Efficiency of grafting of Al2O3, TiO2 and ZrO2 powders by perfluoroalkylsilanes

Page 1

E

JMa

b

h

�

�

�

�

�

a

ARR1AA

KATZSPS

0h

Colloids and Surfaces A: Physicochem. Eng. Aspects 420 (2013) 64– 73

Contents lists available at SciVerse ScienceDirect

Colloids and Surfaces A: Physicochemical andEngineering Aspects

jo ur n al hom ep age: www.elsev ier .com/ locate /co lsur fa

fficiency of grafting of Al2O3, TiO2 and ZrO2 powders by perfluoroalkylsilanes

oanna Kujawaa, Wojciech Kujawskia,∗, Stanisław Kotera, Anna Rozickaa, Sophie Cerneauxb,ichel Persinb, André Larbotb

Faculty of Chemistry, Nicolaus Copernicus University, 7 Gagarina St., 87-100 Torun, PolandInstitut Européen des Membranes, 1919, route de Mende, 34293 Montpellier Cedex 5, France

i g h l i g h t s

Efficient modification of ZrO2, TiO2

and Al2O3 by various perfluoroalkyl-silanes.Modification the hydrophilic surfaceof metal oxides to the hydrophobicone.Characterization of grafting processwith PFAS by various analytical tech-niques.Determination the impact of severalparameters on the grafting effi-ciency.Evaluation and optimization of graft-ing conditions by chemometric sim-plex method.

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 27 September 2012eceived in revised form3 December 2012ccepted 18 December 2012vailable online xxx

eywords:l2O3

iO2

rO2

a b s t r a c t

Metal oxides powders of Al2O3, TiO2 and ZrO2 were modified by two types of perfluo-roalkylsilanes (PFAS) molecules 1H,1H,2H,2H-perfluorooctyltriethoxysilane (C6) and 1H,1H,2H,2H-perfluorotetradecyltriethoxysilane (C12). Studies showed that surface of oxide powders can be efficientlyhydrophobized. Grafting efficiency of oxide powders was determined by TGA and FR-MIR techniques.Strong influence of ratio of amount of PFAS to amount of metal oxide powder, type of grafting molecules,grafting time and concentration of PFAS solution were observed on the grafting efficiency of all powders.

The highest grafting efficiency occurred for alumina, what is related with the highest specific surfacearea of Al2O3. However, zirconia should the lowest grafting efficiency by PFAS molecules.

The mechanism of grafting process was suggested and confirmed by 29Si NMR and FT-MIR techniques.Analyses results revealed three possible types of bonding of the hydrophobic chains to the materials

urface modificationerfluoroalkylsilaneimplex method

surface. It was found that C6 molecules were attached mainly by siloxane bonds, whereas C12 moleculeswere attached by geminal silanol bonds.

The results obtained for alumina modification were additionally approached using chemometric sim-plex method, what allowed to determine the optimal grafting conditions. The highest grafting efficiencyof Al2O3 was obtained using 0.75 mmol C6 g−1 of Al2O3. Concentration of C6 molecules has only a minor

effici

influence on the grafting

∗ Corresponding author. Tel.: +48 56 611 43 15.E-mail address: [email protected] (W. Kujawski).

927-7757/$ – see front matter © 2012 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.colsurfa.2012.12.021

ency.© 2012 Elsevier B.V. All rights reserved.

1. Introduction

Ceramic materials are more stable than other materials espe-cially at elevated temperatures and in more corrosive environment.Among ceramics oxides are the most commonly used materialsbecause they are more stable in the air than non-oxide ceramics

Page 2

J. Kujawa et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 420 (2013) 64– 73 65

Nomenclature

BET Brunauer–Emmett–Teller methodBJH Barrett–Joyner–Halenda methodC6 1H,1H,2H,2H-perfluorooctyltriethoxysilaneC12 1H,1H,2H,2H-perfluorotetradecyltriethoxysilaneDTG Differential Thermal AnalysisFT-MIR Fourier Transform-Middle Infrared

(4000–400 cm−1) SpectrometryNMR CP/MAS Nuclear Magnetic Resonance of Solid State

Cross Polarization/Magic Angle SpinningPFAS PerfluoroalkylsilanePTFE PolytetrafluoroethyleneTGA Thermogravimetric AnalysisXRD X-ray DiffractionCC6/C12 Concentration of C6 or C12 solution (M)DBET Particle size from BET isotherm (nm)Dh k l Particle size from XRD measurements (nm)MeOx Metal oxide (MeOx = Al2O3, TiO2 or ZrO2)SBET Specific surface area from BET isotherm (m2 g−1)tmod Modification time (h)Tmod Modification temperature (◦C)Q Ratio of amount of PFAS to amount of MeOx powder

sfiotp

pcAiisb

bitrfobatactuf

hdpaca

aTh

MeOx(mmol g−1)

uch as nitrides and carbides [1–4]. Oxide ceramics can be classi-ed based on differences in their crystal structure, which dependsn the configuration of the oxygen ions and the stoichiometry ofhe elements. Al2O3, TiO2 and ZrO2 oxides are often utilized for thereparation of ceramic membranes [5–12].

Oxides with a corundum structure display excellent hardnessroperties, corrosion and heat resistance as well as high thermalonductivity. Alumina is a representative oxide of corundum class.

variety of transitional Al2O3 structures can be formed by increas-ng temperature, but all structures are eventually transformedrreversibly to an �-form with a corundum structure of hexagonalystem [1,13]. Nanoporous alumina has a variety of applications iniosensors, biofiltration, and targeted drug delivery [1,14].

TiO2 nanostructures became a focus of interest in the past yearsecause of their unique properties relevant to the applications,

ncluding chemical sensing [15–17], photocatalysis [18] and pho-ovoltaics [2,3]. TiO2-based materials may also be used for theemoval of thiophenic sulfur compounds present in hydrocarbonuels, by a selective adsorption [19]. Modified TiO2-based mixedxides have attracted much interest in area of catalytic applicationsecause of their improved properties. The improvements have beenttributed to modification of the structure, electronic property andhermal stability by doping [4,20]. TiO2 photocatalysis proved to be

promising technology for the purification and treatment of bothontaminated air and water [21]. Several researchers attemptedo combine photocatalysis with membrane separation technique,sing TiO2 powder photocatalyst for removal of organic pollutantsrom water [22].

