Effects of Ionisation on Adsorption from SolutionBY HENRY M. RENDALL“ AND ALEC L. SMITHUnilever Research, Port Sunlight Laboratory, Port Sunlight,Wirral, Merseyside L62 4XNReceived in revised form, 6th August, 1976The adsorption of y-picoline (4-methyl pyridine) at the silica-water interface, as a function of pH,is reported. A general model for the adsorption is developed, which allows for the possibility thatthe ionised (HPic+) and the neutral (pic) forms of the molecule may both adsorb. For the limit ofzero coverage, explicit expressions are derived which give a simultaneous fit of adsorption andelectrokinetic measurements, and the model is extended to consider the complete adsorption isotherms.The magnitudes of the initial slopes of the isotherms, and the Occurrence and position of a maximumin the adsorption with pH, are explained. The two forms of the picoline molecule have specificadsorption potentials of - -4 kT, with the neutral molecule slightly the more strongly adsorbedby - 0.3 kT.The adsorption of the ionised and neutral components of an ionisable moleculehas been widely discussed, both in relation to studies of practical problems involvedin flotation,’ corrosion inhibition,2 and biological activity and from a theoretical~tandpoint.~ It is not normally possible to determine by direct measurement therelative adsorption densities of the two species in the equilibriumBH++H,O + B+H30+in the pH range where both are present in significant quantities.’’ Conway andhis co-workers have studied the separate adsorption of the neutral and the cationiccomponents of organic bases at the mercury-water interface under extreme conditionsof pH.Breuer has explained a number of properties of biological systems on theassumption that only the neutral forms of small organic molecules are adsorbed onproteins and similar hydrophobic surface^.^At the air-water interface, however, Betts and Pethica have suggested a surfacedissociation constant for long chain molecules equal to the value obtained in bulksolution, when allowance is made for the surface potential. This would imply thatboth components have a similar intrinsic tendency to adsorb. It is, therefore, ofinterest to obtain information about the effects of ionisation on the adsorption of asimple molecule at the solid-liquid interface.A study has been conducted of the effect of pH on the adsorption of y-picolineat the silica-water interface, in the presence of NaCl to control the ionic strength.The pK, of y-picoline is 6.02,6 and the isoelectric point (ip) of the silica used wasapproximately pH 3.The adsorption of the picolinium ion onto a neutral andincreasingly negatively charged surface may, therefore, be observed, and comparedwith the adsorption of the conjugate base.In the limit of extrapolation to zero coverage, we have sought the simplest model,based on standard double layer theory, and allowing for the possibility that HPicfand Pic might both be adsorbed, to describe the pH dependence of picoline adsorption.We have used the adsorption affinity so derived for the neutral sptcies to predict theconcentration dependence of its adsorption, and in this way have resolved the10102 EFFECTS OF IONISATION ON ADSORPTIONmeasured adsorption into two components throughout the concentration range.Afew measurements have been made of the adsorption of the methyl quaternaryderivative, MePicf, which is not involved in a reversible ionisation, for comparisonwith the adsorption behaviour inferred for the HPicf ion.EXPERIMENTALMATERIALSAerosil 0 silica was prepared by the method of Taylor, Hockey and Pethica.7 Theprocedure is claimed to produce agglomerates of non-porous primary particles which arespherical, fully hydroxylated (3 to 4.6 of hydroxyl groups) and of about 200 A diameter.This was checked by electron microscopy and nitrogen adsorption (B.E.T.area = 176420 m2 g-’), The aggregates used were mainly in the size range 125-180 pm. It has beenassumed that the results were not seriously influenced by any effect due specifically to theaggregated nature of the adsorbent. Adsorption studies have