4922r Langmuir 1999, 15, 4922-4926

Proton Nuclear Magnetic Resonance Study ofWater in Flocs

A. Fan, P. Somasundaran,. and N. J. Turro

Langmuir Center for Colloids and Interfaces, ColumbiaUniversity, New York, New York 10027

Received JGnuaIj' 15, 1998. In Final Form: April 7, 1999

IntroductionPolymer/surfactant adsorption can lead to colloidal

stability or particle flocculation, both of which are ofimportance in many industrial processes. Floc structureis one of the most important features of particle flocculationand this involves flOC-size, density and strength.1~Spectroscopic techniques to investigate floc structure havebeen mainly limited to small-angle neutron scattering(SANS)7-t and CAT scan.10 SANS is based on the as-sumption that the observed interference pattern of thescattered radiation is determined by the distribution ofseparations between particle centers in the flOCS.6 CAT isa special X-ray instrument that can produce three-dimensional images of an object. Although designed formedical purposes, the CAT has been effectively utilizedto study sedimentation and flocculation systems.1o Inaddition to SANS and CAT scan, computer simulationhas also been used to characterize floc structure.U-16 Thepurpose of this work is to show the potential of using protonnuclear magnetic resonance (NMR) to study water in flocs,and thereby obtain information on floc structure byanalysis of NMR line shapes, intensities, and signalpositions.

Proton NMR of water has been applied extensively tothe study of solid-solute interactions such as water-glass,16-18 water-clay,19,20 protein solution,21 biological

tissues,22 water-hydrated alumina,23 and water-silicapowder .24 A marked decrease of the proton longitudinaland transverse relaxation times, T1 and T2' as well as anincrease in the T11T2 ratio has been observed comparedto the values for pure or "bulk- water. The results wereinterpreted in terms of layers of water molecules withrestricted mobility in the vicinity of an interface. Watermolecules are dynamically exchanged between an envi-ronment in which they relax slowly (free-water phase)and one in which they relax rapidly (bound-water phaseat or near the interface). Under conditions of rapidexchange on the NMR time scale, a single relaxation rateis observed and is the weighted average of the individualrate for each environment.25 Accordingly, the signal onthe spectrum is broader for a higher ratio of bound to freewater.

When the exchange of molecules between sites corre-sponding to different resonances is slow on the NMR timescale, multiple peaks can be observed. Two signals werereported for ion-exchange resin -water26 and silica-watersystems.27 One signal was attributed to bulk water, whilethe other signal corresponds to water inside the resin orsilica pores. Similar phenomena were observed forprotein -water28.29 and carbosil-water30 systems and aseries of gels and macroreticular ion-exchange resins inwater .31.32

The difference in the chemical shifts of the water signalshas been discussed in terms of the number of hydrogenbondings for oxide particles (silica systems) and ion-hydration effect (resins). However, this concept waschallenged when multiple signals were found for systemswithout either donor-acceptor interactions or the ion-hydration effect. The chemical shift difference for watersignals was proposed to result from the contact of waterwith solids.33 It was also found that signals with differentchemical shift. values resulted when different pore sizeswere employed.

In our investigation of alumina and zeolite tlocs, the"H-bonding giving rise to the shift" was shown to beinadequate to explain the observations. When aluminaand zeolites were studied in this investigation, in additionto the signal due to bulk water, an extra downfield-shiftedsignal was also observed. This second signal could not beadequately accounted for using the hydrogen-bondingtheory, and hence it is attributed to the bulk magnetic~

(1) GIugow, L. A CMm. Eng. Prog. 1988,86 (8), 51.(2) IOimpel, R. C.; Hoa, R. J. ColloidInt4rfacc Sci. 1988,113,121.(3) Tambo, N.; Watanabe, Y. Water Bu. 1979, 13,409.(4) MouIil, B. M.; Vuudevan, T. V. Miner. Metall. ProceN. 1989,6,

142.(5) Bache, D. H.; AI-Ami, S. H. Water Sci. Technol. 1988,21,529.(6) Dickinson, E.; ErikIeon. L. Adv. Colloid Interface Sci. 1991,34,

1.(7) WODI, K; Cabane, B.; Duple88jx. R. J. Colloid Int4rfacc Sci. 1988,

123,466.(8) Wong, K; Cabane, B.; So ma8uDdaran, P. CoUoid8 Swf. 1988,30,

356.(9) Cabane, B.; Wong, K; Wang, T. K.; lAfIIma, F.; Duple88ix, R. (22) Finch, E. D.; Homer, L. D. Biophys. J. 1974,14,907.

