June, 1953
MOLECULAR INTERACTION BETWEEN %-PROPYL ALCOHOL AND IRON OR IRON OXIDES
. 591
and vapors through capillary tubes could in general mer5 and for sulfur dioxide by Trautz and WeitzeL6 As far as we know, the only measurement for be safely applied a t temperatures near or even below the critical temperature where deviations the viscosity of ethyl chloride vapor was that by from the gas laws occur. Thus it seems probable Vogel at O o , using the oscillating disc method. that the deviations from Sutherland’s law, which His value is 937 X lo-’ c.g.s. units. It is easy to occur for some vapors a t lower temperatures, could .see that our extrapolated value is in complete be attributed to a probable change in the molecular agreement with that of Vogel. Rankine’ observed that for most gases (T,/C) = field in this range of temperatures. If, now, ( T 3 / z / q ) is plotted against T as 1.14, where To is the critical temperature. For ordinate, Sutherland’s equation (1) gives a straight ethyl chloride vapor T , = 460.3OK.; therefore, line, with the negative intercept on the T axis, ( T J C ) = (460.3/398) = 1.16, in agreement with as Sutherland’s constant C. We see in Fig. 1 Rankine’s rule. the experimental curve is first concave upwards In conclusion the writer wishes t b express his and then becomes straight at higher temperatures. appreciation to Prof. M. Fahmy for his helpful The straight part gives for C the value 398. Similar discussions during the progress of the work. curves were obtained for bromine vapor by Ran(5) 0.Zininier, Vrndlg. P h y s . Ges., 14, 471 (1912). kine,’ Braune and others4; for ethylene by Zim(6) Trauts and W. Weiteel, A n n . PhUs., [4] 7 8 , 305 (1925). (4) H.Braune, R. Basch and W. Wentrel, 2. p h y s i k . Chem., 187, 447 (1928).
(7) A. 0. Rankine, Proc. Roy. SOC.(London), 848, 190 (1911); 86A, 106 (1912).
MOLECULAR INTERACTION BETWEEN n-PROPYL ALCOHOL AND IRON OR IRON OXIDES B Y EDJVaRD
H. LOESER,WILLIAMD. HARKI~YS’ AND SUMXER B. TWISS
Contribution from the Chrysler Corporation of Detroit, Michigan, and the Department of Chemistry, Universitu of Chicago Received November 28, 1968
’
The adsorption isotherms a t 25“ of n-propyl alcohol on reduced iron, untreated iron, Fen03 and FesOr have been determined. The phase changes which occur in the adsorption isotherms a t very low pressures are primarily first order, although one second-order change is observed. The phase changes are more prominent in the alcohol film than in heptane, and the presence of oxides on the iron surface has a marlced effect on the phase changes of adsorbed alcohol. The values of r e , decrease of free surface energy of a solid due to the adsorbed vapor, and WA, work of adhesion between adsorbed liquid and solid, are given. These values, calculated from the adsorption data for n-propyl alcohol on the solids, increase in the order: reduced iron, Fe304,untreated iron and Fe20a. The greater the oxygen in the solid surface, the higher the energy change of adsorption. Abnormal B.E.T. and Huttig plots, which lead t o negative values of “c,” are obtained in several cases. Comparison of the area values for the four solids obtained by the Langmuir, B.E.T., Huttig and H.J. methods indicates that use of n-propyl alcohol data for area measurements gives results of only fair self-consistency and accuracy. Unpublished data for n-propyl alcohol on anatase are included for purpose of area correlation.
