`
`Sangster Research Laboratories, Suite M-3, 12 70 Sherbrooke St. West, Montreal, Quebec, Canada H3G 1 H7
`
`James Sangster
`
`Received July 21, 1988; revised manuscript received January 30, 1989
`
`Octanol-water partition coefficients (log P) for 611 simple organic compounds repre(cid:173)
`senting all principal classes have been retrieved from the literature. Available experimen(cid:173)
`tal details of measurement are documented from original articles. Pertinent thermody(cid:173)
`namic relations are presented, with a discussion of direct and indirect methods of
`measurement. Reported log P data for each compound have been evaluated according to
`stated criteria, and recommended values (with uncertainty) are given.
`
`Key words: octanol-water partition coefficient; organic compounds; hydrophobicity; hydrophilicity
`
`Contents
`
`1111
`1111
`1112
`1112
`1112
`1112
`1113
`1113
`1113
`1113
`1113
`1116
`1116
`1116
`1116
`
`List of tables ............................................................. .
`List of symbols and abbreviations ............................ .
`1. Introduction ........................................................ .
`1.1. General ...................................................... ..
`1.1.a. Definition ...................................... ..
`l.l.b. Scope of this evaluation .................. .
`l.l.c. Need for critical evaluation ............ .
`1.2. Thermodynamics ....................................... .
`1.2.a. General equilibrium relations ....... ..
`1.2.b. Temperature dependence ............... .
`1.2.c. Specific thermodynamic relations .. .
`2. Methods of measurement .................................... .
`2.1. Direct or "Experimental" methods ........... .
`2.1.a. Shake-flask ..................................... .
`2.l.b. Generator column .......................... .
`2.2. Indirect
`or
`"Calculation/correlation"
`methods ...................................................... .
`2.2.a. Methods widely used and/ or having
`some theoretical justification .......... .
`2.2.b. Other correlations ......................... ..
`3. Comparison of methods of measurement or esti-
`1117
`mation ................................................................. .
`1117
`4. Retrieval of data...................................................
`5. Criteriaforevaluation.......................................... 1118
`6. Presentation of data in Tables 3-20 ..................... 1119
`7. References............................................................ 1227
`
`1117
`
`1117
`1117
`
`2. General characteristics of some measurement
`and estimation methods for P ........................... .
`3. Partition coefficients of alkanes ......................... .
`4. Partition coefficients of alkenes and alkynes .... ..
`5. Partition coefficients of aromatics .................... ..
`6. Partition coefficients of cycloalkanes and cy-
`cloalkenes .......................................................... .
`7. Partition coefficients of mixed-type hydrocar-
`bons ................................................................... .
`8. Partition coefficients of ethers .......................... ..
`9. Partition coefficients of alcohols ....................... .
`10. Partition coefficients of aldehydes .................... ..
`11. Partition coefficients of ketones ........................ .
`12. Partition coefficients of acids ............................ .
`13. Partition coefficients of esters ............................ .
`14. Partition coefficients ofhalogcnatcd compounds
`15. Partition coefficients of amines ......................... .
`16. Partition coefficients of nitriles ......................... .
`17. Partition coefficients of nitro compounds ......... .
`18. Partition coefficients of ami des ......................... .
`19. Partition coefficients of sulphur compounds ..... .
`20. Partition coefficients of other compounds ........ ..
`
`1118
`1121
`1123
`1125
`
`1141
`
`1143
`1144
`1150
`1163
`1165
`1170
`1177
`1185
`1192
`1216
`1219
`1221
`1225
`1226
`
`List of Symbols and Abbreviations
`=activity
`= ambient temperature
`= octanol-saturated water phase
`= absorption spectrophotometry
`= organic base
`= concentration
`=heat capacity at constant pressure
`= chemical reaction
`= direct method
`= fluorescence
`= Gibbs energy
`
`a
`AMB
`AQ
`AS
`B
`c
`cp
`CR
`D
`FL
`G
`
`List of Tables
`1. Temperature dependence of Log P of some com(cid:173)
`pounds at room temperature.............................. 1113
`
`@) 1989 by the U.S .. Secretary of Commerce on behalf of the United States.
`This copyright is assigned to the American Institute of Physics and the
`American Chemical Society.