Coll4itl Polym. Sci. 1888,266, 101. (23) (a) Pearson. R. M. J. Catal. 19'71,23, 388. (b) Baker, B. R.;(10) Somuundaran,P.; Huaq, Y. B.; Gt'Yte, C. C.PowderTechnol. Pe8rIOD. R. M. J. CGIol. 1974,33,266.

IN'l, 63, 73. (24) Hanua, F.; Gillia, P. J. Magra. Buon. 11Sf, 59, 437.(11) Di~~. E. CoUoW Brut: lID, 39, 143. (25) Zimmerman, J. R.; Brittin. W. E. J. Ph,YI. CIacm.ll1'7, 61,1328.(12) Dicki_. E. J. Colloid Interface 8ci. 118'7, IlB,286. (26) ~ J. E. J. Ph,.. Chem. 1881, 66, 1160.(13) ADIen. G. C.; DickiD8OD. E. Farodoy Di«UI.. Chem. Soc. 118'7, (27) Cli6ord. J.; Petbjca, B. A H~&nd«l Solvent Syltem6;

83,187; J: CMm. Ph,.. 1188, 86,4079; Ph,.. &V. 118'7, 3M, 2349. Taylor" Francia: LoDdon. 1968; P 169.(14)NaptaDi. T. Ph,.. &V. lID, 4OA, 7286. (28)BeNDdlen.H.J.C.;~C.Ann.N. Y:Amd.. Sci.l_,(15) Y~, Y. B.; Somaaundaran, P. Ph,YI. Rell. 118'7, 3M, 4518. 126,366.(16) Gl88el. J. A; r-, K. H. J. Am. Chem. Soc. 1974, 96, 970. (29) Cli6ord. J.; Sheard, B. Biopol,men 1986, 4, 1057.(17) Fung, B. M.; McGaughy, T. W. J. Magra. Re8On.. INl, 43, 316. (30) Turov, V. V.; Leboda, R.; BOIillo, V. I.; Skubi8zew8kA-Zieba, J.(18) Hirama, Y.;Tlklh8.hi, T.;Hino,M.;Sato, T.J. ColloidInt4rfacc Langmuir 1897,13,1237.

Sci. 1998, 184, 349. (31) Fr8Dk8l. L. S. J. Phy.. Chem. 1971, 75, 1211.(19) Woamer, D. E. J. Ma.-n. Buon. 1980,39, 297. (32) (a) Creekmore, R. W.; Reilly, C. N. Anal. Chem. 1970,42,570;(20) Delvill.. A; Letellier, M.l.a1I6muiJ' 1_,11,1361. AntJ. CJaem.IMO, n, 725.(21)~, S. H.; Brymt, R. 0.; Halleaga, K.; Jacob, G. S. (33) Draao, R. S.; Ferris, D. C.; Burns, D. S. J. Am. CMm. Soc. 1996,

~m~ 1978, 17,4348. 117,6914.

lO.lO2l/1a980068g CCC: $18.00 C 1999 American Chemical SocietyPubJiahed on Web 06118/1999

Langmuir, Vol. 15, No. 14, 1999 4923Notes

susceptibility (BMS)34.36 effect. The basic idea in the BMStheory is that a water molecule inside and outside acompartment will possess two different chemical shiftvalues. This effect was found to be important in an in vivoNMR imaging study of biological tissues33,M and may beuseful for explaining what was observed in this study.According to BMS theory, the downfield shift results fromwater trapped in small compartments: pores in theparticles or microchannels within the flocs depending onwhat kind of particles are used.