shown that the silica surfaceinvolved is accessible to polyvinylpyridines of quite large molecular weights.y-Picoline (BDH AnalaR grade) was distilled once and stored under refrigeration toreduce decomposition. A sample of N-methyl y-picolinium bromide was prepared byquaternisation of 7-picoline under reflux in methanol, and was recrystallised from acetone.The water used for all solutions was prepared by the double distillation of deionisedwater under nitrogen in an all-glass still with alkaline permanganate in the first stage.All other chemicals were of BDH AnalaR grade.ADSORPTION MEASUREMENTThe solutions contained mol dm-3 of NaCI, to maintain the ionic strength, andsufficient hydrochloric acid to obtain the desired pH.For the highest pH studied (pH 8.7)an ammonia/ammonium chloride buffer, or sodium hydroxide, was added. The resultswere in agreement, but in the latter case were rather less reproducible, possibly as a resultof some dissolution of the silica. Measurements at higher pH were avoided because ofthese experimental difficulties, and because an underlying assumption of our treatment, thatthe adsorption affinity of the neutral molecule for the surface is independent of the doublelayer potential, becomes doubtful as the surface potential becomes large.Buffers were employed to control the pH in the adsorption of the quaternised molecule.As far as possible sodium salts were used at lom2 mol dm-3 to keep the conditions comparablewith the picoline adsorption.Pyrex test tubes, with polythene stoppers, and containing known amounts of adsorbentand solution (normally 1 g and 10 cm3), were shaken in a thermostat at 298 K for at least3 h, which was more than sufficient to obtain equilibrium.The solutions were centrifugedto remove suspended silica. The pH of the equilibrium solution was determined tok0.05 pH unit. Concentrations were determined by U.V. absorption. In general, twoseparate dilutions were made for each solution, and the determinations agreed to k0.5 %.The concentration change was usually in the range 10-30 %, so that the error in the amountadsorbed was 2-5 %.ELECTROPHORESISSome measurements of the mobilities of the silica particles as a function of pH atmol dm-3 ionic strength were made with a Rank microelectrophoresis apparatus.Charge reversal was demonstrated with both the HPic+ and the MePic+ ions, and theisoelectric point of the silica in the presence of MePicBr was at pH 5.0 (fig.1).The ions studied here do not form micelles in solution; published experimental resultsshow only repulsive interactions between similar ions, adsorbed at the mercury-solutioninterface, up to high surface coverage. The observed charge reversal may, therefore, beattributed to a specific interaction with the silica, not to the cchemimicelle ” formationpostulated for long chain ions.gmoH.M. RENDALL AND A . L. SMITH 103THEORETICALDOUBLE LAYER MODELThe simplest useful model for the double layer, due to Grahame,lO is illustratedschematically in fig. 2. The application of this model to describe the surface electricalproperties of oxides, including silica, has been discussed by one of us.ll The doublelayer is represented by three planes, the surface proper (assumed to contain the centresof ionised groups), and boundary of the diffuse layer (defined by the distance ofclosest approach to the surface of hydrated counterions) and a plane containing thecentres of specifically adsorbed ions.2 4 6 8 10PHRG.1.-pH dependence of measured potentials and charges of silica in mol dm-3 electrolyte.(a) and (b), (-potentials in NaCl and MePicBr, respectively; (c) and (d), Ud and - uo, respectively,in NaCl.The charge densities (cT), potentials ($), and integral capacities (a,* whoselocation with respect to those planes is shown in fig. 2, are related by the equations(using rationalised electrical units)a O + a p + a d = 0 (1)@o-Il/s = ao/& (2)$B-$d = (3)- ad = (2kT ICE fzeo) sinh(zeo$* J2kT) (4)KT1 = K-13-K-1 1 2 ( 5 )where 1c is the Debye-Huckel reciprocal distance parameter, z is the valence of thecounterion, eo, k, and T are the electronic charge, the Boltzmann constant and thetemperature , respectively .