Materials Science and Engineering C 42 (2014) 280–288

Contents lists available at ScienceDirect

Materials Science and Engineering C

j ourna l homepage: www.e lsev ie r .com/ locate /msec

Plackett–Burman experimental design for bacterial cellulose–silicacomposites synthesis

Anicuta Stoica Guzun a, Marta Stroescu a,⁎, Sorin Ion Jinga a, Georgeta Voicu a,Alexandru Mihai Grumezescu a, Alina Maria Holban b

a University “Politehnica” of Bucharest, Faculty of Applied Chemistry and Materials Science, Polizu 1–3, Bucharest 011061, Romaniab University of Bucharest, Faculty of Biology, Department of Microbiology, 91–95 Spl. Independentei, Bucharest 050095, Romania

⁎ Corresponding author at: Department of ChemicalFaculty of Applied Chemistry and Materials ScienceBucharest, Polizu 1–3, Bucharest 011061, Romania. Tel.: +

E-mail address: marta_stroescu@yahoo.com (M. Stroe

http://dx.doi.org/10.1016/j.msec.2014.05.0310928-4931/© 2014 Elsevier B.V. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 10 March 2014Received in revised form 22 April 2014Accepted 6 May 2014Available online 27 May 2014

Keywords:Bacterial celluloseSilica compositePlackett–Burman designAntimicrobialStaphylococcus aureus

Bacterial cellulose–silica hybrid compositeswere prepared starting fromwet bacterial cellulose (BC)membranesusing Stöber reaction. The structure and surface morphology of hybrid composites were examined by FTIR andSEM. The SEM pictures revealed that the silica particles are attached to BC fibrils and are well dispersed in theBC matrix. The influence of silica particles upon BC crystallinity was studied using XRD analysis. Thermogravi-metric (TG) analysis showed that the composites are stable up to 300 °C. A Plackett–Burman design was appliedin order to investigate the influence of process parameters upon silica particle sizes and silica content of BC–silicacomposites. The statistical model predicted that it is possible for silica particles size to vary the synthesis param-eters in order to obtain silica particles deposed on BC membranes in the range from 34.5 to 500 nm, the signif-icant parameters being ammonia concentration, reaction time and temperature. The silica content also variesdepending on process parameters, the statistical model predicting that the most influential parameters arewater-tetraethoxysilane (TEOS) ratio and reaction temperature. The antimicrobial behavior on Staphylococcusaureus of BC–silica composites functionalized with usnic acid (UA) was also studied, in order to create improvedsurfaces with antiadherence and anti-biofilm properties.

© 2014 Elsevier B.V. All rights reserved.

1. Introduction

Hybrid inorganic–organic composites are versatile materials withhuge application areas from regenerative medicine to industry. Theorganic material is often a synthetic polymer, the use of biopolymersbeing more advantageous because of their biodegradability and bio-compatibility, despite their higher cost. The interaction between theorganic material and the inorganic substrate may have synergeticeffects, which leads to the development of hybrid materials with im-proved properties [1,2]. Hybrid polymer/inorganic materials were alsoused as efficient bio-active surfaces, being suitable for the developmentof stabilizing and controlled release nanosystems [3–5]. Despite theirproved biomedical potential, especially in infection control, severalnatural products, as highly volatile essential oils and their major frac-tions (eugenol, limonene, usnic acid etc.) fail to manifest their fullpotential due to their low stability and high volatility, which leads tothe necessity of being used in high therapeutic doses, which are usuallytoxic [6–8]. Usnic acid (UA) is one of such compound, very efficient to

and Biochemical Engineering,, University “Politehnica” of40 021 402 3870.scu).

eradicate bacterial infections, especially with Gram positive etiologyand it may be considered as an efficient alternative in the therapy ofStaphylococcus aureus infections [9]. However, studies reported thatthis efficient antimicrobial compound has also a significant citotoxicityin vitrowhen it is used at antibacterial doses [9]. For this instance, devel-oping novel, biocompatible and biodegradable nanometric polymericshuttles, which may be used in increasing the efficiency of this com-pound at lower doses, thus reducing its citotoxicity, represent a currentfocus.

Among the numerous candidates for hybrid inorganic–organic com-posites biopolymer/silica nanocomposites were already reported asbeing suitable for the design ofmembranes and coatings, for the entrap-ment of bioactive molecules (like enzyme, drugs, cell markers) or forbiosensors and bioreactors [10–14].