`Reprints available from ACS; see Reprints Lists at back of issue
`
`0047-2689/89/031111-120/$12.00
`
`1111
`
`J. Phys. Chern. Ref. Data, Vol.18, No.3, 1989
`
`CFAD v. Anacor, IPR2015-01776
`ANACOR EX. 2024 - 1/117
`
`
`
`1112
`
`JAMES SANGSTER
`
`= generator column
`GC
`= gas-liquid chromatography
`GLC
`=chromatographic peak height, Eq. (28)
`h
`H
`=enthalpy
`HA
`= organic acid
`HM
`= Henry's law constant
`(RP)-HPLC =(reverse-phase) high pressure liquid chromatography
`= indirect method
`I
`K
`= Kjeldahl method
`= acid ionization constant
`Ka
`1
`=liquid
`m,n
`=correlation constants, Eq. ( 10)
`NS
`= neutral salt solution ·
`ORG
`= water-saturated octanol phase
`p
`=pressure
`p
`= partition coefficient
`= apparent partition coefficient
`= radiochemical method
`=solid
`~entropy
`= shake-flask method
`=temperature (kelvin)
`= temperature of log P measurement (°C)
`=titration
`=volume (general)
`= molar volume (L mol- 1
`= octanol-saturated water solvent
`
`Papp
`RC
`s
`s
`SF
`T
`Temp.
`TN
`v
`v
`w
`
`)
`
`X
`X
`?
`
`= mole fraction
`=solute
`= doubtful Log P value; Code uncertain
`
`Greek
`6.
`r
`"'
`'
`
`tp
`
`Superscripts
`aq
`0
`oct
`org
`sat
`w
`*
`Sublic:ripllj
`app
`fus
`I
`
`org
`tr
`X
`
`= difference in thermodynamic function
`= volume-fraction activity coefficient
`= chemical potential
`= volume fraction
`= solubility (mol L- 1
`
`)
`
`= octanol-saturated water phase
`= standard state
`= pure octanol phase
`= water-saturated octanol phase
`= phase saturated with solute
`= pure water phase
`= Hansch & Leo "selected" Log P value
`
`=apparent (partition coefficient)
`=fusion
`=final
`-initial
`= water-saturated octanol phase
`= transfer process
`=solute
`
`1. Introduction
`
`1.1. General
`
`1.1.a. Definition
`A pure substance may distribute itselfbetween two par(cid:173)
`tially miscible solvents in intimate contact, and the equilibri(cid:173)
`um ratio of solute concentrations in the two phases has come
`to be known as the distribution coefficient or partition coeffi(cid:173)
`cient.1 In preparative organic chemistry, the use of solvents
`of greatly differing polarity (e.g., hydrocarbon and water)
`facilitates the extraction and purification of desired prod(cid:173)
`ucts. In addition, the biological activity of simple organic
`compounds wa.q early found to correlate with their oil-water
`partition coefficients. 2 It became apparent that, for biologi(cid:173)
`cal purposes a partition coefficient based on long-chain ester
`or alcohol solvents was more appropriate. After some delib(cid:173)
`eration, 1-octanol was chosen as the most useful lipophile
`solvent in these applications. Most correlation work has
`been done using the octanol-water pair, and this is the reason
`for its wide use and the existence of a great quantity of data
`on the subject.
`The octanol-water partition coefficient of a substance X
`at a given temperature is, by general consent, 1 represented by
`P and defined by (for reasons explained later, the super(cid:173)
`scripts "org" and '.'aq" are used to denote mutually satu(cid:173)
`rated phases, and "oct" and "w" for the pure solvents.)
`
`(1)
`i.e., the ratio of concentrations (mole/volume) at equilibri(cid:173)
`um; it is therefore unitless. In the interest of standardization
`and precision in interpretation, the partition coefficient is
`defined for the same species on both phases. 1 This is impor(cid:173)
`tant in considering Pof ionizable compounds such as organic
`
`J. Phys. Chem. Ref. Data, Vol. 18, No.3, 1989
`
`acids, amines, and quaternary ammonium salts, which may
`also form dimers or ion-pairs. This is discussed further in
`this Introduction.
`In addition, the solvents represented in Eq. (1) are those
`mutually saturated with each other at the temperature of
`measurement. This is a natural consequence of the classical
`"shake-flask" or extraction method used in experimental
`measurement of P, and is to be taken into account for accura(cid:173)
`cy in measurement and thermodynamic interpretation.