* Note that integral capacities K carry subscripts to distinguish them from the association constantK introduced in eqn (8)1 04 EFFECTS OF IONISATION ON ADSORPTIONFor silica at the experimental ionic strength, in the absence of picoline, i)d (takenequal to the electrokinetic potential), and hence ad, are known.A capacity Kd of20 pF cm-2 was found to give good agreement between titration and electrokineticdata on silica,12 and a potentiometric titration of the silica sample used here gavevalues of a. (fig. 1) virtually identical with those reported by Bolt.13 Hence, inthe absence of picoline adsorption, only one additional parameter is required todefine completely the double layer model.solid solutionpotentials yo yp v,FIG. 2.-Double layer model.pH O F MAXIMUM ADSORPTIONThe adsorption isotherms reported below show a maximum in the adsorption, atfixed solution concentration, as a function of pH. We develop here an approach tofind the conditions under which such a maximum will be observed, for the generalcase in which the ionised and neutral forms of the molecule may be adsorbedsimultaneously .In the limit of extremely low adsorption density, and neglecting ion self-atmospherepotentials we use the Langmuir-Stern isotherm to give an approximate measure ofa non-electrostatic contribution to adsorption,where n, is the amount adsorbed, N' is the density of adsorption sites, assumedidentical for the two species.x is the mole fraction of the species in solution, and<D is the specific adsorption potential of the species, with subscripts +and I? referringto the ionised and neutral molecules respectively.The mole fractions of the two species in solution may be calculated from the totalconcentration c (mol dm-3), the hydrogen ion activity aH and the pK by the equationswhere K is the association constant ( = K; ').ref.(6).CDrwhere A = N,c/55.51.[dn,/d(pH) = O] becomesA convenient tabulation of the fractions ionised as a function of pH is given inWe will write -e,ts/RT = $r (i.e., in units of - 25 mV at 298 K) and -(D/kT =(10)For a constant total concentration the condition for maximum adsorptionThe total adsorption of picoline at low coverage is given byns = A(1 +a&)-' [aHKexp($r) exp(Qr+) +exp(@rJIexp(#r+A@,,) = [I -p(l +a&)]-' (1 1H .M. RENDALL AND A . L. SMITH I05i.e., the amount, in units of kT, by which the positive species where ACDr =is more strongly adsorbed than the neutral species, andP = (e0P.303 kT) W$9/d(PH)l. (12)Froin eqn (1 1) a maximum will never be observed when y = I . In this case theincreased attraction of the surface for the ion, by eqn (6), always compensates forthe decrease in x+ as given by eqn (8). The rate of change of 4'/s with pH will, ofcourse, always be less than that given by the Nernst factor. It is possible for a Nernstrelation to apply to ~ o , not to $@, and in the case of silica even $,, shows considerabledeviations from Nernstian behaviour. In principle, therefore, an adsorptionmaximum under the defined conditions is to be expected for all systems.RESULTS AND DISCUSSIONEXPERIMENTAL ADSORPTION ISOTHERMSThe pH dependence of the adsorption of y-picoline on silica, as a function ofthe initial solution concentration, is shown in fig.3, where S is the ratio of the amountadsorbed (in mmolg-') to the amount remaining in solution (in moldm-3). Theadsorption passes through a maximum with increasing pH for all the concentrationsstudied.2 4 6 8PHFIG. 3.-Adsorption ratio S against pH for y-picoline on silica at the initial solution concentrationsThe pH dependence of the adsorption of the MePic+ ion, at a constant initialconcentration of 6 x mol dm-3, is shown in fig. 4. The total adsorption ofpicoline, obtained at equivalent equilibrium solution concentrations of picoliniuniions, is given for comparison in fig.4. At low pH, where the picoline is mainly inthe ionised form (e.g., at pH 2, HPic+/Pic = lo4), the picoline adsorption parallelsthat of the MePic+ ion, and the higher adsorption of MePic" ion suggests a strongerindicated (mmol dm-3). - - - , zero coverage extrapolation106 EFFECTS OF IONISATION ON ADSORPTIONspecific interaction that for HPic+ with silica. At higher pH the picoline adsorption,however, greatly exceeds that of the MePic+ ion. It is reasonable to conclude thatin this higher pH region the ionised and the neutral forms of picoline must bothmake a significant contribution to the measured adsorption.IPHpicoline adsorption ; - - - , calculated HPic+ adsorption.FIG. 4.