Embed Size (px)
Citation preview
Cm
RWL
h
•
•
•
•
a
ARRAA
KCSMPSS
1
f
0h
Colloids and Surfaces A: Physicochem. Eng. Aspects 447 (2014) 32–43
Contents lists available at ScienceDirect
Colloids and Surfaces A: Physicochemical andEngineering Aspects
j ourna l h om epa ge: www.elsev ier .com/ locate /co lsur fa
harged micelle halo mechanism for agglomeration reduction inetal oxide particle based polishing slurries
ebecca Dylla-Spears ∗, Lana Wong, Philip E. Miller, Michael D. Feit,illiam Steele, Tayyab Suratwala
awrence Livermore National Laboratory, P.O. Box 808, Livermore, CA 94551, USA
i g h l i g h t s
Agglomeration reduction is demon-strated in metal oxide polishingslurries.Stabilization substantially maintainsmaterial removal rate during polish-ing.Stabilization improves filterabilityand reduces roughness of the pol-ished surfaces.Formation of charged micelles isshown to be critical to agglomerationreduction.
g r a p h i c a l a b s t r a c t
r t i c l e i n f o
rticle history:eceived 24 October 2013eceived in revised form 8 January 2014ccepted 17 January 2014vailable online 30 January 2014
eywords:olloidtabilizationicelle
olishing
a b s t r a c t
A method for chemically stabilizing metal oxide polishing slurries to prevent their agglomeration whilemaintaining their surface activity is demonstrated experimentally. Negatively charged ceria, zirconia, andalumina particles are reversibly size-stabilized under low ionic strength conditions at and above theirisoelectric points using anionic surfactants.
Stability is imparted only at surfactant concentrations above the critical micelle concentration andwhen the particle and the micelle have like-signed charges. Zeta potential measurements demonstratethat little adsorption of anionic surfactant occurs under conditions where the particles are negativelycharged. Changes to pH, hydrophobicity, and ionic strength disrupt the surfactant’s ability to size-stabilizethe slurries. These results suggest that the charged micelles electrosterically hinder the agglomerationof oxide particles.
lurryurfactant
Because the stabilization method does not rely on adsorption, the particle surface remains accessiblefor chemical reactions, such as those involved in polishing. Metal oxide slurries stabilized by this methodremove material at a rate comparable to that of unstabilized slurry. In addition, stabilized slurry is easierto filter, which improves the quality of the polished surface. Stabilizing colloids by this method may provevaluable for systems where particle surface functionality is important, such as those used in ceramics
ing, a
processing, optical polish. Introduction
Optimizing the size uniformity of polishing slurries is importantor the optical polishing and chemical-mechanical planarization
∗ Corresponding author. Tel.: +1 925 422 1700; fax: +1 925 423 0792.E-mail address: [email protected] (R. Dylla-Spears).
927-7757/$ – see front matter © 2014 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.colsurfa.2014.01.061
nd chemical-mechanical planarization.© 2014 Elsevier B.V. All rights reserved.
(CMP) industries from the perspective of both reducing rawmaterial consumption and improving polishing outcomes. Agglom-eration of colloidal slurry particles leads to settling, increased slurryconsumption, and reduced filter lifetimes. In addition, polishing in
the presence of large or agglomerated particles at the upper end ofthe size distribution has been shown to degrade the quality of thefinished surface, producing microscratches and increasing surfaceroughness [1–7].
A: Ph
iscwcas
sIaC(apeerfao
smpfebei
ttihcba
aSeafffhsthhtmsAmttiaPpct[
(Saturn Digisizer, Micromeritics) and (2) single particle light obscu-ration (Accusizer 780A, Particle Sizing Systems). Test samples were
R. Dylla-Spears et al. / Colloids and Surfaces
Ceria and zirconia slurries are commonly used for optical polish-ng, as they are two of the most effective polishing compounds forilica based glasses in terms of material removal [8,9]. Commer-ial optical polishing slurries are typically comprised of particlesith average diameters of 0.1–0.5 �m and are used at working con-
entrations of 2–10 wt%. Unfortunately, particle agglomeration is common problem for colloidal slurries, as stability is extremelyensitive to solution conditions [4,10–13].
Agglomeration of colloids is typically prevented using electro-tatic or steric stabilization, either separately or in combination.n electrostatic stabilization, the surface charge of the colloids isdjusted to increase the electrostatic repulsion between particles.eria and zirconia have neutral charge at their isoelectric pointsIEP) of pH 6.6 and 6.2, respectively. Therefore, working particlesre negatively charged at typical optical polishing conditions ofH 7–10 [14]. However, metal oxide slurries have been shown toxhibit shear thinning behavior at pH > IEP, suggesting that agglom-ration is likely even under conditions where the electrostaticepulsion between particles is high, as inter-particle attractiveorces continue to dominate [15]. Thus, electrostatic stabilizationlone is not sufficient to prevent particle agglomeration in metalxide slurries.
Efforts have been made to improve dispersion of metal oxidelurries for CMP and ceramics applications using adsorbed poly-ers, which provide steric stabilization. For example, anionic
olyelectrolytes, such as polyacrylic acid, induce steric repulsiveorces while increasing the surface charge and have been shown tonhance the dispersion of ceria and zirconia [13,16,17]. Care muste taken to properly adjust surface coverage, as low levels of poly-lectrolyte adsorption can also lead to bridging interactions, whichnstead de-stabilize the slurry and cause flocculation [17,18].
Chemical dispersants and slurry additives that tightly bind tohe surface or require high fractional coverage have the potentialo decrease polishing activity [19,20]. Material removal in polishings facilitated by a chemical reaction between the metal oxide andydroxyl groups on the silica surface [8]. Therefore, the degree ofoverage or size of species needed to stabilize the particles maylock chemically active sites on the surface of the polishing particlesnd reduce the material removal rate.
Smaller molecules, such as surfactants, can also preventgglomeration through either steric or electrosteric stabilization.urfactant adsorption on metal oxide surfaces has received consid-rable attention given its applications in froth flotation, detergency,nd separations. Studies of surfactant adsorption have typicallyocused on conditions where there is a clear electrostatic drivingorce for adsorption; i.e., where surfactants have opposite chargerom the metal oxide surface [21–23]. In this regime, researchersave provided rheological and electrokinetic evidence that chargedurfactants adsorb to the surface and reduce inter-particle attrac-ion in metal oxide dispersions. For example, Wei and coworkersave recently demonstrated that the anionic surfactants sodiumexametaphosphate and sodium dodecyl benzene sulfonate reducehe viscosity of nanoceria suspensions [11]. In addition, reduction in
aximum shear yield stress has been achieved by adding chargedurfactants to concentrated ceria and zirconia suspensions [24,25].lthough these experiments have focused over a range of pH, theost substantial rheological effects were observed at a pH where
he particles are either charge neutral or are of opposite charge tohe surfactant. These results all suggest that the charged surfactantsncrease inter-particle repulsive forces, but the studies referencedbove provide no direct evidence of a reduction in agglomeration.alla and Shah used settling experiments to suggest a reduction in
article agglomeration in alumina suspensions with the addition ofharged surfactants; however, their study did not address the prac-ical implications of using such surfactant stabilized suspensions12].ysicochem. Eng. Aspects 447 (2014) 32–43 33
In the present work, we directly demonstrate reduction ofagglomeration in metal oxide particle based polishing slurriesimparted by the use of anionic surfactants under low ionic strengthconditions and where the bare particles are negatively charged(i.e., pH > IEP). We discuss conditions under which colloidal sta-bility is achieved and conditions that reverse this stability. Weaddress practical processing concerns associated with addition of achemical stabilizer, such as maintaining material removal and fil-terability, and discuss the benefits of using these stabilized slurriesfor optical finishing. Finally, we postulate a mechanism which isconsistent with our observations of metal oxide particle stabilityin the presence of charged anionic surfactants. Although demon-strated here for optical polishing slurries, the method is expectedto be broadly applicable.
