Sept.

, 1963 IR’TERPRETATION
OF pH ~IEASUREBIESI-s SOLVEXTS
ISALCOHOL-WATER 1833

solution vs. R In x2 for violet solutions fall on a single
line, but diverge for solutions in which complexes are
formed. A similar divergence may be expected for
gases that complex with a solvent.
The most striking result of this investigation is seen
in Fig. 4, where values of log za for CF,, SFc,CQ2,CH,,
Ar, X,, and Hz are plotted against the solubility param-
eters of a series of solvents ranging from CTFIO,5.85,
to CSZ, 1O.c). We see that these two fluorides diverge
strongly from a simple pattern6to which the other gases
conform remarkably well. The “force constants” of
other gases, e l k , obtained from their second virial co-
eBcients or from their viscosity, also give a pattern
agreeing well with the solubility of other gases. From
these two patterns we would expect SFG, ~jrC = 201,
to have about the same solubilities as ( 3 0 2 , t-/k = 205.
Similarly with CF, and CH4, E/JZ = 153 and 148, re-
(6) Reference 5 , pp. 46, 47.
spectively. The solubilities of the gases of each pair
are nearly the same in the fluorocarbon solvents but the
two fluorides diverge in a common zig-zag manner in
going to solvents with higher solubility parameters
and dive deeply through benzene to carbon disulfide.
This behavior is another and especially arresting ex-
ample of departure by fluorides from relations based
upon the geometric mean “ l a d ’ of attraction betmeeii
molecules of certain different electron types emphasized
by Hildebrand and S ~ o t t . ~
We invite attention to the paper by Hiraoka and
Hildebrand8 on the partial molal volume of SFG i n
some of the same solvents.
Acknowledgment.-This research has been supported
by the Atomic Energy Commission.
(7) Reference 5 , p. 144.
(8) H. Hiraoka and J. H. Hildebrand, J . Phys. Chem.. 67, 1919 (1963).

INTERPRETATION OF pH h1EASUREMENTS IT\: ALCOHOL-WATER SQLVEXTS’

BY ROGERG. BATES,MAYAPAABO,
AND R. A. ROBINSOX

Solution Chemistry Section, iYational Bureau of Standards, Washington, D. C.

Received iVarch 18, 1963

The operational pII numbers for solutions in alcohol-water solvents ohtained from the e.m.f. of the cell with
hydrogen (or glass) electrode, salt bridge, and calomel reference electrode, standardized with aqueous butfer
standards, are subject to no simple interpretation. Because of the indeterminate potential at the junction solu-
tion X (alcohol-HpO) 1 KC1 bridge (as.), they do not lie on a conventional scale of hydrogen ion activity referred
either t o the aqueous standard state ( ~ u H )or to the standard state in the alcoholic medium ( ~ u H * ) . Values of
+
E, log my^^ (where E, is the liquid-junction potential expressed in $3 units and m y r l is the primary medium
effect of chloride ion) were found to be as constant for different buffer solutions in alcoholic solvents of fixed com-
position as for the strictly aqueous medium. Inasmuch as m ~ at ~a given i temperature is dependent only on sol-
vent composition, ihe liquid-junction potential must therefore also be nearly constant for a given solvent medium.
Correction terms E, - log m~~ = 6 have been calculated, and it is shown that pI3 - 6 closely approviniates pan*
under optimum conditions of measurement. An operational scale of pH* (whiCh is related to pa^* in the same
way that pH is related t o pax) is described. Values of ~ U H *were obtained for 12 solutions in methanol-water
solvents (0 to 68.1 wt. yo RfeOH) and for 3 solutions in ethanol-water solvents (0 to looL%EtON).