-Adsorption at equivalent cation concentrations. 0, MePic+ ion adsorption ; 0, totalThe position of the adsorption maximum (fig.3) moves to higher pH as theconcentration of picoline is increased. We may discuss the significance of this shiftin qualitative terms only, since the condition for a maximum, eqn (11), was derivedfor the limit of very low coverage, and since the variation of p with p H and coverageis not well enough known. The term t,br+ACDr represents the total amount by whichthe affinity for the surface of the HPic+ ion exceeds that of the Pic molecule, undergiven conditions. Assuming a,, is independent of coverage, then, by eqn (ll), theaffinity of the cation for the surface must decrease with coverage. Such a decreaseis expected, because the specific adsorption of cations will, in general, make t,bp less25 5 0 7 5 100maximum adsorption/pmol g-'FIG. S.-pH of maximum adsorptionH.M. RENDALL AND A. L. SMITH 107negative. A rapid decrease in apparent adsorption affinity with coverage has beenreported for pyridinium ions at the mercury-solution interface.Throughout the experimental concentration range the pH of maximum adsorptionshowed a linear variation with the amount of picoline adsorbed (25-100 pmol g-')at the maximum (fig. 5). From the intercept, the adsorption maximum at zerocoverage would occur at approximately pH 6.1.LIMITING ADSORPTION RATIO AT ZERO COVERAGEIt is reasonable to assume that the potentials $B appropriate to the limit of zerocoverage are determined by the properties of silica measured in the absence of picoline.Hence if a value is assumed for the integral capacity K,, +b may be calculated, at agiven pH, from $d (fig.6), using eqn (3). The position of the adsorption maximumat zero coverage, as given by eqn (11) with Amr = 0, was found to lie at pH 5.9-6.0for values of K, within physically reasonable limits. This is similar to the pH ofmaximum adsorption suggested by empirical extrapolation of the experimentalresults. The difference is in a direction which would imply that lCDnl > I@+,[ by upto about 0.5 kT.- 1 5 0 -6 aPHFIG. 6.-Electrostatic potentials for silica in mol dm-3 NaCl. (a) #d from electrokineticmeasurements ; (6) $p from $d with K2 = 30 pF cm-2 ; (c) $o from #band uo, with Kl = 60 pFcm-2.0, #p from extrapolated adsorption with AD, = 0 (filled points) and AQr = -0.3 (open points).The magnitude of calculated limiting adsorptions, unlike the position of theadsorption maximum, is very sensitive to the value assumed for K,, and hence for$a.We now seek an extrapolation procedure to obtain values of S at zero coverage.From these we will deduce, for different values of har, the variation with pH of$fl which is required to explain the extrapolated adsorption results. We will thenimpose the condition that t,kB must be related to t+9d (fig. 6) by eqn (3), with a singlevalue of K2. The potentials $b calculated from the adsorption results will be virtuallyindependent of A@ at sufficiently low pH, and will become very sensitive to AQr athigh pH.The procedure outlined should, therefore, yield a unique solution.