DIVISION S-2-SOIL CHEMISTRY

Characterization of Carbonaceous Adsorbents by Soil Fulvicand Humic Acid Adsorption

M. M. Johns,* E. O. Skogley, and W. P. Inskeep

ABSTRACTSix porous carbonaceous adsorbents, selective for nonpolar organic

adsorbates, were evaluated for dissolved soil humic adsorption andtesting. Afnberlite XAD 8 resin was similarly tested. Soil fulvic (FA)and humic (HA) acids were used as models for naturally occurringdissolved organic C in soil. The adsorbents were tightly packed asspherical capsules identical to previous studies using ion exchangeresins for soil adsorption of plant nutrients. Static or passive adsorp-tion from solution was applied. Kinetic studies (0-264 or 0-336 h)indicated a high initial adsorption rate followed by slower adsorption,common for porous adsorbents. Intraparticle diffusion was rate lim-iting. Adsorption differences between adsorbents are probably causedby molecular-size exclusion effects due to pore size. The capsules ofCarboxen-Carbotrap resin mixture were capable of doubling adsorp-tion, requiring no additional time, in response to doubling the FA(43-85 mg C L-') or HA (46-90 mg C L~') concentration. This effectwas seen across the whole adsorption-time range. The constant par-titioning isotherm model described adsorption, indicating linear ad-sorption (r2 > 0.96, P < 0.01) with temperatures ranging from 13 to50 °C, at pH 7.0, and within the concentration ranges of FA (9-126mg C L-1) and HA (11-138 mg C L~'). This suggests a usefulness forthese adsorbents at neutral pH and typical temperatures in waters fordissolved humic adsorption. Subsequently, initial solution concentra-tions of FA and HA were predicted from resin-extractable C, indi-cating a potential for quantifying humic concentrations in solutions.Under identical conditions, XAD 8 resin did not demonstrate the test-ing attributes found with these carbonaceous adsorbents.

TESTING WATER AND SOIL for agricultural and en-vironmental purposes is becoming increasingly

important. Present conventional methods for soilanalysis involve sampling, drying, grinding, sieving,and other physical or chemical procedures unique tothe elements or compounds being tested. In many in-stances, these procedures are costly and time consum-ing, which can place limits on sample numbers. Thisis especially true for sampling and analyzing hazard-ous wastes or pesticides in soils. Interpretations andreliability of the testing results may be inaccurate dueto physicochemical changes in samples brought fromthe field, or because of poor correlation between thetesting method and what is actually occurring in thenatural environment.

An alternative approach to soil testing is based onthe use of ion exchange resins packaged as a sphericalresin capsule to accumulate dissolved constituents inresponse to diffusive movement through the soil, sim-ulating nutrient movement to plant roots (Skogley et

Dep. of Plant and Soil Science, Montana State Univ., Bozeman,MT. Contribution of the Montana Agric. Exp. Stn. Journal no.2761. Received 20 Oct. 1992. *Corresponding author.

Published in Soil Sci. Soc. Am. J. 57:1485-1490 (1993).

al., 1990; Yang et al., 1991). This approach has beenexpanded to include studies with carbonaceous resinsto determine their potential for use in modeling thesoil organic fraction behavior, as well as capabilitiesfor detecting organic compounds of environmentalconcern. Results of initial research indicated the use-fulness of carbonaceous resins placed in saturated soilpastes for determining total soil C and extraction ofsoluble OC constituents (Johns and Skogley, 1991).Advantages of this type of test include elimination ofreagents that present disposal problems (such as di-chromate used in many soil testing laboratories), sim-plicity, potential in situ applications, and use in fieldmonitoring of pollutants (Thakkar and Manes, 1988).

Carbonaceous resins have a polynuclear aromaticstructure similar to graphite, with very few ions orfunctional groups, contain a large number of C-Cbonds, and are porous with high internal surface area(Neely and Isacoff, 1982). The resins have adsorptioncharacteristics similar to those of activated C (Chros-towski et al., 1983), but many of the resin types pos-sess a greater nonpolar surface than activated C. Theseadsorbents possess surface properties classified as TypeI (Osick and Cooper, 1982), where organic com-pounds are adsorbed by nonspecific, physical adsorp-tion. They have been recently used for contaminantsampling of BXTs, straight-chain alkanes, chloroal-kanes, and naphthalenes in air, water, and soil (Betzet al., 1989; Hazard et al., 1991).