`Further, Pis preferably defined as the quantity which is
`independent of concentration, i.e., that value for which the
`solute obeys Henry's law in both solvents simultaneously. In
`practice, this means a P determined at high dilution, or ex(cid:173)
`trapolated to zero concentration. Since P as measured can
`range over many orders of magnitude (10- 2 to 106 ), it is
`usually expressed as its decadic logarithm, log P.
`
`1.1.b. Scope of this evaluation
`This work proposes to have retrieved and evaluated
`most of the significant published experimentally determined
`values oflog P of simple organic molecules. The word "sim(cid:173)
`ple" is taken here to indicate molecules containing no or only
`one polar functional group, i.e., a group having N,O,S
`and/ or halogen atom. A few well-known exceptions ( chlor(cid:173)
`oform, CC14 ) have been included. This limits the number of
`substances involved and, as far as possible, avoids complica(cid:173)
`tions of interpretation due to the presence of neighboring
`polar groups. The compounds are those which are liquids or
`solids at ordinary temperatures and pressures, and no arbi(cid:173)
`trary upper carbon number cutoff limit has been imposed.
`Elements, inorganic, metal-org~nic and unstable species
`have been excluded, as well as quaternary ammonium and
`similar salts.
`
`ANACOR EX. 2024 - 2/117
`
`
`
`OCTANOL-WATER PARTITION COEFFICIENTS OF SIMPLE ORGANIC COMPOUNDS
`
`1113
`
`1.1.c. Need for critical evaluation
`The partition coefficient, as properly defined, is a defi(cid:173)
`,,,,,. equilibrium physico-chemical property of a pure sub(cid:173)
`·.tance under specified conditions. It provides a useful quan(cid:173)
`lllative parameter
`for
`representing
`the
`lipophilic/
`h\drophilic nature of the substance. It is a function of the
`c ; ihhs energy of transfer from water to octanol and hence
`,It-scribes the thermodynamic tendency for the compound to
`partition preferentially in different media. It is not surpris(cid:173)
`•••g, therefore, that it has been widely used in many areas
`-.tu:h as:
`-design of drugs and pharmaceuticals, 4
`-prediction and correlation ofbioconcentration5
`and soil and sediment sorption of organic pollutants,
`-research on medicinal chemicals,
`-modelling of environmental fate of organic chemicals, 6
`-toxicology of substances.
`For many substances, log P has been measured by dif(cid:173)
`ferent laboratories and by difFerent methods. The reported
`log Pvalues of a single substance can sometimes vary a great
`deal; for example, those ofp,p' -DDT cover a range of a factor
`<lf one hundred. 2 Large uncertainties in log Pare undesirable
`in general. The accuracy of the simulation, by calculation, of
`the environmental fate of an organic chemical may become
`quite sensitive to uncertain des of input parameters (e.g.,
`Mirex in Lake Ontario'). The successful development of ad(cid:173)
`ditive-constitutive calculational schemes2
`8 for log P, based
`•
`on molecular structure, requires a database of assessed accu(cid:173)
`racy. Finally, it is difficult, if not impossible, for the unini(cid:173)
`tiated user oflog P data to distinguish accurate and inaccur(cid:173)
`ate data by simple inspection.
`
`1.2 Thermodynamics
`The thermodynamic relationships between log P and
`other quantities will be examined in some detail in this sec(cid:173)
`tion. Many experimental data on log P related thermody(cid:173)
`namic quantities have appeared recently, some of high quali(cid:173)
`ty. Since these were not discussed in any detai,l in former
`compilations and reviews, I.2 the following exposition is
`meant to summarize the important thermodynamic rela(cid:173)
`tions in a concise and rigorous manner.
`
`1.2.a. General equilibrium relations
`The present thermodynamic analysis is a slightly edited
`restatement of the one currently being used to describe the
`two-phase system represented by the octanol-water partition
`38
`10
`34
`coefficient.9
`58 Like the current practice, it uses vol(cid:173)
`•
`•
`•
`•
`ume fractions as composition variable and volume fraction
`activity coefficients. This convention, when used in conjunc(cid:173)
`tion with the (volume-based) partition coefficient, simpli(cid:173)
`fies the thermodynamic argument. An equivalent though in(cid:173)
`is given
`complete, analysis using mole
`fractions
`39 If a liquid substance X is distributed between
`elsewhere. 33
`•
`organic and aqueous phases at equilibrium, we can write for
`eachphase9
`J.tx = J.t~ + RTln ax
`= p.CJc + RTln( YxfPx ),
`
`(2)
`
`(3)
`
`where p. x is the chemical potential of X in solution, p,CJc is the
`chemical potential of pure liquid X, ax is its activity in solu(cid:173)
`tion, rx is the volume-fraction activity coefficient and fPx is
`the volume fraction of X in the solution. From the defining
`Eqs. (2) and ( 3), the activity coefficient is normalized by r x
`-1 astpx-1.