From the biopolymer class, cellulose and bacterial cellulose are alsoversatile materials which could be used to obtain inorganic–organiccomposites. Until now many composite cellulose–silica compositeswere reported using different types of cellulose and different synthesismethods [15–19]. Bacterial cellulose (BC), awell-knownmicrobial poly-mer, could offer more possibilities to obtain biopolymer/silica compos-ites, due to its superior properties in comparison with cellulose fromplants, excepting its higher price. Exhibiting an ultrafine network struc-ture and porous channels, BC could be used as a suitable template for insitu preparation of metal oxides through the sol–gel reaction [20–23].

281A.S. Guzun et al. / Materials Science and Engineering C 42 (2014) 280–288

BC–silica composites were also obtained using the solution impreg-nationmethod. Even if thismethod could improve themechanical prop-erties of the obtained composites, it is difficult to load BC sheets withmore than 10% silica [24–26].

Among the silica synthesis methods, Stöber synthesis is very popu-lar, because it leads to spherical and narrow dispersed silica micro andnanoparticles and because it is also considered simple by involvinghydrolysis followed by polycondensation of the precursor under alka-line conditions [27,28]. Therewith, this method is intensively studiedand the influence of the working parameters on the size and particledistribution is well known. For example, the most effective parametersare: concentration of tetraethoxysilane (TEOS), ammonia and water,polarity of the solvent and the reaction temperature [29–31].

When many quantitative and qualitative variables are involved in aprocess, statistical experimental design methods are widely used foroptimization studies. Among these statistical methods, Plackett–Burmanscreening design is extremely useful in preliminary studies in order toidentify the most important factors that could influence a studied phe-nomenon. The main advantage of this experimental design is to offerthe possibility to screen a large number of parameters using a relativesmall number of experiments. The model doesn't describe the interac-tion between the parameters, because it is based on a first-order poly-nomial model [32].

To our knowledge the Stöber reaction hasn't been used to prepareBC–silica composite membranes and also Plackett–Burman experimen-tal design hasn't been used to appreciate the influence of synthesis pa-rameters upon silica particles' size and on silica content of BC–silicacomposites.

The aim of this study was to depose silica on BC membranes byStöber reaction, to characterize the influence of synthesis parameterson the silica particles' size and silica content by Plackett–Burmandesign and to test the antimicrobial effect of this material functional-ized with UA on S. aureus, in order to create improved surfaces withantiadherence and anti-biofilm properties.

2. Materials and methods

2.1. Materials

Unless otherwise stated, all chemicals and reagents were suppliedby Sigma-Aldrich and were used without further purification.

2.2. Production and purification of bacterial cellulose membranes

Bacterial cellulose membranes were produced by Gluconacetobactersp. strain isolated from traditionally fermented apple vinegar in theMicrobiology Laboratory of Chemical and Biochemical EngineeringDepartment of University Politehnica of Bucharest. The culture wasgrown in a modified Hestrin–Schramm (MHS) medium containing 2%fructose using a static culture at 28 °C for 7 days. The pellicles obtainedwere purified by treatment with 0.5 N NaOH aqueous solutions at 90 °Cfor 1 h to eliminate the bacterial cells and then washed with deionizedwater until the pH of water became neutral. BC pellicles were used aswet membranes.

2.3. Composite BC–silica membranes preparation

Tetraethoxysilane (TEOS) (98% analytical grade) was submitted tohydrolysis/condensation process in situ on the bacterial cellulose wetmembranes. Bacterial cellulose pellicle was immersed in an Erlenmeyerflask containing ethanol, distilled water and ammonium hydroxide indifferent concentration accordingwith Plackett–Burman design. Finally,TEOSwas added and the reactionwas allowed to proceed for one or twohours at two different temperatures (respectively 20 °C and 50 °C) inaccordance with factorial experiments. The BC–silica composites were

then collected, washed with ethanol and distilled water and dried at60 °C, over 24 h.