Soil FAs and HAs are important constituents of dis-solved, naturally occurring C in aquatic and soil en-vironments (Aiken et al., 1985). These acids have anamphipathic character, they can interact with both hy-drophobic and hydrophilic compounds. This includesthe binding (i.e., partitioning) of nonpolar organiccontaminants, which can facilitate contaminant trans-port through soil (Chiou et al., 1986; Enfield et al.,1989). Also, FA and HA can complex both plant-essential and toxic metals (Stevenson and Fitch, 1986).Thus, these acids are appropriate materials for char-acterizing resin behavior in natural systems.

The objective of this research was to characterizesix selected commercial carbonaceous resins throughtheir reactions with dissolved soil FA and HA. Tocontrast adsorption, Amberlite XAD 8 resin (Rohmand Haas Co., Philadelphia, PA) was similarly tested.,because this resin has been recommended in the iso-lation and extraction of aquatic humic substances(Thurman and Malcolm, 1981). The choice of FA andHA as adsorbates does not allow determination of de-tailed adsorbent specificity, but does provide basic

Abbreviations: FA, fulvic acid; HA humic acid; OC, organiccarbon; BXT, benzene-xylene-toluene; MW, molecular weight;CV, coefficient of variation.

1485

1486 SOIL SCI. SOC. AM. J., VOL. 57, NOVEMBER-DECEMBER 1993

Table 1. Physical and chemical properties of hydrophobic carbonaceous and XAD 8 resins.

Resin

Carboxen 569Carbotrap BCarbotrap CXEN 563XEN 564XEN 572XAD 8

Manufacturingprecursor

N/Atcoconut hullscoconut hullsS/DVB*S/DVBS/DVBacrylic ester

Surfacearea

m2g-'48510010

550550

1100160

Porositycm3 g~'

N/AN/AN/A0.600.510.840.48

Mei.i sizemm (inch)

0.84-0.35 (20-45)0.84-0.42 (20-^0)0.84-0.42 (20-^0)0.84-0.? S (20-45)0.84-0.35 (20-45)0.84-0.3 S (20-45)0.84-0.25 (20-60)

t Information not available.$ Styrene/divinylbenze macroeticular resin.

understanding in resin adsorption behavior in naturalenvironments.

THEORETICAL CONSIDERATIONSContinuous flow or completely mixed batch reactor systems

are customarily used to study adsorption of organics. In thisstudy, static adsorption was employed to simulate that de-scribed by Yang et al. (1992) for adsorption of plant nutrientsfrom soils to mixed-bed ion-exchange resin. No distinction ismade between adsorption and sorption (the physical incorpo-ration of the adsorbate into the porous resin). In solutions, aswell as in soils, diffusion is the rate-limiting process for staticadsorption onto resins. Adsorption is assumed to be a processof (i) molecular diffusion of the adsorbate from the bulk so-lution to the stationary boundary layer surrounding each resinparticle, (ii) external mass transfer (film diffusion) across theboundary layer, (iii) intraparticle mass transfer (pore diffusion)from the surface to the interior, and (iv) micropore adsorption,which is considered very rapid (Rosene, 1983). Diffusion frombulk solution to within the tightly packed resin bed (1.1-cm-radius sphere) is assumed to be relatively rapid since interpar-ticle channels are much larger than intraparticle pores.

KineticsA static system such as this will probably be controlled by

either film or intraparticle diffusion during adsorption (Sparks,1989). If film diffusion is rate limiting, the adsorption rate isdirectly proportional to the external adsorbent surface area andis independent of the adsorbent type (Neely and Isacoff, 1982).Therefore, for particles of equal size and shape, adsorbentcomposition and pore structure would not affect the adsorbentrate.