`By definition, in each phase we have
`[X] Yx = fPx,
`(4)
`where Yx is the (partial) molar volume of X in solution.
`(For dilute solutions of liquid nonelectrolytes in water or
`octanol, partial molar volume can be replaced by pure liquid
`molar volume without appreciable error.) At equilibrium,
`p.~rs = p,":. Combining this equality with Eqs. ( 1 ), ( 3), and
`(4) yields9
`log P = log(~/71-ra),
`(5)
`i.e., Pis equivalent to the ratio of the Henrlan activity coeffi(cid:173)
`cients of the solute in the phases. Equation ( 5) has been
`derived using the assumption that ~ ~··
`
`1.2.b. Temperature dependence
`The variation of log P with temperature1 is small~ ap(cid:173)
`proximately ± 0.01 K- 1
`• Table 1 presents experimental
`data of d(log P)/dT for some specific compounds.
`The thermodynamic transfer functions are closely re(cid:173)
`lated to log P:
`atrG= -RTlnP
`d(atrG>Idt = - atrs
`
`(6)
`<7>
`(8)
`As a consequence of the definition of P, these transfer quan(cid:173)
`tities are independent of concentration and refer to the dif(cid:173)
`ference: (function for solute in water-saturated octanol)(cid:173)
`(function for solute in octanol-saturated water).
`The temperature dependence of Atr G and hence of
`log Pcan be represented by AtrHandAtrS throughEqs. (6),
`(7), and ( 8). A van't Hoff plot ofln Pmay be used to obtain
`
`TABLE 1. Temperature dependence of Log P of some compounds at room
`temperature
`
`Substances
`
`Temperature 1000 d(Log P)/dT
`range, •c
`K-•
`
`Ref.
`
`n-propylbenzene
`chlorinated benzenes
`phenol
`phenol
`p-cresol
`phenol
`m-alkoxyphenols
`resorcinol
`substituted phenols
`substituted phenols
`cblorophenols
`hydroxybenzoic acids
`methyl nicotinate
`ephedrine
`methamphetamine
`alkyl amidylpyridines
`methyl acetanilides
`
`10-35
`13-33
`10-60
`20-50
`15-35
`1~35
`15-35
`15-35
`10-60
`20-50
`
`5-25
`1~
`15-40
`20-40
`
`=0
`-29to -5
`-3.4
`-4.9
`-7.5
`-16
`-5to-3
`-8.8
`-8to -1
`-3to + 10
`-8.6 (mean)
`-14 (mean)
`7.4
`8.1
`0.4
`2to6
`=O(mean)
`
`10,11
`12
`16
`17
`18
`18
`19
`19
`16
`17
`20
`20
`14
`15
`15
`13
`20
`
`J. Phys. Chem. Ref. Data, Vol. 18, No.3, 1989
`
`ANACOR EX. 2024 - 3/117
`
`
`
`1114
`
`20 However, Pis a Gibbs
`enthalpy and entropy of transfer. 17
`-
`energy function-as ate solubility and vapor-liquid equilib(cid:173)
`rium-and these functions are usually found to be relatively
`insensitive to temperature. The enthalpy of transfer may be
`more precisely determined21 either by direct experimental
`23 (in which
`measurement in an isoperibol flow calorimeter2
`•
`two immiscible phases are brought into direct contact) or
`indirectly from the calorimetric limiting enthalpies of solu(cid:173)
`tion in the two solvents separately.
`As will be discussed in greater detail in the next section,
`both ll.tr G and ll.trH may be determined from measurements
`on solutions based on the two solvents separately. This is a
`possible route, provided it is realized that solute thermody(cid:173)
`namic functions in pure water or pure octanol may be signifi(cid:173)
`cantly different from those in mutually saturated solvents.
