KEITH J.

LAIDLER

56 1

THE ENERGETICS OF MICELLE FORMATION

BY J. N. PHILLIPS * The Australian National University, Canberra.

Received 14th May, 1954; in final form,4th November, 1954

Two aspects of the energetics of micelle formation have been discussed in this paper. The first deals with the standard free energy change (AGO) involved. This may be calculated from the c.m.c. provided the degree of ionization of the micelle p / N is known. In order to facilitate the calculation of AGO a new definition of the c.m.c. is proposed. It is defined as the point corresponding to the maximum change in gradient in an ideal property-concentration (4 against Tc)relationship, being given by
=0 .

The free energy change per CH2 with reference to a standard state of mole fraction unity is shown to correspond approximately to the energy of adsorption per CH2 at the oil-water interface. The observed relationship between the c.m.c. and chain length in the presence and absence of added electrolyte has been quantitatively accounted for on the basis of a liquid structure for the hydrocarbon chains in the micelle. The second aspect of micelle formation discussed deals with the heat and entropy changes associated with the process. The calculation of the energy, heat and entropy terms involved in the cycle, solid -+detergent in solution -+ micelle in solution -+ solid, is outlined. The energy criteria governing the existence of micelles are discussed and a definition of the Krafft point in terms of these criteria proposed.

Although the aggregation of detergent ions in solution has been the subject of extensive investigations during the past 40 years, there appears to be little quantitative work on the energy relationships associated with the process.1-8 It is the purpose of this paper to enquire into the energy associated with the formation of micelles together with the heat and entropy changes involved in the process. In the first instance two assumptions will be made concerning the aggregation process : (i) that the equilibrium between the single ions and the micelles is a readily reversible one, and (ii) that the micelles are effectively monodispersed. The first assumption is generally accepted, whilst with respect to the second both light-scattering 3 and diffusion measurements 9 s 10 have failed to detect within

* Present address : Dept. of Physical Chemistry, Kings CoIlege, Strand, W.C. 2.

562

MICELLE FORMATION

the limit of experimental error any evidence of polydispersity in a number of micelle systems. Accordingly, for an anionic detergent, the equilibrium between the single ions and the micelles may be represented by eqn. (l), NS(N - Z)Na+ + Mz-, (1) where S- represents the single detergent ions, Naf represents the counter ions, Mz- represents the micelles, N represents the number of detergent ions per micelle, and 2 represents the net charge on the micelle,

and the corresponding equilibrium constant by eqn. (2),

where K , is the equilibrium constant and [a,], [aNa] and [a,] are the activities of the single detergent ions, counter ions and micelles respectively. Owing to the limited quantity of experimental data available, such calculations of the equilibrium constant as have been attempted 1 s 3s 4s 1 1 ~ 1 % 13 have neglected the counter ions or assumed activity coefficients of unity. Both these assumptions would appear to be invalid under experimental conditions and it is not surprising that the calculations have not been extended beyond a semi-quantitative level. However, the recent application of light scattering to the study of micelles by Debye 39 4 provides a means of overcoming these difficulties. As Debye showed, N may be determined by extrapolation of the HC/T against C curve to C = 0, H being the Einstein constant, T the turbidity and C the micelle concentration. Furthermore, as Edsall et a l . 1 4 and Mysels 15 have shown, the effective charge on the micelle may be calculated from the limiting slope of the same curve. These authors found that under ideal conditions the light scattering of colloidal electrolytes could be explained in terms of the Donnan concept and the fluctuation theory, the slope of the curve being related to the total electrolyte concentration and the colloidal charge. Since the Donnan effect is based on ideality, the charge so determined will not be the true charge 2, but rather the effective or equivalent ideal charge p . It is convenient to define the charge on the micelle in terms of this effective charge p which takes into account the total charge N, the number (N-2) of counter ions physically adsorbed on the surface and the number effectively bound in the double layer. It may be interpreted as an activity coefficient and implies that due to double-layer interactions a micellar ion of total primary valence N and actual valence 2 is equivalent in behaviour to an ideal ion of valence p . Hence the term [am] can be replaced by [Cm] and [a,a]N-Z by [C,,lN-P, where [ C , ] and [CNa] are the concentrations of micelles and counter ions respectively. Furthermore, at concentrations at or below the c.m.c., detergents are known to behave as ideal uni-univalent electrolytes so that in this region [a,] may be replaced by [C,],where [C,] is the concentration of the single detergent ions. Eqn. (2) accordingly reduces to
r r
i