Greater than 98% of the surface area for carbonaceous resinsis within pores <500 A diameter. Adsorption occurs predom-inantly within these pores (Neely and Isacoff, 1982). The ex-ternal surface of these spherical resin capsules directly exposedto bulk solution is 1.4 x 10~3 m2, or <0.001% of the totalsurface area. Stirring will have a smaller effect on film dif-fusion for these tightly packed capsules than for a loosely packedor unpacked resin system. Further, it would be impossible tostir the solution within the intraparticle pore space.

AdsorptionAssuming that a simple relationship exists between the resin

capsule and solution phases, the linear or constant partitioningadsorption isotherm was applied;

JsJads [1]

where Qads is the C (mg) adsorbed per kilogram of resin C andCeq is the equilibrium C concentration (mg/kg); K^ is termedthe partition coefficient normalized to organic C. The resin Ccontent is 90%. The use of K^ was analogous to soil modeling,

where the controlling influence for similar adsorption is soilorganic C (Karickhoff, 1981).

MATERIALS AND METHODSResins

The resins employed are all specific for adsorbing nonpolarorganic compounds (Table 1). Carboxen-569 is reported to bespecific for 2 to 5 C length compounds, Carbotrap-B for 6 to12 C compounds, and Carbotrap-C for 12 to 20 C compounds(all from Supelco, Bellefonte, PA). Because no informationwas available on their performance in soils, a mixture of equalproportions (by volume) of all three resin types was used. Amixture should be advantageous for adsorbing polydisperse,humic molecules. The carbonaceous adsorbents Ambersorb 563,564, and 572, and Amberlite XAD 8 (all from Rohm and HaasCo., Philadelphia, PA) were used individually.

Carbonaceous Resin CapsulesSpherical resin capsules of 5-cm3 volumes were constructed

using polyester sieves as described by Yang et al. (1991). Bulkresins were degassed in water under vacuum prior to construc-tion of resin capsules. Equal proportions (by volume) of thecarbonaceous resins Carboxen-569, Carbotrap-B, and Carbo-trap-C for a total wet volume of 5 cm3 were packed into polyes-ter (30-ju.m thread, 140-/j,m openings) cloth and tied withpolyester thread to form a tight sphere. The dry weight of theresin mixture was 2.1 g per 5-cm3 volume. Similarly, resincapsules were made using each of the Ambersorb adsorbentsand XAD 8. Carbonaceous resin beds of 5-cm3 volume (Table1) would posses approximate surface areas of 424 m2 for theCarboxen-Carbotrap mixture, 1177 m2 for Ambersorb 563 or564, and 2354 m2 for Ambersorb 572. All resin capsules werestored under double-distilled water until used. Bleeding (de-composition) of the resins was monitored by OC analysis ofwater aliquots. The OC content never exceeded 0.3 mg L,-1

during the study period.

Soil Fulvic and Humic Acid ExtractionFulvic acid and HA were extracted with 0.25 M Na2CO3

from an Enbar loam (fine-loamy, mixed Cumulic Haploboroll)Ap horizon. The soil has a pH of 6.3 and an OC content of29 g kg-1. After 24 h of shaking, the liquid from the Na2CO3-soil mixture was decanted, centrifuged, and acidified to a pHbelow 2.0 with concentrated HC1. After an additional 24 h,the FA was decanted from the precipitated HA. Excess saltwas removed from the FA solution using repeated dialysis(Spectrapor membrane tubing, 6000-8000 MW cutoff, Spec- "trum Medical Industries, Los Angeles). Excess salt was re-moved from the HA through repeated rinsing (0.01 M HC1)and centrifugation. Both FA and HA were then freeze-dried.The C contents, determined by wet combustion (Synder andTrofymow, 1984) were 420 g kg-1 for FA and 520 g kg-1 forHA. The hydrophobic (48%) and hydrophilic (52%) solute

JOHNS ET AL.: SOIL ACIDS CHARACTERIZE CARBONACEOUS ADSORBENTS 1487

Table 2. Molecular weight determinations by ultrafiltrationfof soil fulvic and humic acids.