`
`1.2.c. Specific thermodynamic relations
`
`Mutually saturated solvents.
`4 the equilibrium
`From liquid-liquid equilibrium data2
`•
`mole fractions of octanol in the two-phase system water/n(cid:173)
`octanol at 25 oc are 7.03X to-sand 0.793. Saturated oc(cid:173)
`tanol thus contains an appreciable amount of water; the mo(cid:173)
`lar volume: uf wc:t octanof5 is 126.6 cm3
`, and the water
`content is equivalent to 1.64 molL -t. The two phases in a
`shake-flask determination of Pare ternary. The question
`whether or not the presence of the other solvent in a phase
`significantly alters the thermodynamic properties of the sol(cid:173)
`ute becomes important in considering recently elaborated
`"activity coefficient" methods of determining the partition
`coefficient.
`For example, Berti eta/. 22 compared the transfer Gibbs
`energies from shake-flask log P values of some common so(cid:173)
`lutes to those found from the directly measured Henrian
`activity coefficients of the same solutes in pure octanol and
`pure water. The differences in ll.trG, 1 to 2 kJ mol- 1
`, is
`equivalent to differences of as much as ± 0.4 in log P, being
`negative or positive or zero, depending on the solute. Calori(cid:173)
`metrically determined enthalpies of transfer of m-alkoxy
`phenols26 in neat and mutually saturated solvents differed by
`up to 1.6 kJ mol 1
`; the same effect is found in the corre(cid:173)
`sponding enthalpies of solution22
`26 from which the transfer
`'
`enthalpies are derived. Again, the magnitude of the effect
`depends on the solute; the transfer enthalpies of n-alkanols27
`for example, are much less sensitive in this respect.
`Platford28
`29 used the isopiestic method to measure the
`•
`limiting activity coefficients of CC14 and benzene in neat and
`mutually saturated solvents, and found no detectable differ(cid:173)
`ence in the results. Henrian activity coefficients in octanol
`for 22 monofunctional compounds were measured by gas
`chromatography58
`; within experimental error the relation
`
`(9)
`
`was valid.
`Relationship with aqueous solubility.
`Thermodynamic considerations have also elucidated
`the relation between log P and aqueous solubility, ;, early
`proposed by Hansch et al. 30 Since both log P and ; may be
`regarded as Gibbs energy transfer functions, an equation of
`the type
`
`J. Phys. Chem. Ref. Data, Vol. 18, No.3, 1989
`
`JAMES SANGSTER
`log P = m log; + n,
`(10)
`might be expected to be valid, where m and n are correlation
`coefficients. A relation like Eq. ( 10), if true, would greatly
`reduce the experimental effort necessary to obtain P. The
`search for refinement and rationalization of Eq. ( 10) has
`been lively, giving rise to at least one polemic exchange in the
`literature.31
`32
`'
`A relation of the form of Eq. ( 10) can be derived from
`thermodynamic first principles. 33
`34 The case for a liquid sol(cid:173)
`•
`ute will be given first, as a solid solute introduces a complica(cid:173)
`tion into the argument. For a liquid solute distributed at
`equilibrium between organic and aqueous solvents (The sol(cid:173)
`vents are assumed to be mutually saturated, in order to keep
`the analysis as close as possible to the conditions in a real
`shake-flask situation.) Eq. ( 3) can be applied to both phases
`to give
`( Yxll'x )org = <rxll'x )aq
`which, by Eq. (4) becomes
`( Yx [X ])org = ( Yx [X ])aq.
`(12)
`Consider now, as a separate system, the solute in saturation
`equilibrium in the aqueous solvent. On the assumption that
`the equilibrium free solute contains no solvent, Eq. ( 3) un(cid:173)
`der these conditions becomes
`
`(11)
`
`or
`
`(13)
`
`(14)
`
`(YxfPx)sat = 1.
`Combined with Eq. ( 4), this becomes
`<rxcxVx)sat = 1,
`(15)
`where Cx has units inverse to that of Vx. Recalling Eqs. ( 1)
`and (5),
`p = [X ]org/[X ]aq = y:;y;s.