0 i, toc water (water trapped infoca). i'*r-toc weier (bioi< water between ftoca)

Inter-floc water 8~1sion

I ntra-#k)c water e~kIn

Experimental Section

Materials. Alumina powder (Praxair Surface Technologies)used had an average size of 0.3 /lm and a BET surface area ofapproximately 15 m2/g, and they aggregate into 2o../lDl microflocain water as measured by Microtrak (Leeds &:. Northrup Instru-ments). Zeolite particles of different crystal sizes were synthe-sized" and had a pore size on the order of 5 A and a cylindricalpore structure; 10 and 2Q-/lDl zeolite particles possess a SiiAlratio of80/20 (16 m2/g), and 100-/lm particles of silicalite (5 m2/g)were employed. Poly(lCJ'Ylic acid) (PAA) of molecular weight10 000 was obtained from Polysciences Inc. Cationic copolymerPercol (Allied Colloids Inc.) was a copolymer of acrylamide anddimethyl aminoethyl acrylate of equal ttaction, the latter beingfully quaternized and therefore positively charged over a widepH range. All the samples were prepared with 0.1 N NaCIsolutions. Triply distilled water was used throughout theexperiment.

Flocculation Test. The method for dual polymer flocculationwas the same as that described before.at The supernatantturbidity was read with a turbidimeter (HF scientific Inc.), andthe sediment was transferred to a cylindrical volumetric tube forthe sediment volume measurement.

NMR Meuurements. Flocs produced from flocculation testswere carefully transferred into a standard 5-mm NMR tube andallowed to settle for a certain amount of time (the typical timescale is a few hours of settling time). The tube was filled so thatthe solid filled the volume of the probe coil. The 1 H NMR spectrumwas obtained using a VXR V arian-400- MHz NMR spectrometer.Spectra were obtained at 25 °C. In all spectra, 64 scans wereacquired without spinning. For all sets of measurements, flocpreparation and NMR measurement were made on a paralleltime scale, which ensured the same settling time for flocs in eachsample.

Figure 1. Schematic representation of int1'at1oc water andinterfloc water and their explusion processes.

Results and Discussion1. NMR spectra of Native Floes. With many of the

samples investigated, separate resonances for the waterprotons were observed for alumina-water and zeolite-water systems, indicating a ,low exchange of moleculesbetween the sites corresponding to these resonances onthe NMR time scale. These separate signals were at-tributed to water in microchannels.(trapped water) andmacrochannels (bulk water) of the flocs as shown in Figure1. In the case of highly porous zeolite, the porous internalsurface within individual particles forms microchannelsin which the water is adsOrbed; while for alumina. themicrochannA~ exist within tJ1e alumina aggregates formedby numerous primary alumina particles. In contrast toprevious NMR reports on colloidal systems describedabove. our research involves in situ study of natural flocs

where the solid-liquid ratio is not quantitatively ma-nipulated. It was proposed37 that two processes of waterremoval can occur with time: relatively fast expulsion ofwater in the macrochannels (bulk water removal) andrelatively slow expulsion of water from within the flocs(trapped water removal). The former is an interfloc processwhile the latter is an intrafloc one (Figure 1). Therefore,when ratios of different water signals are examined, thevalues are for different settling times of the flocs becausewater is extruded as the floc denaifies. Only flocs settledfor the same time can be compared with each other. Thetime effect for a zeolite-water system is shown in Figure2, where the peak for trapped water increases with timeas the zeolite floc settles and denaifies due to bulk waterexpulsion.

Common floccu1ants are of three types: salt, polymer,and surfactant. Salt and small polymer molecules areexpected to induce flocculation mainly by charge neu-tralization, which prod~ very compact and dense flocs.In contrast, bridging is considered to be the mechanismfor large polymer-induced flocculation, and the resul~flocs are much bigger and less dense than those producedby salt or small polymers. Surfactants are less commonlyused as flocculants. Surfactants induce flocculation byimparting hydrophobicity to the particles when thesurfactant species orients with its hydrocarbon moietytoward the aqueous phase.38 Examples of the proton NMRof a compact and a loose floc system are shown in Figure3. On one hand, since the fraction of trapped water isrelated to the floc density, loose flocs produced by theaddition of small amounts of P AA contain an amount of

(34) Chu, S. C.-K.; Xu, Y.; Balacbi, J. A; Springer, C. S., Jr. Ma,.n.&soIL in Medicine; Academic Press: San Diego, CA, 1880; Vol. 13, p239.

