasize the Schottky anomaly in staurolite.

equal to the diference between two Einstein models that are related by two characteristic frequencies fs and 2fs (see,for example, Gopal, 1966).Therefore, the contribution to the heat capacity and entropy of a material from the presence of excited energy states is related to the number of these energy levels and to their degeneracies. The free ferrous ion has 25-fold degeneracy in the ground state that can be reduced or removed, when the ferrous ion is in a crystal, through the combined or separateefects of the local crystal field and cooperative spin-orbit coupling. The spatial distribution of the electronic charge of the free ion is spherically symmetrical. However, when the ion is placed in a crystal the spacial distribution of charges is altered by the distribution of chargesin the neighboringatoms. Chargedistributions (in the form of lobes and other shapes) that are directed toward other atoms are elevated to higher energy levels (splitting of levels) than those that are directed between neighboring atoms. Thus lower symmetry sites produce greater splitting of energy levels. The efect ofthe typical crystal field upon the ferrous ion is to elevate(remove) l0 levels beyond that which could be energetically accessible. The crystal field also exerts a torque upon the orbital momentum resulting in the orbital momentum not being constant in direction and when resolved in Cartesian coordinates to average to zero. This process is called quenching of the magnetic moment of the angular momentum. In some crystal structures the interaction between the energy states of one atom are not independent of the energy states of similar neighboring atoms. In such systems, a mean energy is required to induce population of the different energy states and the associated anomaly (typically a I peak) in the heat capacity is called a cooperative anomaly. Examples of cooperative anomalies of this type may be found in Robie et al. (1982, a,b) for antiferromagnetic ordering in fayalite, tephroite and cobalt olivine. A noncooperativesystem exists ifthe excitation ofthe energy states of the atoms in the structure are independent of the energy levels in the similar neighboringatoms. Where no interdependence exists between the energy levels of similar neighboring atoms, the total energy contributed is equal to the sum of the energies of the independentlevels and theoretical treatment of the system is greatly simplified. The cooperative and noncooperativeanomalieswill be called Schottky anomalies and the anomalies in the entropy or heat capacity will be called Schottky contributions. The Schottky contribution is most easily defined where the energy separation of the different levels is small. In these cases,the contribution is observedat very low temperatureswhere the lattice heat capacity becomes either a Debye-like contribution or an essentiallynegligible contribution of the measured heat capacity. At low temperaturesthe noncooperative Schottky heat capacity anomaly is typically expressedas a bell-shapedpeak that

HEMINGWAY AND ROBIE: GEHLENITE AND STAUROLITE

313

is skewed toward higher temperatures.The temperature of the peak is related to the separationof the levels and the amplitude of the peak is related to the ratio of the degeneracies the levels (e.g., Gopal, 1966).Examples of of this type of heat capacity anomaly may be found in the results of Hill and Smith (1953).It is far more difrcult to recognize Schottky contributions to the measured heat capacitieswhere the energy levels have a large energetic separationand the contribution occurrencesare at higher temperatures.This subject has been excellently discussed by Westrum (1983) and therefore will not be discussed further in this report. The magnetic contribution to the heat capacity of a system may be composed of both a cooperative and a noncooperative contribution. Raquet and Friedberg (1972)concluded that the angular momentum of the 5D ground state of the Fe2+ ion in FeCl2 . 4H2O was quenced and that the ground state of the five spin components was fully split by the combined effects of spin ordering (cooperative) and crystal field splitting (noncooperative). Mdssbauer spectrataken at 4.2Khave been interpreted by Dickson and Smith (1976),Regnard (1976),and Scorzelli et al. (1976) as indicating antiferromagnetic ordering in staurolite. Regnard found an ordering temperatureof 6tl K, and Dickson and Smith obtained7tl K for the ordering temperature. However, the heatcapacity values obtained in the temperaturerange of 5 to 8 in this study do not indicate the strong cooperative exchange that is characteristic of antiferromagnetic ordering. Dickson and Smith (1976) have observed ordering attributable to antiferromagnetic ordering in FeAl2Oain Mdssbauer spectra obtained at temperaturesbelow 8 K. Because of the temperature dependence of magnetic susceptibility data, Roth (1964)suggestedthat antiferromagneticordering developed in FeAl2Oabelow 8 K, but he failed to detect an antiferromagnetic state in FeAl2Oa using neutron diffraction. Roth assumed that FeAl2Oa failed to show antiferromagneticorder in neutron diffraction because of a strong interaction between Fe2* in tetrahedrally and octahedrally coordinated sites that destroyed long rangeorder on the tetrahedral sites. Similarly, long range order in staurolite may not develop. Lyon and Giauque (1949) and Hill and Smith (1953) measuredthe heat capacities of the dilute paramagnetic ferrous salts ferrous sulfate heptahydrate and ferrous ammonium sulfate hexahydrate, respectively. Neither ferrous compound showedan anomaly in the heat capacity that would be consistent with an interpretation of antiferromagneticordering at low temperatures,although two maxima were observed in the low-temperature heat capacitiesof each phase. These maxima were interpreted to be associatedwith a splitting of the ground state of the ferrous ion into two levels of equal degeneracywith a separationof 3 to 6.5 cm-r that gave rise to the specificheat anomaliesat 2 to 4 K and a second group of three