Zirconia can be utilized in various catalytic processes, likeydrogenation of olefins or CO, isomerization of olefins and/orehydration of alcohols [23–25]. Moreover, zirconia and titaniarovide an alternative to silica chromatographic packing materialsnd are characterized by mechanical stability, good separation effi-iency, and chemical inertness at elevated temperatures and over

pH range of 1–14 [26–28].

The above mentioned materials posses hydrophilic character,

s a result of the presence of hydroxyl groups on the surface.he surface modification process can be used to increase theydrophobicity of the ceramic, by either introducing radical groups

Fig. 1. General formula of PFAS.

or hydrophobic polymer chains on the surface. Hydrophobizationof ceramic surface can be done using various methods, howeveralways compounds with a reactive grouping should be used (e.g.methoxy, ethoxy or active chlorine) [5–12]. Perfluoroalkylsilanes(PFAS) (Fig. 1) possessing such reactive grouping can effectivelychange the surface character of a material from hydrophilic tohydrophobic one [5–9]. The reactive groups of PFAS react withhydroxyl groups on the oxide surface, creating a stable covalentbonds (Fig. 2).

The modification process of surfaces is a very interestingproblem in chemistry and many research groups work in thisfield. The modification of surfaces by grafting was investigatedin several laboratories [5,7–9,29–44]. Generally, the research isfocused on the grafting of ceramic membranes by fluoroalkylsilanes(FAS), mainly 1H,1H,2H,2H-perfluorooctyltriethoxysilane (C6) and1H,1H,2H,2H-perfluorodecyltriethoxysilane (C8). However, thereare only a few papers devoted to hydrophobization of metal oxidepowders [9,29–31]. One of these papers focused on the modifica-tion of zirconia oxides powders by C8 molecules [9]. Authors triedto optimize the grafting process with the regards to grafting time,concentration of C8 molecules as well as the ratio of C8 moleculesto the amount of grafted oxide. The optimized parameters wereused for the grafting of tubular zirconia membranes. Subsequentlythese modified membranes were applied in membrane distillationprocess [9]. Recently, Djafar et al. [29] described the surface modi-fication of TiO2 by phosphate and phoshonic acids. However, in thiswork authors focused on the thermal and photocatalitical activityof modified titania powders.

Larbot and co-workers [8,32–35] applied C6 and C8 fluo-roalkylethoxysilanes for the surface modification of �-alumina andzirconia tubular membranes. Khemakhem et al. [36,37] preparedthe microfiltration Tunisian clay membrane and used C8 moleculesto create a hydrophobic layer on the membrane surfaces. The mod-ified membranes were then used in membrane distillation process.

The literature review shows a number of papers presenting var-ious applications of modified membranes in osmotic evaporation[38,39], filtration of organic solvents [32,40], pervaporation [34]and/or membrane distillation processes [5,7,8,36].

Schondelmaier et al. [41] investigated the orientation andself-assembly of fluoroalkylsilanes chains on the modified sur-face. Applying a polarization dependent X-ray absorption nearedge spectroscopy (XANES) and X-ray photoelectron spectroscopy,authors found that most of hydrophobic chains were arranged per-pendicular to the substrate surface.

Picard et al. [7] as well as Koonaphapdeelert et al. [42] inves-tigated the possible ways of attachment C8 [7] molecules and C6molecules [42] to the surfaces. Koonaphapdeelert et al. [42] dis-cussed the possible modes of attachment of C6 molecules to thesurface of alumina hollow fiber ceramic membranes. On the otherhand Picard et al. [7] used membranes made from zirconia. Bothpapers concluded the similar modes of the anchoring of graftingmolecules to the metal oxide surface.

Only recently, C12 molecules are available on the chemical mar-ket and these molecules were used for the modification of TiO2and ZrO2 membranes. Subsequently, the efficiency of metal oxide

Page 3

66 J. Kujawa et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 420 (2013) 64– 73

modi

gi

ofipam

pf

2

2

Z

w

s

2

sstwtmd

pfruwwd

2

2

wdip

with Bruker Avance 300 MHz of magic angle method using a broad-band probe CP/MAS in 7 mm rotor. This technique was used tocharacterize a grafting surface process.

Table 1The conditions of surface modification experiments.

Parameters Tested range

Temperature, Tmod (◦C) 21–50Concentration of grafting solution,

CPFAS (M)C6

0.01–0.100.05

Fig. 2. The mechanism of surface

rafting by C12 molecules was compared to the efficiency of graft-ng by C6 molecules [43,44].

The main aim of this work was a modification of various metalxide powders by perfluoroalkylsilanes. The grafting was per-ormed using two types of PFAS compounds: C6 and C12. Thempact of various parameters on the grafting efficiency, like tem-erature, time of modification, concentration of grafting moleculesnd the ratio of PFAS amount to metal oxide amount was deter-ined and discussed.Moreover, an optimization of the grafting process of metal oxide

owders by applying the simplex chemometric method was per-ormed.

. Materials and methods

.1. Materials

The commercially available metal oxide powders: Al2O3, TiO2,rO2 were purchased from Sigma–Aldrich (Poland).

C6F13C2H4Si(OC2H5)3 and C12F25C2H4Si(OC2H5)3 compoundsere purchased from SynquestLab, USA.

Chloroform stabilized by 1% ethanol, acetone and ethanol wereupplied by POCH (Poland).

.2. Metal oxide powders – physicochemical characteristics

Prior to the grafting process, the specific surface area and poreize of the metal oxides powders were determined. The specificurface area was determined by BET nitrogen adsorption low-emperature analysis (ASAP 20120). The purity of used nitrogenas 99.9995% to ensure the precision of obtained data. In adsorp-

ion measurements 0.1–0.5 g of samples were used. Before theeasurements, non-grafted and grafted samples were degassed

uring 2 h at 200 ◦C and 90 ◦C, respectively.XRD analysis was performed to evaluate the crystal form and

hase composition of metal oxides. The average crystallite sizerom the broadening of reflexes was determined using the Scher-er formula [45]. Rachinger’s method with K�2/K�1 correction wassed to determine full width at half maximum [46]. Measurementsere performed in 2� angle range of 10–80◦, at room temperatureith 0.02◦ step and rate of 3◦/min using Philips X’Pert PW 3040/60iffractometer (K� = 1.5418 Å) with Cu lamp (30 mA and 40 kV).

