Desorption of Glycerol from Clays as a Function of Glycerol Vapor Pressure1

B. F. HAJEK AND J. B. DIXON*

ABSTRACTGlycerol desorption isotherms were developed for montmoril-

lonite, vermiculite, kaolinite, and a mixture of montmorilloniteand vermiculite at 95C and SOC. Application of the Langmuirequation to desorption data for montmorillonite indicated that465 mg of glycerol was necessary for a complete monomolecularlayer on all surfaces of 1 g of 2/i to 0.2ji clay. Although desorptiondata for vermiculite fitted the Langmuir equation, the adsorbedglycerol apparently was held at lower average density than formontmorillonite. Two intersecting straight lines fitted the desorp-tion data for a montmorillonite and vermiculite mixture plottedaccording to the Langmuir equation, indicating that the twominerals could be distinguished quantitatively in the mixture.Multimolecular layers of glycerol appeared to be adsorbed onkaolinite at temperatures and pressures needed for duointerlayer(true monolayer) sorption by montmorillonite. Enthalpies of de-sorption from a Langmuir constant and from the Clausius-Clapeyron equation compared favorably. Enthalpies ranged from40 to 55 kcal/mole of glycerol desorbed from kaolinite and ex-pansible minerals, respectively.

THE SUEFACE AREA of soil clays is of major importance incharacterizing minerals making up this soil fraction.

Often, when soil clays are poorly crystalline or are inter-mediate between mineral types, surface area measurement isespecially useful for determining the amount of area availablefor surface reactions. The importance of this soil propertyjustifies further investigation of conditions necessary for soilsurface area measurement.

Brunauer (5) states that for a given gas and unit weight ofa given adsorbent the amount of gas adsorbed at equilibriumis a function of the final pressure and temperature. The mostfrequently determined experimental adsorption relation is theadsorption isotherm.

Most current theories of adsorption are based in part onearly work conducted by Langmuir and Freundlich. Lang-muir's (11) theoretical kinetic derivation may be written inthe form

v = bPvm/(l + bP)

where v = volume adsorbed, vm = volume necessary to coverthe adsorbent with a monomolecular layer, 6 = constant, andP = equilibrium pressure. Freundlich's equation is empiricalwith no significant theoretical basis (1). Although Langmuirused a kinetic derivation to obtain his isotherm equation, anequation of the same form was obtained thermodynamicallyby Volmer and statistically by Fowler (5).

Brunauer, Emmett, and Teller (4) extended Langmuir's

HAJEK AND DIXON: DESORPTION OF GLYCEROL FROM CLAYS 31

model to multilayer adsorption. This equation, known as theBET equation, is commonly written in the form

ex

in which x = P/P0, P0 = saturation pressure, c = a constant,and the other terms are the same as in Langmuir's equation.

Diamond and Kinter (6) originated a glycerol retentionmethod of surface measurement using the thickness of anadsorbed glycerol layer, determined by X-ray diffraction andbulk liquid density to convert retention to surface area.Mehra and Jackson (12) modified this method by using a lesssaturated glycerol atmosphere to obtain monointerlayersorption. Mehra and Jackson (13) proposed a vacuum methodfor obtaining duointerlayer (a true monomolecular layer)glycerol retention on montmorillonite. Milford and Jackson(14) proposed a further modification using the temperaturegradient in a vacuum oven to control the vapor pressure of afree glycerol source.

Although there are several methods of measuring thesurface area of clay minerals involving retention of polarmolecules (6, 7, 13), nonpolar gases (4, 8, 11), and organiccations or anions (9, 16), no one method is generally used.Glycerol was selected for this study because it satisfactorilydifferentiates montmorillonite from vermiculite. In addition,glycerol is sorbed on interlayer surfaces not accessible tonitrogen, and is sorbed nearer to liquid density than adsorbedglycol (3, 6) (B. F. Hajek, 1964. Determination of surfacearea of layer silicates and heats of glycerol desorption fromequilibrium desorption isotherms. Ph.D. Diss., Auburn Univ.,Auburn, Ala.). Existing glycerol methods often fail to givemeaningful surface area values for clays because the point onthe isotherm from which surface values were obtained has notbeen determined. Surface area values for fine clays by somecurrent methods often appear too high due to excess sorptionof glycerol and/or water vapor contamination.