However, in order to calculate the equilibrium constant it is necessary to know in addition to Nand p , the concentrations of both the single ions and the micelles at any one total detergent ion concentration, but, due to the large exponential term involved in the equilibrium constant ( N being of the order of loo), this is virtually impossible to determine experimentally. This difficulty may be overcome by making use of the fact that many properties of detergent solutions, e.g. con-

J. N. P H I L L I P S

563

ductivity, turbidity, refractive increment, etc., when plotted against the total detergent ion concentration T, show a sharp change in gradient at a point known as the critical micelle concentration (c.m.c.). This point appears independent of whether the property under consideration is ideal or otherwise but it is convenient mathematically f to consider an ideal colligative property and to define the c.m.c. as that concentration corresponding to the maximum change in gradient in the property-concentration (4 - T,) curve,
i .e.

(9)
dTc3
T,
= c.m.c.

=o,

(4)
(5)

where A and B being proportionality constants. The various mass balances may be written Tc = C s NCm

CNa = x

+ Cs + P C ~ ,

(6) (7)

where x is the concentration of added electrolyte. The equilibrium constant K , can be calculated as a function of N, p and the 4 ) , (3, (6) and (7). The complete c.m.c. by simultaneously solving eqn. (3), ( solution is tedious and it is simpler to consider an approximate solution of the two limiting cases of no added electrolyte and a swamping excess of electrolyte. The simultaneous solution of eqn. (3)-(7) must satisfy the condition imposed by eqn. (4)that T , = c.m.c., and since the concentration of micelles [C,] in this region is effectively zero, then eqn. (7) may be replaced by the approximate relationship (7a), CNaN x c,. (74 Let us now consider the case of no added electrolyte, i.e. x = 0. Making the appropriate substitutions from eqn. (3) and (7a) in (5) and (6) we obtain (5a) and ( 6 4 respectively, 4 = A[C,] BKnz[Cs]2N-p (54 and T, = [C,] NKnL[Cs]2N-p. (6a) Differentiating both eqn. (5a) and (6a) with respect to Cs we find d+/dC, = A B(2N - p)Km[C s ]2N -~-l, and dT,/dC, = 1 N(2N - P ) K , [ C ~ ] ~ ~ - P - ~ ,

+ +

+ +

i.e.

In a similar manner this expression may be differentiated twice with respect to T, and equating the resultant differential (d34/dTc3) to zero (see eqn. (4)) and rearranging we have N(2N - p)(4N - 2p - 1) K ; 1 = cs* m-p-1, (8) (2N - p - 2) where Cs* is the concentration of single detergent ions when T, = c.m.c. Now at Tc = c.m.c., eqn. (3) can be written in the form

and equating eqn. (8) and (80) we obtain (2N - p)(4N - 2p - 1) (c.m.c.). "* = (2N- p - 1)(4N - 2p 2) t The choice of an ideal colligative property expressed by eqn. ( 5 ) restricts eqn. (4) to one solution. A non-ideal relationship may result in a number of solutions only one of which corresponds to the c.m.c.

564

MICELLE FORMATION

Substituting for C,, from (8b) in (8) then K i 1 = N(2N - p)(4N - 2p - 1) (2N - p)(4N - 2p - 1) (c.m.c.) (8c) -{(2N - p - 1)(4N--2p+2) (2N - p - 2) Since N is of the order of 50-100 and p of the order of 10, then eqn. (8c) reduces to the approximate expression
____.

r-p-l.