Introduction Practically no difference was found in the adsorption isotherms a t 25’ of a lion-polar vapor, nheptane, on a clean metal and on the same metal coated with oxide.2 Although the values for the decrease of free surface energy of the solid (n) caused by the adsorbed heptane film increase in the order copper, silver, lead and iron, the Ir-value is essentially independent of the presence or absence of an oxide film on the surfaces of these metals. There is evidence3 that the effective cross sectional area of the heptane molecule in the completed monolayer on the surface of non-porous solids is nearly iadependent of the nature and spacing of the solid lattice. The shape of the adsorption isotherms and the n-value for a polar vapor, water, appear to be related to the percentage of ash in graphite.4 The purpose of this paper is to present t8heresults of the adsorption studies of n-propyl alcohol, a molecular species having polar characteristics, on iron and iron oxides. Investigations of the physical adsorp(1) Deceased. (2) W. D. Harkins and E. H. Loeser, J . Chem. PhUs., 18, 556 (1950). (3) E. H. Loeser and W. D. Haikins, J . Am. Chem. Soc., 1 2 , 3427 (1950). (4) W. D. Harkins, G. Jura and E. H. Loeser, ibid., 68, 554 (1946)
tion of two other polar adsorbates, n-amyl chloride on reduced iron and iron oxide, Fez03,and oxygenfree water on reduced iron failed since these vapors reacted chemically with the powders at 25’. Experimental The volumetric apparatus and procedures used in the determination of the adsorption isotherms were described in an earlier paper.6 Stock valves were used instead of stopcocks. The temperature of the bath surrounding the bulb which contained the adsorbent was maintained within lt0.05’. The mercury manometer M was made of 25 mm. tubing which should be large enough to minimize capillary depression errors and to allow the meniscus to move freely. Care was taken that the arms of the manometer were parallel and that the manometer assembly was clamped securely in a vertical position. The heights of the mercury in the arms of the manometer were determined with a traveling microscope of a sensitivity of 0.001 mm. The microscope was mounted on a transit mount; its arc of rotation was limited by means of stops. Thus, measurements were made a t the same location on the Rat section of each meniscus. A fluorescent lamp illuminated the menisci. The following procedure was used for low pressure determinat,ions. Approximately 24 hours were allowed to establish equilibrium between the solid and the vapor. Pressure readings were made after gently tapping the manometer. The absorbent was then isolated from the manometer by closing a Stock valve. After both arms of the manometer had been pumped out, the zero point cor(5) G. Jura and W. D. Harkins, ibid., 66, 1356 (1944).
.
592
EDWARD H. LOESER,WILLIAMD. HARKINS AND SUMNER B. TWISS
3.5
r ' " -
3.0
-
3.0-
z
:
$
2.5
z 0 X
-
5
f m
in
0
-
20
-
1.5
-
1.0
-
K
2.0W
2.5
0
6 K
VOl. 57
0.
i
c1 W
m
K
in
4 in
2
a
0
r 0
0 00
0.002
0.004 ,
0.006
0.008
0.010 5
~
0
I2 RELATIVE PRESSURE, p/pd
RELATIVE PRESSURE
'
Vp,'
Fig. l.-Low pressure isotherms at 25": -A-, n-propyl alcohol on untreated iron: -A-,n-propyl alcohol on reduced iron; -, n-heptane on untreated or reduced iron.
Fig. 2.-Low pressure isotherms a t 25": -0, n-propyl alcohol on Fe2O3;-0-, n-propyl alcohol on Fea04; -, nheptane on Fe20a.
rection was determined. The reversibility of these isotherms was not tested as was done in an earlier investigation.6 However, in several of the systems reported here a large sample of adsorbent was used to determine the extremely low pressure region and a smaller one to determine the remainder of the isotherm. The fact that the curves overlapped in the discontinuous portion indicates the reproducibility but not necessarily the reversibility of the isotherms. The n-propyl alcohol was obtained from the Coleman and Bell Company and was purified by the method of Lund and Bjerrum.' The liquid was thoroughly degassed by repeated solidification and ebullition in a vacuum maintained at 10-6 mm. The vapor tension of the n-propyl alcohol determined a t 25" was 20.86 mm. The iron was 325-mesh electrolytic powder (annealed) and was obtained from Charles Hardy, Inc., of New York. It was guaranteed to be 99.4+ % iron and iron oxide. The existence of an oxide film, formed on the surface of iron exposed to air, is well established. The thickness of the oxide film on this sample, calculated from the amount of water formed on reduction, is about 60 A. Untreated iron is a sample of the unreduced iron, which has been cleaned by heating a t 200" in vacuo for 16 hours. The reduced iron was re ared by passing dry hydrogen through the heated (350') :eta1 powder for eight hours. The entire process was carried out in the adsorption bulb on the vacuum line and the water formed during the reduction waB removed from the reaction zone by means of a liquid NP.trap. The reduced iron was degassed by heating at 200" %nvacuo for 36 hours. The FenO3, of Baker and Adamson rea ent grade, was degassed in vacuo a t 500" for 16 hours. %he sample of FesOc was a 325-mesh powder containing 0.065% Mn, 0.00370 AI, 0.005% Cu Ni and a trace of Si. The Fe,04 content of this sample a t the time of its use is un-
known. An X-ray diffraction analysis, run on the sample about two years after its use in these studies, indicated the presence of 60% Fe301 and 40% Fez03. It was degassed in vacuo at 325" for 16 hours.