2.4. Composite BC–silica membranes as bioactive surfaces

2.4.1. PreparationThe preparation of bioactive surfaces was performed by coating a

1 cm2 section of BC–silica membrane with usnic acid (UA). The layerof the usnic acid on the BC–silica sections was achieved by submergingthe surfaces in 5 mL of phytofluid prepared from usnic acid and chloro-form (UA/CHCl3 1% w/v), followed by extemporaneous drying at roomtemperature. The rapid dryingwas facilitated by the convenient volatil-ity of chloroform. The coated surfaceswere then sterilized by ultravioletirradiation for 15 min, for further utilization in microbiological assay.

2.4.2. Bacterial adherence and biofilm formation assayS. aureus ATCC 25922 strain was used to create an artificial biofilm.

Adherence and biofilm formation were analyzed in 6 multi-well plates(Nunc), using a static model for monospecific biofilm development.Briefly, 1 cm2 pieces of the testedmaterials were placed in 2mLnutrientbroth inoculated with 5 μL of an S. aureus suspension with a density of0.5 McFarland. After 24 h of incubation, adherence was assessed byharvesting the adhered cells, following the viable cell count assay. Thetemporal dynamic of biofilm formation was studied by harvesting thebiofilm developed on the BC–silica surfaces in nutritive broth at 24, 48and 72 h. After incubation, the biofilm developed on the control andtest surfaces was harvested and suspensions were performed in sterilePBS (phosphate saline buffer). The obtained suspensions were furtherten-fold diluted and 5 μL of each dilution were seeded in triplicates onnutritive agar (BioMerieux), incubated for 24 h at 37 °C in order todetermine the viable cell counts and thus, indirectly, to evaluate thenumber of viable bacterial cells embedded in biofilms. All experimentalprocedures were repeated three times and triplicates were used foreach experiment. Fig. 1 presents the schematic illustration of expectedbiofilm development on the coated/uncoated BC–silica composites.

2.5. BC–silica composites characterization

The BC–silica composites were examined on a Jasco FT/IR6200 spec-trometer (ABL& E-JASCO Romania) with Intron μ Infrared Microscopewith ATR-1000-VZ objective. The spectra were the average of 50 scansrecorded at a resolution of 4 cm−1 in a range from 4000 to 500 cm−1

with a TGS detector.For morphological observations a scanning electron microscope

HITACHI S-2600N (Hitachi Romania, Japan) was used, with resolutionin secondary electron image 4.0 nm, electron gun with filament: tung-sten hairpin type; accelerating voltage: 0.5–30 kV, emission current10–12–10–7 A. The work conditions were: accelerating voltage: 25 kV,WD (working distance) 13 mm and beam 30, in a good agreementwith the physical characteristics of the sample. All samples were goldcoated prior to SEM examination.

The X-ray diffraction (XRD) measurements were conducted using aShimadzu XRD 6000 diffractometer (Ni filtered Cu-Kα radiation, 40 kV,30 mA and 0.02° step scan).

The thermal behavior of the composites was tested using thermo-gravimetric analysis on a thermal analyzer (DTG-60-Shimadzu). Theoperating conditions were: temperature range of 20 °C to 1000 °C,with a heating rate of 10 °C/min, and air flow rate of 50 mL/min.

The digital image analysis was carried out using IMAGE J 1.47f soft-ware, a program developed at the National Institute of Health of theUSA and available on Internet (http://image.nih.gov/ij).

2.6. Water vapor transmission rate

Water vapor transmission rate (WVTR) was determined using amodified ASTMmethod (ASTME 96-95, 1989). The test cups were filled

Fig. 1. Schematic representation of the expectedmicrobial biofilm development on the uncoated and coated BC–silica composites: (a) uncoated BC–silica composites; (b) coated BC–silicacomposites.

282 A.S. Guzun et al. / Materials Science and Engineering C 42 (2014) 280–288

with 20 g of silica gel (desiccant) to produce a 0% RH below the film.Silica gels were heated at 180 °C for at least 8 h prior to use for thedetermination. The cups were placed at 30 °C in a desiccator containingdistilled water and a temperature and humidity sensor. The cups wereweighed periodically over a 192 h time interval. To WVTR values weredetermined, by linear regression, as the slopes of the steady state periodof the curves of weight loss as a function of time for all the samples.

WVTR (g m−2 day−1) was calculated as follows:

WVTR ¼ JA

ð1Þ

where: J is the slope of theweight loss curves versus time (g/day) and Ais the aria of the tested films (m2).