2. Materials and methods
2.1. Slurry sample preparation
All experiments were conducted using commercial polishingslurries, which were used without further purification or pre-treatment (e.g. polishing or ultrasonics). Ceria slurries (UniversalPhotonics) were prepared using the optical polishing compoundsHastilite PO (supplied as concentrated slurry) or Cerox 1663 (sup-plied as powder). These were diluted with deionized (DI) water,in accordance with common manufacturing practice, to 4.4 wt%and 7.7 wt% solids, respectively. As-manufactured mean particlesizes were reported by the vendors as 0.2 �m for Hastilite POand 1-2 �m for Cerox 1663. Concentrated zirconia slurry ZOX-PG (Universal Photonics), which has a reported median particlesize of 1.0–1.4 �m, was diluted with DI water to 4.6 wt% solids.Alumina slurries were prepared using Linde 0.3-�m alumina pol-ishing powder (Union Carbide) and DI water to 10 wt% solids. Astir rod and stir plate were used to mechanically mix all dry pol-ishing compounds for a minimum of 2 h. Details of the slurries,including their measured conductivities and IEPs, are summarizedin Table 1.
Indicated concentrations of nonionic, anionic or cationic sur-factants, and other species tested were added to the aqueousslurries, prepared as described above, to test their effects on parti-cle agglomeration. A full list of these species, known critical micelleconcentrations (CMC), as well as the pH conditions and surfac-tant concentrations under which they were tested, is provided inTable 2. For experiments performed at or near pH 10 or pH 4,pH adjustment was made using KOH or HCl, respectively. Mea-surements of pH were made using an Accumet AR60 pH meterwith pH/ATC Double Junction electrode. A few of the surfactant-stabilized ceria samples were titrated with a 5 M NaCl stock solutionto determine the point when stability was lost. NaCl concentrationsfor these samples are indicated in Table 2 in the column labeled“Additional NaCl”.
2.2. Slurry characterization
The degree of particle agglomeration was quantified by directlymeasuring the particle size distributions of the slurry samples.Particle size distributions were measured using two different tech-niques: (1) laser light scattering on the ensemble of particles
pre-diluted 100-500x from their working concentrations with DIwater just prior to PSD analysis and mixed continuously with amagnetic stir bar before each sample was injected into the testchamber.
34 R. Dylla-Spears et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 447 (2014) 32–43
Table 1Properties of metal oxide colloids studied.
Slurry material Trade name Mean size (�m) Reported IEPa Measured IEP Conductivity (�S/cm) Working concentration(wt%)
CeO2 Hastilite PO 0.26.8
6.2 1000 4.4Cerox 1663 1-2 None 550 7.7
ZrO2 ZOX-PG 1–1.4 6.2 6.5 200 4.6Al O 0.3 �m Alumina 0.3 9.0 9.1 140 10.0
utisAsptpwocowoprtt
wuSomacdmt
2
stomlsowtocuumfiatwt
2 3
SiO2 Sillica Sol 0.025 2
a Ref. [14].
Particle settling (or sedimentation) is a well known techniquesed to separate particles by size, either gravimetrically or by cen-rifugation. Settling velocity is influenced by a number of factors,ncluding the difference in fluid and particle densities, particle size,hape, and roughness, as well as interactions between particles.ssessing slurry settling rates [12,13] provided a qualitative mea-ure of the degree of particle agglomeration present in a sample,ermitting comparisons between preparations to be made. Here,he undiluted sample was agitated to ensure particles were sus-ended and then immediately poured into a graduated cylinder,here particles were left to settle under gravity. The initial height
f the fluid column (typically 13–16 cm to the bottom of the menis-us) was determined visually to within 2 mm using the graduationsn the cylinder. As the bulk of the suspended particles drifts down-ard, a distinct interface is formed, with a translucent supernatant
n top and an opaque, white suspension below (cf. Figure S1). Theosition of the meniscus at the translucent-opaque interface wasecorded visually to the nearest 2 mm as it drifted downward overime and then plotted as a fraction of the initial height of slurry inhe column.
Zeta potentials and conductivities of slurries were measured,ithout dilution, using the electrokinetic sonic amplitude methodsing a ZetaFinder Potential Analyzer (Matec Applied Sciences).lurries (200 ml, undiluted) were titrated with either KOH (1 N)r HCl (1 N) to achieve the desired pH just prior to measure-ent. Concentrated aqueous solutions of either anionic surfactant
mmonium lauryl sulfate (ALS) (500 mM) or cationic surfactantetyl tetramethylammonium bromide (CTAB) (82 mM) were addedirectly to 200 ml of 4.4 wt% Hastilite PO ceria at pH 10 toeasure zeta potential as a function of surfactant concentra-
ion.
.3. Filtration
Filtration performance was investigated for both untreated andtabilized slurries with slurry flow isolated from the polisher. Fil-ration experiments were performed at higher solids loading thanther studies reported here (e.g. settling, polishing) to provide aore rigorous test of the filtration equipment and to provide more
atitude for determining mass loss and measuring changes in pres-ure drop within one week of continuous filtration. Two gallonsf 8–10 wt% Hastilite PO ceria slurry, either without surfactant orith 36 mM ALS, was circulated at ambient temperature at 3-4 gpm
hrough a 4′′ CUNO Optima CMP590 (3M Corp.) filter having a cut-ff size of 50 �m or a 4′′ CUNO Optima CMP560 filter having autoff size of 5 �m. Filters were presoaked in DI water prior tose. During the experiments, pressure was continuously monitoredpstream and downstream of the filter. Pressure drops were nor-alized by the manufacturer’s expected pressure drop across the
lter for the given flow rate, filter length, and pore size. Flow rate
nd temperature were also monitored. Slurry samples were cap-ured periodically through the filter vent port. Such samples wereeighed, dried at 80 ◦C for >8 h, and then re-weighed to determinehe solids fraction.
4 100 –
2.4. Material removal and polished surface quality
Round fused silica workpieces (100 mm diameter x 10 mmthick; Corning 7980) were polished using the Convergent Polish-ing configuration [7] (polyurethane pad [IC1000, Rohm & Haas]on a 300 mm granite lap base, 0.3 psi applied workpiece pressure,20 rpm workpiece & lap rotation rates, 0.6 ml/min slurry feed rate(single pass), and slurry Baume 9). A new polyurethane polishingpad was used for each change in polishing slurry type. No pre-treatment, such as diamond conditioning, was performed. Materialremoval rates were determined gravimetrically after polishing for8 h. Note because the slurry is only used on a single pass (as opposedto recirculation), changes to the slurries’ PSD with recirculation arenot considered in this study.