DiskMembrane

XM300YM 100PM 30PM 10YM 3

Molecularweight cutoff

Ha

300 000100 00030000100003000

Cinitial

Fulvic

32858891

retained ofconcentration

Humic

26559393—

t Initial concentrations were 109 mg C L~' for fulvic (pH 7.0) and 107mg C L-1 for humic acid (pH 7.2) and an ionic strength of 0.01 M.Filtration pressure ranged from 55 to 173 kPa.

contents of FA were determined using a XAD 8 resin columnseparation (Leenheer, 1981). The MW distribution of the FAand HA fractions was determined by ultrafiltration (Wershawand Aiken, 1985), using Amicon (W.R. Grace and Co., Bev-erly, MA) disk membrane filters. Approximately 59% of C inthe FA fraction was within 3000 to 100 000 daltons and 67%of C in HA was within 10 000 to 300 000 daltons (Table 2).

Adsorption StudiesAdsorption-time studies were carried out in duplicate using

resin capsules containing the Carboxen-Carbotrap mixture andthe following conditions: pH of 7.2, 24 °C, ionic strength (/)of 0.01 M (NaNO3) at two concentrations of FA (43 and 85mg C L-1) and HA (46 and 90 mg C L-'). Capsules werecompletely immersed in 50 mL of FA or HA solutions incapped 125-mL glass containers at time zero. No shaking wasemployed. At specific times, 1-mL aliquots of the solutionswere removed and analyzed for OC using a Dohrmann DC-80Total Organic C Analyzer (Rosemount Analytical, Santa Clara,CA). About 10 1-mL aliquots were taken during the studytime, ranging from 0 to 264 or 336 h. Total mass of C adsorbedper unit time was calculated by mass balance, taking into ac-count decreasing solution volumes. Similarly, adsorption-timestudies were carried out using the XEN adsorbents from Rohmand Haas (Table 1) at a pH of 7.0, 22 °C, and 7 = 0.01 M.Two concentrations of FA (60 and 120 mg C L-1) and onlyone concentration of HA (55 mg C L~') were studied, withoutreplication.

Film diffusion was evaluated using a shell progressive filmdiffusion equation (Hodges and Johnson, 1987). A plot of therelative fraction adsorbed (CJC0) vs. time should be linear ifadsorption is controlled by film diffusion, where Cs is total C(jug) adsorbed on the resin with time and C0 is total C (jug) insolution at time zero.

Intraparticle diffusion was evaluated using the parabolic dif-fusion equation described by Sparks (1989), where a plot ofCJC0 vs. square root of time should provide a linear relation-ship (Jardine and Sparks, 1984; Aiken et al., 1979).

Adsorption isotherms were determined for the Carboxen-Carbotrap capsules at four temperatures: 13, 23, 36, and 50 °C.At each temperature except 23 °C, a 96-h equilibration timewas used with four FA concentrations ranging from 14 to 102mg C L-1, each at pH 7.0 and / = 0.01 M. At 23 °C, fiveconcentrations of FA and HA were used ranging from 9 to 126mg C L-1 and 11 to 138 mg C L-1, respectively. Adsorptionisotherms were obtained for the XAD-8 resin at 23 °C and apH of 6.3. Two replicates were used for all treatments. Kineticstudies indicated that, at 96 h, the adsorption rate, d(Cs/C0)/d(time), decreased to <10% of the initial rate (first 24 h).Numerous adsorption studies in soils, sediments, organic mat-ter, and activated C (Leenheer and Ahlrichs, 1971; Karickhoff,1984; Miller and Weber, 1984; Kaastrip and Halmo, 1989)are characterized by an initial adsorption rate followed by slower

a. Carboxen-Carbotrap capsules

-Sr Fulvic, 43 mo C/kgFulvic. 88 mg C/kg

-B- Humlo. 48 ma C/kg-0- Humic, 90 mg C/kg

pH • 7.2, 24 C, I • 0.01

100 150 200

HOURS260 300 350

b. Ambersorb 564 capsules

-S- Fulvic, 60 mg C/kgFulvic, 120 mg C/kg

-B- Humic. 58 mg C/kg

100 160 200HOURS

260 300 360

Fig. 1. Relative adsorption (CJCa) of fulvic and humic acidswith time on capsules of (a) Carboxen-Carbotrap resinmixture and (b) Ambersorb 564.

adsorption, which may proceed indefinitely. This has requiredan assigned reaction time (Karickhoff, 1985).