The approach used here was selected so that data could beinterpreted by Langmuir and BET adsorption theories andinterference from moisture could be minimized. Other objec-tives were to evaluate the vapor pressure(s) most favorablefor measuring surface area of clay minerals and to calculateenergies of glycerol desorption.

MATERIALS AND METHODSA sorption apparatus was constructed that is similar to an

apparatus built by Milligan et al. (15). The design was modifiedto include only two silica springs as shown schematically in Pig. 1.Heating tapes were used to prevent condensation of vapor onglass parts not submerged in temperature-controlled baths. Glasstubes containing the silica helix balances were mounted in aconstant temperature water bath. Equilibrium vapor pressure ofglycerol was controlled by changing the temperature of the waterbath (Fig. 1) in which the glycerol-source container was mounted.Insulation around both water baths and their covers is omittedfor clarity. The change in spring length, measured with a catheto-meter, was a direct measure of sample weight and thus the amountof sorption. Sensitivity of weighing was ± 0.3 mg or better.

Glycerol vapor pressure values were obtained from the Landoltand Bornstein tables (17) and a linear plot of log of vapor pres-sure versus the reciprocal of the absolute temperature. Vaporpressure values for 30, 40, and SOC were obtained by extrapola-tion to lower temperatures of the stated linear plot.

A. S i l ica Spr ing Ba lances and AluminumB u c k e t s M o u n t e d i n T u b e s

B. H e a t i n g T a p eC . A d s o r b a t e S o u r c e C o n t a i n e rD. O u t l e t to D r y i n g T r a p , McLeod G a u g e .

T h e r m o c o u p l e G a u g e a n d V a c u u m P u m p

E . W a t e r L e v e lFig. 1—Schematic drawing of adsorption-desorption apparatus.

The following sample materials were the sources of clays used:(i) crude bentonite, from Mowrey beds, Crook County, Wyoming(Baroid Division of National Lead Co., Houston, Tex.); (ii)vermiculite from Llano County, Texas (National Clay Con-ference field trip, October 1961, alternate stop 11); and (iii)poorly crystalline kaolinite (Georgia Kaolin Co., Elizabeth,N.J.). The coarse clay fractions (2 to 0.2/t) used were obtained bydispersing in Na2CO3 and centrifuging (10). After fraetionationthe samples were saturated with Mg, washed free of salts, anddried from acetone.

The spring length with an attached passivated Al-bucket wasdetermined in an evacuated sorption tube at the isotherm tem-perature. Approximately 100 mg of a clay sample was added tothe sample bucket. The sample and apparatus were out-gassedat 95C by continuous evacuation at < 20ju Hg pressure for atleast 24 hours. The bath was thermostated at the isotherm tem-perature and the spring length determined for sample and bucket.The sample bucket was removed. 1.5 ml of 5% (by weight)glycerol and water solution was added, and the wetted samplewas allowed to stand for at least 8 hours. Excess amounts of waterand glycerol were removed by heating the sample at about 85Cthen at HOC. The sample was then returned to the apparatus toequilibrate at isotherm temperature and various vapor pressures.The time required for samples to reach equilibrium depended onthe pressure and initial amount adsorbed. In general, 24 to 36hours were required to reach equilibrium after the initial excesswas removed. Sorption values reported are averages of duplicatedeterminations. A minor correction for sorption on the containerwas made.