levels centered at 38 cm-r or less for the anomalies observedin the 15 to 20 K region. Thus, the ferrous ions in staurolite may develop small separationsof the ground state similar to those seen in the dilute ferrous salts. Raquetand Friedberg (1972)measuredthe heat capacity of a ferrous salt (FeCl2 . 4H2O) of greater ferrous concentration than those studied by Lyon and Giaugue (1949)and Hill and Smith (1953). Raquet and Friedberg observed a sharp )r-type peak consistent with low-temperature antiferromagnetic ordering. However, the exchange energy associated with the antiferromagnetic ordering was an order of magnitudesmaller than the zerofield splitting of the ground state of Fe2+, that is, the )r peak for the cooperative processwas at a lower temperature than the noncooperative anomaly. Our heat-capacity results do not extend to a low enoughtemperatureto rule out a cooperative interaction in staurolite nor do they rule out a small separationof a lower doublet of the Fe2* ion. Bearing in mind the limitations of theory discussedabove and the ambiguity associatedwith the chemistry of our staurolite sample, we shall use several simple methods in an attempt to estimate that portion of the magnetic heat capacity and, consequently,entropy not defined by our measurement. In the preceding section, an empirical case was presentedin which the entropy of staurolite was shown to be adequatelyapproximated by the summation of entropies ofthe model 2 componentsbut not as favorably predicted by the summation of entropies of the model I components. In this section, we shall examine the theoretical aspects of the magnetic entropy of staurolite, using a modified version of model 2 to approximate the staurolite lattice contribution to the entropy. Staurolite is a paramagneticsalt. The low symmetry of the several sites in which the Fe2* ion may be found (Smith, 1968, and Takuchi et al., 1972) and the low concentration of Fe within several of those sites (Tak6uchi et al.) allow us to initially assume an ideal state for each Fe2+ ion in which the ion interacts only with its immediate crystal field and not with other Fe ions. The same assumptioncan be made for Fe3+ and Mn2*. The iron transition group paramagnetic salts yield experimental magneton values consistent with calculations assuming a quenched angular momentum. Consequently, the contribution of the electronic system of the iron transition group ions to the entropy is Rln(2S + 1), where S is the spin quantum number for which S : 2 for Fe2* and S : 512 Mn2* and Fe3+ (e.g., see Kittel, for

r976t.
Severalempirical approaches have been used to extract the magnetic contributions to the entropy and/or heat capacity (e.g., Lyon and Giauque, 1949;Osborne and Westrum, 1953;Stout and Catalano, 1955;and Friedberg et al., 1962). Osborne and Westrum assumed that the lattice heat capacity could be approximated by the heat capacity of an isomorphousdiamagneticphasecontaining a similar cation. Lyon and Giauque, and Stout and