Introduction in cell I is adequately constant for aqueous solutions of

The operational pH value of a solution (X) is defined
by the equation
where E x and Es represent the e.1n.f. of the cell
1%; H B , sola. X or S 1 concentrated KCl(aq.); refer-
ence electrode (I)
containing, respectively, the “unknown” solution (X) or
an aqueous pH standard (S).2 The hydrogen elec-
trode in cell I may be replaced by other electrodes, for
example, glass or quinhydrone, that are reversible to
hydrogen ions.
I n eq. 1, pH(S) is the assigned pH, on a conventional
activity scale in water, of a standard solution (S).
The latter is derived from e.m.f. measurements of
hydrogen-silver chloride cells without liquid junction.3
It is generally agreed that the liquid-junction potential
(1) Presented before the Division of Analytical Chemistry a t the 144th
National Meeting of the American Chemical Society, Los Angeles, California,
April 2, 1963.
(2) See R. G. Bates and E. A. Guggenheim, Pure A p p l . Chem., 1, 163
(1960).
(3) R. G. Bates, J . Res. Nail. Bur. Std.. 668, 179 (1962).
ionic strength not in excess of 0.1 and pH between 2.5
and 11.5, and that the operational pH thus can be
identified with -log mHyH with an uncertainty of not
more than a few hundredths of a pH unit.
We are now concerned with what meaning, if any,
can be attached to operational pH values for solutions
in alcohol-water media. The hydrogen electrode be-
haves satisfactorily in alcohol-water solvents, and a
comparison of the e.m.f. obtained with that given by a
glass electrode shows that the response of the glass
electrode is often substantially unimpaired a t solvent
compositions below about 90 w t . % alcohol. The
liquid-junction potential is large but appears to be suf-
ficiently reproducible that the replacement of the aque-
ous reference electrode by an alcoholic reference elec-
trode does not appear either necessary or warranted.
A pH meter therefore is capable of giving reproduc-
ible e.m.f. values in alcohol-water solvents. The pH
numbers read from the instrument, however, are sub-
ject to no simple, clear interpretatioil in terms of chemi-
cal equilibrium.
In order to obtain an accurate measurement of the
relative acidity or difference of p H in a given solvent,
the potential at the junction solution X(nonaq.1 IKCl
(coned. as.) must remain constant when the nature
1834 ROGERG. BATES,JIAYA
PAABO,
of the solute is altered. Electromotive force measure-
ments of solutions of strong acids and salts in partially
aqueous media4 suggest that the liquid-junction poten-
tial of the pH cell is, as a first approximation, a functioii
only of the composition of the solvent and largely
independent of the nature of the solute. There was
heretofore, however, too little evidence that the liquid-
junction potential remains constant for buffer solutions
over a wide range of pH in these solvents. We have
now been able to demonstrate that satisfactory con-
stancy is indeed maintained and that a useful interpre-
tation of pH data for dilute solutions in alcohol-water
solvents is possible. A suitable scale of pH for these
media is described and methods for the practical stand-
ardization of pH measurements in these media are out-
lined.
Method
In the work of T’an Uitert and Haasla with dioxane-
water mixtures, the constancy of the liquid-.junction
potential could be confirmed directly only for solutions
of strong acids, where the hydrogen ion concentration
was known. In the study by Gutbezahl and Grun-
~ v a l dvalues
,~ of m ~ the~ primary
, medium effect6for the
hydrogen ion, were derived from an analysis of solvent
effects on the dissociation constants of uncharged acids
and cation acids. The results were used to evaluate the
potentials at the junction between electrolytic solutions
in ethanol-water solvents and the aqueous potassium
chloride solutioii of the reference electrode. The poten-
tials obtained from data for hydrochloric acid solu-
tions, for carboxylic acid buffers, and for aiiilinium
buffers agreed satisfactorily, at each solvent composi-
tion.
deLigny and his associates7 have determined pa^*,
referred to the standard state in each mediums
XSD R. A. ROBIXSOS
Vol. 67
of the cell
P t ; Hn(g, 1 atni.), buffer soln. X (in alcohol-water),
C1-, AgCl; Ag (11)
by the equation
where ,“Eo is the standard e.m.f. of cell I1 (aqueous
standard state, molal scale) and m is molality. The
second quantity was the operational pH, which is de-
rived from the e.m.f. of cell I by eq. l. The KBS phos-
phate buffer, 0.025 m KH2P04,0.025 m Na2HP04,was
chosen as the primary standard. The assigned value
of pH(S) is 6.865 at 25°.3 Hydrogen electrodes were
used and the reference electrode was calomel; Hg in a
3.0 111 or saturated solution of potassium chloride.
The pH(S) value of the aqueous standard in cell I
is defined by
PH(S) E -log aH(C01117.)= P ( ~ H ~ c I log
) YCI (4)
where ycl is fixed by convention.2 The value of pH,
but for the residual liquid-junction potential, would fall
on this same scale and would represent the ~ U (coiiv.)
H
of the solution referred to the standard state of aqueous
solutions. Thus
pH - I?, = -log aH(conv.) = pa13 (5)
where E, is the error (in pH units) caused by the re-
sidual liquid-junction potential, namely EJ(v.)/0.05916
a t 2 5 O . In the remainder of this paper, (conv.) can be
omitted from these symbols.
The difference between the two experimental quanti-
ties then is

pall* = -log (nZH.,yH) (2) PH - ~ ( ~ H Y C I=) (G + log m ~ c i )+ log SYCI (6)