RESULTS AND DISCUSSIONEquilibrium desorption data obtained for montmorillonite,

vermiculite, kaolinite, and a mixture of montmorillonite andvermiculite at two temperatures and several vapor pressuresare shown graphically in Fig. 2. The 80 C glycerol-mont-morillonite isotherm is of the Langmuir type; however,glycerol desorbing at 95 C yields an isotherm that approachesa type IV shape. Isotherms of the latter type have beenexplained by a capillary condensation theory (5).

Relatively small amounts of glycerol desorbed from vermic-ulite and kaolinite in the pressure ranges studied. Theadsorption isotherm of the 50% mixture of montmorillonite

32 SOIL SCI. SOC. AMEE. PROC., VOL. 30, 1966

and vermiculite at 95 C is intermediate between the isothermsof the pure minerals at the same temperature.

The Langmuir equation in linear form was applied to thedesorption data. The equation was written in the followingform:

P = J_ P_' W bWm Wm

whereP = equilibrium vapor pressure,W = mg glycerol sorbed per gram clay at pressure P,b = constant for a given temperature, adsorbate and

adsorbent, andWm = mg glycerol for monomolecular coverage of 1 g of

clay.A plot of P /W vs. P should yield a straight line and the twovalues b and Wm can be evaluated from the slope (l/Wm) andintercept (i/bWm).

Desorption from MontmorilloniteApplication of Langmuir's equation to desorption of

glycerol from montmorillonite at 80 C yields a satisfactorystraight line (Fig. 3). From the slope of this line, it wasdetermined that 464 mg glycerol adsorbed per gram of claygave monomolecular coverage at 80 C. The 464-mg valueapproximately corresponds to .sorption for the horizontal partof the 80 C isotherm. Data obtained at 95 C did not fit asingle straight line (Fig. 3). The curve can be treated differ-entially as described by Brunauer (5), i.e., two straight linescan be drawn, each passing through or near four points withone point common to both lines. Partial collapse of themontmorillonite structure accompanying removal of one ofthe two molecular layers of glycerol explains the fit of thesedata to straight lines. From the slope of each line, it wasdetermined that 465 mg of glycerol formed a completemonolayer (duointerlayer) and that 238 mg of glycerol should

completely cover 1 g of clay with clay surfaces on two sidesof adsorbed molecules in the interlayer spaces. Assuming thatthe amount sorbed on external surfaces is the same whenmono- and duointerlayers are sorbed, these data indicate that11 mg of glycerol sorbed was necessary for monolayer coverageon external surfaces and 227 mg of glycerol for monointerlayercoverage of 1 g of montmorillonite. The 11-mg value issomewhat larger than the value calculated by Jackson (10)and agrees reasonably well with values determined onreference minerals (6).

A surface area of 1,742 m2/g of adsorbed glycerol wascalculated, based on the Wm value of 465 and a total surfacearea of 810 m2/g for the montmorillonite. This value agreeswell with 1,765 m2/g of glycerol reported by Diamond andKinter (6) and reasonably well with 1,910 m2/g of glycerolcalculated by Jackson (10) on the basis of molecular layerthicknesses of 4.5A and 4.15A, respectively.

A surface area of 26.7A2 per glycerol molecule was calculatedfrom the Wm of 465, using a montmorillonite surface areaof 810 m2/g, the molecular weight of glycerol, and Avogadro'snumber. This agrees with the value of 26.9A2 per moleculeproposed by Diamond and Kinter (6).

Desorption from VermiculiteWhen the Langmuir equation was applied to data for

vermiculite desorption at 80 C, the points fell in a straightline as shown in Fig. 4. The point at the highest degree ofsaturation is off the line indicating that in excess of a mono-layer had been adsorbed. The Wm value determined for thisvermiculite was 185 at 80 C and 178 at 95 C (Fig. 5). It wasestimated that the vermiculite in this study had a specificsurface of 606 m2/g. This value was determined assuming thatthe 238 mg of glycerol sorbed on 1 g of montmorillonite(monointerlayer coverage at 95 C) represents the sorptionsurface of vermiculite at the same temperature and glycerolvapor pressure.