K,,, _N 4N2(c.rn.c.)*N-f-1. (9) It can be shown that the ratio (Cs*/Cm*) of the molar concentrations of the single ions to that of the micelles at the critical micelle point is given by N(2N - p)(4N - 2p - 1) c s * __ NC,* __ __ N 4N2. (94 C,* (c.m.c.) - C,* (2N - p - 2) This justifies the approximation (see eqn. (7) and (70)) that at the c.m.c. the concentration of the micelles can be neglected relative to that of the single ions. N x) it Similarly in the case of a swamping excess of added electrolyte (CNa can be shown that
K k l N 2N2 (c.m.c.)N-IxN -P

(96)

and
C,*/C,* N 2N2. (9c) For all practical purposes, K , may be represented over the whole range of electrolyte concentration by K; = 3N2 (c.m.c.)N-l[x (c.m.c.)lN-P (10) Whilst a maximum error of f 33 % in K , may appear to be very great the exponential factors are so large that, for example, in the calculation of the standard free energy of formation of the sodium dodecyl sulphate micelle the error is rt 0.05 % which is well within experimental error. The calculation of the standard free energy change AGO per molecule associated with micelle formation follows directly from the equilibrium constant and is given by

In (c.m.c.) (N - P) In [x (c.m.c.)l. (12) N N N The value of AGOdepends on the standard states chosen and it is convenient for purposes of comparison to express concentrations in mole fractions so that AGO will be with reference to a state of mole fraction unity. The physical significance of this standard state will be discussed later. It may be noted that the term p/N involved in the calculation of AGOmay be determined either from light scattering, as already mentioned, from a combination of osmotic and conductance data by an incompletely published method due to Philippoff 16 ( p / N = a), or from conductance data and the aggregation number.17 These various methods of determining p / N facilitate the calculation of AGO for a number of detergent micelles (see Philippoff,l6 table 1, for collected data). Values of AG"/kT have been calculated at various salt concentrations for sodium dodecyl (lauryl) sulphate dodecylamine hydrochloride and decyl-, dodecyl-, and tetradecyl-trimethylammonium bromide, the results being summarized in table 1. The qualitative relationship between the concentration of single ions, micelles and free counter ions has been predicted from general mass action considerations.lL12,13 It has been postulated that up to the c.m.c. the concentration of micelles is negligible whilst beyond this point it increases rapidly, with the concentration of free detergent ions remaining almost constant. As Murray and Hartley 12 pointed out, however, since the number of bound counter ions can
-.

i.e.

AGO
kT

- In 3 -

+ 2 1 n N +-( N -

AG"
1)

(kT/N) In K,,

(1 1)

J. N. PHILLIPS

565

never exceed the number of detergent ions in the micelle, then, beyond the c.m.c., the concentration of free detergent ions will tend to decrease slightly and that of the free counter ions to increase slightly with increasing total detergent ion concentration. Such calculations can now be carried out quantitatively and their applicability to the study of the effect of salt on the solubility of detergents has been recently discussed.19 It is interesting to note that according to the definition of the c.m.c. adopted in this paper, i.e.

( 9 )

=o,

dTc3 T~ = c.m.c. the ratio of the number of free detergent ions to the number in the micellar form at the c.m.c. is of the order 3N, the corresponding mole ratio being 3N2. The calculated magnitude of the free energy gained per ion entering a C12 micelle, (14-16kT), suggests that the interior of the aggregate is liquid in character. The work of adsorption per CH2 group at an oil-water interface has been shown to be 1-37kT,% 21 which for a C12 chain amounts to 16-4kT. The close agreement between the calculated values and 16.4 k T indicates the liquid-like character of the interior of the micelle. If this be so the gain in free energy per additional CH2 should be equal to 1*37kT, other factors being constant. Inspection of table 1 reveals that this is approximately true for the quaternary ammonium bromide series, which shows an increase in - AGO of 6.3 kT between Clo and C14 the calculated increment being 5.5 kT. It will be noted that A G O is virtually independent of electrolyte concentration -particularly for sodium dodecyl sulphate. This suggests that the electrical contribution to the overall free energy change is small compared with the cohesive contribution, a fact which may be alternatively deduced from the agreement between A G O and 16.4kT. The effect of the nature of the polar head on A G O is more striking. The results in table 1 suggest that the quaternary ammonium micelle is more highly ionized than the sulphate micelle-assuming the lower AGOto be associated with the greater electrical repulsion. Accepting a liquid structure for the interior of the micelle and treating other factors as constant, one may write - A G ) = (1.37kT)~ B', (1 3) where n is the number of CH2 groups per detergent chain and B' is a constant. In the absence of added electrolyte it may be deduced from eqn. (12) that 2.3 kT(logl,3 2 loglo N ) 2-3kT (2N - p - 1) AGO = loglo (c.rn.c.). (14) 3N N

TABLE 1
detergent solution

c.m.c.