for an oxalate buffer solution and a succinat’ebuffer solu- In eq. 6, y c l (referred to the aqueous standard state) has
tion in methanol-water solvents. These values, ob- been replaced by its equivalent, m y C l ‘ s Y C l r namely, the
ta,ined from the e.m.f. of cells without liquid junction, product of the medium effect and the “salt effect,” the
then were compared with the practical pH values for latter being the activity coefficient referred to the
the same solut’ionsdetermined by the pH meter. The standard state in the alcoholic medium concerned.
differences, which they called the “liquid-junction po- Although the medium effect for chloride ion cannot be
tential,” were nearly the same for the two buffer solu- evaluated, it is a constant for a fixed temperature and
tions provided that the solvent composition was the solvent composition and is independent of the concen-
same.g tration of dissolved solutes.
In order to examine in a more direct way the degree The validity of a relative measurement of log a~
of constancy in buffer solutions in alcohol-water sol- in alcoholic media thus requires that the quantity en-
vents, we have used two experimental quantities. closed in parentheses in eq. 6 be constant, at a given
The first was p(a~yc1)derived’O from the e.m.f. ( E ) temperature, for a variety of acid-base systems in a
solvent of fixed composition
(4) (a) L. G. Van Uitert and C. G. Haas, J . Am. Chem. SOC.,‘71,451
(1953); (b) A. G. Mitchell and W. F. K. Wynne-Jones, Trans. Faraday Soc..
51, 1690 (19.56): (c) I. T . Oiwa. Sei. Rept. Tohoku Unin., First Ser., 41, 129 pH - P ( a ~ y c i )- log SYCI= +
(1957); see also A. L. Bacarella, E. Grunwald, H. P. Marshall, and E. L. log m ~ C =
l const. (7)
Purlee, J . Org. Chem., 20, 747 (1955).
( 5 ) B. Gutbezahl and E. Grunwald, J . A m . Chern. Soc.. ‘75, 666 (1953). I n
this and related papers, the medium effect is called the “degenerate activity The ionic strength ( I ) , on the molality scale, of the
coefficient’’ and designated .f. solutions selected for a test of eq. 7 was low (usually
(6) B. B. Owen, ibid., 54, 1758 (1932). about 0.02), so that log s y ~ could
l be estimated with
(7) C. L. deligny, P. F. hI. Luykx, 51. Rehbach, and A . A . Wieneke,
Rec. Irau. chim., 79, 699. 713 (1960). suficient accuracy by the Debye-Huckel equation in
(8) The subscripts w and s to the left of a symbol indicate which standard the form2
state is used, namely t h a t in water (w) or in solvent (9) of the composition i n
question. The subscript may be omitted Then the aqueous standard state (1.823 X 106)(~T)-S’Zd12
is used. The subscript m has a different meaning: , , , ~ i is the (primary)
medium effect on the species i, t h a t is, the ratio , v - , i / s ~ i .
(9) C. L. deLigny and M. Rehbach. Rec. trav.chim.. ’79, 727 (1960).
-log SYCl =
1 + (8)
1.5(78.3/~)~/~dId~
(10) This quantity was called pwH b y R . G. Bates, Chem. Rec., 42, 1 In this Laboratory and elsewhere The change is consistent n i t h recent
(1948), and this symbol has been used in many subsequent papers originating leeommendations concernmg pH symbols and terminology; see ref. 2.
 Sept., 1963 INTERPREThTIOlU O F pH AIEASUREJIESTS
Ih' AhLCOHOL-\J7ATEH. SOLVENTS 1835