3t4

HEMINGWAY AND ROBIE: GEHLENITE AND STAUROLITE

Catalano using a similar approach, corrected the heat capacity of the isomorphous diamagnetic phase using a corresponding-states argument. Friedberg et al. assumed that the lattice heat capacity behavedlike that ofa Debye solid having Cp n T3 and that the magnetic contribution above the Schottky anomaly should go to zero as ?-2. Friedberg et al. plotted CpTz as a function of T5 and fit the data with a straight line. The slope of this line represented the best estimate of the constant for the Debye lattice approximation. The use of an isomorphous diamagnetic phase as a model for the lattice heat capacity of a paramagnetic phasecan be theoretically justified only if we assumethat the Einstein and Debye temperaturesof the model differ little from those of the true lattice and we recall the additive properties of the Schottky anomaly given in the first paragraphof this section. None of these procedures is directly applicable to an interpretation of the Schottky anomaly in staurolite. No heat-capacitydata exist for a diamagneticphase isomorphous with staurolite. Furthermore, the approachusedby Friedberg et al. (1962)assumesthat all the paramagnetic ions experiencethe same zero-field splitting, that is, that all the paramagnetic ions reside in similar lattice sites having essentially identical crystal fields. Smith (1968) and Takuchi etal. (1972)analyzedgamma-rayresonance spectrafor staurolite and showed that the Fe site is only 77Vo occupied by Fe2+. They have further shown that only 80Voof the Fe2* ions are located in the Fe sites. Similar results are given by Bancroft et al. (1967). The structure analysis of staurolite by Smith (1968) locates Fe2* in the Fe site (tetrahedra)and in the A(3A), A(3B), U(l), and U(2) octahedra(using Smith's notation). The U(1) and U(2) octahedra are larger than the Al(3A) and A(3B) octrahedra. Manganeseis located in the U(1) and U(2) octahedra. Smith noted that the Fe3+ reported in chemical analysesof staurolites may represent oxidation during the analysis procedure because the Mcissbauer pattern does not contain ferric iron peaks. Assuming that the electric fields produced by the neighboring atoms at the A1(3A) and Al(3B) sites are essentiallyidentical, and making similar assumptionsfor U(1) and U(2) and for the tetrahedral Fe sites, we conclude that the Fe2* ions interact with a minimum of 3 different crystal fields which may yield a different distribution of split sublevels.This effect would not be seenin the results of Krupka, Hemingway, and Robie (1978, unpublisheddata) for enstatite becausethe crystal fields of the M1 and M2 sites would be nearly equivalent (Morimoto, 1959).Consequently,we would expect the maximum in the Schottky anomaly found by Krupka et al. (1978,unpublisheddata) and shown schematically in the model 2 data set to be more clearly defined and to represent a smaller energy spectrum than that seen in staurolite, as may be seen in Figure 2. Somemay ask, "If we know the approximate contribution of the iron group transition ions to the entropy,

Rln(2S * 1), and, because at room temperature we are presumably at a temperature sumciently greater than the maximum in the Schottky anomaly to assumethat there is an equal distribution among the spin states and hence a constant magneticcontribution to the entropy, why do we worry about the specifics of the lattice and magnetic entropies at low temperatures?". The answer lies in the energeticsof the system. Should a cooperative interaction exist, then the zero-splitting should be large compared to the exchange energy, and the specific heat of staurolite should exhibit a cooperative anomaly at a Nel point at some temperature less than the maximum in the Schottky anomaly, as was shown in the work of Friedberg et al. (1962)for FeCl2 . 4HzO (also see Kittel, 1976, Chapt. 14 and 15). Alternatively, should the magnetic contribution of the Fe2* ion to the entropy of staurolite arise from a splitting of the magnetic sublevels into an upper triplet and lower doublet as is commonly seen in Fe2+paramagneticsalts (Friedberg et al., 1962and Lyon and Giauque, 1949),then a substantialcontribution to the entropy could arise below 4 K from a splitting of the lower doublet. Without a reasonably valid model for the magneticheat capacity distribution, we cannot determine what portion of the magnetic heat capacity has been measureddirectly with the calorimeter. We may calculate the magnetic entropy for the iron transition group ions to be 44.1 J/mol . K or 43.2 J/mol . K if we assumeall the iron as the Fe2+ion based upon the corrected chemical analysisfor the sample.The lattice entropy may be estimated in two ways, neither of which is very rigorous. A first approximation to the lattice heat capacity may be obtained from a plot of CrT2 vs. T5 following, for example, Friedberg et al. (1962).Using the roughly linear trend for the experimental data in the 20 to 30 K region, we obtain equation(l) c p = 5 8 9 0 r - 2+ 8 l . 2 5 x l o - 5 ? 3 (l)

from which we estimate the lattice heat capacity as CL = 81.25xl0-5T3.The heat capacityderivedfrom this equation exceeds the measured heat capacity of staurolite above 35 K, and the calculated magnetic entropy (measured entropy less the estimated lattice entropy of 26.5 J/mol . K) is about half of that predicted from theory. The magneticheat capacity approximation derived from equation (l) and designatedas model 3 is shown graphically in Figure 3. We are not surprised that the approach followed by Friedberg et al. (1962)fails for staurolite. The Schottky anomaly reported by Friedberg et al. was located at about 3 K where materials behave more like Debye solids than they do at the 20 to 30 K temperature range of the staurolite anomaly. Where multiple Schottky anomalies can be expectedto be superimposedupon each other, as in the case of staurolite, a component of the lower temperature Schottky anomalies may contribute to the