.3. Grafting process

.3.1. Grafting procedureThe grafting process of metal oxide powders (Al2O3, TiO2 i ZrO2)

as performed by applying a multistep procedure. Initially, pow-ers were rinsed in acetone, ethanol and distilled water for 10 min

n each solvent. Then, the oxides were dried at 110 ◦C for 3 h. Allowders were stored in a desiccator, in order to maintain the same

fication of metal oxide powders.

conditions of temperature and humidity. Every single experimentconsisted of adding a precise volume of grafting solution (in ml) tothe given amount of the metal oxide powder (in grams). Becauseof the risk of polycondensation of C6 and C12, which might occurin the presence of humidity from air, the preparation procedureof grafting solutions and the grafting process required an ambientatmosphere of argon. All grafting experiments were performed in20 ml glass vials closed with PTFE/butyl septa. Roller mixer (StuartSRT6D, Bibby Scientific Limited, UK) was used to ensure uniformcontact of samples with the grafting solution. After grafting pro-cess completed, the modified powders were rinsed sequentially inacetone, ethanol and distilled water and dried at 140 ◦C for 3 h andstored in the desiccator in Ar atmosphere. Modified metal oxidepowders were then analyzed by TGA, FT-MIR and 29Si NMR tech-niques.

Metal oxide powders were modified at different conditions, inorder to determine the influence of various parameters on the effi-ciency of grafting process (Table 1).

2.3.2. Analytical methodsNon-grafted and grafted metal oxide powders were analyzed

by thermogravimetric analysis on the Simultaneous TGA-DTA SDT2960 Thermal Analysis apparatus. The TGA measured the lossof sample weight as a function of temperature and time. TGAmeasurements were performed in the nitrogen atmosphere andthe temperature range of 25–1000 ◦C, with the heating rate of20◦C/min.

Pristine and grafted metal oxides powders were also analyzedusing FT-MIR analysis on the PerkinElmer Spectrum 200 apparatusin the range of 4000–400 cm−1. Samples were prepared in a formof a thin film on KBr plate.

The TGA and FT-MIR techniques were used to evaluate the effi-ciency of grafting process.

29Si NMR CP/MAS (Nuclear Magnetic Resonance of solid-statecross polarization/ magic angle spinning) analysis was performed

C12Quotient of amount of PFAS to amount

of MeOx powder, QMeOx (mmol g−1)0–25

Grafting time, tmod (h) 1–300

Page 4

J. Kujawa et al. / Colloids and Surfaces A: Phys

Table 2The crystal form of oxide powders used to grafting process.

Metal oxides Crystal form Crystallographic system

Al2O3 gamma rhombohedral

3

3

wtp

irZ

m

loop can be observed for all registered isotherms. The amount of

TiO2 anatase tetragonalZrO2 baddeleyite monoclinic

. Results and discussion

.1. Characterization of metal oxide powders

According to XRD analysis, crystal forms of metal oxide powdersere identified using the position of the diffraction lines were used

o identify the crystalline phase by comparing with the referenceattern (Philips data base). The results are presented in Table 2.

Based on the location of diffraction lines of alumina oxide wasdentified as a gamma-form (76%). At room temperature, mainlyeflections from the anatase (78%) were detected for titania sample.

rO2 was identified as a monoclinic zirconia (82%).

The adsorption/desorption isotherms of nitrogen provide infor-ation about a given porous material, its specific surface area, pore

Fig. 3. Nitrogen adsorption/desorption isotherms for �

icochem. Eng. Aspects 420 (2013) 64– 73 67

size and pore size distribution. The adsorption-desorption nitrogenisotherms of non-grafted metal oxide powders are shown in Fig. 3.Analysis proposed by Lippens and de Boer named as a t-plot methodwas applied to confirm the presence of micropores [47]. The rela-tion proposed by Harkins and Jura was used as a standard referencet-curve for calculating the micropores volume [48]. Additionally,Barrett– Joyner–Halenda (BJH) fitting procedure allowed to indi-cate the presence of mesoporous in the structure of powders [49].

On each registered isotherm the separate adsorption anddesorption branches can be observed (Fig. 3). Brunauer–Emmett–Teller (BET) method was used to determine specific surfacearea. Low-pressure parts of the isotherms correspond to the gasadsorption on the powder surface. The relative pressure rangeP/Po = 0.05–0.3 (magnified in Fig. 3) can be used to determinethe specific surface area by a BET model. The analysis of theisotherms suggests that micropores are filled at the first instanceand a monolayer of adsorbate is created on the surface. Then thecapillary condensation process occurs in mesopores. The hysteresis

adsorbed gas (N2) varied for each type of metal oxide powders.The amount of adsorbed gas increased in the following sequence:ZrO2-B < TiO2 < ZrO2-A < �-Al2O3.

-Al2O3 (A), TiO2 (B), ZrO2-A (C) and ZrO2-B (D).

Page 5

68 J. Kujawa et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 420 (2013) 64– 73

Table 3Textural characteristics of pristine metal oxide powders.

Adsorbent SBET (m2 g−1) DBET (nm)

�-Al2O3 103 15TiO2 15 95ZrO2-A 83 12ZrO2-B 5 Dhkl = 42a

SBET – specific surface area from BET isotherm.DBET – particle size from BET isotherm.

a Dhkl – particle size from XRD measurements.

Table 4Textural characteristics of modified metal oxide powders.

Adsorbent SBET (m2g−1) DBET (nm)

scmaiptIpm

w(

lospdopfir7

�-Al2O3 modified 97 16ZrO2-A modified 76 13

Isotherms (Fig. 3A–C) for the �-Al2O3, TiO2 and ZrO2-A showimilarity to type IV adsorption isotherm, according to IUPAClassification. Type IV isotherm is characteristic for mesoporousaterials, in which capillary condensation and the multilayer

dsorption (for higher pressures) take place. In addition, thesesotherms show similarity to isotherms characteristic for meso-orous materials with micropores addition [49]. On the other hand,he nitrogen isotherm for ZrO2-B (Fig. 3D) was similar to typeI isotherm. Surface textural properties of pristine metal oxideowders determined from nitrogen adsorption isotherms are sum-arized in Table 3.Nitrogen adsorption isotherms after the modification process

ere registered for two types of oxide powders �-Al2O3 and ZrO2-AFig. 4).