.
+
(6) G. Jura, E. H. Loeser, P. R. Basford and W. D. Hsrkins, J . Chsm.
Phys., 14, 117 (1946). (7) H.Lund and J. Bjerrum, Ber., 64, 210 (1931).
Phase Changes in Adsorbed Films.-The low pressure region of the adsorption isotherms exhibited in Fig. 1 indicates the presence of two extensive discontinuities in the curve for n-C3H,0H on untreated Fe. If the oxide coating is removed from the iron powder by reduction, the amount of adsorption is decreased and the discontinuities nearly disappear. The curve for n-heptane on untreated Fe, shown in Fig. 1 for purposes of comparison, exhibits no discontinuity. The curves in Fig. 2 indicate a single discontinuity in the adsorption isotherms of n-C3H70H on Fe304or Fe203at 25'. Although the pressure a t which the discontinui€y occurs is the same for the two oxides, Fe203 adsorbs approximately twice as many alcohol molecules per unit area before leaving the discontinuous portion of the curve. The length of the discontinuity in the alcohol on Fe203 curve is many times that in the heptane on Fe203 curve. The heterogeneous nature of the solid surfaces makes the interpretation of the experimental data difficult. The discontinuities commonly observed in this study are of the type in which the amount of vapor adsorbed varies without any change in pressure. They may be interpreted as first-order phase transitions in the adsorbed films.s The two firstorder changes in the film of n-C3H70Hon untreated Fe may be due t o the presence of two different
.
June, 1953
M OLECULAR INTERACTION BETWEEN n-PRoPYL ALCOHOL AND IRON OR IRON OXIDES 593
the presence of the oxide coating on the iron. In Fig. 5 are shown the complete isotherms a t 25' of n-propyl alcohol on Fe203 and on Fe304. For p/po values below 0.5, the number of alcohol moledules adsorbed per unit area on Fez03is much larger than adsorbed on Fe304. The shape of the curve of the alcohol on untreated iron is similar to that of the alcohol on Fez03. The values of the free surface energy changes were calculated from the adsorption data by the method described by Jura and HarkimS Figure 6 exhibits the curves for the lowering of the free surface energy ).( of the solids as a function of the relative pressure of n-propyl alcohol. The alcohol's effectiveness a t any p / p , value in lowering the free surface energy decreases in the order, FezO3, untreated Fe, Fe304 and reFig. 3.-The pressure-area relationships in the low pressure region of n-propyl alco- duced Fe. hol films a t 25': -A-, untreated iron; -A-,reduced iron; -0, FezOo; -0-,Fes04. Table I presents the values the adsorbed phase of alcohol on reduced iron. of r e ,the free energy decrease bf the solid caused It is not known whether the first-order change in by the adsorbed film, and WA, the work of adthe alcohol film on FeaOd is due to contamination hesion for the various systems. The r e and WA by FezO3. values for n-heptane on the untreated iron are The relation between T , the film pressure and I u, the area per adsorbed molecule, for n-C3H70H on untreated Fe and iron oxides is shown in Fig. 3. The low pressure region only is shown. The transitions take place a t film pressures ranging from 0.2 to 2 ergs ern.-+. The values of ?r begin to rise again whe? the area occupied by the molecule is about 33 A.2 on untreated Fe, 42 on Fe203and 86 on Fe304. These values are in reasonable agreement with the area occupied in a completed monolayer, urn,as given in Table 11. Free Surface Energy Relations.-The adsorption $ I2 isotherms a t 25" of n-propyl alcohol on untreated a and reduced iron for the entire pressure range are 0 exhibited in Fig. 4. The presence of the thin coat 9 IO m of oxide on the untreated iron increases the amount a cn of alcohol adsorbed. The curve which represents 0 the adsorption of a hydrocarbon, n-heptane, on 4 a either untreated or reduced iron is given also in a Fig. 4. The number of heptane molecules adsorbed per unit area of solid is less than that of p 6 propyl alcohol at all pressures and is unaffected by
crystal surfaces which act as nuclei for cluster formation. Removal of the oxide coating changes the solid surface since only a very short first-order transition and a second-order change are found in
I '\
I
.