--- , #o calculated from a modified Nernst equation with 0, = 0.0001 and P.Z.C. at pH 3.2108 EFFECTS OF IONISATION ON ADSORPTIONFor low coverage by a single species, the values of In S, by eqn (6), and (7), givea direct indication of adsorption energy. Even for very low adsorption densitieswhich would correspond to the " initial slope " region of a Langmuir-type isotherm,the experimental values of 1nS at low pH decreased rapidly with coverage. Acomparable decrease with coverage in the apparent adsorption energies of similarcations at the mercury-water interface has been reported by Bockris l 5 and Conway.2We have used a linear extrapolation of In S against n,, to evaluate the limiting initialslopes (g) of the isotherms.In the pH range where both species contribute significantly to the measuredadsorption, S may be subdivided asS = aS++(l-a)S, (1 3)where 01 is the fraction ionised.Although, in this case, In S does not provide asimple measure of adsorption energy, the same extrapolation procedure was used toobtain the approximate values for the initial slopes of the isotherms, which are shownin fig. 3.By eqn (6) the limiting slope p+, for cation adsorption at the ip of the silica givesa measure of @+, and for zero coverage at other pH values we have(14) - S+ = g; exp - (eo$s/kT) ;similarly - S , = 5; -A@r.From the extrapolated limiting slopes we may calculate $b using eqn (13), (14)and (15).Consistency with the values obtained from $d using eqn (3) was obtained(fig. 6) with A@, = -0.3 (Lee, the neutral molecule more adsorbed than the cation)and with K2 = 30 pF cm-2. Smaller values of a, gave potentials I$s[ which increasedrapidly at high pH and could not be reproduced by the substitution of any value ofKz (exemplified in fig. 6 by the case A@, = 0), whereas larger values of On madedecrease at high pH.It was confirmed that, using the values of t,bB and hence of p given by eqn (3)for K2 = 30 pF cm-2 (fig. 6), and assuming AQr = -0.3, eqn (11) predicts anadsorption maximum at pH 6.15, which is consistent with the zero coverage extra-polation (fig. 3 and 5).Usingthe experimental values of cro for this ionic strength, the pH dependence of @o wasobtained from eqn (2), as shown in fig.6. The deviation from the Nernst equationof the calculated surface potentials $o is consistent with theoretical predictionsFrom eqn (5), Kl = 60 pF cm-2 for Kd = 20 pF cm-2, K2 = 30 pF cm-2.(fig. 6).ADSORPTION AT FINITE COVERAGEA Langmuir-type adsorption isotherm, with a pre-exponential term modified onthe basis of Flory-Huggins statistics, has been proposed by Parsons.16 This maybe written in a general form,-- - x exp (- AG,,,/kT) 0r(l-Oywhere r is the ratio of the area occupied by adsorbate and solvent molecules, and 8the fractional coverage of the surface by the adsorbate. An exactly equivalentstatistical term has been given by Levine, Mingins and Bell.14 Eqn (16) reduces, atvery low coverage, to a simple linear form as given by eqn (6) and (7)H .M. RENDALL AND A. L. SMITH 109We now consider briefly the complete measured adsorption isotherms, assuminga competitive adsorption of the two species. Values of x, may be obtained for eachexperimental point using eqn (9), and the corresponding nso obtained from eqn (16),but with the total measured adsorption included in the site-blocking term. Bysubtraction, adsorption isotherms for the HPic+ ion may be inferred. isnot sufficiently well known here we are unable to make a detailed comparison withtheoretical predictions. We may, however, compare the calculated HPic+ adsorptionwith the measured MePicf adsorption.As was observed by Conway at the mercurysolution interface, the substitutionI' = 1 (the Langmuir form of the adsorption isotherm) would give a satisfactoryaccount of the experimental results only if CD,, was allowed to decrease with coverage.This could imply either that there is a lateral repulsion between the adsorbed neutralmolecules, or that I' # 1.Since the solute and solvent molecules clearly differconsiderably in size, it is reasonable to examine the latter possibility.Since5 10 15 20concentrationlmmol dm-3FIG. 7.--Observed (filled points) and calculated (open points) cation adsorption. HPici ion at 0,pH 2 ; 'I, pH 4 ; A, pH 4.7; a, pH 5.9; 0, pH 6.8; 0, pH 7.6. MePic+ ion at ($, pH 6.9.