HEMINGWAY AND ROBIE: GEHLENITE AND STAUROLITE

3t5

at 170 K is 42.4 J/mol . K. This value represents about 96% of the theoretical magnetic entropy. We present two additional estimates of the magnetic heat capacity of staurolite in Figure 3. In the first case, 14 : o the lattice heat capacity estimatesfrom equation (2) were E subtractedfrom the observed heat capacity of staurolite. t In the secondcase, following the procedures outlined by Friedberg et al. (1962), we may estimate the averaged separationof an assumedupper triplet of the ground state of the Fe2+ion as 38.8 cm-l from the coefficientof T-2 \ term in equation (l) and, following the equationsgiven by Lewis and Randall (1961, p. 423), we calculate the \ magnetic entropy and heat capacity by assumingall iron as Fe2+. For convenience,this shall be designatedmodel o 20 ''o ''{o 160 'o 4. Also following Friedberg et al., one may calculate an ::"*Jr*.,:'-,,;i: averageseparationof 30.7 cm-r from the temperatureof Fig. 3. Estimates the magnetic of heatcapacity staurolite. the maximum in the Schottky anomaly. of Thediamonds represent values the obtained assuming lattice the Becauseof the complexity of the staurolite structure, heatcapacity(model3) obtained (l). from equation The open we cannot expect to be able to extract physically mean(model squares wereobtained the assuming latticeheatcapacity ingful data regarding the splitting of the ground state of 2) obtained from equation(2). The triangles and circleswere (1961), the iron group transition ions in a particular site in calculated from the equations Lewis and Randall of assuming uppertriplet anda lowerdoubletwith a separation staurolite from the measuredheat capacity data, because an the heat capacity is an average of all such effects. It is of 38.E cm-t and30.7-trespectively, model as 4. unlikely that either of the postulated energy levels is completely degenerateas we have assumed. However, slope derived from the data fit as Ce'F vs.?6, causing an Friedberg et al. (1962)have shown that, even though the error in the estimate of the lattice component. number oflevels and their distributions cannot be uniqueAn alternate approximation to the lattice heat capacity ly determined, one can eliminate some arrangementsby may be made using the model 2 summation discussed examining the variation of the maximum of the Schottky above combined with the version of the corresponding- anomaly, C.u*, with the various arrangementsof levels statesargument presented by Lyon and Giauque (1949). and distributions. Gopal (1966)presented a similar arguLyon and Giauque (1949) found the ratio of the heat ment. For staurolite, we find that C-.* from each approxcapacity of FeSOa . 7H2Oto the heat capacity of diamag- imation is close to the predicted C."* : 0.62R for the netic ZnSOa . 7HzO, both taken at the sametemperature, model of energy levels used above. to vary linearly between 65 and 200 K. If we substitute Adequate measurements of paramagnetic resonance the heat capacity of the synthetic MgSiO3(Krupka et al., and susceptibility for staurolite have not been found in 1978,unpublished data) for the calculated heat capacity the literature. Runciman et al. (1973) have calculated a of FeSiO3used in the model 2 approximation, then the scaled model 2 c,anbe used as an estimate of the heat | 006 capacity of a diamagnetic phase for staurolite in the procedure described by Lyon and Giauque. ,t' r oos In Figure 4, we present a plot of the ratio of the heat 4_ capacity of staurolite, Cp.. to the heat capacity of the u t oo,o scaledmodel2 summation,Cp,2. Between170and 350K, the ratio varies linearly. We can approximate the lattice d loo3 heat capacity of staurolite from equation (2) if we assume U that the model 2 summation is a reasonableapproximarooz * U tion to the lattice heat capacity at room temperatureand if ? roor we assumethat the ratio maintains the same relationship at lower temperatures.
( q

ii

= cp.Jcp.z 1.000+ 3 x l0-5(T-

160.6)

(2)

t0L t50

IN IEMPERATURE, KETVINS

The construction of this model requires that the total thermal contribution to the excess heat capacity arising from the electronic system be fully developed below the temperature at which the ratio of Cp,JCp,z becomes linear. The magnetic entropy calculated from this model

Fig. 4. Ratio of the measuredheat capacity of staurolite to the heat capacity calculated at the same temperature from the modified version of the model 2 approximation discussedin text. The linear equation derived from this ratio is usedto estimatethe lattice heat capacity of staurolite below 100 K.