where e is the diele'ctric constant of t,he solveiit,ll do

its density,12 and T the temperature in degrees Kelvin.
Measurements of cells I and I1 a t 25' have been
made for 12 selected solute systems iiz methanol-water
solvents varying in methanol content from 0 to 68.1 wt.
yo. I n order to avoid appreciable ion pair formation,
higher methan.01 concentrations were not used. The
solutions chosen were (a) st'rong acids or acid-salt
mixtures, (b) weak acid-salt buffers, and (c) weak base-
salt buffers, arid the met'hanol-free mixtures varied in
+
pH from 1.8t'o 10.1. Values of gj log my^^ computed
from the data in methanol-wat,er solvents and from tlie
related e.m.f. measuremenhs of Bates and Schwarzen-
bach13 in ethanol-water solvents display about the
same order of constancy as in pure water. In addition
to the well known abnormality in tlie liquid-junction
potential for strong acids, unexpected deviations, pos-
sibly indicative of ineomplet'e ionization of ainnionium
salts, mere found in the weak base-salt buffer solutions
at all solvent #compositions.
Experimental
Methanol, reagent grade, was distilled once and the first
and last 10% fract,ionswere rejected. Commercial reagent grade 2 4 6 6 10
chemicals were used as solut,es for the preparation of the buffer pH IN WATER,
solutions with appropriate drying, but without further purifica-
tion except as specified below. Fig. 1.-a, + ~ 7l ) for 12 solutions in water and in
log m ~ (eq.
XBS Stsandad Samples were used to prepare the phosphate five methanol-water i3olvents: a, water; b, 8.17, MeOH; c,
and borax solutions. Potassium chloride was recrystallized once 16.3% MeOH; d, 33.37, MeOH; e, 52.1% MeOH; f, 68.17,
from water. 4-Aminopyridine was recrystallized once from a MeOH; C,HC1 and HCl-KCl solutions; 0, acid-salt buffers;
mixture of ethanol and benzene and twice from water. Citric 0 , base-salt buffers.
acid (anhydrous) was a specifially purified sample furnished by trations appear to give identical results for the pH. Details
Chas. Pfizer and Co., Inc. Tris-(hydroxymethy1)-amincmethane of the measurements in ethanol-water solvents have been given
of primary standaird grade was used. elsewhere.I3
The succinate buffer solutions were prepared by adding the
necessary amounts of succinic acid and disodium succinate Results
direct,ly to the methanol-water mixtures. The solution of hydro- The values of p1-I and p ( a ~ y ~ computed
1) from the
chloric acid was prepared by dilution of a, twice-distilled sample. e.m.f. of cells I and I1 by eq. 1 and 3 , respectively, are
It was analyzed gravimetrically for chloride content. A sodium
hydroxide solution was made by dilution of the 50% solution summarized in Table I (methanol-water solvents)
from which the carbonate has been removed by filtration through and Table I1 (ethanol-water solvents). Ion pair for-
a sintered steel crucible. The diluted solution was standardized mation is negligible in the methanol-water solvents
by titration of potassium hydrogen phthalate, NBS Standard studied, but appears sultficieiit to change p(a~yc1)for
Sample.
Glacial acetic acid was purified by a single fractional freezing. solution A' by about 0.1 at 100% ethanol and by less
The acetate buffer solutions were prepared by combining the than 0.01 in 85.4% ethanol.
calculated aniount,s of standardized solutions of acetic acid and
sodium hydroxide. The solution of ammonia was made by
+
Values of El log myC1, derived from eq. 7, for the I 2
solutions in methanol-water solvents are plotted in
bubbling nitrogen through a concentrated solution of ammonia in Fig. 1 as a function of the pH of the solution in water.
a gas-washing bottle and conducting the gas into a flask con-
taining conductivity water. It was standardized by titration The average values of +
log m y C l for methanol-
with standard hydrochloric acid. water mixtures are listed in Table 111. Similar data
The 0.025 m standard phosphate buffer solution was prepared for the ethanol-water solvents appear in Table ITT.
by dissolving 3 3 9 g. of pohssium dihydrogen phosphate and In computing the averages in methanol-water sol-
3.53 g. of disodium hydrogen phosphate in water and diluting to a
voume of 1 1. All of the osther solutions were prepared by weight. vents, only the results for weak acid-salt buffers (which
The measured e.m.f. of cell I1 was corrected to B partial fall on the horizontal lines of Fig. 1) have been used.
pressure of 760 mm. of hydrogen with the aid of the known vapor The omission of the two points for solutions of hydro-
pressures of methanol-water mixtures . I 4 The liquid junction chloric acid is consistent with the recognized abnor-
cell (I) differed from that already described's in that fritted glass mality of the liquid-junction potential at low pH. l6
disks 10 mm. in diameter gave enhanced mechanical stability to
the interface b e h e e n the bridge solution of potassium chloride The poiiits for the three base-salt buffers studied lie
below and the buffer solution above. The liquid-liquid bound- below the horizontal line by amounts varying from 0.01
ary was formed just above the surface of the disk. The bridge unit in the aqueous medium to 0.05 unit in 68.1 wt. yo
solution was 3.0 /If KCI for the measurements in methanol- methanol. These results may signify either a specific
water solvents and saturated KC1 for the measurements in
ethanol-mater media. Bridge solutions of t,hese two concen- alteration of the liquid-juiiction potential or incomplete
ionization of the chlorides of the nitrogen bases.I7
(11) Dielectric conntants of the methanol-wat,er mixtures were taken
from L. J. Costing a n d P. S. Albright, J . Am. Chem. Soc., 68, 1061 (1946),
The medium effect for hydrochloric acid in methanol-
and those for ethanol-water mixtures from G. ftkeri.of, ibid., 54, 41:!5 (1932). water solvents and in ethanol-water solvents, namelg
(12) I. T. Oiwa, Sci. R s p t . l'oholu Univ., Firsf Ssr., 41, 47 (1957). or ( m y ~ . m y ~measures
~ ) 1 ~ 2 ,the free energy of transfer
(13) R. G. Bates and G. Schwarzenbach, Helu. Chim. Acta, 38, 689 (1956).
(14) J. A. V. Butler, D. W. Thomson, and W. 13. Maclennan, J . Chen.
of a mole of hydrochloric acid from the standard state
Soc., 674 (1933). (16) (a) D. I. Hitchcock a n d A. C . Taylor, J. Am. Chem. Soc., 59, 1812
(15) R. G. Bates, "Electrornetric pH determinations," John WiIey a n d (1937); 60, 2710 (1938); (b) R. G. Bates, G. D. Pinchmg, a n d E. R. Smith,
sons, Inc., New York, N. Y . , 1!364, p. 59. J . Rss. A'atl. B w .Std.,45, 418 (1950).
1836 AKD R. A. ROBINSON
ROGERG. BATES,MAYAPAABO, Vol. 67