The isotherms after grafting process were characterized byower volume amounts of adsorbed gas compared with non-graftednes (Fig. 4). Shapes of adsorption isotherms of grafted powder-were very similar with isotherms of non-grafted powders. Table 4resents the textural properties of modified powders. Comparingata from Table 3 and Table 4, it can be observed the influencef the grafting process on the specific surface area as well as thearticle size of powders. The decreasing of specific surface area

or both grafted powders was observed. Alumina was character-zed by 103 and 97 m2 g−1, before and after modification process,espectively. Specific surface area for zirconia was equal to 83 and6 m2 g−1, before and after grafting. These measurements enabled

Fig. 4. Nitrogen adsorption isotherms for �-Al2O3 (

Fig. 5. The water drop on the modified alumina powder.

also determination of particle size of metal oxide powders [48].Results obtained on the base of isotherms indicate the increase ofaverage particle size from 15 nm to 16 nm for alumina and from12 nm to 13 nm for zirconia, respectively. The increase of aver-age particle size confirms that PFAS molecules were grafted onthe metal oxides surface. On the other hand, this fact does notprovide enough information on the possible conformation of thefluoro-carbon chains on the surface and within the pores of oxidespowders.

3.2. Mechanism of modification process

The surface modification of metal oxides occurs as an effectof condensation reaction between reactive group of PFAS andhydroxyl groups on the oxide surface. The reaction leads to thecreation of stable covalent bonds [5–11]. Moreover, the modifica-tion changes the character of surface of metal oxide powders fromhydrophilic to hydrophobic one (Fig. 5).

In order to confirm the occurrence of surface modification aswell as to determine the efficiency of grafting, FT-MIR spectraand thermal analyses were realized. FT-MIR spectra (Fig. 6) wererecorded before and after the modification of alumina powders by

C6 and C12 molecules.

Bands in the region 1300–900 cm−1, which are characteristic foralkylsilanes, are highlighted in Fig. 6 and additionally summarizedin Table 5. After inspecting spectra presented in Fig. 6 it can be also

A) i ZrO2-A (B) before and after modification.

Page 6

J. Kujawa et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 420 (2013) 64– 73 69

Fig. 6. FT-MIR spectra of non-grafted and grafted �-Al2O3 by C6 and C12 solu-tions. Conditions of modification: QAl2O3

= 1.50 mmol g−1, CC6,C12 = 0.05 M, tmod = 72 h,Tmod = 21 ◦C.

Table 5Infrared absorption bands characteristic for alkylsilanes. Conditions of modification:QAl2O3

= 1.50 mmol g−1, CC6,C12 = 0.05 M, tmod = 72 h, Tmod = 21 ◦C.

Wavenumber (cm−1) Group

1242 CF2, CF3

ccbhi(tt

mli

1210, 1196, 1145, 1068, 1018 Si O CH2CH2R1119 Si O

oncluded that the specific vibration bands of the perfluorinatedhains at frequency range 1242–1018 cm−1 are observed. As it cane seen from Fig. 6. bands of C12 chains were characterized by aigher intensity than those of C6. The vibration of the Si O bond

s found at frequency ca. 1119 cm−1. The presence of the Si O XX = Al, Ti or Zr) bond is the result of a chemical reaction betweenhe Si(OCH2CH3)3 moieties and the available hydroxyl groups onhe surface of the powder materials.

29

NMR analysis was performed to gain more information onechanism of grafting process (Fig. 7). NMR spectra were col-

ected for titania oxide. CP/MAS NMR technique was employed toncrease the signal resolution and sensitivity of the silicon atoms in

Fig. 7. 29Si NMR spectra of TiO2 grafted by C6 (A) and by C12 (B). Grafting con

Fig. 8. Tn structure of silicon on the 29Si NMR spectra of modified titania powder.

proximity to protons [50,51]. To better comparison as well as to rec-ognize the mechanism of modification process, NMR spectra wereperformed before and after surface modification of titania oxidewith C6 and C12 molecules. On the 29Si NMR spectrum obtainedfor pristine titania powder no peaks of silicon atoms were detected(Fig. 7A and B). Three distinct peaks were observed in the 29Si NMRspectra (Fig. 7) of the fluoroalkylsilane-modified titania oxide withboth grafting molecules indicating three distinct silicon chemicalenvironments. The peaks at chemical shifts of approximately inranges −42 to −51, −53 to −59, and −62 to 69 ppm are representa-tives of silicon atoms connected to T1 (single silanol; OSi(OH)2R), T2

(geminal silanol; O2Si(OH)R), and T3 (siloxane; O3SiR) structures,respectively (Fig. 8) [50,51]. The presence of Tn bands suggestscovalent linkages between the organic group and the backboneof PFAS molecules (i.e., Si O bridges). The ratio of peak areas fortitania grafted by C6 were equal to 3:6:10 for T1:T2:T3 structures,

respectively. Based on this results it can be concluded that C6molecules were bonded by three oxygen atoms on powder surfacewith the highest efficiency. On the other hand, grafting mechanismof C12 molecules on the titania oxide surface was different. The

ditions: QTiO2= 0.75 mmol g-1, CC6/C12 = 0.05 M, tmod = 72 h, Tmod = 21 ◦C.

Page 7

70 J. Kujawa et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 420 (2013) 64– 73

Table 6Characterization of grafting efficiency of metal oxide powders. Grafting conditions: CC6,C12 = 0.05 M, QMeOx = 0.75 and 1.50 mmol g−1, tmod = 72 h, Tmod = 21 ◦C.