`
`(16)
`
`(17)
`
`Introducing Eq. (15),
`P= 1/(r':8b~Vx)
`logP= -log~x -log(rx Vx)org,
`(18)
`Eq. (18) has the same form ofEq. (10). Clearly, data for
`liquid solutes would all fall on a common linear plot oflog P
`vs log ~' with the following provisos:
`ASSUMPTION 1 :tx8, the Henrian activity coefficient
`for liquid solutes in organic solvent,
`is the same for all solutes.
`ASSUMPTION 2:;x re(ers to water saturated with oc(cid:173)
`tanol.
`ASSUMPTION 3:the solute obeys Henry's law for con(cid:173)
`centrations up to saturation in the
`aqueous solvent.
`ASSUMPTION 4:the free liquid solute, as an equilibri(cid:173)
`um phase, contains no solvent (i.e.,
`its activity is unity).
`For solutes which are solid at temperature of measure(cid:173)
`ment, fundamental equations such as Eq. (2) are valid. In
`this case, however, the reference state for Jl~ cannot be the
`pure solid, since it is desirable to keep the same Raoult's law
`convention for the activity coefficients. The approach is then
`
`ANACOR EX. 2024 - 4/117
`
`
`
`OCTANOL-WATER PARTITION COEFFICIENTS OF SIMPLE ORGANIC COMPOUNDS
`
`1115
`
`through solid-liquid equilibrium. For a component in equi(cid:173)
`librium between solid and liquid phases,
`(19)
`J.lx (s) = J.lx (1 ),
`p~(s) +RTlnax(s) =p~(l) +RTlnax(l), (20)
`p~ ( 1) - p~ (s) = RTln ax (s)lax (1 ).
`(21)
`The quantity on the left-hand side of Eq. (21) is the Gibbs
`energy of fusion of the solute at temperature T, arus ~- If the
`equilibrium solid phase is pure solid, then
`
`'(22)
`
`For temperatures below the normal melting point of the sol(cid:173)
`ute, afus(t>T refers tO the prOCeSS (solid-+SUpercooled liq(cid:173)
`uid). 34·35 Since the solution is saturated,
`afuGG~ = - RTln(rxfPx)sat.
`(23)
`The thermodynamic argument represented by Eqs. ( 12 )(cid:173)
`(18) can be repeated again, with Eq. (23). The result is
`log p_... - arusG~/2.303RT-log~x
`
`-log(Yx Vx )org,
`(24)
`Eq. ( 24) is the same as Eq. ( 18), with the addition of the
`Gibbs energy of fusion term. The validity of Eq. (24) is of
`course subject to the same. assumptions attached to Eq. ( 18).
`For correlation purposes, the Gibbs energy of fusion to
`the supercooled liquid state may be expressed in terms of the
`usual fusion quantities:
`
`- arusG~/RT
`= (afusH0/2.303R)( T- Trus )/TTrus
`+ (afusC~)(ln(T/Trus) + (Trus- T)/T).
`
`(25)
`
`In Eq. (25), the enthalpy and heat capacity quantities refer
`to corresponding changes at the normal melting point
`( Trus ). [The heat capacities of solid and liquid solute have
`been assumed to be independent of temperature; They are
`not so in general, and Eq. ( 25) could be modified to take this
`into account.] Eqs. (24) and (25) together are identical to
`Eq. ( 14) of Miller et a/. 34 Eq. ( 25) may be simplified by
`putting arus C~ = 0 and36 arusH 0 = Trus arusS0 = 56.5Trus
`
`J mol- 1• The result is an expression equivalent to Eq. ( 26)
`of Mackay eta/. 33
`Apart from these simplifications, the four assumptions
`quoted above are of varying importance. The Henrian activ(cid:173)
`ity coefficient in octanol, y;:t, has been estimated by indi(cid:173)
`rect33 and direct37·38 methods. It is clear, particularly for hy(cid:173)
`drophobic compounds,
`that
`this activity coefficient
`increases with molecular weight (or molar volume). The
`water solubility bx is usually taken as that in pure water. The
`solubilities of organic compounds in octanol-saturated wa(cid:173)
`•38·39 The difference increases
`ter are measurably different. 26
`with molecular weight. It is difficult at present (if not impos(cid:173)
`sible) to define precisely the individual errors introduced
`into Eqs. ( 18) and ( 24) by these assumptions and simplifi(cid:173)
`cations. 33'38'39
`Ratio of solubilities.