TABLE
I
VALUES O F pH, p(UHyCl), AND pUa* I N hfETHANOL-lvbTER SOLVENTS AT 25"
W t . % MeOH mi PH P (aHYcI) PaH* JVt. % MeOH ml PH PbHYCl) pas*
Soln. A: HCl(m1) Soln. G: KH2P04(ml),NarHPOl(mz), KCl(ma)
0 0.01992 1.826 1.816 1.756 ml = m2, ma = 0.01000
33.3 ,02113 1.866 1.474 1.756
0 0.002496 7,015 7.071 7.011
68.1 ,02283 1.918 1.012 1.776
8.1 .002541 7.198 7.170 7.193
16.3 .002578 7.387 7,261 7.377
Soln. B: HCl(ml), KCl(m2) 33.3 ,002642 7.815 7.482 7.765
m2 = 0.01000 52.1 ,002752 8.339 7.722 8.211
68.1 .002869 8.826 7.940 8.707
0 0.01002 2.089 2.113 2.053
8.1 .OlO15 2.087 2.021 2.045 Soln. H : borax(ml), KCl(m2)
16.3 ,01032 2.084 1.928 2.044
ml = 0,004996, rn2 = 0,009992
33.3 .01059 2.143 1.775 2.058
52.1 .01101 2,220 1.583 2.072 0 9.171 9.231 9.174
68.1 . 01 156 2.204 1.307 2.074 8.1 9.195 9,165 9.189
16.3 9.235 9.117 9.233
33.3 9.378 9.040 9.324
Soh. C: citric acid(ntl), KCl(m2)
52.1 9.565 8.941 9.432
VZI = 0.01000, vzz = 0.01762 68.1 9.640 8.747 9.517
0 2.623 2,680 2.620 Soln. I: NaHCOa(ml), XazCOa(mz), KC1(via)
33.3 2.943 2.611 2.895
68.1 3,362 2.466 3.236 m1 = m2, m3 = 0.01000
0 0.002514 10.087 10.143 10.083
Soln. D: succinic acid(ml), S a H succinate(mz), KCl(ma) 16.3 .002590 10.591 10.462 10.578
52.1 .002771 11.454 10.829 11.318
ml = 0,01999, m2 = m3 = 0,009995 68.1 ,002895 11.645 10.745 11.512
0 3.833 3.890 3.830 .002500 11.540 10.636 11.406
8.1 3,960 3.938 3.962
16.3 4.108 3.991 4.107 Soln. J: tris-hgdroxymethgl-aminomethane hydrocliloride(ml),
33.3 4.482 4.144 4.428 tris-( hydroxymethy1)-aminomethane( m2)
52.1 4.939 4.318 4.809 ml = m2 = 0.002000
68.1 5.330 4.443 5.213 0 8.124 8.201 8.141
8.120 8.202 8.142
Soln. E: HOAc(ml), NaOAc(m2),KC1(m3) 8,126" ... ...
ml = m2, 1n3 = 0,01000 8.1 8.091 8.088 8.112
16.3 8.064 7.972 8,088
0 0.01067 4,699 4.754 4 694
33.3 8.036 7.728 8.012
8.1 ,01094 4.802 4.779 4.801 52.1 8,045 7,460 7.951
16.3 .OllOO 4.934 4.820 4.935 68.1 8.062 7.223 7.993
33.3 ,01120 5.289 4.959 5.241
52.1 .01160 5.760 5.148 5.636 S o h K : 4-aminopyridinium chloride(m), 4-aminopyridine(nz?)
68.1 .01211 6.173 5.295 6.061 m2 = 0.04000 - 1121