316

HEMINGWAY AND ROBIE: GEHLENITE AND STAUROLITE

separation of 1.5 cm-l for a lower doublet for Fe2* in octahedral coordination in enstatite. They reported unpublished paramagneticspectrum data in support of their interpretation. Similarly, Runciman et al. (1973)calculated a splitting of 0.27 cm-r for Fe2* in octahedral coordination in olivine. Friedberg et al. (1962)suggesta splitting of less than 0.8 cm-l for the lower doublet for Fe2* in octahedral coordination in FeCl2 . 4H2O. Consequently, it is reasonableto assumethat the crystal field of staurolite would causea small separationof the lower doublet of the ground state of Fe2+. Because our specific heat measurements staurolite show a continuous decrease for (within the limits of experimental accuracy) both in the absolutevalue and in the estimatedmagneticcontribution below 15 K, we can safely assume that if a split lower doubletexists,the separation mustbe lessthan 0.8 cm-r. If model 4 were a fair representationof the behavior of Fe2* in staurolite, then an entropy contribution of Rln 2 per mole of Fe2+ would be developedbelow I K. This entropy contribution would be added to the values calculated from the model 2 and 3 approaches,yielding 60.9 and 45.0 Jimol . K, respectively, for the magnetic entropy. The model 4 heat capacities are not a good representation of either the model 2 or 3 estimatedheat capacities. Although the peak temperatures and maxima values as derived from the model 2 and 3 approximations are generallyconsistent with these values derived from model 4, the agreementin estimated heat-capacity values is particularly poor at temperaturesbelow the maximum in the Schottky anomaly where the lattice heat capacity contributions becomenegligible. Even when a separation of 30.7cm-r lsee discussion model 3 above)is considof ered, the slope of the Schottky anomaly at temperatures below the maximum is not consistent with the theoretical values calculated from model 4 (see Fig. 3). Although several explanations for the difierence between the theoretical and experimental estimated curves could be presented, we think that the mismatch reflects the antiferromagnetic ordering observed by Scorzelli et al. (1976),Dickson and Smith (1976),and Regnard(1976). The lack of a pronouncedanomaly in the low temperature heat capacity of staurolite is not inconsistent with this interpretation when one considers the concentration of chemicalimpurities and the non-ideal distribution of Fe2+ in natural staurolite, where both effects would contribute to a broadeningofthe peak associatedwith spin ordering. Regnardand Scorzelliet al. have noted that at4.2Kthe quadrupole interaction is of the same order of magnitude as that produced by magnetic interactions. A calculation of the magnetic entropy of staurolite based upon the model 2lattice approximation discussed above yields a value of about 42 Jlmol . K at temperatures sufficiently larger than the temperatureof the maximum in the Schottky anomaly to allow us to assumethat the magneticcontribution is constant. Although no claims are made herein for the absolute accuracy of our lattice model. we think that the model is sufficientlv accurate to

