You are on page 1of 14

THERMODYNAMICS OF AQUEOUS UREA SOLUTIONS

[Manuscript received February 17, 19671 Summary


A compilation of heat capacity, heat of dilution, and activity data for aqueous urea solutions is made from published sources. The existing apparent volume data are supplemented by new measurements a t 0, 40, and 50'. The thermodynamic data are interpreted in terms of association models, and it is found that the use of Flory-Huggins statistics for the entropy of mixing of the various associated species and the solvent gives a much better description than mole-fraction statistics. Heats of association and equilibrium constants are evaluated.

The thermodynamic properties of aqueous urea solutions have been intensively studied and are probably better known than those of any other aqueous nonelectrolyte. High-quality data are available for : heat capacities1r2 (2-40') ; heat of d i l ~ t i o n (25") ~,~ ; and activity coefficients (25" and F.P.); density (25" and 30)).8Vi~cosities,~ diffusion coefficients,1 and self-diffusionlo coefficients (25")have also been measured. Activity data for mixed urea-NaC14 and urea-sucrose4 solutions have been obtained. S o l u b i l i t i e ~ are ~ ~ known. To complete the thermodynamic data only the densities a t temperatures far from room temperature are needed. Density measurements a t 0, 40, and 50' are reported here. The data have been fitted to suitable equations and used to calculate osmotic and activity coeffioients, heats of dilution, and apparent molar volumes in the range 0-40". These are given in Tables 7-11. A previous compilation of some of these properties by Kresheck and Scheraga12 extended only to 2 . 5 ~ .

* Department of Physical and Inorganic Chemistry, University of New England, Armidale,


N.S.W. Gucker, F. T., and Ayres, F . D., J. A m . chem. Soc., 1937, 59, 2152. Egan, E. P., and Luff, B. B., J. chem. engng Data, 1966, 11, 192. Gucker, F. T., and Pickard, H. B., J. A m . chem. Soc., 1940, 62, 1464. Ellerton, H. D., and Dunlop, P. J., J . phys. Chem., 1964, 70, 183. Scatchard, G., Hamer, W. J., and Wood, S. E., J. A m . chem. Soc., 1938, 60, 3061. Bower, V. E., and Robinson, R. A., J. phys. Chem., 1963, 67, 1524. Stokes, R. H., J. phys. Chem., 1966, 70, 1199. Gucker, F. T., Gage, F. W., and Rloser, C. E., J. A m . chem. Soc., 1938, 60, 2582. O Gosting, L. J., and Akeley, D. F., J. A m . chem. Soc., 1952, 74, 2058. lo Albright, J. G., and Mills, R., J. phys. Chem., 1965, 69, 3120. l1 Miller, F. W., and Dittmar, H. R., J. A m . chem. Soc., 1934, 56, 848. l2 Kresheck, G. C., and Scheraga, H. A., J. phys. Chem., 1965, 69, 1704.

'

Aust. J . Chem., 1967, 20, 2087-100

R. H . S T O K E S

Solvent and solute components ; solvent activity; activities of monomer, dimer . . . forms of solute (similar subscripts are used on other quantities) ; molarity ; heat capacity; enthalpy of formation of one mole of solute-solute bonds; molality-scale dimerization constant ; thermodynamic dimerization constant ; molarity-scale dimerization constant ; molality (moles solutelkg solvent) (e.g. m = 12m); number of moles of solute-solute bonds present a t equilibrium; numbers of moles of solute in monomeric, dimeric . . . forms a t equilibrium ; total number of moles of actual species present a t equilibrium; stoicheiometric numbers of moles of solvent and solute; numbers of moles of solute-solute bonds per stoicheiometric mole of solute;

VB/VA E @:/Vi;
entropy of mixing; volume of solution; partial molar volumes of solvent and solute; partial molar volumes of monomeric, dimeric, . . . solute; molecular weight of solvent ; mole fraction of species i ; m/(1+0.15 m); thermal expansion coefficient ; molal activity coefficient ; (unsubscripted) molal osmotic coefficient ; osmotic coefficient of a solution of unassociated solute obeying volume-fraction statistics ; general symbol for apparent molar quantities; subscript denotes quantity; superscript zero refers to infinite dilution. relative apparent molar enthalpy.