0 0.02006 9.177 9.245 9.185

Soln. F: S a H succinate(ml), Sa2 succinate(m~),KCl(m8) .02004 9.174" ... ...
ml = m2,m3 = 0.01200 52.1 ,02186 8.646 8.058 8.546
0 0.001997 5.532 5.593 5.533 68.1 .02302 8.407 7.568 8.332
8.1 .002058 5.673 5.647 5.670 Soln. L: NH&l(ml), Irr"a(m2)
16.3 .002063 5 . 836 5.721 5.837
33.3 ,002123 6.263 5.927 6.210 ml = m2
52 1 ,002193 6.822 6.203 6.693 0 0.02016 9.297 9 . 3'ilb 9.311
68.1 ,002301 7.326 6.444 7.212 0,02025 9,294" 9.372' 9.312
a Capillary liquid junction. Int'erpolated from d a h of Bates and Pinching.17

in water to the standard state in solvents. It can be from E, + ~ , values of E, - log m~~ given
log m ~ C the
derived therefore from the values of the standard e.ni.f. in the last columns of Tables 111 and I V were obtained.
of cell I1 in the two media +
The average gJ log m y C l and EJ- log r r i in ~ meth-
~

anol-water mixtures and in ethanol-water mixtures are

By subtracting the values of log (mYH'mYC1) so obtained1*
(17) The ion sizes necessary t o fit activity coefficient terms of the form log
(YBH+YCi-/*,B) for weak nitrogenous bases B to the Debye-Hiickel formula
are usually improbably small, often zero, and sometimes negative; see for
example, R. G. Bates and G. D. Pinching, J Res Katl. Bur. Std., 4%, 419
(1949). The results in solutions of piperidine and piperidinium chloride were
explained in terms of ion pair formation b y R. G. Bates and V. E. Bower,
s b z d , 67, 153 (1956).
(18) The standard e.m.f. in the mixed solvents was interpolated on a plot
of EO as a function of wt. % alcohol. The data for methanol-water mix-
tures were those of H. S. Harned and H. C. Thomas, J . Am. Chem. SOC..67,
plotted as a function of the solvent composition in Fig.
2 and 3, respectively. The dots indicate the values of
Ej - log m~~ derived from the data of deligny, et al.
(methanol-water mixtures) 9 and the results given by
Gutbezahl and Grunwald (ethanol-water mixtures) .6
1666 (1935); 68, 761 (1936); I. T. Oiwa (ref. 10); J. M. Austin, A. H. H u n t ,
F. A. Johnson, and H. K. Parton, unpublished work cited by R. A. Robinson
and R. H. Stokes, "Electrolyte Solutions," 2nd Ed., appendix 8.2, Academic
Press, Inc., Iiew York, N. Y., 1969; and R. G. Bates and D. Rosenthal,
J . Phys. Chem., 67, 1088 (1963). For ethanol-water mixtures, the data
were those given b y G. J. Janz in D. J. G. Ives and G. J. Janz, Ed., "Refer-
ence Electrodes," Academic Press, New York, N. Y., 1961, Chapter 4.
Sept., 1963 INTERPRETATION O F pH 31EASUREJIESTS IN ALCOHOL-WATER
SOLVESTS 1837

TABLE
I1
VALUES O F pH, ~ ( U H ~ C I AND
), PUR*I N ETHANOL-WATER SOLVENTS I I
AT 25'
Wt. % EtOH ml PH PbHYCl) PaH*
Soln. A': HCl(m,), NaCl(m2); m2 = 4ml
0.
16.2
0.002006
.002058
2. 750
2.755
2,798
2.593
2.753
2.739
/ B I
33.2 ,002116 2,871 2.438 2.769
52.0 .a102210 3.017 2,222 2 , 787
73.4 .a102338 3.070 1.791 2,881
I I

-
85.4 ,002422 2.975 1.381 3.012 \\\

100 .a02548 2.193 0.162 5.081

Soln. B': HOAc(ml), NaOAc(mz), XaCl(m8)
m2 = 0.5nzl; m3 = 0.25m1 -It

0. 0.0201 4.405 4.462 4.408 0 50 '+O

75
WT, PERCENT METHANOL,
16.2 0206 4.649 4.509 4.644
+
I

33.2 " 0212 5.081 4.673 4.992 Fig. 2.-& log myci (upper curve) and ej - log ,,,YH
52.0 .0221 5.622 4.851 5.400 (lower curve) in methanol-water solvents: 0, this investigation;
73.4 .0234 6.266 4.998 6.065 0 , data of deLigny and Rehbach.Y
85.4 .a242 6.636 5.049 6.652
100 .0:255 7.054 5.088 9.950
Soln. C': triethanolammonium chloride(rnr ),
triethanolamine( m 2 )
m2 = 0.5m,
0. 0 . 00401 7.500 7.531 7.501
16.2 ,00412 7.355 7.204 7,367
33.2 ,00423 7.246 6.822 7.176
52.0 .00442 7.136 6.327 6.923
73.4 ,00468 6.946 5,613 fi ,745
85.4 ,00484 6.697 5.056 61.743