PIN

sodium dodecyl sulphate 18

water *02MNaCl -03M NaCl -10M NaCl -20M NaCl *40MNaCl dodecylamine water hydrochloride 4 .0157M NaCl -0237M NaCl -0460M NaCl decyltrimethyl-amwater monium bromide 4 -013M NaCl dodecyltrimethyl-am- water monium bromide 3 ~ 4 so13M NaCl tetradecyltrimethyl-am- water monium bromide 3- 4 ,013M NaCl
3 9
3 9

-0081 M *00382M -00309M -00139M -00083M -00052M -0131M -0104M -00925M .00723M -0680M -0634M *0153M -0107M -00302M -00180M

so
94 100 112 118 126 56 93 101 142 36 38 50 56 75 96

0.18 0.14 0-13 0.12 0.14 0.13 0.14 0.13 0.12 0.09 0.25 0.26 0.21 0-17 0.14 013

+ 15.8 + 16.0 + 16-2 + 15-8 + 15.7

AGo/kT

4-15-21

+ 14-3 + 14.6)' + 17.9 + 17.3)'

14*5 17'6

566

MlCELLE FORMATION

Since N is large, eqn. (14) reduces to (15), AGO= 2-3 (2 - p/N)kTloglo (c.m.c.). (15) Accepting a value of 0.2 for p / N (see table 1 and Philippoffs 16 calculated value of a (a = p/iV)) and coupling eqn. (13) and (15) leads to loglo (c.m.c.) = 0.33n B, where B = Bj4.6 kT. This predicts a linear relationship between log c.m.c. and rz which is in fact found. Moreover, the data collected by Harkins22 and Klevens23 for acid, sulphonate and quaternary ammonium halide series indicates that the effect of chain length is quantitatively given by eqn. (16), the c.m.c. being increased by a two-fold factor for each CH2 group removed * (antilog 0.33 = 2.14). It may be pointed out that Debye 3, 4 using Harkins data calculated an energy gain of 2kT per CH2 entering the micelle (1200 cal/mole) as compared with the value of 1.37 kT (810 cal/mole) implied in the above calculations. The theory on which Debyes calculation has been based has been criticized 8 due to its neglect of the entropy of the system and of the surface energy, and the value may be accordingly discarded. In the presence of excess electrolyte it may be deduced from eqn. (12) that
-

AGO=

+ Since N is large and x is

2*3kT(loglo3 -4 2 loglo N) , 2.3 kT ( N - 1) -_____ loglo (c.m.c.) T N N 2.3 kT ( N -- p ) loglo [ x 4-(c.m.c.)].

> (c.m.c.), eqn. (17) reduces to

AGN 2-3kT loglo (c.m.c.)

+ 2.3 (1

- p / N ) lCTlogl0 x,

(18)

and coupling eqn. (13) and (18) gives - loglo (c.m.c.) c !0.6n where

+ B ,

Again a linear relationship is predicted between loglo (c.m.c.) and rz, although the slope should be greater than that in the absence of electrolytes. This relationClo-, ship has been confirmed experimentally 25 in our laboratory for sodium CS-, C12,- C14-sulphates in 0.16 M NaCl at 25 C and 35 C where the average slope was found to be 0.50 Zt .03 in reasonable agreement with the calculated value of 0.6. The c.m.c. data suggests that the energy gained per CH2 is slightly less than 1-37kT, a value of 1.2 kT fitting the data better. Such a value gives a slope of 0.29 for the log (c.m.c.) against n relationship in the absence of added electrolyte and 0.52 for that in the presence of excess electrolyte. The lower value of the energy increment per CH2 entering a micelle (1.2 kT) as compared with its value at an oil-water interface (1.37 kT) can be attributed to the fact that when aggregation occurs the amount of hydrocarbon surface exposed to the aqueous phase is merely reduced whereas when adsorption at an oil-water interface takes place it is completely eliminated. The heat and entropy changes involved in the formation of ionic aggregates have been discussed in some detail by Stainsby and Alexander.6 The heat change AH, determined from the effect of temperature on the c.m.c. was found to be small and negative (N- 1.0 kcal/mole) at 25-35 C for a number of ionic detergents

* For the perfluoro-acids the c.m.c. changes by a factor of 3.1 per CF2 group.24 Assuming a value of 0.2 for PIN this leads to a calculated energy gain of 1300 callmole for the CF2 group as compared with the value of 800 cal/mole for the CHZgroup. This value of 1300 callmole is in reasonabIe agreement with the adsorption energy of a CF2 group at the air-water interface (N 1470 cal/mole) recently determined in this laboratory by Dr. J. T.Da v i s and Dr. H. B . Klevens.