The surface morphology and roughness of selected polishingsamples were characterized using atomic force microscopy (AFM).The AFM (Digital Instrument Dimension 3100) sits on a vibrationisolation table and is operated in a temperature and air-flow con-trolled room. High aspect ratio silicon tips with small tip radius(Veeco OTESPAW) were used to ensure the accuracy of the AFMresults and to minimize any tip convolution of the shape mea-sured. Fresh AFM tips were used for each sample, and the sharpnessof each tip used in the study was characterized against a tita-nium roughness sample. A height calibration standard was usedperiodically to ensure the accuracy and reproducibility of the mea-surements. Four 50 �m × 50 �m scans with roughly ∼10 nm lateralresolution were obtained for each polishing sample at a scan rateof ∼0.15 Hz. 2D surface roughness analysis was performed usingthe raw AFM images obtained. Four measurements were used tocompute the average root mean square (RMS) roughness of eachsample.
3. Results
3.1. Effect of surfactants on settling and particle size distribution
Hastilite PO is a commonly used ceria polishing slurry, havinga reported mean particle diameter of 200 nm. When the slurryis prepared at a typical polishing concentration[9] of Baume 9(approximately 4.4 wt%) at pH 6.6, using the method recommendedby the manufacturer, it settles within ten minutes as shown bythe filled squares in Fig. 1a. Analysis of settling times for singleceria particles in water using Stoke’s law reveals that such a shortsettling time is inconsistent with 200-nm particles; instead, the set-tling time is more indicative of 10-�m particles. The rapid settlingsuggests that the slurry is agglomerating following initial manufac-ture or upon dilution. Changing the pH of the suspension above theIEP of ceria increases the electrostatic repulsion between particles;however, as the open squares in Fig. 1a show, the settling time doesnot improve when the ceria slurry is prepared at pH 10.75. Electro-
static stabilization alone is not sufficient to overcome the attractiveforces between particles.Many studies have focused on the adsorption of charged sur-factants and polymers to oppositely charged metal oxide surfaces,
R. Dylla-Spears et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 447 (2014) 32–43 35
Table 2Surfactant properties, experimental conditions, and observed results. The last column provides the interface height of the treated slurry after settling for 1 h normalized bythe interface height of slurry at the same pH and solid concentration, without addition of the indicated chemical species. Values near 1 indicate the presence of agglomeration,while values � 1 suggest reduction in particle size (improved stability), as particles remain suspended significantly longer than without treatment. Items marked with (*)exhibited multiple settling fronts during experiments; the rate for the slowest-moving interface is indicated in the table.
a Ref. [26].b Ref. [27].c Ref. [28].d Ref. [29].e Ref. [30].f CMC and structure information not provided by manufacturer.
36 R. Dylla-Spears et al. / Colloids and Surfaces A: Ph
Fig. 1. (a) Observed settling behavior of Hastilite PO ceria slurry (4.4 wt%) withpH change and surfactant addition. (b) Combined PSDs for Hastilite PO ceria slurry(4.4 wt%) at pH 8 with 36 mM ALS and without surfactant. Data for particles < 0.5 �mwoo
wpwbltroetbtcNptrs
wDpo
as obtained using laser light scattering, while data for particles > 0.5 �m wasbtained using laser light obscuration, which is more sensitive to small numbersf particles.
hich can provide a steric barrier to agglomeration [31–34]. Table 2rovides details of the additives assessed in the present study asell as a summary of their observed effects on the agglomeration
ehavior of ceria, zirconia, and alumina slurries under particu-ar solution conditions. All of the anionic and cationic surfactantsested greatly increased the settling times of the metal oxide slur-ies in particular pH ranges when compared to untreated samplesf the same slurries, indicating that they reduced particle agglom-ration. Representative settling curves are shown in Fig. 1a forhe cationic surfactant CTAB and for the anionic surfactant ALS,oth of which prevented complete settling of Hastilite PO for morehan 4 days, but only under pH conditions where the slurry parti-les were either charge-neutral or of like charge to the surfactant.onionic surfactants Triton X-100 and Tween 20 and unchargedolymers polyethene glycol (PEG-200, PEG-20K) were not effec-ive at preventing agglomeration at the concentrations tested. Aepresentative settling curve is provided in Fig. 1a for the non-ionicurfactant Triton X-100.
Measurements of the slurry particle size distributions (PSDs)
ere also made to quantify the degree of agglomeration. A Saturnigisizer II (Micromeritics), which uses the Mie solution to calculatearticle size distributions based on light scattered by an ensemblef particles, was used to measure the full distribution of particleysicochem. Eng. Aspects 447 (2014) 32–43
sizes ranging from 40 nm to 2.5 mm. To detect small changes in thelarge end of the distribution, measurements of particle size weremade using an Accusizer 780AD (Particle Sizing Systems), which ismore sensitive than light scattering techniques to small numbersof large particles. In this method, individual particles > 0.5 �m aresized by the single particle light obscuration technique, i.e., indi-vidual particles are passed through a laser beam, with the amountof light blocked corresponding to particle size. This method doesnot detect particles < 0.5 �m in diameter.
PSDs from the two techniques can be combined into a singledistribution, with the ensemble method providing data for par-ticles < 0.5 �m and the single-particle method providing data forparticles > 0.5 �m. Fig. 1b shows representative combined PSDs for4.4 wt% Hastilite PO ceria stabilized using 36 mM ALS at pH 7.5 andwithout surfactant. Note that the fractional particle count for thecombined distribution spans 9 orders of magnitude, such that par-ticles in the upper end of the distribution are present at ppb levelsrelative to the average particle. Particles larger than 0.5 �m rep-resent just 0.2–0.6% of the total number of particles. These largerparticles are considered to be agglomerates, as the slopes of thetails of the PSD of a given slurry can be reversibly shifted by alter-ing the solution conditions, and these tails correlate with observedsettling behavior.
For both the stabilized and unstabilized slurries, the majorityof particles (>99%) is smaller than 0.5 �m and the mean par-ticle size is around 0.1 �m, which is in reasonable agreementwith the manufacturer’s reported mean particle size of 0.2 �m.Brunauer–Emmett–Teller (BET) analysis of the stabilized HastilitePO yields a surface area of 7.49 m2 g−1, consistent with 0.1-�maverage-diameter particles. Several important differences, how-ever, are observed in the upper ends of the distributions that wouldbe missed using data from only the light-scattering method. First,the size range of agglomerated particles is reduced from 1–10 �mto 1–4 �m upon addition of ALS. Second, the size of the averageagglomerate is reduced from 2 �m to 1 �m with addition of thesurfactant, suggesting that agglomerates are very large, comprisedof 103 to 104 average-sized particles. Third, using the surfactantreduces the number of average-sized agglomerates 100 fold, from1 agglomerate in 104 particles to 1 in 106 particles. Thus, the addi-tion of ALS to the Hastilite PO ceria slurry reduces both the numberand average size of agglomerates.
The effected stabilization evident from the combined PSDs inFig. 1b is not restricted to this particular surfactant and slurry pair.Fig. 2a shows a reduction of the number of large particles > 0.5 �min Hastilite PO ceria upon the addition of the anionic surfactantsALS, Triton H-66, and dodecylbenzene sulfonate (DBS) as wellas the cationic surfactant CTAB. Solution conditions for the sam-ples are described in Table 2. These data suggest that significantimprovements can be made to the upper end of the PSD for neutralto negatively charged ceria using negatively charged surfactantshaving either phosphate, sulfate, or sulfonate head groups. Thereduction in large particles can also be achieved for neutral to posi-tively charged ceria using positively charged surfactants having anammonium head group.