2.7. Factorial design

In this paper a Plackett–Burman designwas applied for twelve trialsin order to evaluate the significance of seven variables grouped asformulation parameters (ratio between water-TEOS, ammonia concen-tration, ratio between ethanol–water, the presence of an emulsifierhexadecyltrimethylammonium bromide (CTABr)), and as processparameters (agitation rate, temperature and reaction time) upon silicacontent and silica mean particles' size, deposed on BC membranes.Each parameter was tested at two levels coded as (−1) for lowerlevel and (+1) for higher level, as it is depicted in Table 1. TheSTATISTICA statistical package software trial version 10.0 (Stat Soft

Table 1Factors and levels used in Plackett–Burman design matrix for BC–silica compositesobtaining.

Factors Levels

Low (−1) High (+1)

(X1)-ratio H2O/TEOS (mol/mol) 25 100(X2)-NH3 concentration (mol/L) 1 4.5(X3)-ratio ethanol/H2O (mol/mol) 0.35 3.5(X4)-CTABr concentration (mol/L) 0 10−3

(X5)-stirring rate (rpm) 50 100(X6)-temperature (°C) 25 50(X7)-reaction time (h) 1 2(X8; X9; X10; X11)-dummy factors – –

Inc., Tulsa, USA) was used for experimental design analysis and dataprocessing.

3. Results and discussion

3.1. Microstructure and FT-IR spectra of BC–silica composites

The SEM images of the studied compositeswere differentiated aboutthe particles size. So, in Fig. 2 are presented samples for which the par-ticles' sizes were 462 ± 55.75 nm (S1) and 385.4 ± 63.97 nm (S9).Spherical individual particles could be observed entrapped betweenBC fibrils. SEM pictures for trials S2 and S3 presented in Fig. 3 are ratherdifferent. In these cases the particles look like buds on the BC fibrilswhich are also coated with silica. It is evident that the silica grains aredecreasing in size in comparison with samples S1 and S9.

Fig. 4 presents the FT-IR spectra of BC and of composite samples S1,S9, S2, and S7. At first sight one could observe that the spectra of com-posite materials exhibit both BC and silica absorption peaks and alsopossible overlapping between some strong bands of the two compo-nents: BC and silica. BC spectrum presents characteristic bands like3341 cm−1 (stretching of hydroxyl groups involved in intramolecularhydrogen bonding), 2894 cm−1 (\CH2 stretching vibrations),1645 cm−1 (due to deformation vibrations of the absorbed water mol-ecules), 1160 cm−1 (C\O\C asymmetric stretching vibration from theglycosidic ring), 1106 cm−1 (C\O and C\C stretching vibrations, ring(polysaccharides)) 1054 cm−1 (C\O symmetric stretching of primaryalcohol), and 900 cm−1 (CH out of bending vibrations). There aresome notable differences between BC spectrum and the spectra of com-posite materials. Composite spectra present a decrease of the intensityof C\H absorbance group, while the intensity of broad absorbanceband in the 3300–3400 cm−1 shifts to higher wavenumber in compar-isonwith BC spectrum. Broaden peak for the BC–silica samples at wave-number 1078 cm−1 could be attributed to an overlapping effectbetween silica and bacterial cellulose. As it is already known the FT-IRspectrum of colloidal silica particles shows absorption band arisingfrom asymmetric vibration of Si\O at 1090 cm−1 and BC also has char-acteristic peaks in the same region. For composite spectra it is also visi-ble at peaks 956 cm−1 (S2) and 961 cm−1 (S7)which could be assignedto silica (asymmetric vibration of Si\OH) [24,33–35]. Similar spectrawere obtained for cellulose–silica composites and for BC–silica compos-ites [15,24,36].

Fig. 2. SEM pictures for trials S1 and S9 at different magnifications.

283A.S. Guzun et al. / Materials Science and Engineering C 42 (2014) 280–288

3.2. X-ray diffraction analysis

Fig. 5 shows XRD patterns for BC and BC–silica composites S1, S2, S5,S9 and S12. Native BC shows three strong Bragg peaks observed at14.44°, 16.77° and 22.7°. The same peaks are observed for all the studiedcomposite sampleswhich confirm the fact the deposited silica nano andmicroparticles apparently are not affecting BC crystalline structure. A

Fig. 3. SEM pictures for trials S2 and

suggestion of Yano et al. 2008 could be mentioned in relation with ourresults [26]. The aforementioned authors have obtained BC–silica com-posite by adding silica sol in the BC fermentation medium and haveobserved on the XRD patterns that the height of the peak observed at16.77° decreased as the amount of silica was increased and also thatthe ratio of the 16.77° peak to the 22.7° peakdecreaseswith the increaseof silica content. The explanation of this behavior is the possibility of

S3 at different magnifications.