J. N. PHILLIPS

567

and appeared to become more negative with increasing temperature. On the other hand, theoretical estimates of the heat and entropy changes, based on the energy factors involved, led to considerably higher values (from - 4 . 5 to - 7.0 kcal/mole at 25" C). An alternative approach to the problem is set out below. The free energy, heat and entropy changes involved in the aggregation process may be expressed, relative to a given standard state, by eqn. (20), AG" = AH" - T A P . (20) It has proved convenient to adopt unit mole fraction as the reference standard state. Such a reference state in the system under consideration is purely hypothetical and can best be interpreted physically in terms of a completely hydrated state. Fig. 1 sets out the relationship between the solid detergent (standard state 1) and a solution of the detergent at the critical micelle concentration (standard state 2).
1

detergent (solid)

-+detergent

- - -'- - -+micelle f (hydrate)

4 FIG. 1.

AH4=0

3 t- - - - - - micelie detergent (solution) + (solution) AG3=0

Stage 1 represents the complete hydration of the solid detergent to form the hypothetical hydrate which is identified as the standard state of mole fraction unity of the reactant. Stage 2 represents the dispersion of the hydrate into a solution of the detergent at the critical micelle concentration. Since complete hydration is postulated in stage 1, no heat change will be involved in this process (AH2 = 0) and the energy change will be an entropy effect only. Stage 3 represents the equilibrium in the solution between the single ions, the counter ions and the micelles and accordingIy the free energy change will be zero (AG, = 0). Stage 4 represents the formation of the micelle hydrate, which may be regarded as the standard state of mole fraction unity of the product, the heat change again being zero (AH, = 0). Stage 5 represents the conversion of the single ions to the micelle, with reference to a standard state of mole fraction unity. The determination of some of the free energy, heat and entropy changes involved in the system can be effected as follows : (a) AGs has been identified with the standard free energy change with reference to mole fraction unity and is accordingly given by eqn. (21)
AG~=AG"NRT 2---

&)

In (c.m.c.1.

Since PIN is usually 0.2 & 0.05, then as a first approximation AG, will be given by AG, N (1-8 f 0.05) RTln (c.m.c.). (22) (b) The heat change involved in the equilibrium between the single ions and the micelles (AH3) can be calculated from the effect of temperature on the equilibrium constant and hence on the c.m.c., as in eqn. (23),*

* The value of

and Ubbelohde 26 differ from those calculated here by the factor 2

AH, determined by both Stainsby and Alexander 6 and by Flockhart - 1 due to

neglect of the gegen-ion term in their calculation. by eqn. (25) differs from that calculated by Stainsby and Alexander by a factor of 20, again due to neglect of the gegen-ion term.

-c -), ( " Similarly the value of AH1 determined

568

MICELLE F O R M A T I O N

(c) The free energy change involved in the transfer of the detergent from the solid to the hydrated state (AG,) will be given by AGl = - RTln K,, (24) where Ks is the solubility product expressed in mole fractions. ( d ) The heat change involved in the transfer of the detergent from the solid to the hydrated state (AH,) will be given by

(e) The free energy of dispersion of the detergent hydrate AG2 will be given by AG2 = 2RTln (c.m.c.),