Anionic surfactants were also effective at stabilizing slurriesother than Hastilite PO ceria at pH ≥ IEP. Fig. 2b and c shows thePSDs for particles > 0.5 �m for zirconia (ZOX-PG, Universal Photon-ics) and alumina (Union Carbide) slurries respectively, both withand without 36 mM ALS. A second ceria slurry (Cerox 1663) wasalso successfully stabilized with 36 mM ALS (Figure S2). The num-ber of larger particles was reduced with the addition of ALS for allfour metal oxide particle based slurries tested. Limited conclusions
should be drawn from the data in Fig. 2, as the solution conditionsare not uniform across these samples and no attempt was made tooptimize the conditions for a particular slurry-surfactant system.Despite this, the results described here suggest that the presence
R. Dylla-Spears et al. / Colloids and Surfaces A: Ph
Fig. 2. PSDs for particles > 0.5 �m for three different metal oxide slurries in theabsence (squares) and presence of various surfactants: (a) ceria (4.4 wt% HastilitePO), (b) zirconia (4.6 wt% ZOX-PG), and (c) alumina (10 wt%).
ysicochem. Eng. Aspects 447 (2014) 32–43 37
of charged surfactants, at the indicated concentrations and pH con-ditions, is sufficient to overcome inter-particle attractive forces andreduce the agglomeration of ceria, zirconia, and alumina particles.
Measured IEP and conductivities at working concentrations ofthe colloids used in this study are provided in Table 1. As the com-positions of commercial polishing slurries are typically proprietary,inclusion of such detail is not practical. However, except for Cerox1663, the close agreement of slurry IEP with reported literaturevalues suggests a low degree of chemisorption of any unreportedorganic additives. We have also measured pHs and confirmed inde-pendently that the ionic strength of each slurry is less than ∼25 mMNaCl at working concentrations, performed inductively coupledplasma mass spectrometry (ICP-MS) impurity analyses to verifycompositions, and discussed with vendors the potential presenceof impurities and organic additives in the samples provided as slur-ries. In addition, the main results discussed are common among theslurries prepared from either powders (alumina and Cerox 1663ceria) or slurries (Hastilite PO ceria and zirconia), and among dif-ferent manufacturers, suggesting that the behavior seen here isgeneral, rather than being specific to a particular slurry compo-sition.
3.2. Effect of surfactant on material removal
A second important practical consideration for polishing appli-cations is that the dispersant must not interfere with the surfacechemistry such that it reduces the slurry’s polishing efficacy.Table 3 summarizes the material removal rates (MRR) from pol-ishing experiments in which fused silica substrates were polishedboth with unstabilized slurries and with slurries stabilized usingseveral surfactants at various pH. MRR for the unstabilized slur-ries vary with pH and are measured to be 0.58–0.85 �m/h forHastilite PO ceria and 0.55 �m/h for ZOX-PG zirconia. Note thatboth the ceria and zirconia slurries containing anionic surfactantshave removal rates that are greater than 70% of the removal rates oftheir unstabilized equivalents across the pH range where agglomer-ation reduction is observed (pH ≥ IEP). Therefore, all of the anionicsurfactants tested appear to both reduce agglomeration and sub-stantially maintain MRR under the conditions tested.
In contrast, MRR for the cationic surfactant CTAB are lessthan 20% of that without added surfactant across the pH rangewhere agglomeration reduction was observed (pH ≤ IEP). Thissuggests that the cationic surfactant interferes with the conden-sation/hydrolysis reactions required for chemical removal of silica,which could occur if a layer of adsorbed surfactant were to coateither the ceria particle and/or the surface of the silica work piece.Zeta potential measurements on 25-nm silica particles, which arenegatively charged at pH > 2, confirm that cationic surfactant CTABadsorbs to the silica surface at the pH 4–10 (Figure S3). There-fore, despite its potential for reducing particle agglomeration, thecationic surfactant CTAB does not make a suitable dispersant formetal oxide particle based slurries in applications such as silicapolishing, where the surface activity is critical.
3.3. Additional processing benefits
The addition of anionic surfactant to prevent agglomeration inmetal oxide polishing slurries at pH ≥ IEP results in many otherprocessing benefits. A common source of large-particle contamina-tion in industries where slurries are used, e.g., in optical fabrication,arises from the formation of hard, chemically bonded agglomeratesas the slurry dries. Samples of Hastilite PO ceria slurry, stabilized
with 36 mM ALS and untreated (no surfactant), at pH 7 were driedat 110 ◦C and then re-suspended with DI water at pH 7. Results areshown in Fig. 3a. Following re-suspension, the PSD for the untreatedceria slurry demonstrates that both more and larger (up to 30 �m)
38 R. Dylla-Spears et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 447 (2014) 32–43
Table 3Material removal rates (MRR) in the presence and absence of additives. Data are reported for fused silica samples polished with 4.4 wt% Hastilite PO ceria slurry and 4.6 wt%ZOX-PG zirconia under pH conditions where stability was observed upon surfactant addition. Errors in MRR are ±0.11 �m/h.
Removal rate (�m/h) Hastilite PO + 36 mM ALS Hastilite PO + 0.3 vol%Triton H-66
Hastilite PO + 0.6 mMSulfonic 100
HastilitePO + 8.2 mM CTAB
ZOX-PG + 36 mMALS
pH 4 pH 7 pH 10 pH 7 pH 10 pH 7 pH 10 pH 4.8 pH 6.7 pH 7
85
62
abFi
Fa((4
Without additive 0.63 0.58 0.85 0.58 0.With additive 0.67 0.60 0.62 0.73 0.
gglomerates are present in the sample. In contrast, ceria slurry sta-ilized with ALS can be re-suspended with no change to the PSD (cf.ig. 3a). Thus, the use of anionic surfactant in metal oxide polish-ng slurries at pH ≥ IEP reduces the formation of large agglomerates
ig. 3. (a) Effect of drying and re-suspension of slurry (4.4 wt% Hastilite PO ceria withnd without surfactant) on PSDs of particles > 0.5 �m. (b) Effect of filtration of slurry4.4 wt% Hastilite PO ceria stabilized with 36 mM ALS) on PSDs of particles > 0.5 �m.c) AFM images (50 �m × 50 �m area) on fused silica surface after polishing with.4 wt% Hastilite PO ceria slurry without surfactant (left) and with 36 mM ALS (right).
0.58 0.85 0.63 0.58 0.550.863 0.93 0.012 0.10 0.4
on drying, which would likely reduce rogue particle contaminationduring processing.