suggestthat the measuredheat capacity of staurolite and the extrapolation of these data to 0 K underestimatethe magneticcontribution to the entropy of staurolite. On the basis of the evidence and observationspresentedabove, limits can be placed upon the error in the entropy of staurolite attributable to unresolved magnetic entropy below 5 K as 10t10 J/mol . K. Chemical site-configurational contributions to the entropy of staurolite Ulbrich and Waldbaum 0976\ have calculated the chemical site-configurationaland the magnetic contributions to the entropy of staurolite. These calculations are based upon a simplification of the occupanicesreported by Smith (1968).For a chemical composition for staurolite as HaFeaAllsSi6Oas, Ulbrich and Waldbaumgave53.5 J/(mol ' K) for the magnetic entropy (therefore assuming all iron as Fe2*) and 23.0Ji(mol . K) for the chemical siteconfigurational entropy. Smith (1968)and Tak6uchi etal. (1972\have shown that the formula for staurolite adopted by Ulbrich and Waldbaum (1976) too idealized,as it requiresnearly halfthe is iron to be in the ferric state for electroneutrality, which is at variance with experimental results showing iron to be predominantly in the ferrous state. N6ray-Szab6 and Sasv6ri (1958) have given the formula HzFe+AheSisO+a for staurolite which requires all iron to be in the ferrous state for electrostatic balance. Smith noted that this formula conflicts with the water contents reported by Juurinen (1956).Takuchi et al. reported a substantially lower water content for the staurolite they studied than that found by Juurinen, but they still require three hydrogensper 48 oxygens. On the basis of an analysis of the location of hydrogen in staurolite and eight chemical analyses for staurolite, Takduchi et al. (l972'l have shown that the ideal staurolite should have between 2 and 4 hydrogens per 48 oxygens. Substantial substitutions of divalent Mg for Al and of Al for Si can lead to higher water contents in staurolite. Thus, we may take the two formulations of Takduchi et al. cited above to represent the two limiting cases for ideal staurolite. Smith (1968) has shown that Al and Fe are found in slightly higher concentrations in the A1(3A) sites as compared to Al(3B). The Al(3) octahedra cannot accommodate concurrent occupancy of an octahedral cation and hydrogen (Takduchi et al., 1972). This limitation would imply a correspondingly higher occupancy of hydrogen in the A(3B) octahedra(the P(lB) sites of Takduchi et al.). Tak6uchi et al. found nearly identical occupancy of hydrogenin the P(lB) and P(lA) sites(whereP(lA) is located in the Al(3A) octahedra). Consequently, we may treat the Al(3) octahedra as being identical, without causinga significanterror in our estimatesof the chemical site-configurational contribution to the entropy of staurolite. Following the procedures and assumptionsoutlined by

HEMINGWAY AND ROBIE: GEHLENITE AND STAUROLITE

317

Ulbrich and Waldbaum (1976)and using the site occupancy data of Smith (1968)as a guide, we may estimate 34.6 and 121.0J/(mol . K) for the chemical site-configurational entropy contribution for staurolite having the compositions H2Al2FeaAll6Si6Oas and (H3Al1.1rFe.6.io) tn"3.t, (Mg6.aaAl15.26)Si6Oas, FeSj4Tio.gEMno.ozAlr.rs) respectively. Entropy of staurolite The entropy of staurolite may be calculated through a summation of the calorimetrically determined entropy, the chemical site-configurationalentropy, and additional magnetic entropy not extracted through the measured heat capacities, or obtained through an analysis of reversed phase equilibrium data. Ultimately, equilibrium data must be used to evaluate the accuracy of our estimates of the additional magnetic and chemical siteconfigurationalentropies. Our best estimate of the entropy of staurolite as (H3Ah.rsFeo2.to)(Fez2.izFe3.1+Tro.oaMno.ozAlr.rs) (Mgo.s Alr5.26)Si8O4E 298.15 K and I bar is ll0l.0-+12 at J/(mol ' K); for H2Al2FeaAl16SfuOa6 best estimate is our 1019.6+ J/(mol . K). An additional10J hasbeenadded l2 to the entropy of the natural staurolite composition on the basis of our analysis of the magnetic entropy. We think that the assumptionsthat have been made in calculating the entropy of the latter composition will yield a value that may be consideredto representthe minimum entropy for this ideal staurolite. Relatively few experimental phase equilibrium data involving staurolite exist in the literature (Richardson, 1966, 1968; Hoschek, 1967, 1969; and Ganguly, 1968, 1972;Rao and Johannes 1979;Y ardley, l98l; Pigageand , Greenwood, 19E2). those data, many reactionsinvolve Of phaseslike almandine, chloritoid, and Fe-cordierite for which we have no good estimates of the entropy. The equilibriurn data are not reversed for reactions involving staurolite and iron phases,like magnetite,for which good estimates of the entropy exist. Consequently, we are unable to further examine the accuracy of our estimates of the entropy of staurolite at this time. Acknowledgments provided the gehlenitesamplefrom JamesA. Woodhead material synthesized characterized his PhDdissertahe and for tion. E-an Zen providedthe staurolite sample and supervised Jane H. Hammarstrom who preparedand characterized the sample. FORTRANprogram A usedto calculate Schottky the thermalfunctionswas kindly providedby RobertChirico.We thankour U.S. Geological Surveycolleagues, TrenHaselton H. andSusan Kiefer for their helpfulcomments. W. References Bancroft, M., Maddock, G. and Burns,R. G. (1967) G. A. Applications the Mrissbauer of effectto silicate mineralogy-I. Iron silicatesof known crystal structures.Geochimica et Cosmochimica Acta. 31 2219-2246. .