`It has been stated that P is equivalent to the ratio of
`solute solubilities in the two solvents, 40 or is well approxi(cid:173)
`mated by this ratio. 37 The thermodynamic justification for
`
`(26>
`
`liquid and solid solutes is derived from Eqs. ( 1 ) , ( 15), and
`(22) for the case of two phases saturated with solute. The
`result is
`P= ;o;g;;a:~;o;t;;x
`As before, the important qualifying assumptions apply here
`also. There may be some fortuitous cancellation of effects.
`Y alkowsky eta/. 37 have tested Eq. ( 26) using solubilities in
`neat solvents for 36 solid compounds and found reasonable,
`though not exact, correlation.
`Henry's law.
`The Henry's law constant, like log P, is a limiting Gibbs
`energy quantity and its usefulness overlaps that of log P.41
`Henry's constant for a solute on the mole fraction scale, H M•
`may be defined as41 (where x is the mole fraction of solute)
`Iim(p!x) = HM,
`x-0
`where p is the partial pressure of solute above the solution.
`Henrian behavior of solutes has been exploited" 2 in the
`"head-space gas chromatographic method" for measuring
`P. The principal feature of this method43 is the sampling and
`quantitative analysis, by gas chromatography, of the vapor
`mixture above a liquid solution. In a measurement of P, the
`vapor above an unsaturated aqueous solution (volume v1 ) of
`the solute is sampled and the gas chromatographic peak
`height ( h 1 ) of the solute is obtained. A volume v2 of octanol
`is added, and after equilibration the vapor is sampled and
`analyzed as before (h2 ). The partition coefficient is then
`P= v1(h 11h2 - 1)/v2•
`(28)
`Equation ( 28) assumes that Henry's law is obeyed by the
`solute in the aqueous phase; that H M is independent of the
`presence of co-solvent; that there is a strict mass balance for
`the solute; and the vapor behaves ideally.
`
`(27)
`
`1.3. The case of ionizable solutes
`In the present work, two types of organic compounds
`may ionize in aqueous solution, viz., acids (HA) and amines
`(B):
`
`HA~H++A-,
`
`(29)
`(30)
`BH+~B+H+,
`where A- is the acid anion. [Water should appear on both
`sides of Eqs. (29) and (30), but since in dilute solution its
`activity is practically unity and does not change, it may be
`omitted from the thermodynamic analysis.] The thermody(cid:173)
`namic dissociation constant isdefined as
`Ka = a(H+)a(A -)/a(HA)
`after Eq. (29), and
`Ka = a(H+ )a(B)/a(BH+)
`after Eq. ( 30). Since Pis defined only for the same ( undisso(cid:173)
`ciated) species in both phases, the apparent partition coeffi(cid:173)
`cient Papp (sometimes called distribution coefficient) mea(cid:173)
`sured in the presence of appreciable ionization according to
`Eqs. (29) and (30) will differ from P. It can be shown44 that
`Papp and P for an acid are related by
`P=Papp[1 + 10<pH-pKa)]
`
`(31)
`
`(32)
`
`(33)
`
`J. Phys. Chern. Ref. Data, Vol.18, No.3, 1989
`
`ANACOR EX. 2024 - 5/117
`
`
`
`1116
`
`JAMES SANGSTER
`
`and for an amine by
`p = Papp [ 1 + 10<PKa-PH>].
`(34)
`For weak acids (pKa > 7) or weak bases (pKa <7) in water,
`there is negligible ionization (Papp = P). For other com(cid:173)
`pounds under some experimental conditions, there will be
`appreciable ionization. [For example, phenol is a very weak
`acid (pKa = 9.9) and is not appreciably ionized in neutral
`solution. For pentachlorophenol, however, the alcoholic hy(cid:173)
`drogen atom is rendered more labile (pKa = 4.8). The mea(cid:173)
`sured values of P over the pH range 1.2-13.5 vary by more
`than three orders ofmagnitude.45
`] In these cases, Papp may
`be corrected according to Eqs. ( 33) and ( 34), or a buffer of
`suitable pH may be used as the aqueous phase in order to
`suppress ionization. Some compounds (e.g., acids) may
`form dimers or other associated species in the organic phase.
`'l'his source of error may be avoided by the use of sufficiently
`dilute solutions, which is usual practice.