(35) SPrlDler, C. S., Jr. In NMR in Ph)'.~ and Biomedicine;Academic Pr8s: San Diego, CA. 1994; Chapter 5.

(36) Zhao, D.; Qiu, S.; Pang, W. z.oliteI1tt8, 13,478.(37) Harris, C. C.; Somaaundaran, P.; JeD8eD, R. a; Powder Technol.

19'11, 11, 75.(38) Somaaundaran, P.; Healy, T. W.; FuerlteDau, D. W. J. Colloid

l11terfa« Sci. 1N8, 22, 599.(39) Somaaundaran,P.;Fuentenau,D. W.J. Ph,... Chem.1986, 70,

00.

Notes4924 Langmuir, Vol. 15, No. 14,1999

I!lOOJl I \./. Silcalte

\,.-."'..;.../ "--

20J'

/ZSM-S (Si:Al- 80:20)r '

~

~~

/ \.-J '--- ~--~ . , , . b ' . . , ~ -~ ' , . . b ' . , , ~

Chemical Shift (ppm) Chemical Shift (ppm)

Figure 2. Time effect on NMR peak ratio.

Alumln. Zeolite

PM

f\II \

\./,.-~ "' ,.

-~ I I I I b 'I I , ~

Cbealical Shift (ppm)

Figure 3. Floca produced by three kinds of flocculanta: salt.polymer, and surfactant.

trapped water that is 80 small that only one bulk waterpeak is observed; on the other hand, the addition ofappropriate amounts of sodium dodecyl sulfate (SDS)enh~~ the floc packing density.

For a fixed time scale, the particle size can be expectedto affect the packing. Figure 4 shows the NMR spectra forzeolite-water systems: with particle sizes varying by afactorofl0. It can be seen that the peak ratio of the trappedwater to bulk water increases with decreasing particlesize, indicating a denser packing of smaller particles anda greater fraction of trapped water for two ~ -5 zeolitesdiffering in particle size but not in pore size or polarity.The three kinds of zeolite particles possess the sameinternal pore size and pore structure, thus, in agreementwith BMB theory I result in the same frequency shift forthe trapped water. It is important to note that thecomposition of the l00-/lm silicalite particles (Si:Al > 1000)is different from those of the other two (Si:Al = 80:20).N everthelesa, the downtield shifts of water signals are

very similar for all of the zeolite samples. This againindicates a compartmental effect wtead of donor-acceptor interaction with the ~cle surface.

Packing density can be changed also with temperature.This effect can be seen from an examination of Figure 5.For hydrophilic alumina ~cles, the trapped water peakdecreases markedly with an increase in temperature,which means that the packing is less dense at highertemperature. For 10-fm1 zeolite (Si:Al = 80:20) which ismuch less hydrophilic than alumina, however, the tem-perature effect is quite different. The trapped water peak,instead of decreasing without changing its originalreso~ frequency, moves toward the bulk water peakas temperature is increased. This is associated with thefact that water molecules are expelled from the smallhydrophobic pores as the temperature is raised, and/orexcPange between water inside and outside the poresbegins to take place.

2. SurfactaDt-lnduced Floes. The adsorption ofsodium dodecyl sulfate (SDS) on alwnina has been well-studied, with an adsorption isotherm which can be dividedinto four regions- (Figure 6). The surfactant adsorbsnoncooperatively and electrostatically as individual ionsin region I and associate into solloids (hemimicelles) inregion n of the isotherm. In the solloids, the surfadantsare oriented with their charged headgroups toward the

Notes Langmuir, Vol. 15, No. 14, 1999 4925

I'"/ ,

solid surface, while the hydrocarbon chains protrude intothe aqueous phase, thus forming hydrophobic patches onthe surface. Further adsorption results in an increasingnumber of surfactant aggregates, with surfactant mol-ecules adsorbing with an opposite orientation once thesurface is neutralized by the oppositely charged surfactant.Finally, in the plateau region, region IV, the adsorptionlayer possesses the structure of a bilayer with the adsorbedsurfactant presenting the external surface with a changedstructure.