Dickinson, B. L. and Smith, G. (1976)Low-temperature optical absorption and Mossbauer spectra of staurolite and spinel. CanadianMineralogist, 14, 206-215. Friedberg, S. A., Cohen, A. F. and SchellengJ. H. (1962)The specific heat of FeCl2 ' 4H2O between 1.15 and 20 K. Journal of the Physical Society of Japan, 17, 515-517. Ganguly, J. (196E)Analysis of the stabilities of chloritoid and staurolite and someequilibria in the system FeO-AlzOr-SiOr H2M2. American Journal of Science, 26f,277-29E. Ganguly, J. (1972)Staurolite stability and related parageneses: Theory, experiments, and application. Journal of Petrology, 13.335-365. Copal, E. S. R. (1966) Specific Heats at Low Temperatures. Plenum,New York. Hemingway,B. S., Krupka, K. M. and Robie,R. A. (l9El) Heat capacities of the alkali feldspars between 350 and 1000K from differential scanning calorimetry, the thermodynamic functions of the alkali feldspars from 298.15 to 1400K, and the reaction quartz + jadeite = analbite. American Mineralogist, 66,1202-1215. Hill, R. W. and Smith, P. L. (1953)The anomalousspecific heat offerrous ammonium sulphate. Physical Society of London, Proceedings 66, 22E-232. A, Hoschek, G. (1967)Untersuchungenzum stabilitdtsbereichvon chloritoid und staurolith. Contributions to Mineralogy and Petrology, 14, 123-162. Hoschek, G. (1969)The stability of staurolite and chloritoid and their significancein metamorphismof pelitic rocks. Contributions to Mineralogy and Petrology, 22,2OE-232. Juurinen, A. (1956) Composition and properties of staurolite. Annals of the Academy of Science Fennical, Series A. III Geol.,47, l-53. Kittel, C. (1976)Introduction to Solid State Physics.John Wiley & Sons,Inc., New York. Klotz, I. M. (1950) Chemical Thermodynamics. Prentice-Hall, Inc. Englewood Clifis, New Jersey. Krupka, K. M., Robie, R. A., and Hemingway, B. S. (1979) High-temperatureheat capacitiesof corundum, periclase, anglass, muscovite, pyrophyllite, KAlSirOs orthite, CaAlzSi2OE glass,grossular,and NaAlSi3Osglass.American Mineralogist, 64. E6-101. Revised Thermodynamics. Lewis, G. N., and Randall,M. (1961) by Pitzer, K. S. and Brewer, L., McGraw-Hill Book Co., New York. Lyon, D. N. and Giauque, W. F. (1949)Magnetismand the third law of thermodynamics.Magnetic properties of ferrous sulfate heptahydratefrom I to 20 K. Heat capacity from I to 310 K. Journal ofthe American Chemical Society, 71,1647-1656. Morimoto, N. (1959)The structural relations among three polymorphs of MgSiO3-enstatite,protoenstatite, and clinoenstatite. CarnegieInstitution of Washington Year Book, 58, 197. N6ray-Szab6, I. and Sasv6ri, K. (195E) On the structure of staurolite, HFezAlqSLOz+.Acta Crystallographica, ll, E62865. Osborne, D. W. and Westrum, E. F., Jr. (1953) The heat capacity of thorium dioxide from l0 to 305 K. The heat capacityanomaliesin uranium dioxide and neptuniumdioxide. Journal of Chemical Physics, 21, 1884. Pankratz, L. B. and Kelley, K. K. (1964)High-temperatureheat contents and entropies of akermanite, cordierite, gehlenite, and merwinite. U.S. Bureau of Mines Report of Investigations. 6555.