Fig. 4. FT-IR spectra of bacterial cellulose (BC) and of composite samples S1, S2, S7, and S9.

Fig. 6. Thermogravimetric curves for BC and composite samples S3, S10, and S12.

284 A.S. Guzun et al. / Materials Science and Engineering C 42 (2014) 280–288

silica particles to disrupt the bundling ofmicrofibrils into ribbons. In ourexperiments silica particleswere formed by conducting the Stöber reac-tion on BC wet membranes in different experimental conditions, so it isalso possible that suchmechanism of disruption to take place in specificreaction conditions. In Fig. 5 one can observe that there are differencesbetween the intensity of the 16.77° peaks of the BC–silica composites,but we have not found a significant statistical correlation between theintensity of this peak and the silica content of the samples.

3.3. Thermal analysis

The thermal behavior of the composite was studied using thermo-gravimetric analysis (TG). Typical plots of weight loss versus tempera-ture for the studied samples and BC are presented in Fig. 6. These TGcurves indicated a distinct total weight loss between different samplesof BC–silica composites. The highest weight loss was obtained for pure

Fig. 5. XRD patterns of BC and BC–silica composites (samples S1, S5, S9, S2, and S12).

BC. Three loss stages were observed for all the studied samples. Thefirst onewas between 30 and 200 °C (about 2–6% from the total weightloss) andwas due to thewater removal or sample dehydration. The sec-ond stage which is more significant (about 20–50% from the totalweight loss) was between 200 and 400 °C and the third was between400 and 600 °C, the mass decrease being less than in the second stage(about 10–30% from the total weight loss). The two last stages couldbe attributed to the thermal degradation and decomposition of BC inthe hybrid composites. The residual ash after heating above 600 °C con-sists of SiO2. Differential thermal analysis (DTA)was also applied for thesame samples and the DTA profiles reveal two exothermic peaks overtwo temperature ranges: 200–400 °C and 400–600 °C. These peakscould be attributed to the decomposition of the organic part of the com-positeswhich is the biopolymer (BC). Themaximumtemperature peaksobtained from DTA are presented in Fig. 7. Our results are similar tothose obtained for cellulose silica composites and which were alreadyreported [15,36].

3.4. Water vapor permeability of BC–silica composites

Moisture permeability through a polymer or a compositematerial isan important characteristic useful in different applications. For example,in some applications, high barrier to migration of gases and vapors isdesirable, while for others it is undesirable. The permeability of a mate-rial to gases is dependent on a large number of intercorrelated factors. In

Fig. 7. DTA maximum temperature peaks for BC and composites S3, S10 and S12.

Fig. 8. WVTR values for BC–silica composite samples S1–S12.

285A.S. Guzun et al. / Materials Science and Engineering C 42 (2014) 280–288

a brief enumeration the following are among the most important ofthem: temperature, pressure gradient, relative humidity of the environ-ment film thickness, density and composition, pore structure and size,crystallinity and hydrophilic–hydrophobic ratio of the film componentsfor composite films [37,38]. In this study, WVTR was measured in orderto understand the effect of silica contents on composite films. Fig. 8presents the WVTR values of all composite films. For pure BC we havemeasured aWVTR value of 867.203 g/m2 day. From the values obtainedone could observe that for all the composite samples the WVTR valuesare lower than for BC, which is known as a very hydrophilic biopolymer.This result is in a good agreement with other results already reported inliterature, but for cellulose–silica composite. For example, Pinto et al.(2008) have measured values of weight percentage of water retentionfor cellulose–silica composite lower than those for cellulose, thedecrease being higher with the increase of silica content [16]. Inorder to verify whether this assumption is also true for our sampleswe have tested if a statistical correlation exists between the mea-suredWVPR values and silica content. Because we have not obtaineda good correlation, we have tried to see if a statistical correlation ispossible between the measured WVTR values, silica content and parti-cle mean diameters measured from SEM images. In this case we havetested a second order polynomial model using STATISTICA softwareand we have obtained a value for coefficient of determination R2 of0.9234, which indicates that 92.34% of the variability of the responsecould be explained by the model.