u)
-I
W
TABLE I FREEENERGY OF INTERACTION BETWEEN SOLIDS AND ADSORBED
4
VAPORSAT 25"
Solid
Liquid
r e ,decrease of free surface energy, ergs/cm. - 2
Fe (untreated) Fe (reduced) FeaOc Fez03 Fe (untreated) Fe (reduced) Reference 2.
n-PrOH n-PrOH n-PrOH n-PrOH n-C7H16 n-C7H16
102 73 86 122 54" 53"
W A ,work of adhesion, ergs cm.?
148 119 132 168 94" 93"
2
0
0.2
0.4
0.6
0.8
3
RELATIVE PRESSURE,
'Po.
Fig. 4.-Adsorption isotherms at 25': -A-, n-propyl alcohol on untreated iron; -A-,n-propyl alcohol on reduced iron; -, n-heptane on untreated or reduced iron.
EDWARD H. LOESER,WILLIAMD. HARKINS A N D SUMNER B. TWISS
594
0
02
0 4
06
VOl. 57
OB
10
R E L A T I V E PRESSURE pD , o.
Fig. B.-Lowering of free surface energy (T) for films of n-propyl alcohol on various solids as a function of the relative pressure (pip").
twenty-one non-porous solids was found to be conLivingston'2 has assigned "best" values of 0 02 04 06 08 IO urn for the adsorbed molecules of twenty-three RELATIVE P R E S S U R E , p chemical compounds. The applicability of the /io' Fig. 5.-Adsorption isotherms at 25": -0-,n-propyl alco- B.E.T. and Huttig theories to the n-propyl alcohol hol on Fez03;- 0 - ,n-propyl alcohol on FesOd; -, n-propyl isotherms presented here may be questioned for alcohol on untreated iron. two reasons. The first is the presence of discontinuities. According to the B.E.T. and Huttig the same as those for heptane 01) the reduced iron. theories there are no significant intermolecular In the case of the alcohol, however, the oxide film forces between the adsorbed molecules in an incomproduces a large increase in the re and W;1 values. plete monolayer. This mould rule out phase These values appear to increase with the percent- changes. Secondly, several of the B.E.T. and one age of oxygen present in the solid lattice. This of the Huttig plots given in Figs. 8 and 9 have inis also in agreement with earlier results for polar tercepts on the abscissa rather than on the ordiliquids adsorbed on untreated tin and tin oxide, nate. These are the first such curves t o be reported where r e and WA were proportional t o concentra- and would yield negative values of the parameter c tion of polar sites on the solid surfaces. of the B.E.T. equation. Surface Area Relationships.-Several methods Because of this anomaly, the experimental prohave been utilized to calculate the areas of solids cedures used and the calculations made in this infrom adsorption data. The LangmuirJs B.E.T.s vestigation were re-examined. The correction for and HuttiglO methods provide the means for the the change in density of mercury with temperature calculation of u,, the volume of vapor required to was applied to the pressure readings. A correction form a completed monolayer on the solid. In ad- for adsorption of propyl alcohol on the glass surface dition, the average cross-sectional area, urn, oc- of the apparatus was not made. It was probably cupied by tthe adsorbed molecule is needed for area negligible since the largest area of glass, the bulbs calculations by these methods. The Harkins- of the buret, was not used until a p/po value of 0.7 Jura" (H.J.) method uses the adsorption data and was reached. The perfect gas law was assumed to a constant, k . The value of k for each vapor at a hold for n-propyl alcohol a t 25'. To test this, the given temperature is evaluated by the use of a solid data for n-propyl alcohol on anatase was recalcuof known area, anatase. The area of the anatase lated using the Berthelot equation is 13.8 m.2g.-1 as determined by the Harkins and 9 PT Jura absolute method. pv = RT [l + 128 p,T (1 In general, the vapors used for area determinations have been non-polar, e.g., nitrogen or hydro- where the critical temperature and pressure, T , and carbons. The value of a,, for n-heptane on p,, are the constants in the equation. The recalculated values indicate a very slight deviation from (8) I. Langmuir, J . Am. Chem. Soc., 40, 1361 (1918). (9) S. Brunauer, P. H. Emmett and E. Teller, ibid., 60, 309 (1938). the perfect gas law and no. change in the negative (10) 8. Ross, THISJOURNAL, SS, 383 (1949). intercept of the B.E.T. plot. In spite of the ob(11) W. D. Harkins and G. Jura, J . Am. Chem. Sac., 66, 1366 1
T)]
(1944).