The addition of anionic surfactants to stabilize slurries may alsoimprove their filterability. This result is important, as other disper-sants have been shown to negatively impact filtration of ceria [35].In this study, the pressure drop across a 4-inch cartridge filter wasmeasured as a function of the number of times the total contents ofthe slurry tank passed through the filter and was normalized by theexpected pressure drop across a clean filter under the same exper-imental conditions. Results for filtration of Hastilite PO ceria bothwith 36 mM ALS and without added surfactant are shown in Fig-ure S4a. Particle agglomeration creates the need for frequent filterchanges with unstabilized Hastilite PO ceria as indicated by a rapidand continual rise (1.6 × 10−3 pass−1) in the pressure drop with thenumber of slurry passes through the filter (50-�m cutoff). Dur-ing filtration, unstabilized ceria solids were lost at a rate of 0.01 %pass−1 (Figure S4b), either due to settling in the reservoir or depo-sition in the filter. In contrast, for the stabilized Hastilite PO ceria,the relative pressure drop across a 5-�m cutoff filter remained rel-atively steady, rising 1.3 × 10−5 pass−1, indicating little uptake ofparticles by the filter and leading to improved filter lifetime evenat a reduced pore size. Filtration of stabilized Hastilite PO ceriawas accomplished with a solids loss rate of only 0.0017 % pass−1,a nearly 10× improvement over the unstabilized Hastilite PO. Byreducing the slurry’s propensity to agglomerate, continuous filtra-tion to remove large-particle contaminants is accomplished withless loss, even at a smaller filter cutoff, which also has the potentialto reduce defect density and scratching [3,6,7,36].
Improvements in filtration performance shown here forHastilite PO ceria are expected to vary with the slurry composi-tion and polishing process parameters. Because slurry flow wasisolated from the polisher in this study, neither the mechanical northe chemical effects of polishing process were evaluated for theireffects on filter lifetime. As a limiting case, if the primary sourceof filter failure is due to debris generated during polishing (ratherthan as a result of agglomeration due to inter-particle attractiveforces), stabilizing the slurry would not improve filter lifetimes.When agglomeration does contribute to filter failure, it is possiblethat some agglomerates in the unstabilized sample may be bro-ken down during the act of polishing, decreasing the load on thefilter and reducing the disparity in filter lifetime between the unsta-blized and stabilized slurry samples to some degree. However, theresidence time of slurry in the polisher is typically much smallerthan is its residence time in the holding tank, and it is unlikelythat the breakup of agglomerates at the polisher can completelycompensate for continuous agglomeration occurring in the tank.
The ability to continuously filter the stabilized slurry at smallerfilter cutoffs (Fig. 3b) enables further refinement of the PSD byremoving particles in the upper end of the distribution. Large par-ticles may be inherent to the stock slurry or may be introducedduring processing (e.g., as a result of stress-induced agglomeration[4]), by changes in solution chemistry induced by the introduction
of glass products, or from environmental contamination. The capa-bility to tailor the distribution using filtration is significant, as ithas been shown that it is the slope of the large particle tail of thePSD—not the mean particle size—that affects the micro-roughness
A: Physicochem. Eng. Aspects 447 (2014) 32–43 39
osAta0taspiaeq
4
4
HfPwpcnectoat
aaisioahsPaaf
tdcflnssttta
ccop
Fig. 4. (a) Zeta potential as a function of pH for 4.4 wt% Hastilite PO ceria. Vertical
R. Dylla-Spears et al. / Colloids and Surfaces
f polished parts [7]. Fig. 3c shows AFM images of fused silicaurfaces polished with Hastilite PO ceria slurry both with 36 mMLS and without surfactant. Under the same polishing conditions,
he slurry containing ALS produced polished surfaces having aver-ge RMS roughness (0.1–50 �m) of 0.65 nm ± 0.08 nm, compared to.99 nm ± 0.1 nm with the unstabilized slurry. Recall from Fig. 1bhat the addition of ALS reduces the number and size of agglomer-tes but does not change the mean ceria particle size. These resultshow that reducing agglomeration in metal oxide particle basedolishing slurries using anionic surfactants at pH ≥ IEP may lead to
mproved surface quality. Suratwala et al. have proposed a mech-nism describing how the loaded fraction of particles in the uppernd of the PSD contributes to removal, ultimately affecting surfaceuality [7].
. Discussion
.1. Surfactant adsorption
The effect of surfactant addition on the zeta potential of theastilite PO ceria slurry was examined to probe the mechanism
or stabilization. Fig. 4a shows the zeta potential of the HastiliteO ceria at working concentration as a function of pH. Hastilite POith no added surfactant shows the expected charge reversal atH 6–8, consistent with reported literature values for the IEP oferia [14]. The addition of 36 mM ALS to the Hastilite PO ceria doesot change the zeta potential of the ceria much at pH > IEP; how-ver, ALS reverses the charge of ceria particles at pH < IEP whenompared to the sample without surfactant. These results suggesthat, as expected, significant anionic surfactant adsorption on ceriaccurs when the particle charge is positive (pH < IEP), whereas only
small amount of adsorption occurs when particle charge is nega-ive (pH > IEP).
At pH > IEP, the metal oxide particles possess an overall neg-tive charge; hence the electrostatic driving force for surfactantdsorption is negligible. Despite this, several groups have producedsotherms demonstrating low levels of adsorption of both anionicurfactants and anionic polyelectrolytes on metal oxides, includ-ng ceria and zirconia, at pH ≥ IEP [17,22,23,37,38]. Small numbersf anionic surfactant molecules may bind either by electrostaticttraction to isolated positively charged sites that exist on theeterogeneously charged particle surface or by chemisorption, asuggested by a shift of the observed IEP for ALS-stabilized HastiliteO (Fig. 4a) [39]. Any chemical affinity promoting adsorption in thebsence of electrostatic attraction is likely to be non-specific, asgglomeration reduction has been observed in the present studyor different surfactant heads and tails and on multiple slurries.
Fig. 4b shows the effect of increasing the surfactant concentra-ion when starting with the untreated Hastilite PO ceria slurry atifferent pHs. Results are shown for both ALS and CTAB. When theeria starts out positively charged (pH < IEP) and the cationic sur-actant CTAB is added (diamonds), the zeta potential changes veryittle and levels off near 1 mM, close to the “clean” (i.e., aqueous,o particles) CMC for CTAB (0.9 mM) [30]. Similarly, when the ceriatarts out with a negative charge (pH > IEP), addition of the anionicurfactant ALS (circles) causes very little change in the zeta poten-ial. A slight change in slope is observed at [ALS] ≈ 11 mM, very nearhe clean CMC for lauryl sulfate (8.2 mM) [28]. The results confirmhat, at a fixed pH, surfactants of like charge to the solid do notdsorb appreciably to the solid surface.
In contrast, the zeta potential for oppositely charged species
hanges significantly with the addition of surfactant. When theeria has a slight positive charge (pH ≈ IEP), increasing the amountf negatively charged ALS (triangles) gradually decreases the zetaotential until it begins to level out when [ALS] approaches 20 mMsolid line represents the measured IEP (pH = 6.2). (b) Zeta potential as a function ofsurfactant concentration for 4.4 wt% Hastilite PO ceria. Dashed vertical lines markthe CMCs of CTAB and ALS.
(2.4× the clean CMC). When the ceria starts out highly nega-tively charged (pH > IEP), the addition of positively charged CTAB(squares) causes a rapid charge reversal, with zeta potential level-ing off when [CTAB] approaches 5 mM (5× clean). Adsorption to thesolid surface should plateau when the solution reaches the CMC, asbeyond the CMC any additional surfactant would form micelles inthe bulk phase rather than adsorb to the surface [32].
The observed CMC of a given surfactant in the presence of ametal oxide surface is dependent on how much surfactant adsorbsat the metal oxide surface, which varies with electrostatic driv-ing force (i.e., pH) and surface area available for adsorption, amongother factors. It is therefore reasonable to assume that the surfac-tant concentrations corresponding to saturation in zeta potentialare actually the CMCs for each surfactant at particular solids load-ings and pH.