(29

where the c.m.c. is expressed in mole fractions. The remaining terms may be calculated directly. Since AH2 = AH4 = 0, then AH^ = AH,, - TASz = AG, and - TAS4 = AG4. Furthermore, since AG, = 0 then AG5 = AG2 1- AG4. Finally, TASl = AH1 - AGl, and TAS5 = AH, - AG5. The energy relationships may be used to define the condition under which micelles are formed as being given by AGl AG2 < 0. For simple uncharged long chain compounds, e.g. alcohols, ketones, etc., the free energy of solution (AG1) is so positive that (AG1 AG2) > 0 and no micelles are formed. AG1 may be reduced either by an ionic polar head or by a highly hydrated polar head of the poly-glycol or ethylene oxide type. However, charged detergents do not form micelles under all conditions, in particular at low temperatures or high electrolyte concentration. The point at which micelles begin to form, which in the case of the temperature relationship is known as the Krafft point, is given by ~ 0. This definition of the Krafft point is equivalent the condition A G ~ A G = to that commonly used, viz., " the temperature at which the c.m.c. equals the saturation solubility ". The calculated values of AG, AH and TAS for the sodium decyl suIphonate system at 25" C (Krafft point = 22.5" C), based on the solubility measurements of Tartar and Wright 27 are shown schematically in fig. 2, the values being expressed in kcal/mole. The c.m.c. has been taken as 0.04 mole/l.,28 and PIN as 0.345.19

CloH21S03- Na+

AH, = TAS1 =

+ 14.6

6.2

AH, = - 0- 5 TASs = -I- 6.6 AGz = - 8 . 5 AHz=O

AG4 =

AH4 = 0 TAS4 = - 1.4

+ 1.4

AG3 = 0 AH3 = - 0.5 TAS3 = - 0 5

FIG.2.
It will be observed that there is a considerable gain in entropy associated with the formation of the detergent hydrate (AS1 = 20.7 cal/mole deg.), and a further entropy gain (AS, = 22 cal/mole deg.) in proceeding to the micelle

J. N . P H I L L I P S

569

hydrate. Whilst it is tempting to interpret such entropy changes in terms of structural configurations the many complications associated with this type of system render any such interpretation dubious. The author wishes to acknowledge the interest and encouragement of Prof. Sir Eric Rideal and Prof. Karol Mysels and to express his appreciation to the Colgate-PalmolivePeet Company for providing a fellowship, and to the Australian National University. Part of this paper was presented in an abridged form at the annual meeting of the American Chemical Society (Colloid Section), Chicago, September 1953.
Hartley, Aqueous Solutions of Parafin-Chain Salts (Hermann et cie,, Paris, 1936). Ward, Proc. Rov. SOC. A , 1940, 176, 412. Debye, Ann. N. Y. Acad. Sci., 1949, 51, 575. 4Debye, J. Physic. Chem., 1949, 53, 1. 5 Hobbs, J , Physic. Chem., 1951, 55, 675. 6 Stainsby and Alexander, Trans. Farada-y SOC., 1950, 46, 587. 7 Nakagaki, J. Chem. Sac. Japan (Pure Chem. Section), 1951, 72, 113. 8 Ooshika, J. Colloid Sci., 1954, 9, 254. 9 Lamm, Kolloid-Z., 1942, 98, 45. IoVetter, J. Physic. Chem., 1947, 51, 262. Jones and Bury, Phil. Mag. A, 1927, 6, 841. 12 Murray and Hartley, Trans. Faraday Soc., 1935, 31, 183. l 3 Corrin, J. Colloid Sci., 1948, 3, 333. 14 Edsall, Edelhoch, Lontie and Morrison, J . Amer. Chem. SOC.,1950, 72,4641. l5 Mysels, J. Physic. Chem., 1954, 58, 303. 16 Philippoff, Faraday Sac. Discussions, 1951, 11, 96. l7 Hutchinson, J. Colloid Sci., 1954, 9, 190. 18 Mysels and Phillips, in preparation. 19 Phillips ; paper presented at 1st World Congress on Detergency (Paris, 1954). 20 Davies, Trans. Faraday Soc., 1952, 48, 1052. 21 Crisp, Surface Chemistry (Buttenvorths, London, 1949), p. 65. 22 Harkins, J. Amer. Chem. Soc., 1947, 69, 1428. 23 Klevens, J. Physic. Chem., 1948, 52, 130. 24 Klevens and Raison, J. Chim. Phys., 1954, 51, 1. 25 Taylor, private communication. 26 Flockhart and Ubbelohde, J. Colloid Sci.,1953, 8, 428. 27 Tartar and Wright, J. Amer. Chem. SOC., 1939, 61,539. 28 Wright, Abbott, Sivertz and Tartar, J. Amer. Chem. Soc., 1939, 61, 549.
1
2 3