Schellman13 discussed the thermodynamic properties of urea solutions in terms of a series of association equilibria

u, +Ul+ u, U,+U, + U,,

(1)

etc.
=

(2) -2.09 kcal

and estimated the heat change in the association reaction (1) as AH


Sohellman, J . A., 0. r. Trav. Lab. Carlsberg, 1955, 29, 223.

THERMODYNAMICS O F AQUEOUS UREA SOLUTIONS

2089

by a method involving the limiting slope of the heat of dilution and the osmotic coefficient. He also obtained K , = 0.041 for reaction (1) a t 25", by considering the deviation of the osmotic coefficient from unity a t low concentrations. This treatment was extended to other temperatures by Kresheck and Scheraga,12 who found AH values from -2.36 kcal a t 0" to -2 11kcal a t 40". Implicit in Schellman's argument is the assumption that an ideal solution would have a molal osmotic coefficient of unity. The inexactitude of this assumption has been previously pointed out by the present writer.14 Expansion of the expression for the molal osmotic coefficient of an ideal aqueous non-electrolyte solution yields

The osmotic coefficient of urea a t l m and 25" is 0.962; it makes a considerable quantitative difference to our conclusions whether we measure the non-ideality by 1-4 = 0.038, or by 1.009-4 = 0.047. Furthermore since the osmotic coefficient tends to linearity in rn a t low concentrations, the difference will not be eliminated by extrapolation to zero concentration. This is because in non-electrolyte solutions we are dealing with short-range forces (in contrast with electrolytes, where the longa t low concentrations, and we can range forces lead to a Z/rn dependence of adequately test theories of the limiting slope). The borderline between "physical" interactions such as the electrostatic interaction of dipoles and "chemical" interactions resulting in the formation of a new chemical species is very indistinct, and there has been much argument about criteria of types of interaction, notably in the field of ion association and complex formation in electrolyte solutions. From a thermodynamic point of view, there is no need to distinguish between physical and chemical association, provided the product is stable enough to be regarded as a distinct entity in the solution, persisting through a considerable number of encounters with neighbouring molecules. As Guggenheim15 has pointed out, the mere fact that two molecules happen to be adjacent does not mean that they are associated. To apply an association model to the thermodynamic properties of a solution we require that (a) the heat of association should be negative and several times R T ; (6) the configurational entropy of mixing of the equilibrium mixture of the various associated and free species should be calculable for the model; (c) the number of separate equilibrium constants or heats we invoke should be as small as possible; otherwise we may have so many parameters that anything can be explained by any model. Condition (b) above is of great importance. The usual assumption is that the entropy of mixing follows the ideal solution law

AS,,,

= -R

C Xi ln X ,
i

where the X , refer to the actual species present a t equilibrium (as distinct from the thermodynamic components, which may be simply the solvent and the solute). I t is by no means certain that (4) is the best available expression for the entropy of mixing. For long-chain solute species in a small-molecular solvent, the limiting
l4

Stokes, R. H., J. phys. Chem., 1965, 69, 4012. Guggenheim, E. A,, Trans. Faraday Soc., 1960, 56, 1159, especially p. 1161.

2090

R. H. STOKES

Flory-Huggins16 expression :

is known to be superior to (4). For more or less globular molecules of substantially different sizes the position is less well known. Hildebrand17 considers that the evidence from solubility data favours (4), but vapour pressure measurements in progress in this laboratory favour (5). We shall therefore examine association models using both alternatives for the case of aqueous urea solutions. EQUATIONS FOR
A

SOLUTE FORMING A SERIESO F NON-CYCLIC POLYMERS

Consider a solution prepared from nA moles of solvent A, which we suppose to be monomeric throughout, and n, moles of solute B, which is subject to a series of polymerization equilibria as in (1) and (2). We assume that the molecular interactions responsible for the polymerization are of short range, so that the enthalpy change in the process Bn+B, is

= Bn+1

(6)

AH=h
independent of n. Provided that no closed-ring polymers are formed, every mole of B-B "bonds" formed results in the absorption of heat h. For the model to be applicable at all, h must be negative. The formation of a "bond" may well modify the interactions of the solute with the solvent; the heat h will then include a contribution from this origin.
@ , The apparent molar heat of dilution of the solute, per mole of solute for the process

is the enthalpy change

solution+aq.