- - -, theoretical Pic adsorption isotherm, eqn (16) with AG = -4.2 kT.Taking the areas occupied per molecule of picoline and of water as 40A2 and10 A2, respectively, and assuming that the whole B.E.T.area is available for adsorp-tion, substitution of the value of ST into eqn (16) gives @+ = -3.9 kT. WithA@r = -0.3, this gives (Q, = -4.2 kT. Using these values of Qn and of r in eqn(16), as described, the HPic+ ion adsorption, obtained by subtraction of the calculatedPic adsorption, paralleled that of the MePic+ ion in its dependence on pH (fig. 4)and on concentration (fig. 7). Lower values of an gave too high HPic+ion adsorption,as judged by comparison with the MePic+ ion, whereas higher values increased thecalculated adsorption of the neutral molecule, for small a, above the experimentaladsorption level. The value of AQr required to give a reasonable description of theadsorption at finite coverage was, therefore, the same as that which fitted the extra-polated zero coverage values.A difference (AQr) of this order is predicted by thesimple " image charge " term proposed by Frisch and Stillinger l7 on the basis ofthe Onsager-Samaras model. 110 EFFECTS OF IONISATION ON ADSORPTIONCOMPARISON WITH METHYL PICOLINIUM ADSORPTIONThe measurements which we have reported of the adsorption of the MePic+ ionmay be used to give a more quantitative test of the consistency of the model, and ofthe extrapolation techniques, as follows.1. The displacement of the ip of a surface with the concentration of an ion whichis adsorbed at the interface provides a sensitive measure of the specific adsorptionp0tentia1s.l~ By eqn (3) and (4), $B = 0 at the ip, and eqn (1) and (2) then give-0s = $oKi.(17)If it is assumed that $o (fig. 6) is unchanged in the presence of adsorbed MePic+ ion,os may be calculated from the reported ip shift. Substitution into eqn (16) gives aAG,,,(= @+ since t,hB = 0) of -4.5 kT for the MePicf ion.2. The adsorption isotherm for MePic+ at pH 6.9 was extrapolated, as 1nSagainst coverage, and t,br (froni fig. 6) subtracted from the zero coverage interceptto give ST for the MePic+ ion. Substitution of gz into eqn (16) gave a+ = -4.5 kTfor the MePicf ion.3. The difference (0.6 kT) between the specific adsorption potentials of HPic+and MePic+ would require that approximately twice as much HPic+ as MePic+ wouldwould be needed in solution to achieve the same level of adsorption under identicalconditions.Within experimental error this was found for the MePic+ adsorptionand the observed or calculated HPicf adsorption where a reasonable comparison waspossible, i.e., at pH 2, 4, 4.6 and 6.9. Only in the latter case was this comparisonsignificantly dependent on the calculation of Pic adsorption.We thank Dr. P. J. Anderson, who initiated this project, and Mr. J. R. Brownfor the preparation of the materials.H. C. Li and P. L. de Bruyn, Surface Sci., 1966,5,203.'B. E. Conway, R. G. Barradas, P. G. Hamilton and J. M. Parry, J. Electroanalyt. Chem.,1965, 10, 485.M. M. Breuer, Physico-Chemical and Biophysical Facfors Afecting the Activity of Pesticides(S.C.I. Monograph No. 29, London, 1968), p. 54.G. M. Bell, R. E. Chapman and M. M. Breuer, J. Colloid Interface Sci., 1968, 27, 161,J. J. Betts and B. A. Pethica, Trans. Faraahy Sue., 1956, 52, 1581.A. Albert and E. P. Serjeant, Ionisation Constants of Acids and Bases (Methuen, London, 1962). ' J. A. G. Taylor, J. A. Hockey and B. A. Pethica, Pruc. Brit. Cerarn. Soc., 1965, 5, 133. * J. D. Wagner, Ph.D. Thesis (CNAA, 1974).D. W. Fuerstenau, J. Phys. Chem., 1956, 60,981 ; P. Somasundaran, T. W. Healy and D. W.Fuerstenau, J. Phys. Chem., 1964,68, 3562 ; S. G. Dick, D. W. Fuerstenau and T. W. Healy,J. Colloid Interface Sci., 1971, 31, 595.lo D. C. Grahame, Chem. Rev., 1947, 41,441.l1 A. L. Smith in Dispersions of Powder in Liquids, ed. G. D. Parfitt (Applied Science Publishers,'' B. Lynskey, Ph.D Thesis (Liverpool Polytechnic, 1973).l3 G. H. Bolt, J. Phys. Chem., 1957, 61, 1166.l4 S. Levine, J. Mingins and G. M. Bell, J. Electroanalyt. Chem., 1967, 13, 280.l5 E. Blomgren and J. O'M. Bockris, J. Phys. Chem., 1959, 63,1475.l6 R. Parsons, J. Electroanalyt. Chem., 1964, 8, 93.l7 H. L. Frisch and F. H. Stillinger, J. Phys. Chem., 1962, 66, 823.l8 L. Onsager and N. N. T. Samaras, J. Chem. Phys., 1934,2,528.l9 H. M. Rendall, A. L. Smith and L. A. Williams, to be published.2nd edn., 1973), p. 93 ; S. Levine and A. L. Smith, Disc. Farczday SOC., 1971, 52, 290.(PAPER 4/297