4.2. Conditions for stability
Fig. 5a is a stability map describing the concentrations of anionicsurfactant ALS required to increase the settling time of HastilitePO, at a fixed pH of 7, as a function of ceria surface area per unitmass of slurry, which was adjusted by increasing the solids fraction
40 R. Dylla-Spears et al. / Colloids and Surfaces A: Ph
Fig. 5. Stability phase maps for Hastilite PO ceria in the presence of surfactant ALS at(a) fixed pH = 7 and (b) fixed solids concentration = 4.4 wt%. Symbols indicate con-ditions where ceria particles settled (red squares) or remained suspended (greentriangles) after 1 h. The black diamond indicates observation of a mix of suspendedaa
orppsberc“pccr
tnic[mAdt
solution containing > 30% by volume of ethanol [30,42]. Further,
nd agglomerated particles after 1 h. Shading indicates regions of instability (red)nd stability (green).
f Hastilite PO ceria. The values of the points depicted in Fig. 5aepresent a relative settling time scale in which the height of thehase-separation interface of a column of unperturbed slurry sam-le after 1 h was normalized by the interface height of the untreatedlurry (same solids fraction without surfactant) after 1 h. Settlingehavior was chosen as the observable for samples with differ-nt solids concentrations because measurements of PSD wouldequire dilution of all samples to approximately the same parti-le loading, which may affect the observation. Slurries consideredstable” exhibited reduced settling relative to the untreated sam-les (i.e., relative settling times � 1). Note from Fig. 5a that a criticaloncentration of surfactant is required to reduce attractive parti-le interactions and prevent particle agglomeration and that theequired concentration increases with increasing surface area.
In fact, the settling data in Fig. 5a shows that the concentra-ion of ALS required to reduce agglomeration at pH ≈ IEP occursear or above the reported CMC of lauryl sulfate in water (8.2 mM)
n the absence of slurry [28]. Above the CMC, surface adsorptioneases, and the remaining surfactant forms micelles in solution32]. These data suggest that surfactant molecules begin to form
icelles before the particle stabilization is achieved at pH ≈ IEP.
bove the point where stability is achieved (i.e., where settling timeramatically increases), there is no observed benefit to adding addi-ional surfactant, suggesting either that the presence of micelles isysicochem. Eng. Aspects 447 (2014) 32–43
sufficient to achieve stability or that there is an optimal ratio ofmicelles per particle.
Several reference curves supporting the involvement of micellesin the stabilization of the slurry are shown in Fig. 5a. For example,the stable region is approximately 10× above the ALS concentra-tion that would be required to achieve 100% monolayer coverageon particles if the individual ALS molecules were fully extendedand adsorbed in a hexagonal close packed (hcp) array. ALS wasmodeled as a cylinder with radial cross sectional area of 25 squareAngstroms [21]. This scenario does not accurately predict the loca-tion of the transition between aggregated and dispersed phases. Itis also unlikely to occur given the lack of electrostatic driving forceand absence of change in zeta potential at pH ≥ IEP.
Actually, the transition to stability occurs almost an order ofmagnitude higher than expected for monolayer coverage, but quiteclose to the clean CMC for ALS, and it increases with surface area. Anempirical CMC is represented by the dotted curve, which is a linearfit between the literature value for the CMC of lauryl sulfate in DIwater at pH 7 and the apparent CMC at [ALS] ≈ 20 mM at pH 7 inthe presence of 4.4 wt% Hastilite PO ceria (0.33 m2 g−1) (cf. Fig. 4b).The dashed curve shows the concentration of ALS that would berequired to achieve the clean CMC and then cover the entire surfaceof the ceria particles with an hcp array of 2.4-nm diameter, sphericalALS micelles containing 60 monomers each—representing approx-imately 6000 micelles per slurry particle [28]. Both of these curvesdescribe the transition correctly within the experimental error, andthis suggests that either the presence of micelles or some degreeof micellar coverage around the particles reduces particle-particleinteraction and prevents agglomeration.
Fig. 5b shows the stability phase map for 4.4 wt% Hastilite POceria slurry (0.33 m2 g−1) as a function of pH. For pH ≥ IEP, theconcentration of ALS required to prevent agglomeration is againsignificantly higher than would be required to completely coverthe particles with individual surfactant molecules and is close tothe clean CMC. At pH 11, the ALS concentration required for stabil-ity does drop below the clean CMC of ALS at pH 7, which does notchange much across the pH range if total ion content remains fixed[40]. However, research has shown that the CMC drops sharply andthat micelle size increases with ionic strength, which was increasedduring pH adjustment with KOH [28,41]. This could be responsi-ble for the steeper slope in [ALS] with pH observed at pH > IEP. AtpH < IEP, note that the slurry is only partially stable upon addition ofALS. Although agglomeration is reduced compared to the unstabi-lized samples, particle–particle interactions appeared to be higher,as these samples settle faster than stabilized samples at pH ≥ IEP.
4.3. Proposed mechanism
The results of this study suggest that the particles are not sta-bilized by traditional steric or electrosteric stabilization effectedby adsorption of surfactant/polymer molecules onto the particlesurface. Instead, experimental results suggest that both (1) micelleformation and (2) micelle-micelle repulsion are critical for prevent-ing agglomeration of metal oxide particles. Hastilite PO ceria slurrystabilized at pH ≥ IEP using [ALS] > CMC (cf. Fig. 4) remain unag-glomerated long term (>2 years) at room temperature. However,the stabilizing effect of the surfactant on the slurry is immedi-ately reversible by diluting the sample below the CMC. In addition,no reduction in agglomeration is observed when 36 mM ALS isadded to 4 wt% ceria slurry in a 50/50 ethanol/water mixture, likelybecause micelles do not form under such conditions. SurfactantCMCs are known to increase substantially in the presence of a
the anionic surfactant dodecyl benzene sulfonate (DBS) stabilizesthe Hastilite ceria slurry above its CMC, but the hydrotrope ammo-nium xylene sulfonate (AXS) has no effect at similar concentrations,
R. Dylla-Spears et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 447 (2014) 32–43 41
Fig. 6. Schematic of the proposed “charged micelle halo” stabilization mechanism for metal oxide particles: (a) anionic surfactant in negatively charged slurry; (b) cationics slurryw rce fo
dsctAop
ecMsmdtnTk
scoadsano
ipamd
urfactant in a negatively charged slurry; (c) anionic surfactant in positively charged
hen the surfactant concentration exceeds the CMC and the electrostatic driving fo
espite its having the same anionic head group (Table 2). Likeurfactants, hydrotropes have both hydrophobic and hydrophilicharacter and will form structures to solubilize fats in aqueous solu-ions; however, they do not self-assemble in solution. The fact thatXS does not reduce agglomeration suggests that the self-assemblyf surfactant into micelles is critical to the dispersion of metal oxidearticles.
The mere presence of micelles does not promote stability, how-ver. At pH 7, stabilized Hastilite PO ceria slurry (36 mM ALS > CMC)an be returned to an agglomerated state by adding 200 mM NaCl.icelles remain at this condition [28], but the addition of ions
creens charge and reduces the electrostatic repulsions betweenicelles and particles. In addition, the nonionic surfactants tested
o not promote stabilization, even at concentrations well aboveheir CMCs. Uncharged micelles formed by nonionic surfactants doot repel each other to the same degree as do charged micelles [43].hese observations imply that electrostatic repulsive forces play aey role in the stability mechanism.