s infinitely dilute solution

Since a t infinite dilution there are no solute-solute bonds left, we have

where n is the number of moles of bonds in a solution containing 1 kg of solvent. Since each bond formed results in the "disappearance" of one kinetic entity from the solution, the total number of such entities a t equilibrium is

where W A is the molecular weight of the solvent. At this point we must make our choice of expressions for the entropy of mixing:
la Flory,

P. J., J. chem. Phys., 1941, 9 , 660; 1942, 10, 51. Huggins, M. L., J. chem. Phys.,

1941, 9, 440.
l7

Hildebrand, J. H., and Scott, R. L., "Regular Solutions." (Prentice-Hall: New York
1962.)

THERXODYNAXICS OF AQUEOUS UREA SOLUTIONS

2091

(i) Entropy o f Mixing Given by Equation (4) (Mole-Fraction Htatistics) On this basis the activity of the solvent is given by

On substituting from ( 7 ) and (8), this gives

Thus the value of h is determinable from the heat of dilution and the solvent activity. If the model is realistic, h should be a constant independent of concentration, though not necessarily independent of temperature. (ii) Entropy o f Mixing Given by Equation ( 5 ) (Volume-Fraction or Flory-Huggins Xtutistics) On this model the solvent activity is that of an "athermal" mixture of the various equilibrium species and the solvent. The entropy of mixing of a solution containing Xi moles of each of several non-reacting species i is given by ( 5 ) , where 7, is the partial molar volume of species i. The volume of the solution is

v = 2 XIV,
V, nipi lna, = l n - + 1 - - - - ~ n j

(1%)

and if there is no volume or heat change on mixing, partial differentiation of (5) with respect to n, gives for the activity of species i

where the sum in the last term extends over all the species present. I n the present case, though there are chemical equilibria among the solute species, these do not alter the number of moles of solvent (nA)present. The heat changes involved in reaching equilibrium do not affect the calculation of the solvent activity by (13) except through their effect on (En,) in the last term, which now becomes the number of distinct kinetic entities present a t equilibrium, i.e. n,,,,,. Hence

The osmotic coefficient is given by (14), ( 7 ) , and (8):

2092

R. H. STOKES

The first two terms of (15) are what may be called the "unperturbed volumefraction statistics osmotic coefficient", i.e. the osmotic coefficient of a solute which undergoes no association but which obeys the Flory-Huggins statistics. Denoting these two terms by C$v, we have

According to this model, h can be evaluated if the heat of dilution, osmotic coefficient (or solvent activity), and molar volumes are known. Again, h should be a constant independent of concentration if the model is correct. I t will be seen that neither equation (11) nor equation (16) requires any specific assumptions about the number of polymerization equilibria involved, nor about the relative magnitudes of the successive equilibrium constants describing the equilibria. Some possibilities for these will be discussed later; first we shall test the two equations for constancy of the heat of bond formation h. Table 1 shows that the mole-fraction statistics assumption (i) fails badly.

In testing equation (16) we put:


r = @O,/

7;

(17)

The h values obtained from equation (16) using the data of Tables 7, 9, and 11, are shown in Table 2, and are remarkably constant up to the highest concentrations a t which the necessary data are available, in striking contrast with Table 1 based on equation (11). The temperature dependence of the h values corresponds to a positive value of AC, for the association reaction.
FOR EQUILIBRIUM CONSTANTS THE

ASSOCIATION PROCESS

The first stage of association may be taken as the dimerization (1). Using equation (13) for the activities of the monomer 1 and dimer 2 a t their actual equilibrium concentrations, we obtain for the thermodynamic equilibrium constant

where

V , is

the volume of the monomer and

v2that of the dimer.

The relation

THERMODYNAMICS OF AQUEOUS UREA SOLUTIONS

2093

between these volumes is not accessible from experiment, though the change of the apparent molar volume with concentration suggests that there is some volume change in the association reaction. However, we shall make the approximation 7, = 27, so that the last term of (18) can be dropped to give

whereas before we have put 7, = @. Equation (19) is (within the limits of the approximation 7 , = 27,) a relation between the thermodynamic equilibrium constant and the apparent volume-scale equilibrium constant (without activity
TABLE2
H E A T O F FORMATION O F T H E UREA-UREA BOND I N AQUEOUS SOLUTION CALCULATED B Y EQUATION