(12) H. K. Livingston. J . Colloid Sci., 4, 447 (1949).
June, 1053
I\qOLECUL.4R INTERACTION BETWEEK ?+PROPYL
ALCOHOL AND I R O N OR I R O N
OXIDES
595
tained in this Laboratory. The p / p o ranges over which the linear relationships hold and u,]~values are also given. 4 200 The latter were calculated from the equation, urn = X / N , where 2 is the surface 150 area of the solid, determined 3 in this case from n-C7H16 or N, adsorption data, and N is PI> the number of adsorbate 2 io0 molecules necessary to complete the monolayer. N may be calculated from urn. The results indicate that the value I 50 of unl for the alcohol is not constant but varies from solid to solid. The Larigmuir O F I o value is less variable than the 0 I 2 3 4 5 6 7 B.E.T. or Huttig values. p . m m HGi The ,a value calculated by Fig. 7.-Langmuir plots for n-propyl alcohol a t 25": -A-, untreated iron; -A-] re- L ~ method ~is apduced iron; -0-,Fez03; -V-, TiOz. Droximatelv the same as that jections to the use of the B.E.T. and Hiittig theo- calculated for closeli packed "liquid molecules ries, the results obtained by these methods were in- while the B.E.T. and Huttig values are about 1.4 cluded in the correlations between urn and Z values times larger. The p / p o ranges over which the Hiittig relation holds are more extensive than obtained by various methods. In Tables IIA, B and C are listed the urn values obtained by the other methods. Figures 7, 8 calculated by the Langmuir, B.E.T. and Huttig and 9 show the Langmuir, B.E.T. and Huttig methods from the adsorption data of n-propyl alco- plots for n-propyl alcohol 0'1 TiOn, Fe203, unhol on five non-porous solids. The results for ana- treated iron and reduced iron. The plots for FelOd tase were calculated from unpublished data ob- mere omitted from Figs. 7, 8 and 9 because they 5
700
'
- 18
1-
A
t
600 6oo
30
25
- I6
- 14
-
20
C1 -;
16
IO
05
p/Po.
Fig. %--BET. plots for n-propyl alcohol a t 2 5 " : -Auntreated iron; -A-,reduced iron; -o-,FeeOt; -V-, TiOg.
PRESSURE
p , m m . HG
Fig, 9.-Huttig plots for n-propyl alcohol a t 25": -A-, uritreated iron; -A-]reduced iron; -O-] FesOa; -V-, TiOs.
~
. Vel. 57
EDWARD H. LOESER,WILLIAMD. HARKINS AND SUMNER B. TIVISS
596
TABLES IIA, B AND C CALCULATED VALUESOF u FOR n-CsH70H AT 25" OBTAINED FROM VARIOUS THEORIES Area,= urn, cc. g.-1 Solid
Range of validity, p / p o
(S.T.P.)
m.Zg.-L
Calcd.
TiOz
25.6 24.6
..