We cannot completely rule out the involvement of depletiontabilization, in which a deficit of non-adsorbing polymer segmentoncentration between particles creates a repulsive force; however,ur results do not appear completely consistent with this mech-nism [44,45]. For example, we do not observe a correspondingepletion flocculation upon addition of small amounts of anionicurfactants, as would be predicted by a depletion mechanism. Inddition, we were unable to stabilize the slurry with uncharged,on-absorbing polymer species, even at higher molecular weightr at higher concentrations.
We hypothesize instead a “charged micelle halo” mechanismn which the metal oxide polishing slurries are stabilized by the
resence of micelles that are loosely associated with each particlend prevent particle agglomeration due to the presence of suchicelles in the interstitial fluid between individual particles. This isirectly analogous to the “nanoparticle halo” colloidal stabilization
; and (d) cationic surfactant in positively charged slurry. Agglomeration is preventedr adsorption is low, i.e., (a) and (d).
mechanism proposed by Tohver et al. [46]. The “charged micellehalo” concept is depicted schematically in Fig. 6a for the case wherethe particle and surfactant are both negatively charged. Althoughsome surfactant may adsorb to the particle surface, particle stabi-lization is not achieved until the surfactant concentration exceedsthe CMC for a given solids loading. The micelle-micelle repulsionis stronger than the micelle–particle repulsion, which causes themicelles to surround the metal oxide particles to maximize theirdistance from one another in the continuous phase. The result isa halo of charged micelles surrounding each slurry particle. Theparticle–particle interaction is effectively blocked by the chargedmicelles, which repel one another electrostatically, thereby reduc-ing or eliminating metal oxide particle agglomeration. Insufficientrepulsion between micelles, the lack of micelles at concentrationsbelow the CMC, or an insufficient number of protective haloswould leave the metal-oxide particles vulnerable to attractiveinteractions.
Of course, interactions between the surfactant and the particledepend upon their charges at the pH of interest (see Fig. 6b–d).When the surfactant charge is the opposite of the metal oxideparticle, surfactant adsorption is likely favored, dominated by elec-trostatic attraction of the oppositely charged species (Fig. 6b,c).In the present study, fully stabilized ceria slurry was not achiev-able with anionic surfactant when ceria was positively charged(pH < IEP), even at [ALS] � CMC. Surface charge neutrality andcharge reversal due to surfactant adsorption can actually reduceelectrostatic repulsion between particles and may lead to aggrega-tion [32,47]. Alternatively, bridging between adsorbed surfactantmolecules, as illustrated in Fig. 6b, may contribute to the observedagglomeration. In contrast, stabilization of ceria was achieved using
the cationic surfactant CTAB at pH ≤ IEP (cf. Figs. 1 and 2). This resultis depicted schematically in Fig. 6d, where the particle and surfac-tant are both positively charged. Thus, charged micelle stabilizationappears to be most effective under conditions where surfactant
42 R. Dylla-Spears et al. / Colloids and Surfaces A: Ph
Fig. 7. Effect of the addition of silica nanoparticles on the PSDs of Hastilite PO cerias
atstepfapo
awpaTcpnmlmppwf
usTlrwpcfiosfebso
Wiley, 1993.[10] E.J.W. Verwey, J.Th.G. Overbeek, Theory of Lyophobic Colloids, Dover, New York,
NY, USA, 1999.
lurry particles > 0.5 �m.
dsorption is low, i.e., where micelle and particle charges are ofhe same sign. When surface functionality is critical, however, theelection of surfactant should also involve a consideration of poten-ial surfactant adsorption onto other surfaces in the system. Forxample, the scenario depicted in Fig. 6d resulted in poor MRR,resumably due to CTAB adsorption on the silica work piece sur-ace impeding chemical reaction. Therefore, for silica polishingpplications, the combination of negatively charged metal oxidearticles (pH > IEP) and anionic surfactants (Fig. 6a) yields the bestutcome.
These results recall those of a glass polishing study by Cumbond coworkers, in which slurries of alumina, zirconia, and ceriaere found to be stable to agglomeration in the presence of glassolishing products when slurry pH ≥ IEP but formed agglomeratesnd increased roughness of the polished surface at pH < IEP [20].hese researchers termed the phenomenon the “slurry-charge-ontrol effect” and recommended that optical polishing slurries berepared at pH ≥ IEP such that slurry particles and silica both haveegative charge. A mechanism similar to that proposed in Fig. 6ay also explain these previous results. In this case, the molecu-
ar networks of negatively charged silica removed during polishingay themselves act as the stabilizing species for the metal oxide
articles at pH ≥ IEP. When the metal oxide particles are insteadositively charged at pH < IEP, the negatively charged silica net-orks may contribute to charge reversal and facilitate agglomerate
ormation.Fig. 7 shows that 25-nm silica nanoparticles[48] can, in fact, be
sed to stabilize the Hastilite PO ceria particles at pH > IEP, pre-umably by the same nanoparticle halo mechanism proposed byohver and coworkers. At pH 10, agglomeration reduction equiva-ent to that imparted by the anionic surfactant ALS is observed at aatio of ∼20 silica nanoparticles per ceria particle, under conditionshere both silica and ceria particles were negatively charged, onar with the ∼60 25-nm silica spheres that would be required toover a 100-nm ceria particle in an hcp array. This experiment con-rms the viability of the charged micelle halo mechanism for metalxide particles under the conditions presented in this study. Theuccessful stabilization of metal oxide particles with charged sur-actant micelles also implies that the haloing phenomenon can bextended to a broader range of conditions than originally reportedy Tohver et al., including smaller aspect ratios (particle:haloing
pecies) and conditions where the particle and haloing species aref the same charge.[[
ysicochem. Eng. Aspects 447 (2014) 32–43
5. Conclusions
A stabilization technique amenable to reducing agglomerationin metal oxide polishing slurries while maintaining their chemicalactivity has been demonstrated experimentally. The method hasbeen demonstrated using ceria, zirconia, and alumina slurries atpH ≥ IEP in combination with several different anionic surfactants.The key results are as follows. (1) The upper ends of the particle sizedistributions are significantly improved relative to unstabilizedslurries. (2) Polishing rates (material removal rates) are main-tained. (3) Stabilized slurries are easily filtered and less materialis lost, even using smaller pore size filters, when compared tounstabilized slurries. (4) Work pieces exhibit significantly lowermicro-roughness when polished using stabilized slurries. (5) Driedstabilized slurries can be easily re-dispersed without introducingagglomerates.
We offer a “charged micelle halo” stabilization mechanism toexplain why agglomeration reduction is observed where micellesform and where the surfactant and slurry particle charges are ofthe same sign. Micelles are thought to repel one another in thebulk, enveloping the particles and creating an electrosteric bar-rier between slurry particles. Because the micelles are not directlybound to the particle surface and free surfactant adsorption islow, the particles are able to maintain their surface functionality.Although the data presented in this study focuses on polishing, thisstabilization method is expected to be broadly applicable for thestabilization of metal oxide colloidal particles. We note that thecharged micelle halo stabilization mechanism may be of partic-ularly high value in applications where steric stabilization usingadsorbed species may unfavorably alter the particle size or mayinterfere with reactivity or availability of the surface, such as inceramics processing, chemical-mechanical planarization, colloidalsurface functionalization, or catalysis.