Powder Mass loss (% m/m) Surface concentrationof PFAS (mol m−2)

Mass loss (% m/m) Surface concentrationof PFAS (mol m−2)

QMeOx = 1.5 (mmol g−1) QMeOx = 0.75 (mmol g−1)

Al2O3 grafted by C6 20.6 5.3 × 10−6 16.6 4.3 × 10−6

Al2O3 grafted by C12 35.5 5.1 × 10−6 26.1 3.8 × 10−6

TiO2 grafted by C6 9.3 16.6 × 10−6 9.6 17.0 × 10−6

0−6 −6

10−6

0−6

rf(orc

3

cc

3

gwweAulg

putat2Cmcc

FG

efficiency is presented in Fig. 10. Alumina powders were graftedduring 72 h by C6 solution in concentration range of 0.01–0.10 M. Itshould be mentioned that QAl2O3

value was constant and was equalto 1.00 mmol g−1. It can be seen, that with increasing concentration

TiO2 grafted by C12 35 34.6 × 1ZrO2-A grafted by C6 2.4 0.76 ×ZrO2-A grafted by C12 10.5 1.9 × 1

atio of peak areas for titania modified by C12 is equal to 2:7:1or T1:T2:T3structures, respectively. The geminal silanol groupsO2Si(OH)R) were bonded with the highest efficiency. This kindf behavior of C12 molecules in comparison with C6 is probablyelated with steric hindrance effect of the C12 molecules. The longerhains of C12 has a limited access to the titania surface.

.3. The influence of various parameters on the grafting process

In the subsequent sections, the more detailed discussion of thehosen parameters (Table 1) on the grafting efficiency process effi-iency will be presented.

.3.1. The impact of a type of grafting moleculeThermal analysis was carried out to determine the efficiency of

rafting by C6 and C12 molecules (Table 6). Prior to experimentsith grafted powders, the TGA analysis for non-grafted powdersas done as well. The mass losses for non-grafted samples were

qual to 6.1 ± 0.2%, 3.6 ± 0.1%, 1.2 ± 0.06% and 0.3 ± 0.01% and forl2O3, TiO2, ZrO2-A and ZrO2-B, respectively. These values weresed as a correction in the calculation of the amount of grafted

igands. This procedure allowed also determining amounts of OHroups available on the metal oxides surface.

An exemplary thermogram for the grafted alumina powder isresented in Fig. 9. The thermogram can be divided into two zones:p to 205 ◦C and above this temperature [52]. In the lower tempera-ure zone, the mass loss is relatively small (up to 1.5%) and it can bessociated with the loss of water molecules followed by degrada-ion of unbonded hydroxyl groups during grafting process. Above05 ◦C, the weight loss for the alumina oxide powders grafted by

12 (35.5%) was almost two-fold higher than that for the powderodified by C6 (20.6%). The differences in mass loss can be asso-

iated mainly with the differences in the length of fluoro-carbonhain.

ig. 9. The mass loss for non-grafted and grafted alumina by C6 and C12 solutions.rafting conditions: CC6,C12 = 0.05 M, QAl2O3

= 1.50 mmol g−1, tmod = 72 h, Tmod = 21 ◦C.

23.1 22.8 × 100.7 0.2 × 10−6

5.8 1.0 × 10−6

Grafting efficiency was expressed mainly by the mass loss.Moreover, the grafting efficiency was expressed by surface con-centration of PFAS on the unit surface (in mol m−2). The resultsobtained after modification experiments are presented in Table 6.Based on these results, it can be concluded that surface concen-trations of C6 and C12 on the Al2O3 was quite similar. Titania wascharacterized by the highest value of surface concentration. It wasrelated with the largest value of available OH groups on the unit sur-face. Moreover, titania grafted by C12 was characterized by almosttwo-fold higher value of surface concentration than modified by C6,at both QTiO2

= 1.5 and 0.75 mmol g−1. On the other hand, surfaceconcentration was quite similar for TiO2 modified by C6 at QTiO2

=1.5 and 0.75 mmol g−1. For ZrO2-A with decreasing QZrO2 value from1.5 to 0.75 mmol g−1 a decrease of grafting efficiency was noticed.The largest difference in grafting efficiency with decreasing of QZrO2value was observed for zirconia modified by C6. Additionally, thelowest value of surface concentration was detected for ZrO2-A whatwas caused by the smallest amount of available OH groups on thezirconia powder.

3.3.2. The impact of concentration of grafting molecules andgrafting time

The influence of grafting solution concentration on the grafting

Fig. 10. The influence of the concentration of C6 solution on the grafting efficiency.Grafting conditions: QAl2O3

= 1.00 mmol g−1, tmod = 72 h, Tmod = 21 ◦C.

Page 8

J. Kujawa et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 420 (2013) 64– 73 71

Fd

octCCt

Iiotsww

3o

o(owrs0

bmtf

fCtaw

2TQ

ig. 11. The influence of the grafting time on the grafting efficiency. Grafting con-itions: QTiO2

= 1.00 mmol g-1, CC6 = 0.01 M, Tmod = 21 ◦C.

f grafting solution, a slight monotonic increase in grafting effi-iency was observed. Efficiency of surface modification process forhis C6 concentration range was from 4.1% (0.01 M) to 8.7% (0.1 M).hanges can be related with better availability of OH groups for the6 molecules on the alumina surface with increasing C6 concen-ration, at constant QAl2O3

ratio.The influence of grafting time was also evaluated (Fig. 11).

n this case, titania powder was modified by C6 solution. In thenvestigated time range, an increase of the grafting efficiency wasbserved up to 200 h of grafting (Fig. 11). However, during first 72 hhe highest increase of grafting efficiency can be observed. Duringubsequent modification only a small increase of grafting efficiencyas noticed. According to these results, the grafting time of 72 has chosen for all other experiments.

.3.3. Influence of the ratio of amount of PFAS to amount of metalxide powder

The results of powders modification efficiency as a functionf the ratio of C6 amount (in mmol) to metal oxides (in g)denoted as QMeOx) are presented in Fig. 12. The measured rangef QMeOx was equal to 0.15–3.00 mmol g−1 for �-Al2O3 and TiO2,hereas for ZrO2-B experiments were performed in the wider

ange 1.00–25.00 mmol C6 g−1. The concentration of C6 graftingolution during these experiments was constant and equal to.01 M. Modifications were performed at 21 ◦C for 72 h.

According to the obtained results, it can be stated that relationetween grafting efficiency and QMeOx values moved through theaximum for all metal oxides. On the other hand, the location of

he maximum of grafting efficiency was at different range of QMeOxor various metal oxides.

The highest mass loss was found for alumina oxide. Maximumor Al2O3 was located at 0.75 mmol C6 g−1 and was equal to 6.7% (forC6 = 0.01 M). Titania powder was modified at the same ratio of C6o metal amount range (0.15–3.0 mmol g−1) and C6 concentrations alumina one. Maximum of grafting efficiency for this powderas also found at QTiO2

= 0.75 mmol C6 g−1 and was equal to 4.7%.