318 Pigage, C. and Greenwood,H. J. (1982) L. Internallyconsistent estimates of pressure and temperature: The staurolite problem. American Journal of Science. 282.943-969. Rao, B. B. and Johannes,W. (1979)Further data on the stability of staurolite + quartz and related assemblages. Neues Jahrbuch fiir Mineralogie Monatshefte, 437-M7. Raquet, C. A. and Friedberg, S. A. (1972) Heat capacity of FeCl2'4H2O between0.4 and 5.3 K. PhysicalReview B, 6, 4301-4309. Regnard, J. R. (1976) M<issbauerstudy of natural crystals of staurolite.Journalde Physique,ColloqueC6, 3'7,797-800. Richardson,S. W. (1966)Staurolite, CarnegieInstitute of Washington Year Book, 65, 24E-252. Richardson, S. W. (1968) Staurolite stability in a part of the system Fe-Al-Si-O-H. Journal of Petrology, 9, 467-488. Robie,R. A. and Hemingway,B. S. (1972) Calorimeters heat for of solution and low-temperatureheat capacity measurements. U.S. Geological Survey ProfessionalPaper 755. Robie, R. A., Hemingway, B. S. and Takei, H. (1982)Heat capacities and entropies of Mg2SiOa,Mn2SiOa,and Co2SiOa between 5 and 380 K. American Mineralogist, 67, 470482. Robie, R. A. and Hemingway,B. S., and Wilson, W. H. (1976) The heat capacities of Calorimetry Conference copper and of muscovite KAlr(AlSi3)Oro(OH)2, pyrophyllire Al2Si4Oro(OH)r,and illite K:(AlzMg) (Si1aAl2)Oa6(OH)s between 15 and 375 K and their standardentropies at 298.15K. U.S. GeologicalSurvey Journalof Research, 4,631-644. Robie,R. A., Hemingway,B. S. and Wilson, W. H. (1978) Lowtemperature heat capacities and entropies of feldspar glasses and of anorthite. American Mineralogist, 63, 109-123. Roth, W. L. (1964)Magnetic properties of normal spinels with only A-A interactions. Journalde Physique,25,507-515. Runciman,W. A., Sengupta,D. and Marshall, M. (1973)The polarized spectra of iron in silicates. L Enstatite. American Mineralogist, 58, 444450. Runciman,W. A., Sengupta, D., and Gourley, J. T. (1973) The polarized spectra of iron in silicates. II. Olivine. American Mineralogist, 58, 451456. Schottky, W. (1922)The rotation of atomic axes in solids (with magnetic, thermal and chemical applications). Physikalische Zeitschrift. 23. 448-455. Scorzelli,R. B., Baggio-Saitorvitch, and Danon, J. (1976) E. Mcissbauerspectra and electron exchangein tourmaline and staurolite. Journal de Physique, Colloque C6, 37, 801-805. Smith, J. V. (1968)The crystal structure of staurolite. American Mineralogist, 53, I 139-1 155. Stout, J. W. and Catalano, E. (1955) Heat capacity of zinc fluoride from 1l to 300 K. Thermodynamic functions of zinc fluoride. Entropy and heat capacity associatedwith the antiferromagnetic ordering of manganousfluoride, ferrous fluoride. cobaltous flouride. and nickelous fluoride. Journal of Chemical Physics, 23, 2013-2022. Takduchi, Y., Aikawa, N., and Yamamoto, T. (1972) The hydrogen locations and chemical compositions of staurolite. Zeitschrift fiir Kristallographie, 136, l-22. Ulbrich, H. H. and Waldbaum, D. R. (1976)Structural and other contributions to the thirdlaw entropies of sificates. Geochimica et Cosmochimica Acta. 40. l-24. Weller, W. W. and Kelley, K. K. (1963)Low-temperature heat capacitiesand entropies at298.15 K ofakermanite, cordierite, gehlenite, and merwinite. U.S. Bureau of Mines Report of Investigations6343. to Westrum,E. F., Jr. (1983) Lattice and Schottkycontributions the morphology of lanthanide heat capacities. Journal of ChemicalThermodynamics, 305-325. 15, Westrum, E. F., Jr., Essene,E. J. and Perkins,D., III (1979) Thermophysical properties of the garnet grossular: 11, Ca3Al2Si3O12. Journalof ChemicalTbermodynamics, 5766. Woodhead, J. A. (1977) The crystallographic and calorimetric effectsof Al-Si distribution of the tetrahedral sites of melilite. Ph.D. dissertation, Princeton, University Microfrlms, Ann Arbor, Michigan. Yardley,B. W. D. (1981) note of the composition and stability A of Fe-staurolite. Neues Jahrbuch fiir Mineralogie Monatschefte. H3. 127-132. Zen, E-an (1981)Metamorphic mineral assemblages slightly of calcic pelitic rocks in and around the Taconic allochthon, southwest Massachusettsand adjacent Connecticut and New York. U.S. GeologicalSurvey Professional Paper ll13.

Manuscript received, March 3, 1983; accepted for publication, October 3, /983.