-
103 .00510 6.075 4.075 9.048

TABLE
I11
VALUES OF Ej + LOG my^^ AND -
Ej LOG m ~ IN ~ pH, UNITS,FOR
hfETIIAiYOL-~$TATER SOLVEKTB AT 25'
(Solutions C t o I)
Wt. % MeOH ej + log mYCl sj - log m,yE( = s
0 0 . IO02 =I= 0.002" 0
8.1 .lo89 =I= .003 0.002
16.3 .187 & .005 .004 I U
0 50 75 100
33.3 .412 f .002 .051 WT. PERCENT ETHANOL.
52.1 .'714 =I= .003 .130
68.1 1.004 f ,006 .121 Fig. 3.-Ej + log r a y G I (upper curve) and ej
- log my^
(lower curve) in ethanol-water solvents: 0, this investigation;
a Mean deviation from the average value. 0 , points derived from the data of Gutbezahl and Grunwald5;
X, data of W. J. Gelserna and C. L. deLigny (unpublished meas-
TABLE
IV urements ).
VALVES O F Ej + LOG m y c l AKD Ej - LOG m?H, IN pH UNI:TS, FOR
ETkIANOL-~TATER SOLVENTS .AT25"
from the results of this investigation that the potential
(Solutions A', B', and C') across the junction between alcohol-water solutions
Wt. % EtOH e, + log rnYCl Ej - log m Y E = 6 and a saturated or 3.0 M aqueous solution of potas-
0 -0.002 0 sium chloride is as reproducible as that at the junction
16.2 +O.ZOl 0.00:: between aqueous solutions and the salt bridge. The
33.2 0.484 ,086
useful interpretation of pH measurements in alcohol--
52.0 0.876 .221
73.4 1.419 .19e1
water solvents thus becomes a matter of importance.
85.6 1.771 -0.032 The quantity Ej -- log m~~ is of special significance
100 2.22 -2.91 in the determination of pH values in partially aqueous
and .t-ionaqueouljnzedia; it will be designated 6
Discussion
It is well known that the residual liquid-junction 6 ZSZ Ej - log m Y H (10)
error involved in pH rneasurements in dilute aqueous For a given alcohol-water solvent, one can write
solutions may amount to 0.02 to 0.0'5 unit, when the
pH is less than 2.5 or inore than ll.5.l9 In the inter- paH = pn%H- log m')" - log sYH = paH* -
mediate pH range, the error a t 25' appears to be less
than 0.02 unit when the ionic strength of the "un- log m Y H (11)
known" is in t'he range 0.005 to 0.2. It would appear where CZH and a ~ are* activities referred, respectively,
(19) See, for example,, ref. 16E8. to the standard staters in pure water and in the alcoholic
1838 J. C. AXDERSOK
AND 81. SCHIEBER
solvent. By eq. 5 , 10, and 11, these acidity functions
can be related to the experimental quantity pH
pa^ = pH - Ej (3
pmH = pH - 6 + log s ~ a (12)
pas" = pH - 6 (13)
Inasmuch as log s~~ in eq. 12 cannot be obtained for
"unknowns" and the liquid-junction potential error EJ
(eq. 5 ) cannot be derived independently of m ? ~ ,PWZH
and pa^ are not suitable units for interpreting the
measured acidity (pH) of alcohol-water media. On the
other hand, subtraction of a constant (6) from pH
gives an approximate value of pa^* (eq. 13).
I t is recognized that pH in the aqueous medium ap-
proaches the PUH scale of the standards oiily under
ideal conditions; similar restrictions apply to the inter-
pretation of pH - 6 in terms of pa^". It is therefore
suggested that the "practical" unit, related to pa^* in
the may that pH is related to pa^, be designated pH*.
The operational scale of pH* is thus defined as
where pH*(S) is identified with the conventional pall*
values of certain selected suitable standard buffer solu-
tions. The standard values can be assigned in the
same way as the KBS standards for pH.20
This procedure for calculating pa^*, which involves
essentially the addition of the logarithm of a coiiven-
tioiial single ion activity coefficient to the experimental
~ ( a g y c J *has
, ~ ~already been used to determine pa^*
values for certain citrate-chloride and phosphate-
chloride buffers in 10 and 20% methanolz2and for an
oxalate buffer and a succinate buffer oyer the entire
(20) Reference 15, chapter 4.
(21) Compare eq. 4. P ( ~ H ? C ~ )is* calculated by eq. 3, using sEO instead of
wEo (see footnote 18).
(22) R. L. Paiks, H. D. Clockford. and 9 . B. Knight, J . EEasha Mztchell
Set. Soe., 73, 289 (1957).
CRYSTAL GROWTH I N THE SYSTEM
E I T H I I X OXIDE-BORON TRIOXIDE-FERRIC OXIDE
BY J. C. A