Adsorbed C was removed by vacuum extraction (Model 24,Centurion International, Lincoln, NE) using 50 mL of 2 MNaOH with an extraction time of 1 h. Adsorbed OC was de-termined on a Dohrmann DC-80 Total Organic C Analyzerafter 5-mL aliquots were acidified (H3PO4) to pH 2.0 and spargedwith O2 for 5 min. Due to a limited supply of the Carboxenand Carbotrap resins, these capsules were regenerated by twosuccessive vacuum extractions using 50 mL of 2 M NaOHeach extraction, and then a final 10 mL of 1M HC1 extraction.Capsules were then rinsed with double-distilled water until freeof Cl. Analysis of variance between new (unused prior) resincapsule blanks and regenerated capsules indicated no signifi-cant differences in background C (P = 0.47).

RESULTS AND DISCUSSIONKinetics

Both FA and HA were adsorbed by the resin capsuleswith a high initial adsorption rate (slope) that decreasedwith time (Fig. 1). Ambersorb 563 and 572 are not shownsince their plots could almost coincide with Ambersorb

1488 SOIL SCI. SOC. AM. J., VOL. 57, NOVEMBER-DECEMBER 1993

O

o

a. Carboxen-Carbotrap capsules

s 10(HOURS)*

b. Amberaorb 564 capsules

o 6 10(HOURS)*

Fig. 2. Relative adsorption (CJCJ vs. time"2 for adsorptionof fulvic (FA) and humic acids (HA) on capsules of (a)Carboxen-Carbotrap resin mixture and (b) Ambersorb 564.Regression lines for first 97 to 100 h.

564 (Fig. Ib). Ambersorb 572 showed slightly less (8-10% of Cs/C0) adsorption than 563 or 564 during thetotal time period. These curvilinear adsorption-time re-sponses are commonly observed (i.e., batch adsorption-time studies) with porous adsorbents, where the adsorp-tion rate is limited by intraparticle diffusion (Wu andGschwend, 1986). These were static adsorption-timestudies, however, requiring longer reaction periods using50-mL volumes of FA and HA. Observed linearity ofCS/CQ vs. time172 for the initial 97 to 100 h of adsorption(Fig. 2) suggests that intraparticle diffusion was indeedrate limiting (Sparks, 1989; Leenheer and Ahlrichs, 1971).Curvature in Fig. 1 indicates that film diffusion wasprobably not adsorption rate limiting (Hodges and John-son, 1987). In general, the kinetic behavior seen withFA and HA adsorption on these resin capsules is similarto many low-water-soluble organic adsorbates and po-rous adsorbents, such as activated C (Weber and Smith,1989), soil organic matter (Leenheer and Ahlrichs, 1971),and polymeric resins (Aiken et al., 1979).

An adsorbent's surface area is predominantly due tosmaller pores (i.e., micropores of <2 nm) (Neely andIsacoff, 1982). This can explain why the Carboxen-Car-

botrap mix had greater adsorption of FA and HA thanthe Ambersorb resins with all treatments at all times(Fig. 1). It is likely that this resin mix has a lower per-centage of smaller pores than the Ambersorb resins, basedon their surface areas (Table 1). Pore diameter charac-teristics were not available for the Carboxen-Carbotrapresins. Aiken et al. (1979) deduced that extreme stericrestrictions to penetration of FA should occur in poresof < 10 nm. Less steric hindrance of these humic ma-cromolecules due to a greater percentage of larger poreswill increase intraparticle diffusion rates and result ingreater adsorption. Consequently, HA in all resin typeswas slightly less adsorbed than FA (Fig. 1 and 2) due togreater size exclusion of HA's larger mass (Table 2).Among the Ambersorb adsorbents, the pore diameterdistribution ranges from < 2 to > 50 nm, with Ambersorb572 possessing the greatest porosity volume of < 20-nmpores (according to information from Rohm and HaasCo.). Ambersorb 572 (not shown) exhibited the lowestadsorption rate, which might be the result of the resin'sgreater volume of smaller pores. A simple comparisonof relative fraction adsorbed (CJC0) of FA for the initial97 to 100 h seen in Fig. 1, and for Ambersorb 563 and572 (not shown), indicated that adsorption decreasedCarboxen-Carbotrap > Ambersorb 563 = Ambersorb564 > Ambersorb 572.