(16)

* The recent data of Egan and Luff2 make possible the extension of the test a t 25' to 2 0 ~ ; even a t this extreme ooncentration h remains in the range - 1.67 to - 1.70 kcal mole-1.
coefficients), K c . We may conclude quite generally that in solutions where the Flory-Huggins type of mixing statistics hold, the "elementary" form of the equilibrium constant expression

Kc = lI(c products)/II(c reactants)

(20)

should be a true isothermal constant provided the volume change in the reaction considered is negligible. Differentiation of equation (19) with respect to temperature gives for the heat of association

To connect Kc or K 2 with the measured thermodynamic properties we have to make some assumptions about higher stages of association. The assumption that

2094

R. H. STOKES

h is independent of the degree of association does not mean that the successive K values are also equal, for, as has often been pointed out, in the first stage of the association two molecules of monomer lose some orientational entropy, while subsequent stages involve only the addition of one more monomer molec~le to an already existing complex. This tends to make K for the later stages larger than for the dimerization step. Some preliminary analysis indicated that the relation Kc(,, = 2Kc
TABLE3
E Q U I L I B R I U M CONSTANTS

K O FOR

T H E DIMERIZATION PROCESS

TEMPERATURE D E P E N D E N C E 08

Kc

VALUES EXTRAPOL-4TED TO ZERO CONCENTRATION

K O , (extrap.)
0.065, 0,062, 0~060, 0.055,

2 5 10 20

K c (eqn. (25))*
0.0652 0.0633 0.0602 0.0549

t
25O 30 40

Kg (extrap.)
0.052, 0.050, 0.047,

K c (eqn. (25))* (0.0527) 0.0507 0.0473

* Values calculated from t h e 25'

value using the heat d a t a of Table 2 and t h e volume data.

for all stages after the dimerization would be a reasonable first approximation in the present case. With this assumption one obtains the following relation between q, the number of moles of solute-solute bonds per stoicheiometric mole of solute, and Kc :

Subject to our assumptions, q is given by (14):


nmm = % A + ~ B-4) U

so that on conversion to the molality scale we have

Equations (22) and (24) have been used to calculate Kc using the data of Tables 7 and 11 below for the osmotic coefficients and partial volumes. The Kc values

THERilIODYNAMICS OF AQUEOUS UREA SOLUTIOSS

2095

(Table 3) are more concentration dependent than one could wish, though the variation of 5-10% over the range to 12m is not unacceptable in view of the approximations made in representing the complex series of equilibria involved. Extrapolation of Kc to zero concentration should give a valid dimerization constant, and the results (Table 4) are consistent with the temperature dependence required by equation (21). This is shown by the agreement between the extrapolated values of Kc with those computed from the 25" value by integration of equation (19), using the h value of Table 2 and values of In@$ interpolated from Table 11 below:

The essential features of the successful model are: (a)For each bond formed the number of separate kinetic entities in the solution is reduced by one. (b) The heat of formation of a bond is independent of the stage of polymerization, but depends on temperature. ( c ) The entropy of mixing of the various solute species with each other and the solvent is given by the limiting form of the Flory-Huggins statistics (equation (2)). (d) All equilibrium constants for stages of association beyond the first are equal, but greater than the first by a constant factor (2 in Kc in the present case). The model gives a surprisingly exact relation between the osmotic coefficient and the heat of dilution over the whole range of concentration and temperature. It also yields equilibrium constants which vary by only a few per cent over the concentration range to 12m, and which are consistent with the heat and apparent volume data.

It should however be emphasized t h a t this quantitative success is not a final proof of the correctness of the model. Independent evidence about the extent of association is still needed; for example, if the fraction of monomer present could be found from spectroscopic data (as has beea done with alcohols in inert solvents), a valuable check would be provided. It may also be asked why the model works so well in spite of ignoring the fact that the solvent water is also an associated substance.

The Temperature Dependence of the Apparent Molar Volume of Aqueous L'reo


Urea shown in previous work7 to be of satisfactory purity was used. Solutions were prepared by weight close to the round molalities quoted in Table 1, and used on the day of preparation. Density measurements at 0' were made in an iceldistilled water bath, and those below On in a stirred Dewar flask filled with a suitable freezing-mixture. Those a t 40' and 50' were made in a water bath controlled to &0.005 at 40 and & 0 . 0 l 0 at 50'. Temperatures were measured with calorimeter thermometers calibrated in situ against a platinum resistance thermometer and Mdler bridge.