50

0.2

• M o n t m o r i l l o n i t e• V e r m i c u l i t e• K a o l i n i t e« 507. V e r m i c u l i t e - 50% M o n t m o r i l l o n i t e

Fig. 2—Glycerol desorption isotherms forthree clay minerals and a mixture.

P (mm Hg)Fig. 3—Application of Langmuir equation

to desorption of glycerol from mont-morillonite at 95C and SOC.

P (mm Hgl

Fig. 4—Application of Langmuri equationto desorption of glycerol from vermi-culite at SOC.

HAJEK AND DIXON: DESORPTION OF GLYCEROL FROM CLAYS 33

Since vermiculite obtained from different sources varieswidely in purity, it is not always possible to compare datareported by various workers. X-ray and differential thermalanalysis (DTA) indicated no contaminating minerals. Onlytraces of potassium and sodium were detected by flamespectrophotometry. X-ray spectrographic analysis indicatedthat about 0.2% K20 was present. This would represent about2% mica (12). Cation-exchange capacity was determined tobe 208 meq/100 g, (Ca replaced by Na at pH 7). Thesedeterminations indicate that the vermiculite used in thisstudy was essentially pure, although small amounts of mineralssuch as quartz, chlorite, and pyrophyllite could have beenpresent.

The purity of the vermiculite and the structural similarityof vermiculite and montmorillonite suggests that the surfacearea should have been nearly the same for both minerals.Adsorbed water could cause some error; however, this errorwas reduced by taking the base dry weight in an evacuatedsystem. Incomplete filling of the vermiculite interlayer spaceby glycerol as reported for water by Walker (18) appears tobe a reasonable interpretation of these data. The differencesbetween determined and theoretical values for surface areaand cation exchange of vermiculite (assuming mica layer-charge density) are of the same magnitude which suggeststhat glycerol was not sorbed where exchange sites were blockedor neutralized.

Desorption from Montmorillonite-Vermiculite MixtureDesorption of glycerol from a mixture of 50% montmoril-

lonite and 50% vermiculite indicated that a differentialapplication of the Langmuir equation (two straight lines)could be used to estimate the quantity of montmorillonite inthis mixture. The difference between the two Wm values(Fig. 6) should be because of loss of one molecular layer ofglycerol from the montmorillonite interlayer space due tolowering of glycerol vapor pressure. This interpretation

assumes desorption of montmorillonite and vermiculite in themixture analogous to that in the "pure" samples (Fig. 3 and5). In the glycerol vapor pressure range employed the vermic-ulite lost very little glycerol and montmorillonite lost onemolecular interlayer. However, the determination of thepercent montmorillonite and vermiculite in the mixture is notin complete agreement with the sorption of the individualminerals. The 340 mg and 227 mg of glycerol obtained fromthe slopes of the two lines (Fig. 6) are higher than 322 mg and208 mg of glycerol per gram of clay calculated from Wm valuesobtained (Fig. 3 and 5).

Desorption from Kaolinite

Kaolinite desorption data could not be represented by theLangmuir equation over the entire range examined (Fig. 7).Apparently multilayer glycerol adsorption does occur atrelative pressures > 0.4 or 0.5 for the 80 C isotherm (Fig. 2).This sorption, which is in excess of that predicted for amonolayer by particle size, density, and geometric considera-tions, could introduce error into the measurement of soilsurface areas when this mineral is abundant in the clayfraction. The greater range of vapor pressures where amonolayer was sorbed at 95 than SOC indicates the highertemperature would be better suited for studying mixtures ofkaolinite and other minerals which desorb over a wide vaporpressure range. Application of the Langmuir equation at lowpressures indicated that 33 mg of glycerol at 80 C and 24 mgat 95 C gave monolayer coverage on 1 g of kaolinite.

Desorption Evaluated with BET EquationIn general, the BET equation fitted desorption data from

all three clay minerals at relative pressures < 0.35. At theselow pressures, a straight line passes so near the origin that theslope can be considered as l/Wm since the constant c of the

O O . O Z 0.04 0.06 0.08 ' 0.10

P (mm Hg)

Fig. 5—Application of Langmuir equationto desorption of glycerol from vermi-culite at 95C.