One observation with the use of these resin capsuleswas that there was essentially no change in the relativefraction adsorption rate, (Cs/C0)/d(time), due to a dou-bling of initial FA concentration (Fig. 1). This supportsour belief that film diffusion was not rate limiting, sinceunder film diffusion limitations an approximate doublingof initial FA concentration should result in a proportionalreduction in CJC0, which did not occur. Rather, for theinitial 97 to 100 h of adsorption, a doubling of initialconcentration (C0) resulted in a nearly constant CJC0.This suggests that more than sufficient adsorption siteswere available. A constant CJC0 can be useful in soil-water testing, analogous to a constant partition coeffi-cient used to quantify adsorbate masses between twophases. These spherical resin capsules contained a rela-tively large adsorbent mass (e.g., 2.1 g for the Car-boxen-Carbotrap mix). In none of the resin types didthe mass of FA or HA adsorbed (based on C) exceed0.2% of the adsorbent mass, indicating that the capsuleload capacity was never exceeded. Many carbonaceousadsorbents (e.g., activated charcoal) have saturation ca-pacities of > 10% of adsorbent mass for adsorbate mass.

AdsorptionCapsules composed of the Carboxen-Carbotrap mix-

ture, as well as some with Amberlite XAD 8, were usedsolely in these adsorption studies. A lack of sufficientquantities of the Ambersorb resins during the time ofthese experiments restricted their use for these particularstudies. Present commercial availability and lower costof the Ambersorb resins resulted in greater emphasis onthese resins in direct soil studies.

The constant partition adsorption isotherm was appliedto the adsorption of FA and HA on Carboxen-Carbotrapcapsules at different temperatures (Fig. 3). Regressionlines were highly significant (r2 > 0.96, P< 0.01). Theslopes of the regressions are the partition coefficients

JOHNS ET AL.: SOIL ACIDS CHARACTERIZE CARBONACEOUS ADSORBENTS 1489

2600 T

30 40C,, (mg C/L)

Fig. 3. Adsorption isotherms using Carboxen-Carbotrap resincapsules and fulvic acid (FA) at four temperatures and humicacid (HA) at 23 °C.

normalized to organic (e.g., resin) C, ATOC. Less adsorp-tion of HA at 23 °C occurred than for FA (23 °C) acrossthe isotherm range. A 96-h reaction time was assignedas expression of adsorbed-state concentrations (localequilibrium assumption). This was done to allow esti-mation of adsorbent capsule properties because porousadsorbents of large surface areas display indefinite ad-sorption (Karickhoff, 1985). All isotherms can be de-scribed as indicating linear adsorption (constant slopes)within the concentration ranges used. This condition re-sults from the presence of an excess of relatively ho-mogeneous adsorption sites (Dragun, 1988). Based oninformation from previous studies (Betz et al., 1989;Hazard et al., 1991), these adsorption surfaces bind bynonspecific interactions (i.e., dispersion forces). Thisprobably differs from activated C, which possesses amore heterogeneous surface of numerous functional groupsand, in many cases, a curvilinear adsorption isotherm(Suffet and McGuire, 1980). The adsorption isothermsshown in Fig. 3 were at pH 7.0. In comparison, the XAD8 resin at pH 6.3 showed no adsorption for either FA orHA at lower concentrations (Fig. 4). A pH of 6.3 isbeyond the operational pH range (pH 2.0) reported forXAD 8 resins (Aiken et al., 1979). This result suggeststhat carbonaceous adsorbents are more useful for ad-sorbing humic compounds at neutral pHs in soils andwaters, while XAD 8 is not. Since >50% of FA wasadsorbed (Fig. 1), both the hydrophobic (48%) and hy-drophilic (52%) fractions were being adsorbed to someextent.