R. H. STOKES
At 40, Ostwald pyknometers of 25-ml capacity were used. For the other temperatures dilatometers of 80-120-ml capacity, having stems calibrated within 0.001 ml, were used. Calibrations with water were made a t each temperature. Appropriate vacuum corrections were applied to all weighings. The reproducibility of the density measurements was 1-3 x 10-5 ml-1. , is estimated as kO.02 ml mole-1 The data are presented in Table 5. The uncertainty in @ a t l m and is inversely proportional to molality. Table 5 includes the smoothed data of Gucker, Gage, and Mosers a t 25" and 30" for completeness.

A P P A R E N T MOLAR VOLUMES O F U R E A

@ , in (ml mole-l)

From data of Gucker, Gage, and M o ~ e r . ~ Extrapolated.

Fig, 1.-Concentration and temperature dependence of the apparent molar volume of urea in aqueous solution.

I
0

I 2

I
4

I
6

I
8

!
10

Molality

The results show two features of interest: (a) The concentration dependence of @ , becomes rapidly more pronounced as the temperature is lowered (Fig. 1). This probably reflects the increasing importance of solute-solute interactions a t low temperatures, in that associated solute molecules disturb the water structure less than free solute molecules. (b) The limiting apparent molar volumes at infinite dilution, @$, are strongly temperature dependent. From the slope of the curve in Figure 2 we may estimate the thermal expansion coefficient of the limiting apparent molar volume, defined by

THERMODYXAMICS OF AQUEOUS UREA SOLUTIONS

2097

(This quantity must not be confused with the thermal expansion coefficient of the solution, which of course becomes equal to that of water a t infinite dilution.) Table 6 compares these thermal expansion coefficients with some others of interest. The thermal expansion coefficients of both normal and associated pure liquids increase with temperature, more rapidly as their boiling points are approached. This is also true of their apparent molar volumes a t infinite dilution in "normal" solvents. We see from Table 6 that a(@.) for urea shows the opposite

Fig. 2.-Temperature dependence of the apparent molar volume of urea in aqueous solution a t i d n i t e dilution and at 8m.

THERXAL EXPANSION

O F

UREA AT

TABLE6 COEFFICIENTS O F THE LIMITING APPARENT IxFINITE DILUTION, C O I M P A B E D WITH THOSE O F LIQUIDS AND O F SOLID UREA Values in a/(degc)-I

MOLAR VOLUME

VARIOUS

PURE

a(@), urea a(HCONH,) a(Et0H) a(%O) a(so1id urea)

0.0025 0~00076 0.00105 -0.00007 0~00016

0.0015 0.00076 0~00108 0.00026 0.00016

0~0011 0.00076 0~00110 0.00039 0~00016

behaviour; it is abnormally large a t low temperatures and decreases towards a normal value as the temperature is raised. I t is at all temperatures an order of magnitude larger than that of solid urea. 1 8 The thermal expansion of pure liquid urea is unknown, because of decomposition a t the melting point, but it might well be similar to that of formamide. Since we are now considering infinite dilution values, the increase in (@) a t lower temperatures cannot be ascribed t o solute-solute interactions but must arise from solute-solvent interactions. The forces between the urea dipoles and water dipoles do not seem likely to provide a sufficiently strong temperature dependence on any simple electrostatic model. The effect possibly arises from structural changes in the water near the urea molecules: this would be greatest a t low temperatures where the structural pecularities are most marked. Whatever the process may be, it is sufficient to cause a large decrease ( 2 . 4 5 ml mole-1) in the limiting apparent molar volume between 25' and 0".
l8

"International Critical Tables." Vol. 111, p. 45. (WIcGraw-Hill: New York 1926.)