21.2 27.8
B. B.E.T. 13.8 1.3809 0.07 to0.25 37.2 .05 to .25 30.9 7.51 0.8976 .1816 .04 to .24 36.9 1.83 0.187 ,02161 .05 to .23 32.6 .182 .01870 .03 to .28 38.2
TiO8
Fda FeaOc Fe (untreated) Fe (reduced)
13.8 7.51
FesOd
1.83
Fe (untreated)
0.187
Fe (reduced)
AREASOF
C. Huttig Ti02 13.8 1.3176 0.09 t o 0 . 7 6 39.1 Fee08 7.51 '0.9684 .05 to .50 28.9 Fea04 1.83 .1733 .01 to .17 39.3 Fe (untreated) 0.187 .02253 .06 to .55 30.9 Fe (reduced) .182 .02053 .O05 to .55 33.0 a Area values calculated by B.E.T. and H.J. methods using Nz or n-C7H16data. * No linear relationship found.
k Value
k1 = kz = k3 = kz = ks =
Fe203
,182
AT
25"
Range of validity,
Area, 2 . mag. -1
Solid
urn,A.2
A. Langmuir 13.8 2.006 0.03 t o 0 . 3 6 7.51 1.136 .003to .17 b 1.83 ......... 0.187 0.02874 .01 to .23 .182 ,02439 ,004to .19
TiOz Fez03 FeaO4 Fe (untreated) Fe (reduced)
TABLE IV k V.4LUES FOR n-CaH7OH HARKINS-JURA
AS
k, = k4 = k3 = kl = k3 = k3
4.31 5.13 8.93 5.14 8.79 12.82 11.51 8.42 4.14 8.50 8.21 11.7
P/PO
0 . 0 3 t00.18 . 1 8 t o .43 .58t0 .76 .01 to .30 . 4 2 t 0 .88 .01 to .01 . 0 6 t 0 .52 .52t0 .92 .05 to .23 .47t0 .82 . 0 9 t o .54 .75 to .87
kr = TABLEV SOLIDSCALCULATED BY H.J. METHOD, k = 8.93
TiOz FezOI Fe304 Fe (untreated) Fe (reduced)
13.8 7.51 1.83 0.187 0.182
.
13.8 7.57 1.91 0.199 0.206
..... -0.8 -4.4 -6.4 -13.2
TABLE I11 AREASOF SOLIDSCALCULATED BY LANGMUIR, B.E.T.
AND
HUTTIGMETHODS
Solid
0
0.05
0 IO
0 20
0.15
0.25
035
03 0
I.50
-
IO0
,j
050-
S
=I n 0
4
0.00 -
-0.50
-
-1.00
0
.
.
.
.
.
0.5
.
.
.
.
.
.o
I
.
.
.
.
.
1.5
.
.
.
culated area values may deviate from the true values. This is shown in Table 111. The area values obtained from the alcohol data by the Langmuir method are more self consistent and are closer to the probable values detennined by nitrogen adsorption; the B.E.T. and Huttig values show deviations from values determined by nitrogen adsorption, which we consider to give the most probable value of the area. If the adsorption data are plotted with log p vs. l/v2, one or more linear sections are obtained. According to the H.J. method, the square .
June, 1953
THECis- AND hYJnS-cYCLOHEXANEDIOL-1,2
specified temperature. The value of k for n-C3H70Hwas determined from the slopes of these H.J. lines for the vapor adsorbed on solids of knowii areas. Figure 10 exhibits the H.J. plots for n-C3H70H on TiOt and Fe2O3. It is apparent from these plots and the values given in Table IV that several k values may be calculated for each isotherm. Although the IC values appear to fall into groups of approximately the same magnitude, the p / p o range of applicability of each is not the same for the various solids. The areas of four solids: Fez03, FesOd, Fe (untreated) and Fe (reduced) were calculated using k3 = 8.93, which was one of the values obtained for n-C3H70H on TiOz. The results given in Table
SYSTEM
597
V show fairly good self consistency. The determination of area by the H.J. relative method using n-C3H70H data is not recommended since there appears to be no means of determining the p / p o region of the isotherm over which a given k value is valid. It is apparent from the lack of agreement between these methods for calculating area, and the divergent results obtained with the polar vapor, npropyl alcohol, as compared with the non-polar vapors, nitrogen and n-heptane, that the safest procedure is to use non-polar gases for accurate area measurements. This becomes increasingly important when polar solids are involved, such as the iron oxides used in this study.