Zirconia was modified in the wider range of QZrO2 from 1 to

5 mmol C6 g−1 and was characterized by the smallest mass losses.he maximum grafting efficiency was equal to 1.2% and located atZrO2 = 5 mmol C6 g−1.

Fig. 12. The influence of the ratio of amount of PFAS to amount of oxide powder onthe grafting efficiency. Grafting conditions: CC6 = 0.01 M, tmod = 72 h, Tmod = 21◦C.

However, if comparing the grafting efficiency expressed as a sur-face concentration of PFAS molecules on the powder surface, theefficiency decreased in following order: TiO2, Al2O3, ZrO2-B andZrO2-A. Such a large difference between modification efficiencywas related with different specific surface area of used powders(Table 3) as well as with different amounts of available OH groupson the powders surface.

3.3.4. Impact of specific surface areaThe impact of specific surface area on the grafting efficiency was

already noticed for all oxide powders. However, to eliminate theinfluence of the type of specific metal oxide, experiments wereperformed with one type of metal oxides, but characterized bydifferent specific surface areas (Fig. 13). For these experiments zir-conia powders with two different specific surface areas of 83 and5 m2 g−1 were used in a wide range of QMeOx (0.15–25 mmol g−1).

The results presented in Fig. 13 allow to conclude. The graftingefficiency expressed by the mass loss was much higher for metaloxide powders characterized by a greater specific surface area. Zir-conia with bigger specific surface area (ZrO2-A) was characterizedby around 1% higher grafting efficiency than ZrO2-B. On the otherhand, expressing the grafting efficiency as a surface concentration(moles of PFAS/specific area of metal oxide powders), zirconia witha smaller specific surface area was characterized by a higher mod-ification efficiency. Regardless the specific surface area of zirconiaoxides the maximum of grafting efficiency was observed at QZrO2equal to ca. 2.5 mmol C6 g−1 (Fig. 13).

3.3.5. Effect of grafting temperatureThe temperature influence on the grafting efficiency was eval-

uated and results are presented in Fig. 14. The modifications wereperformed for all oxide powder at four different temperatures inthe range 21–50 ◦C. Powders were modified during 72 h by 0.05 Mof C6 molecules at QMeOx = 0.75 mmol C6 g−1. In the temperatures

range 21–50 ◦C, an increase of grafting efficiency with increasingtemperature can be observed only for �-Al2O3 and TiO2. At 21 and30 ◦C, the higher efficiency was observed for �-Al2O3. The massloss were equal to 6.3% and 7.1% at temperature 21 and 30 ◦C,

Page 9

72 J. Kujawa et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 420 (2013) 64– 73

Fig. 13. The influence of specific surface area and the amount of PFAS too=

rhlTtto

3

m[sst

Ft

xide powder on the grafting efficiency. Grafting conditions: CC6 = 0.05 M, QZrO2

0.15–25.00 mmol g−1, tmod = 72 h, Tmod = 21 ◦C.

espectively. However, at elevated temperatures (40 and 50 ◦C) theigher grafting efficiency was observed for TiO2 powders. The mass

oss was equal to 10.2% and 10.7% at 40 ◦C and 50 ◦C, respectively.he changes of grafting efficiency for ZrO2 were much smallerhan for alumina and titania. Moreover, practically no tempera-ure impact on the grafting efficiency for ZrO2-A and ZrO2-B wasbserved (Fig. 14).

.4. Chemometric analysis

The obtained results were additionally analyzed by a chemo-etric simplex method following the Nelder–Mead’s algorithm

53]. It is a sequential method based on a regular search for response

urface and selection of new parameters, and then repeating thisequence to obtain the parameters producing the optimal value ofhe objective function [43,54].

ig. 14. The influence of the temperature on the grafting efficiency. Grafting condi-ions: CC6 = 0.05 M, QMeOx = 0.75 mmol g−1, Tmod = 72 h.

Fig. 15. Simplex optimization results. Grafting conditions for experimental part:CC6 = 0.05 M, tmod = 72 h, Tmod = 21 ◦C.

The grafting results were analyzed by using the simplex methodto determine the optimal grafting conditions. Because the most ofresults were obtained for alumina oxide powder modified by C6molecules, these results were chosen for the analysis by a simplexchemometric method. The weight loss obtained in thermogravi-metric analysis was chosen as an objective function, whereas theratio of C6 to metal oxide powders (QMeOx) and concentration ofC6 molecules (CC6) were chosen as control variables. Based on thesimplex model, the optimal conditions were found and then wascompared with experimental values. Results of the optimizationare presented in Fig. 15.

The chemometric analysis allowed concluding that QAl2O3equal

to 0.75 mmol g−1 was the most efficient parameter for modificationof alumina oxide powder. According to results of simplex analysis,the grafting efficiency at these conditions was equal to 35.7%. Theobserved value of grafting efficiency achieved during experimentswas equal to 34.6% and was consistent with the predicted valueshowing the good validity of the simplex method.

Moreover, as it is seen from Fig. 15, the grafting efficiency ofalumina oxide is only slightly dependent on the concentration ofgrafting solution.

4. Conclusions

In presented work three types of metal oxide powders were effi-ciently modified. The mechanism of grafting process was proposedand described by a 29Si NMR and FT-MIR techniques. The followingobservations and conclusions can be drawn out from the obtainedresults:

• The efficient change of metal oxide powders character fromhydrophilic into hydrophobic was proven using various analyticaltechniques (Fig. 5).

• A type of PFAS molecules, ratio of PFAS to the amount of metaloxides as well as modification time had a strong influence on thegrafting efficiency.

• Strong impact of temperature on the grafting efficiency wasobserved for titania and alumina oxide powders.

Page 10

: Phys

•

•

A

f

w

R

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

J. Kujawa et al. / Colloids and Surfaces A

The highest grafting efficiency was observed for alumina oxide,it was related with the highest specific surface area equal to103 m2 g−1.The simplex method can be efficiently used to define the optimalcondition of grafting of metal oxide powders.

cknowledgements

This research is supported by MNiSzW nr NN 209 255138 grantrom the Polish Ministry of Science and Higher Education.