The model equation is:

WVTR ¼ 248:998þ 0:5853Y1þ 9:329Y2þ 0:00018Y12

þ 0:03477Y22 þ 0:01807Y1Y2 ð2Þ

where: Y1 is themean diameter of the particles (nm) and Y2 is the silicacontent (%)w/wof the studied samples, these values being presented inTable 2.

Table 2Plackett–Burman design matrix with response values: silica particle diameter (Y1) and silica c

Exp. run X1 X2 X3 X4 X5 X6 X7

1 1 1 1 −1 −1 −1 12 −1 −1 1 −1 1 1 −13 1 1 −1 −1 −1 1 −14 1 1 −1 1 1 1 −15 1 −1 1 1 −1 1 16 1 −1 1 1 1 −1 −17 −1 −1 −1 −1 −1 −1 −18 −1 1 1 −1 1 1 19 −1 1 1 1 −1 −1 −110 1 −1 −1 −1 1 −1 111 −1 1 −1 1 1 −1 112 −1 −1 −1 1 −1 1 1

Even if it is hard to explain the obtained results some remarks couldbe done. In the Nielsen tortuosity model, which is very popular forpermeation, the tortuosity is the main factor which is modified wheninorganic substances are embedded in the polymer matrix [39]. Also,the nanoparticulate fillers could produce changes to the polymermatrixitself at the interfacial region and as a consequence influence the barrierproperties [40]. We presume that for BC–silica the both mechanismsare present. Hydrogen interactions between silica hydroxyls with OHgroups of BC are also possible [25].

3.5. Factorial design results

In Table 1 the factors, codes, low and high levels in Plackett–Burmandesign were already presented. Analyzing SEM pictures of the samplesS1–S12 using Image J Software, we have measured the mean diameterof the particles (Y1) for all the composite films. More than 200 particleswere analyzed from SEM images of every sample of BC–silica compos-ites. The results obtained are presented in Table 2 and were used asdependent variable Y1 in the Plackett–Burman design. The secondresponse used in PB design was silica content (Y2) determined as theresidual ash after heating above 600 °C and which was considered toconsist of SiO2. The TG data for all the samples were used to obtainthe percentage of silica in the initial hybrid composites after also takinginto account the total weight loss for a sample of silica powder (TGcurve for silica powder is not shown). This silica content was used asthe second response (Y2) in the PB design. PB design matrix with re-sponse values: Y1 (silica particles mean diameter) and Y2 (silica con-tent) are depicted in Table 2.

The analysis of variance (ANOVA) applied to test the availability ofPB design for the response Y1 (mean diameter of silica particles) andthe results are shown in Table 3. The P-values less than 0.05 indicatethat the model terms are significant. Analysis of P-values depicted inTable 3 showed that among the variables tested, ammonia concentra-tion, time and temperature had a significant effect on the silica particlediameter. The model coefficient of determination (R2) has a value of0.9677, which indicates that 96.77% of the variability of the responsecould be explained by the model and about 3.23% of the total variationcannot be explained by this model. The F-ratio was found to be fargreater than the theoretical value with very low probability (P =0.007782) for the regression model, indicating that the regressionmodel is significant.

The polynomial model which predicts the silica particles diameterobtained from PB regression analysis is expressed in terms of codedfactors as:

Y1 ¼ 212:2201−18:8974X1þ 130:0982X2þ 19:4034X3þ 0:744X4þ 41:5857X5−61:2249X6þ 63:2426X7: ð3Þ

The Pareto chart could be also used to identify the significant factors.As it could be seen in Fig. 9a three parameters (ammonia concentration(X2), time (X7) and temperature (X6)) have a confidence level greater

ontent (Y2).