P (mm Hg)

Fig. 6—Application of Langmuir equation' to desorption of glycerol from a prepared

mixture containing 50% montmoril-lonite and 50% vermiculite at 95C.

0.02 0.04 0.06 0.08 0.10

P (mm Hg)

Fig. 7—Application of Langmuir equationto desorption of glycerol from kaoliniteat 95C and SOC.

34 SOIL SCI. SOC. AMEH. PROC., VOL. 30, 1966

BET equation would be large. This is apparent when theBET equation is written in linear form, i.e.,

x I (c - l)x+ -W (I - x) cTT- cWn

Table 1—Isosteric heats of desorption of glycerol from clayminerals calculated by two methods

________________Mineral_________adsorbed_____kcal/mole*_________

Clausius~Clapeyron equation

When c»l, (c — 1) is approximately equal to c and the lastterm is approximately equal to x/Wm. Thus, the BETequation becomes analogous to the Langmuir equation; how-ever, different variables are plotted (P0 is included). Mono-layer amounts evaluated by this method were about equal tothe percent sorbed at a relative pressure of 0.2, which isconsidered by many to be a reasonable estimate of thesorption necessary for monolayer coverage.

Hysteresis is known to occur in clay-adsorbate systems (2)and no doubt occurs with clay-glycerol. Adsorption of glycerolfrom the vapor phase was attempted; however, the study wasnot practical because of the slow rate of adsorption at 95 C.

Enthalpy DeterminationThe Clausius-Clapeyron equation was used to calculate

enthalpy of desorption (1) by using positions of equal adsorp-tion on the 80 and 95 C isotherms. The enthalpy of desorptionalso can be calculated from the Langmuir constant b. Adamson(1) has stated that

b = K exp (Q/RT).

The symbols Q, R, and T refer to the enthalpy of desorption,gas constant, and absolute temperature, respectively. If theslight temperature variation of K is neglected, Q can readily becalculated from two values of b obtained at two isothermtemperatures.

The Langmuir model restricts the heat of adsorption to aconstant. This would indicate that the heat obtained byevaluating the constant b should be the heat of desorptioncorresponding to monolayer coverage because the slope isI/Wm and the intercept of this line at P = O is l/bWm.

Heats of desorption are given in Table 1. More values werenot calculated because W ranges were narrow where aconstant percent of glycerol was sorbed at two isothermtemperatures. Desorption energies were nearly equal formontmorillonite and vermiculite at amounts adsorbed rangingfrom 180 to 390 mg/g. However, if the energy calculated fromthe Langmuir constant is comparable to the Clausius-Clapey-ron value desorption energy decreased at true monolayercoverage (duointerlayer) on montmorillonite.

Desorption energy was less for kaolinite but this may bedue to multilayer adsorption at 30 mg/g adsorbed.

The desorption energies determined are 2 to 3 times greaterthan the heat of vaporization, 21 kcal/mole, determined fromthe slope of the straight line obtained by plotting log of vaporpressure versus the reciprocal of the absolute temperature.Energy values of sorption which are 2 to 3 times heat ofvaporization have been considered within the energy rangeof physical adsorption; however, values greater than 10kcal/mole have been considered in the chemisorption range.It seems that the 10-kcal value cannot be used in this casesince the normal heat of vaporization is greater. Theseenthalpy values are presented to show differences betweenminerals and at different amounts sorbed although the

Montmorillonite

VermiculiteKaolinite

Montmorillonite

Vermiculite

3930183

Langmuir constant, b46. 5 (95C)46.4 (SOC)17. S (95C)18.5 (SOC)

56555440

47

53

*Glycerol heat of vaporization = 21

reversibility of the process has not been confirmed for condi-tions employed.

f _. _ ~ _ _ _ _ _