Special thanks to M.Sc. Karolina Jarzynka for her kind assistanceith the text editing.

eferences

[1] P.K. Rohatgi, B.C. Pai, S.E. Panda, Preparation of cast aluminium-silica particu-late composites, J. Mater. Sci. 14 (1979) 2283.

[2] X. Quan, S.G. Yang, X.L. Ruan, H.M. Zhao, Preparation of titania nanotubes andtheir environmental applications as electrode, Environ. Sci. Technol. 39 (2005)3770–3775.

[3] K.I. Hadjiivanov, D.G. Klissurski, ChemInform abstract: surface chemistry oftitania (anatase) and titania-supported catalysts, Chem. Soc. Rev. 25 (1996) 25,61.

[4] S. Shingo Watanabe, X. Ma, C. Song, Characterization of structural and surfaceproperties of nanocrystalline TiO2−CeO2 mixed oxides by XRD, XPS, TPR, andTPD, J. Phys. Chem. C 113 (2009) 14249–14257.

[5] S. Cerneaux, I. Struzynska, W.M. Kujawski, M. Persin, A. Larbot, Comparisonof various membrane distillation methods for desalination using hydrophobicceramic membranes, J. Membr. Sci. 337 (2009) 55–60.

[6] W. Kujawski, S. Krajewska, M. Kujawski, L. Gazanges, A. Larbot, M. Persin, Per-vaporation properties of fluoroalkylosilanes (FAS) grafted ceramic membrane,Desalination 205 (2007) 75–86.

[7] S. Krajewski, W. Kujawski, M. Bukowska, C. Picard, A. Larbot, Application offluoroalkylsilanes (FAS) grafted ceramic membranes in membrane distillationprocess of NaCl solutions, J. Membr. Sci. 281 (2006) 253–259.

[8] A. Larbot, L. Gazanges, S. Krajewski, M. Bukowska, W. Kujawski, Water desali-nation using ceramic membrane distillation, Desalination 168 (2004) 367–372.

[9] S.R. Krajewski, W. Kujawski, F. Dijoux, C. Picard, A. Larbot, Grafting of ZrO2 pow-der and ZrO2 membrane, by fluoroalkylsilanes, Colloids Surf., A: Physicochem.Eng. Aspects 243 (2004) 43–47.

10] J.R. Miller, W. Koros, The formation of chemically modified �-alumina micro-porous membranes, Sep. Sci. Technol. 13 (1990) 1257–1280.

11] W. Yoshida, Y. Cohen, Removal of methyl tert-butyl ether from water by per-vaporation using ceramic-supported polymer membranes, J. Membr. Sci. 229(2004) 27–32.

12] P. Blain, M.N. Wayde, F.L. Ray, Surface modification of alumina nanofibres forthe selective adsorption of alachlor and imazaquin herbicides, J. Colloid Inter-face Sci. 360 (2011) 132–138.

13] R. Krishnan, S. Dash, R. Kesavamoorthy, C. Babu Rao, A.K. Tyagi, B. Raj, Laser sur-face modification and characterization of air plasma sprayed alumina coatings,Surf. Coat. Technol. 200 (2006) 2791–2799.

14] O.K. Varghese, D.W. Gong, M. Paulose, K.G. Ong, C.A. Grimes, Hydrogen sensingusing titania nanotubes, Sens. Actuators, B 93 (2003) 338–344.

15] O.K. Varghese, D.W. Gong, M. Paulose, K.G. Ong, E.C. Dickey, C.A. Grimes,Extreme changes in the electrical resistance of titania nanotubes with hydrogenexposure, Adv. Mater. 15 (2003) 624–627.

16] G.K. Mor, M.A. Carvalho, O.M. Varghese, M.V. Pishko, C.A. Grimes, A room-temperature TiO2-nanotube hydrogen sensor able to self-clean photoactivelyfrom environmental contamination, J. Mater. Res. 19 (2004) 628–634.

17] S. Livraghi, A. Votta, M.C. Paganini, E. Giamello, The nature of paramagneticspecies in nitrogen doped TiO2 active in visible light photocatalysis, Chem.Commun. 4 (2005) 498–500.

18] G.K. Mor, K. Shankar, M. Paulose, O.K. Varghese, C.A. Grimes, Enhanced photo-cleavage of water using titania nanotube arrays, Nano Lett. 5 (2005) 191–195.

19] T. Lopez, R. Fernando, R. Katz, F. Galindo, F. Balankin, A. Buljan, Porosity, struc-tural and fractal study of sol–gel TiO2–CeO2 mixed oxides, J. Solid State Chem.177 (2004) 1873–1885.

20] A. Hagfeldt, M. Gratzel, Light-induced redox reactions in nanocrystalline sys-tems, Chem. Rev. 95 (1995) 49–68.

21] W.M. Xi, S.U. Geissen, Separation of titanium dioxide from photocatalyticallytreated water by cross-flow microfiltration, Water Res. 35 (2001) 1256–1262.

22] R. Molinari, L. Palmisano, E. Drioli, M. Schiavello, Studies on various reactorconfigurations for coupling photocatalysis and membrane processes in waterpurification, J. Membr. Sci. 206 (2002) 399–415.

23] T. Yamaguchi, J.W. Hightower, Hydrogenation of 1,3-butadiene with 1,3-cyclohexadiene and molecular deuterium over zirconium dioxide catalysts, J.Am. Chem. Soc. 99 (1977) 4201–4203.

24] Y. Nakano, T. Iizuka, H. Hattori, K. Tanabe, Surface properties of zirconium oxideand its catalytic activity for isomerization of 1-butene, J. Catal. 57 (1979) 1–10.

[

icochem. Eng. Aspects 420 (2013) 64– 73 73

25] J. Nawrocki, M.P. Rigney, A. McCormick, P.W. Carr, Chemistry of zirconia andits use in chromatography, J. Chromatogr., A 657 (1993) 229–282.

26] G. Srinivasan, M. Pursch, L.C. Sander, Müller FTIR studies of C30 self-assembledmonolayers on silica, titania, and zirconia, Langmuir 20 (2004) 1746–1752.

27] B. Jongsomjit, J. Panpranot, J.G. Goodwin Jr., Effect of zirconia-modified aluminaon the propertiesof Co/�-Al2O3 catalysts, J. Catal. 215 (2003) 66–77.