2098

R. H. STOKES

Compilation o f Thermodynamic Properties o f Aqueous Urea Solutions Note on Curve Fitting.-Most of the thermodynamic properties of urea solutions, when plotted against molality, show (especially a t low temperatures) a marked curvature in the dilute region and tend to straighten out a t high molalities. Curves of this form are not well suited to representation by power series in the molality. For example, Ellerton and Dunlop4 found it TABLE7
MOLAL OSMOTIC COEFFICIENTS

O F AQUEOUS UREA

SOLUTIONS, 2-40*

* At

25' the following equation reproduces the osmotic coefficients to 20111:

TABLE 8
APPARENT
@ ,,
=
MOLAR HEAT

CAPACITIES

@8,+am(l+O.l5m)+bm

Dc, in (cal deg-l mole-1)


S.D. 0.08 0.08 0.07 0.06 0.09 0.19 Range of m 0.3- 8 0.3-10 0.3-10 0.3-13 0.3-17 0.3-17

necessary to use a sixth-power polynomial in m to describe their 25" osmotic coefficients up to 20m. At 20m the successive terms of this series are of alternating sign and all larger in absolute value than the quantity they describe. Such expressions can lead to serious errors if the last term is inadvertently dropped (as did in fact occur in the case quoted, where the term in ma was omitted from the equation given by Ellerton and Dunlop), and also demand a large number of significant figures in all coefficients.

THERMODYNAMICS OF AQUEOUS UREA SOLUTIONS

2099

I n the case of urea solutions the author has found the use of the variable IL. = m/(1+0.15 m) to yield much more convenient equations in analysing the heat capacity and apparent volume data. I n computing the data below, the first step was to fit the 25' osmotic coefficients4 to equation (26) (Table 7). This reproduces the data to 20m with an average deviation of 0.0005 in 4; for the 50 points, the only deviations greater than 0.001 in 4 are for two points a t 0.2101m and 0.9365m (64 = 0.0018 and 0.0022 respectively).

RELATIVE APPARENT MOLAR

TABLE9 ENTHALPIES:

THE

COEFFICIENTS A

AND

B (28) (29)

At 25", up to 12m ~t to

@ , ( , , )

= -85.87m+6.815m2-0.4569maf

0.01470m4

@ L ( ~ )= @L(Z5) f A m l ( 1S

O . 15m)+Bm

Next the apparent molar heat capacities1 in the range 2-40" were fitted to equation (27). From the reproducibility of the heat capacity measurements it was concluded that the uncertainty in @ ,, was of the order of (0. llm) cal deg-I mole-1. Accordingly it seemed best to omit the measurements a t 0.093 and 0.164m from the analysis. The coefficients of equation (27) and the standard deviations are given in Table 8. The heats of dilution measured by Gucker and Pickarda a t 25" are well represented by their equation (equation (28) in Table 9). Since these measurements* extend only to 12m, this is the highest concentration a t which @ ,, and hence 4 and y, can be calculated a t temperatures

* The more complete 25' measurements of ref. 2 became available after the calculations were done. They make possible an extension of the oalcultttions in the range 26-40' to 17m, but not a t lower temperatures as the heat capacity data of ref. 1 remain a restriction.

2100

R. H. STOKES

other than 25'. The coefficients of equation (27) were interpolated at 1-deg intervals from 2" t o 40' by four-point Lagrangian interpolation, and numerical integration of equation (27) with respect to temperature and combination with equation (28) led to equation (24). The coefficients A and B of equation (29) were obtained a t 1-deg intervals (not all reported here). Conversion t o partial molar enthalpies of the solvent, and a further numerical integration with respect to temperature (of.') along with equation (27), yielded the osmotic coefficients to 12m over the range 2-40' given in Table 7. The activity coefficients y (Table 10) were then obtained via the Gibbs-Duhem equation by another numerical integration. The apparent molar volumes are listed in Table 11. TABLE11
APPARENT MOLAR VOLUMES

cD,
t
0

@~+am/(1+0.15m).Data from Table 5 and ref. 8

@"V
41.80 44.17 45.08 45.57
4

a 0.460 0.212 0.125 0.127

S.D. 0.03 0.04 0.01 0.01

25 40 50

The experimental work and a large part of the calculations were carried out during the tenure of a Xational Science Foundation Senior Foreign Scientist Fellowship a t the University of Wisconsin during 1965-66. The author thanks the N.S.F., and Professor L. J. Gosting and J. W. Williams for their hospitality. The work was supported in part by Research Grant AM-05177 from the National Institute of Arthritis and Metabolic Diseases. The guidance of Dr E. J. Burr of the University of New England in programming the calculations for the 1.B.N 1620 computer is gratefully acknowledged.

You might also like