Increasing temperature increased adsorption of FA andincreased intraparticle diffusivities (Fig. 3). For this tem-perature range (13-50 °C) about a 10% increase in dif-fusivity would probably be expected (Dragun, 1988).More than a 50% increase in FA adsorption occurred forthis temperature range, seen by the ATocs in Fig. 3. Thismight suggest some increase in adsorption with increas-ing temperature. These isotherm observations were notsufficiently precise, nor was the choice of polydispersehumic adsorbates, to distinguish between influences ondiffusion and on Kx. One notable example for increasedKx with increasing temperature is with adsorption oflarge macromolecules on polymer surfaces, where ad-sorption is entropy driven (i.e., hydrophobic adsorption)

40 100 120 14060 80Ce, (mg C/L)

Fig. 4. Adsorption isotherms using Amberlite XAD 8 resincapsules and fulvic and humic acid.

coupled with adsorbent surface-related dispersion forces(Adamson, 1990).

DesorptionThe fact that initial solution concentrations of FA and

HA can be predicted from resin-extractable C (Fig. 5)indicates a usefulness for carbonaceous resin capsules inwaters containing dissolved humics. The correlation ofinitial solution C with resin-extractable C (2 M NaOHstripping) was highly significant (r2 = 0.99, P < 0.001)for both acids. It suggests a composite constant parti-tioning for each FA and HA fraction. Constant parti-tioning may apply to single adsorbates as well. A muchearlier study (not shown) using citric and ascorbic acids,at pH 2.0 to prevent deprotonization, indicated an iden-tical predictive capability.

Though resin adsorption and extraction were success-ful in the mentioned predictions, further studies are re-quired for achieving higher recovery of adsorbates fromsolution. Use of 0.1 M NaOH as the stripping solventresulted in an average recovery of 25.5% for adsorbedFA (CV = 45.2%) and 37.9% for HA (CV = 27.9%).

160

140 o Fulvlo Acid X Hum 0 Aold

0 100 200 300 400 600 600 700 800 900 1000Resin Extracted Carbon (ug C)

Fig. 5. Predicting initial concentrations of fulvic and humicacid from Carboxen-Carbotrap resin capsules (mean valuesof extracted C).

1490 SOIL SCI. SOC. AM. J., VOL. 57, NOVEMBER-DECEMBER 1993

The use of 2 M NaOH resulted in an average recoveryof 24.2% for adsorbed FA (CV = 17.9%) and 21.7%for HA (CV = 22.0%). The extraction times for 0.1 MNaOH were double (2 h) those used for 2 M NaOH.Recovery was similar with both 0.1 and 2 M NaOH, butCVs were lower with the latter. Thus, the more concen-trated solution to date appears to be a better choice.Methanol is recommended by the resin supplier, Su-pelco, while for volatiles and semivolatiles (e.g., gasadsorption), thermal desorption is recommended. Meth-anol is not suitable for numerous OC quantifications,however.

Desorption from porous C adsorbents often results inincomplete recovery of the adsorbate (Adamson, 1990).Extreme tortuosity, steric restriction, and multiple sur-face attraction can prevent movement of adsorbate outof small pores even though the adsorbate is not irrevers-ibly, chemically bound to the surface. This hysteresiseffect is probably significant, considering that FA andHA molecules are large. For example, the study usingCarboxen-Carbotrap resins adsorbed with citric or as-corbic acids resulted in higher recovery (2 M NaOH) of54.5% (CV = 5.8%) for citric and 37.9% (CV = 5.8%)for ascorbic acid.

ACKNOWLEDGMENTSMajor support for this research was provided by a loan from

the Montana Science and Technology Alliance, Helena, MT.Resin capsule manufacturing technology is licensed to UNI-BEST, Bozeman, MT (E.O. Skogley, pres.).