Acknowledgements
The authors gratefully acknowledge Nan Shen for perform-ing AFM experiments and Joshua Kuntz for helpful discussions.This work was performed under the auspices of the US Depart-ment of Energy by Lawrence Livermore National Laboratory underContract DE-AC52-07NA27344 within the LDRD program LLNL-JRNL-644902.
Appendix A. Supplementary data
Supplementary data associated with this article can befound, in the online version, at http://dx.doi.org/10.1016/j.colsurfa.2014.01.061.
References
[1] G.B. Basim, B.M. Moudgil, J. Colloid Interface Sci. 256 (2002) 137–142.[2] G.B. Basim, Adv. Powder Tech. 22 (2011) 257–265.[3] Y.J. Seo, S.Y. Kim, Y.O. Choi, Y.T. Oh, W.S. Lee, Mater. Lett. 58 (2004) 2091–2095.[4] F.C. Chang, P. Kumar, R. Singh, K. Balasundaram, L. Jaeseok, L. Jinhyung, R.K.
Singh, Colloids Surf. A 389 (2011) 33–37.[5] R.K. Singh, R. Bajaj, MRS Bull. 27 (2002) 743–751.[6] T. Suratwala, R. Steele, M.D. Feit, L. Wong, P. Miller, J. Menapace, P. Davis, J.
Non-Cryst. Sol. 354 (2008) 2023–2037.[7] T. Suratwala, M.D. Feit, R. Steele, L. Wong, N. Shen, R. Dylla-Spears, R. Desjardin,
D. Mason, P. Geraghty, P. Miller, S. Baxamusa, J. Am. Ceram. Soc. 97 (2014)81–91, http://dx.doi.org/10.1111/jace.12631.
[8] L.M. Cook, J. Non-Cryst. Sol. 120 (1990) 152–171.[9] H.H. Karow, Fabrication Methods for Precision Optics, New York, NY, USA,
11] Q. Wei, Y. Meng, J. He, G. Tang, J. Rare Earths 28 (2010) 478–481.12] J. Palla, D.O. Shah, J. Colloid Interface Sci. 223 (2000) 102–111.
A: Ph
[
[[[[[
[
[
[[[[
[
[[[
[
[
[
[
[
[[[[[[
[[
[
[[[
R. Dylla-Spears et al. / Colloids and Surfaces
13] H.M. Kim, R.P. Venkatesh, T.Y. Kwon, J.G. Park, Colloids Surf. A 411 (2012)122–128.
14] G. Parks, Chem. Rev. 65 (1965) 177.15] J.P. Hsu, A. Nacu, J. Colloid Interface Sci. 274 (2004) 277–284.16] Y. Sehgal, Lalatonne, J.F. Berret, M. Morvan, Langmuir 21 (2005) 9359–9364.17] H. Guldberg-Pedersen, L. Bergstrom, Acta Mater. 48 (2000) 4563–4570.18] S.B. Johnson, G.V. Franks, P.J. Scales, D.V. Boger, T.W. Healy, Int. J. Miner. Process.
58 (2000) 267–304.19] L. Wang, B. Liu, Z. Song, W. Liu, S. Feng, D. Huang, S.V. Babu, Electrochem. Sol.
State Lett. 14 (2011) H128–H131.20] M.J. Cumbo, D. Fairhurst, S.D. Jacobs, B.E. Puchebner, Appl. Opt. 34 (1995)
3743–3755.21] P. Somasundaran, D.W. Fuerstenau, J. Phys. Chem. 70 (1966) 90–96.22] X. Fan, P. Somasundaran, N.J. Turro, Langmuir 13 (1997) 506–510.23] D. Bitting, J.H. Harwell, Langmuir 3 (1987) 500–511.24] M.J. Solomon, T. Saeki, M. Wan, P.J. Scales, D.V. Boger, H. Usui, Langmuir 15
(1999) 20–26.25] Y.K. Leong, P.J. Scales, T.W. Healy, D.V. Boger, R. Buscall, J. Chem. Soc., Faraday
Trans. 89 (1993) 2473–2478.26] Sigma Product Information Sheet, Tween 20, Sigma–Aldrich Co., 2003.27] Sigma Product Information Sheet, Triton X-100, Sigma–Aldrich Co., 2003.
28] M. Bergstrom, J.S. Pedersen, Phys. Chem. Chem. Phys. 1 (1999)4437–4446.29] M. Ashokkumar, T. Niblett, L. Tantiongco, F. Grieser, Aust. J. Chem. 56 (2003)
1045.30] L. Wei, M. Zhang, J. Zhang, Y. Han, Front. Chem. China 1 (2006) 438.
[
[[
ysicochem. Eng. Aspects 447 (2014) 32–43 43
31] N.G. Hoogeveen, M.A. Cohen Stuart, G.J. Fleer, J. Colloid Interface Sci. 182 (1996)133–145.
32] J.F. Scamehorn, R.S. Schechter, W.H. Wade, J. Colloid Interface Sci. 85 (1982)463–478.
33] S.G. Dick, D.W. Fuerstenau, T.W. Healy, J. Colloid Interface Sci. 37 (1971)595–602.
34] J.E. Gebhardt, D.W. Fuerstenau, Colloids Surf. 7 (1983) 221–231.35] Dietrich, A. Neubrand, Y. Hirata, J. Am. Ceram. Soc. 85 (2002) 2719–2724.36] Y.H. Kim, S.K. Kim, J.G. Park, U. Paik, J. Electrochem. Soc. 157 (2010) H72–H77.37] S. Kim, J.H. So, D.J. Lee, S.M. Yang, J. Colloid Interface Sci. 319 (2008) 48–52.38] M.R. Bohmer, L.K. Koopal, Langmuir 8 (1992) 2649–2659.39] K.N. Han, T.W. Healy, D.W. Fuerstenau, J. Colloid Interface Sci. 44 (1973)
407–414.40] A. Rahman, C.W. Brown, J. Appl. Polym. Sci. 28 (1983) 1331–1334.41] P.C. Pavan, E.L. Crepaldi, G.D. Gomes, J.B. Valim, Colloids Surf. A 154 (1999)
399–410.42] S. Javadian, H. Gharibi, B. Sohrabi, H. Bijanzadeh, M.A. Safarpour, R.
Behjatmanesh-Ardakani, J. Mol. Liq. 137 (2008) 74–79.43] L. Marszall, Colloids Surf. 25 (1987) 279–285.44] R.I. Feigin, D.H. Napper, J. Colloid Interface Sci. 74 (1980) 567–571.45] J.Y. Walz, A. Sharma, J. Colloid Interface Sci. 168 (1994) 485–496.
46] V. Tohver, J.E. Smay, A. Braem, P.V. Braun, J.A. Lewis, Proc. Natl. Acad. Sci. U.S.A.98 (2001) 8950–8954.47] W. Fuerstenau, Faraday Discuss. 59 (1975) 157–168.48] T.I. Suratwala, L. Carman, I. Thomas, UCRL-ID-154626, 2003,
http://dx.doi.org/10.2171/15005259.