Figure 6 shows the water signals measured for aluminaflocs with different coverages ofSDS. The flocculation datacould be correlated with the NMR spectra. The SDScoverage is illustrated by indicating its position on thecorresponding adsorption isotherm. As described above,the hydrophobicity of alumina increases with SDS ad-sorption in adsorption regions I and n and starts todecrease from region ill until it reaches a constant valuein region IV where adsorption is saturated. Therefore,particles in sample c are most hydrophobic and have thestrongest tendency forfl<K:culation (Figure 6). Accordingly,the separation between the two peaks varies and showsa clear trend. The separation for sample c is the largestand decreases toward the two ends of the adsorption

isotherm. Thia change in peak separation indicates achange of the compartmenta18tn1cture in the tlocs. A moredownfield-shifted peak denotes better screened and smallerCompartments for trapped water. As the compartmental8tn1cture becomes bigger and more open, the two peaksstart to approach each other.

The more hydrophobic the particles are, the strongercan be their tendency to aggregate in water. However, itis interesting to note from the NMR spectra that tlocsresulted from the more hydrophobic particles have aslightly higher ratio of free/trapped water, which is alsoin agreement with their higher specific sediment volume(Figure 6). In other words, although better tl.occu1ationcan be achieved with more hydrophobic particles, lessoverall packing density is achieved. This is attributed tothe stronger attraction between hydrophobic particlescausing a larger viscous force working against thegravitational driving force of better packing of tlocs.InWtloc water (bulk water) expulsion is therefore retardedafter ftocs settle down.

3. Polymer-Flocculated Particlee. Polymers are thecommonly used ftocculants. We have studied dual polymerflocculation with the system Percol-poly(acrylic acid)-

4926 Langmuir, Vol. 15, No. 14, 1999 Notes

A structure of polymer-bridging-induced tlocs, the ratio ofthe trapped-to-bulk water signal is much smaller thanthat for particles with an adsorbed surfactant and bareparticles. The amount of trapped water becomes smallerwith time and eventually disappears, with the signalintensity for bulk water staying the same. The disap-pearance of the trapped water suggests intratloc waterexpulsion accompanying polymer rearrangement. Thisphenomenon occurs more readily with polymer-inducedloose tlocs where the bulk water removal is hindered bythe presence of the polymer.

7 dayS

- - ; / ~

AI 1'-'

I.,P ,' J

~-/

1\

10099

at 98. 81

8885M83828180 ,.

0 1 2 3 4 5 8i--~ ' "'-- *"-. day

-~ . . . . b ' , . . ~

Cheallcal Sbift (ppm)

Figure '1. Time effect for flocs induced by dual polymers.(PAA)-alumina.40 Fot1Iti8 study, 5 ppm of the cationicpolymer (Percol) and 5 ppm of the anionic polymer (P AA)were uaed to produce very good flocculation. In tJ1is system,P AA is first adsorbed on alumina and served as the anchorfor Percol which bridges the ~cles together.

NMR spectra were taken at different times after theflocculation. As shown in Figure 7, because of the loose

\I ConclusionThe feasibility of using conventional proton NMR

spectroscopy for studying floc structure and floc sedi-mentation is demonstrated. It provides a quick andstraightforward method to follow the interfloc and intraflocwater expulsion processes and to examine the floc packingand structure. Some new information was gained onsurfaCtant- and polymer-induced flocs.

Acknowledgment. The authors thank the Environ-mental Protection Agency (EPA R823301-QI-Q) and theNational Science Foundation for the financial support.They also thank Wei Li in Turro's group for the synthesisof the zeolite samples.

(40) Fan. A.; Turro N. J.' SomasundaraD. P. Submitted to CoUoid8Surf; 1917. .. LA98OO68G