Department of Electrzcal Engzneerzng, Iinperzal College of Sczence urd Technolugy, London SW 7 , England
Received lilarch 18, 2963
Phase equilibria in the ternary system Li?O-B20a-Fe2Q3 in the temperature range 950-1200" were studied and
approximate phase stability boundaries m r e established. The study was carried out by slowly cooling (2"/hr.)
ternary melts from the above temperatures to 800'. The crystals were identified by chemical, X-ray, and micro-
scopic analyses. From these data, recommended mixtures, soaking temperatures, and detailed procedures are
given for producing LiFeOz, LiFea08, and a-FeeOs single crystals.
Introduction
The flux-melt method for growing single crystals,
especially of ferrite materjals, is widely used. The com-
ponents of the crystal required are held in a molten
flux, in which they dissolve, and are slowly cooled over
an extended period. Under suitable conditions nuclea-
tion and growth of the desired crystals occurs.
The most commonly used flux for growth of iron-
bearing crystals such as ferrites and garnets is lead
TTol.67
range of water-methanol solvei~ts.~Different forms
of the Debye-Huckel equation were used in the two
investigations to evaluate ycl in eq. 4.
Values of pa^* have been calculated likewise from the
data given in Tables I and 11; they are listed in the
last coluinns of these tables. Equation 8 TT-as adopted
as the convention by which the activity coefficient of
chloride ion in each medium was evaluated. The pa^*
values in Tables I and 11, together with those from the
l i t e r a t ~ r e2,2~can be considered as provisional stand-
ards. pH*(S), for a scale of pH* in alcohol-water
media.
The pH* becomes equal to pH when the medium is
diluted continuously with pure water. Furthermore,
the pH* determined under ideal conditions (low ionic
strength, pH* neither too high nor too low) plays a well-
defined role in chemical equilibria, for equilibrium con-
stants in alcoholic solvents usually are also referred to
the standard state in each particular medium. As eq.
11 shows, pH* is, under ideal conditions of measure-
ment, actually a pmH value corrected for "salt effects,"
EYH.
The pH* is, however, a plurality of scales, different
for each solvent composition, rather than a single uni-
versal scale of acidity. Solutions of equal pH* in
different media, therefore, may prove quite different
with respect to acid-base behavior. A single scale for
all alcohol-water mixtures, based on pal$ (ey. l l ) , will
become possible only when the validity of an inde-
pendent nonthermodynamic evaluation of log m~~ or
EJ,such as that of Gutbezahl and Grunrvald5or that of
I z m a i l o ~ .is, ~generally
~ accepted.
(23) The simplest means of obtaining pH* mould be to measure pH and
subtract the appropriate value of 6 for the particular solvent composition and
temperature Some 1 ariahihty in the as) mmetry potential of the glass
electrode probablr can be expected a h e n the electrodes are transferred
from the aqueous standard solution t o the alcoholic medium, hence, st'nd-
arda of the same sollent composition as the' unknowns'' are t o be preferrede
If difficulty is evperienced i n adjusting the pH meter to the standard p H
value a t high alcohol concentrations, a n aibitiaiy refeience point can b*
chosen for measurement of the difference p H x ( X ) - pH*(S)
(24) K -1 Izinailoi, Dokl Akad Nauk SSSR,127, 104 (1959).
4AKD M.SCHIERER
~ ~ ~ ~ ~ ~
oxide.1,2 This suffers from the disadvantages of loss of
flux through volat,ilization, corrosion of the platinum
crucible by the moltsenmaterial, and the high density of
the flux, which leads to the crystals tending to float to
the surface. Various methods such as the addition of
PbF2, T120a, or P 2 0 5 to the flux have been tried3-5 to
(1) S. P. Remeika, J . Am. Ckem. Sne.. 78,1259 (1956).
(2) S.IT. Nielson and E. F. Dearborn, J . P h y s . Chem. S o l i d s , 5, 602 (1958).
(3) J. W.Nielson, J . A p p l . Phys., 31,515 (1960).
(4) H. Makrom and R . Iirishnan. Compt. rend., 254, 3340 (1962).
( 5 ) D. G. Wickham, J . BppE. P h ~ s .38,, 3597 (1962).