A STUDY OF THE SYSTEM cis- AND trans-CYCLOHEXANEDIOL-I,2 BY W. J. SVIRBELY AND SAMUEL GOLDHAGEN'J Contribution from the Department of Chemistry, University of Maryland, College Park, Md. Received December 6, 1961
A phase study of the cis- and trans-cyclohexanediol-l,2system was made using a micro hot stage in conjunction with a chemical microscope. X-Ray diffraction patterns were determined for the cis-diol, trans-diol and several solutions ranging from 5% to 25% trans-diol in order to help establish the phase diagram. The resulting diagram.shows: (1) a eutectic, melting a t 71.9", a t 56.6% trans-diol; (2) a solid solution field for mixtures containing more than 55% cis-diol; (3) three different crystalline forms for the pure cis-diol.
The use of the cis- and trans-cyclohexanediols in determining mechanism has been widespread. In two of the instancesJ3v4data have been obtained from which the temperature-composition curve of the cis- and trans-cyclohexanediol-l,2 system could be constructed. The diagrams obtained from those studies are not in agreement. In the present work, temperature-composition data have been obtained chiefly from experiments carried out on a "hot stage" in conjunction with a polarizing microscope. The results have been used to construct the phase diagram of the cis- and trans-cyclohexanediol-1,2system. Experimental Procedure cis- and trans-Cyclohexanediol-l,2.-The crude mixture of isomers was prepared by hydrogenating catechol in ethanol over Raney nickel a t 195-200". The isomers were separated from the crude mixture by use of various procedures. The m. p. of the cis-diol and the trans-diol were 98.5-99.5' (uncor.) and 103-104' (uncor.), respectively. Mixed melting points of the cis- and h-ans-diols with each other and with catechol (m.p. 104') showed a marked lowering. Apparatus and Method.-At the start of this work, the classical, cooling curve technique was attempted. The results obtained could not be intcrpreted; therefore, a procedure involving a chemical microscope with a micro hot stage was used. Crossed nicols were used to make observations when anisotropic crystals on a fused slide were not easily discernible. Isotropic substances were observed directly. If some doubt existed, for a particular sample, as to whether an isotropic or an anisotropic solid resulted, then a firstorder red plate was used. The temperature of the hot stage was raised slowly so that the temperature at which the last crystal disappeared could be determined. If cooling were (1) Abstracted in part from the M.S. Thesis of 6. Coldhagen. (2) Presented at the Atlantic City Meeting of the American Chernical Society, September, 1949. (3) B. Rothstein, Ann. Chem., 14, 461 (1930). (4) S. Winstein and R. E. Buckles, J . Am. Chena. Soc., 64, 2780 (1842).
then started immediately by decreasing the heat input, super-cooling was avoided in most cases. The samples were prepared as follows: weighed amounts of cis- and trans-diols, usually amounting to 0.4 g. for the mixture, except when the mixture approached one of the pure components in composition, were placed in a vial. The mixture was melted and the melt stirred thoroughly with a wire stirrer. After cooling, the solid mixture was pulverized and again mixed thoroughly. A small portion of the powder, 10-20 mg., was placed on a 1" X 1.5" microscope slide, covered with a cover glass and slowly melted over a low flame. The cover glass was then squeezed tightly agaipst the slide to decrease the amount of the solid to a minimum. The excess solid was removed and the slide wiped clean. The slide was placed on the hot stage and the microscope was focused on an interior portion of the material. Actual observations as to the changes taking place during the slow heating and slow cooling were thus easily obtainable. In the above work, a Leite petrographic microsco e with a 16-mm. objective and a 1OX eye iece was used. TEe light, source was a Spencer Type 370 Emp with a blue, frosted glass. The hot stage was an electrically heated, Kofler micro unit. The current input was controlled with a variable rheostat supplied with the hot stage. The thermometer belonging to the hot stage was calibrated using fused slides containing pure compounds whose true melting points were in turn obtained by standard cooling curve technique. In the latter case, Anschiite thermometers calibrated in 0.2" and checked against similar thermometers calibrated by the Bureau of Standards were used.
Experimental Results Temperature data were gathered for the disappearance of the last crystal (f'p') and the appearance of the first crystal (fp). Anywhere from three to eight determinations involving new slides and used slides were made on each mixture or pure compound. In two cases, namely, pure trans-diol and 77.6% trans-diol, supercooling prevented the obtaining of (fp) data. Otherwise, the disappearance of the last crystal and .its reappearance were found to be reproducible. All data obtained for