28] R.S. Deinhammer, M. Ho, J.W. Anderegg, M.D. Porter, Electrochemical oxidationof amine-containing compounds: a route to the surface modification of glassycarbon electrodes, Langmuir 10 (1994) 1306–1313.

29] L. Djafer, A. Ayral, B. Boury, R. M. Laine, Surface modification oftitania powder P25 with phosphate and phosphonic acids – effecton thermal stability and photocatalytic activity, J. Colloid Interf. Sci.,http://dx.doi.org/10.1016/j.jcis.2012.11.002

30] N. Stevens, S. Tedeschi, K. Powers, B. Moudgil, H. El-Shall, Controlling uncon-fined yield strength in a humid environment through surface modification ofpowders, Powder Technol. 191 (2009) 170–175.

31] Y. Ouabbasa, A. Chamayou, L. Galet, M. Baron, G. Thomas, P. Grosseau, B. Guilhot,Surface modification of silica particles by dry coating: characterization andpowder ageing, Powder Technol. 190 (2009) 200–209.

32] C. Picard, A. Larbot, F. Guida-Pietrasanta, B. Boutevin, A. Ratsimihety, Graftingof ceramic membranes by fluorinated silanes: hydrophobic features, Sep. Purif.Technol. 25 (2001) 65–69.

33] C. Picard, A. Larbot, E. Tronel-Peyroz, R. Berjoan, Characterisation of hydrophilicceramic membranes modified by fluoroalkylsilanes into hydrophobic mem-branes, Solid State Sci. 6 (2004) 605–612.

34] W. Kujawski, S. Krajewska, M. Kujawski, L. Gazanges, A. Larbot, M. Persin, Per-vaporation properties of fluoroalkylosilanes (FAS) grafted ceramic membrane,Desalination 205 (2007) 75–86.

35] J. Lu, Y. Yu, J. Zhou, L. Song, X. Hu, A. Larbot, FAS grafted superhydrophobicceramic membrane, Appl. Surf. Sci. 255 (2009) 9092–9099.

36] S. Khemakhem, R.B. Amar, Modification of Tunisian clay membrane surface bysilane grafting: application for desalination with Air Gap Membrane Distillationprocess, Colloids Surf., A: Physicochem. Eng. Aspects 387 (2011) 79–85.

37] S. Khemakhem, R.B. Amar, Grafting of fluoroalkylsilanes on microfiltrationTunisian clay membrane, Ceram. Int. 37 (2011) 3323–3328.

38] F. Gabino, M.-P. Belleville, L. Preziosi-Belloy, M. Dornier, J. Sanchez, Evaluationof the cleaning of a new hydrophobic membrane for osmotic evaporation, Sep.Purif. Technol. 55 (2007) 191–197.

39] A. Vargas-Garcia, B. Torrestiana-Sanchez, A. Garcia-Borquez, G. Aguilar-Uscanga, Effect of grafting on microstructure, composition and surface andtransport properties of ceramic membranes for osmotic evaporation, Sep. Purif.Technol. 80 (2011) 473–481.

40] G.D. Bothun, K. Peay, Sh. Ilias, Role of tail chemistry on liquid and gas transportthrough organosilane-modified mesoporous ceramic membranes, J. Membr.Sci. 301 (2007) 162–170.

41] D. Schondelmaier, S. Cramm, R.J. Klingeler Morenzin, Ch. Zilkens, W. Eberhardt,Orientation and self-assembly of hydrophobic fluoroalkylsilanes, Langmuir 18(2002) 6242–6245.

42] S. Koonaphapdeelert, K. Li, Preparation and characterization of hydrophobicceramic hollow fibre membrane, J. Membr. Sci. 291 (2007) 70–76.

43] J. Kujawa, W. Kujawski, S. Koter, K. Jarzynka, A. Rozicka, K. Bajda, S.Cerneaux, M. Persin, A. Larbot, Membrane distillation properties of TiO2

ceramic membranes modified by perfluoroalkylsilanes, Desalin. Water Treat.,http://dx.doi.org/10.1080/19443994.2012.704976

44] J. Kujawa, W. Kujawski, S. Cerneaux, K. Jarzynka, M. Persin, A. Larbot, Trans-port and selective properties of surface grafted ceramic membranes in air-gapand direct-contact membrane distillation process, Procedia Eng. 44 (2012)437–439.

45] S. Desset, O. Spalla, P. Lixon, B. Cabane, From powders to dispersions in water:effect of adsorbed molecules on the redispersion of alumina particles, Langmuir17 (2001) 6408–6418.

46] B.C. Lippens, J.H. de Boer, Studies on pore systems in catalysts: V. The t method,J. Catal. 4 (1965) 319–323.

47] G. Jura, W.D. Harkins, The contact angle between water and a monolayer of eggalbumin on glass as a function of film pressure, J. Colloid Sci. 1 (1946) 137–140.

48] C. Scherdel, G. Reichenauer, M. Wiener, Relationship between pore volumesand surface areas derived from the evaluation of N2-sorption data by DR-, BET-and t-plot, Microporous Mesoporous Mater. 123 (2010) 572–575.

49] K. Albert, E. Bayer, Characterization of bonded phases by solid-state NMR spec-troscopy, J. Chromatogr., A 544 (1991) 345–370.

50] S. Huh, J.W. Wiench, J. Yoo, M. Pruski, V. Lin, Organic functionalization and mor-phology control of mesoporous silicas via a co-condensation synthesis method,Chem. Mater. 15 (2003) 4247–4256.

51] M.P. Rigney, E.F. Funkenbusch, P.W. Carr, Physical and chemical characteriza-tion of microporous zirconia, J. Chromatogr., A 499 (1990) 291–304.

52] J.A. Nelder, R. Mead, A simplex method for function minimization, Comput. J. 7(1965) 308–313.

53] R. Wódzki, J. Ceynowa, Simplex design method for planning the optimum

experiments, J. Pol. Chem. Soc. 30 (1976) 337.

54] J.A. Mendiola, P.J. Martin –Alvarez, F. Javier Senorans, G. Reglero, A. Capodicasa,F. Nazzaro, A. Sada, A. Cifuentes, E. Ibanez, Design of natural food antioxidantingredients through a chemometric approach, J. Agric. Food Chem. 8 (2010)787–792.