`
`2. Methods of Measurement
`Values of P reported in the literature have been deter(cid:173)
`mined by many methods. Mention is made in this section of a
`t·ather large number of these, but only the more reliable or
`soundly based will be discussed in some detail. For the pur(cid:173)
`poses of evaluation, the methods have been classified into
`two groups (direct and indirect). This division is made for
`convenience of discussion, and does not necessarily imply
`fundamental or far-reaching theoretical differences.
`
`2.1. Direct or "experimental" methods
`
`2.1.a. Shake-flask method
`This classic extraction procedure is widely used and,
`with due attention to experimental conditions, manipulation
`and range of applicability produces reliable results. It has
`been described briefly46 and necessary precautions have been
`3 The case of weakly ionized so(cid:173)
`discussed in some detail. 1
`•
`lutes has been given special attention. 47 In essence, the meth(cid:173)
`od is simple. A small amount of the solute is dissolved in
`either aqueous or organic phase, equilibrium partition is ob(cid:173)
`tained by agitation, the phases are separated and one or both
`phases are analyzed for solute. Apart from requirements al(cid:173)
`ready mentioned in Sec. 1 of this review, a few of the othta·
`important precautions may be mentioned here.
`Purity of chemicals.
`Depending on the analytical method used. the presence
`of partitionable impurities in solvents or sample may lead to
`erroneous measured solute concentrations.
`Mutually saturated solvents.
`It is often the practice in this method to prepare a solu(cid:173)
`tion of known initial solute concentration c; in one solvent,
`and equilibrate a definite volume v1 of this with a definite
`volume v2 of the other solvent. The final solute concentra(cid:173)
`tion cf is measured in the first solvent and the partition
`( c; - c f) v 1 I
`coefficient
`is
`given
`by
`the
`ratio
`c fv2 or its inverse. Since the densities of neat octanol and
`water-saturated octanol are measurably different, the use of
`presaturated solvents avoids any error through changes in
`solvent volume upon equilibration.
`
`J. Phys. Chem. Ref. Data, Vol.18, No.3, 1989
`
`Mixing and separation.
`Several different methods of agitation are used to
`the solvents into intimate contact. In general, prolong
`violent shaking is not necessary and tends to cause emt
`formation. The phases separate under normal gravit)
`they are usually centrifuged to accelerate t:Qe separati
`the smallest droplets.
`Sampling.
`Ideally, both phases are analyzed. This is not aJ
`done, for reasons of convenience and time. If only one 1
`is analyzed, a mass balance between the phases is ass1
`and it must be established that no solute has been lc
`adsorption on glass, rubber stopper or other material U!
`manipulation. Since many solutes preferentially par
`into the organic phase, care must be taken to ensun
`sampling devices introduced into the phases do not ina
`tently carry over one phase into the other.
`Analysis. , ·
`Absorption spectrophotometry and gas-liquid
`matography are often used. There may be intermedia·
`traction and/or concentration steps, or one involv:
`chemical reaction. The usual precautions in quanti1
`analysis apply here.
`Volatile solutes.
`If the solute has an appreciable vapor pressure it m
`necessary to consider the amount of vapor space in the
`libration vessel and details of manipulation during sam]
`analysis, etc.
`Since the shake-flask method is at times tediow
`time-consuming, a number of automated or simplifie'
`sions have been used. Most are closed-loop flow device
`-counter-current distribution (engineering de~
`Centrifugal partition chromatography49 may b
`scribed as the addition of a high gravity field t
`counter-current method.
`·
`-The AKUFVE system50 uses continuous centri
`tion to separate the phases, while the rapid mix,
`probe51
`52 uses different membrane filters.
`•
`-the segmented flow device53 is a miniaturized cot
`current flow system.
`-a three-phase partition system54 is a kind of cor
`current method using two aqueous phases:
`AQ 1/0RG/AQ2,
`where AQ 1 and AQ2 are different aqueous bt
`This device is used principally for investigating 1
`ics of partition.
`-the exponential concentration change method55
`be regarded as a multi-step extraction procedure
`
`2.1.b. Generator column method
`The ordinary liquid chromatographic column c
`has been adapted for the measurement of partition c
`11 The solid support is usually silanized dial
`cients. 10
`•
`ceous silica. The column is loaded by pulling an unsatu
`solution of the solute of interest at a known concentrat
`water-saturated octanol through the column. The sol
`eluted with octanol-saturated water, and the effluent i:
`lyzed