X8 X9 X10 X11 Y1 (nm) Y2 (%) (w/w)

−1 1 1 −1 456.18 ± 35.75 11.4 ± 0.371 1 1 −1 70.27 ± 10.01 41.5 ± 0.351 1 −1 1 89.1 ± 17.02 24 ± 1.01−1 −1 1 −1 270 ± 13.38 38.7 ± 0.661 −1 −1 −1 52.096 ± 8.05 42.5 ± 1.08−1 1 −1 1 72.56 ± 13.38 19.5 ± 0.35−1 −1 −1 −1 43.3 ± 6.82 36.2 ± 0.58−1 −1 −1 1 390 ± 33.97 55.8 ± 0.571 −1 1 1 348.63 ± 39.37 30 ± 0.851 −1 1 1 220 ± 35.75 20.1 ± 0.501 1 −1 −1 500 ± 63.97 45.4 ± 0.29−1 1 1 1 34.5 ± 7.12 50 ± 1.55

Table 3Analysis of variance (ANOVA) for the regression model of silica particle diameters usingthe Plackett–Burman design.

Term Sum of squares DF Mean square F P

Model 325646.4 7 46520.92 17.11942 0.007782a

X1 4285.3 1 4285.3 1.5770 0.277545X2 203106.7 1 203106.7 74.7420 0.000985a

X3 4517.9 1 4517.9 1.6626 0.266764X4 6.6 1 6.6 0.0024 0.962926X5 20752.5 1 20752.5 7.6368 0.050669X6 44981.9 1 44981.9 16.5531 0.015241a

X7 47995.5 1 47995.5 17.6621 0.013667a

Error 10869.7 4 2717.4Total SS 336516.2 11R2 = 0.9677 Adj R2 = 0.9111

a P b 0.05 is considered significant.

286 A.S. Guzun et al. / Materials Science and Engineering C 42 (2014) 280–288

than 95% and could be considered as significant. The results are notsurprising because it is already known that the concentration of catalystand temperature are oneof the key parameterswhich influence the par-ticles' size in Stöber reaction [11,12,27,28]. Arantes et al. have obtainedsimilar results regarding the influence of ammonia concentration uponsilica particle size in a factorial experiment for synthesis of colloidalsilica nanoparticles [41].

The influence of reaction temperature upon particle size is alsowell-known in Stöber reaction and there are more explanations for thisresult. The first one is that if the temperature increases the saturationof ammonia decreases, which then produces a lower dissolution ofTEOS and as a consequence a reduced hydrolysis rate which, in turn,decreases particles' size. The second is that the ammonia concentrationcould decrease when temperature increases by evaporation. The thirdexplanation sustains that the increase of temperature also increasesthe nucleation rate and as a consequence the particles' size is reduced[12,28]. Our results showed that there is a significant influence of tem-perature, but the minus sign in Eq. (3) suggests that a low temperatureis more favorable for the process. This result could be explained by theinfluence of BC matrix upon the synthesis of silica particles. As it isalready stated, the inorganic material is deposed in the biopolymermatrix forming hydrogen bonding between organic and inorganicphases [17,24]. These interactions are powerful enough to influencethe particle size and for this reason the increase of temperature is notnecessary. But, in the same time, we also agree with the conclusion ofBarud et al. [34], who consider that the BC structure influences thesize and size dispersion of silica deposed on this biopolymer in a waywhich is not clear enough.

Fig. 9. Pareto chart of estimated effects of seven basic factors on: a) mean diam

The second response value which was analyzed was the silica con-tent as itwas determined from thermogravimetric analysis. The polyno-mial model which predicts the silica content of BC–silica compositesobtained from PB regression analysis is expressed in terms of codedfactors as:

Y2 ¼ 32:633−8:85X1−0:1666X2−1:8333X3þ 3:0833X4þ 3:5166X5þ 6:5166X6þ 2:6166X7: ð4Þ

The analysis of variance (ANOVA) applied to test the availability ofPB design for the response Y2 and the results are shown in Table 4.

Analysis of P-values depicted in Table 4 showed that among thevariables tested H2O/TEOS ratio, and temperature had a significanteffect on the silica content of BC–silica composites. As shown inTable 4 the model coefficient of determination (R2) has a value of0.960, which indicates that 96% of the response variability could beexplained by themodel and the goodness of fit of themodel was con-firmed. In Fig. 9b the Pareto chart for variable Y2 is also presentedwhich confirms that only two parameters are very significant: water-TEOS ratio and temperature. These parameters are also reported to bevery important in silica and cellulose–silica hybrid synthesis [12]. Inthis case theminus sign for the variable X1 (water/TEOS ratio) suggeststhat the lower ratio is recommended. It is also known that the ratiobetween water and TEOS is a crucial factor in Stöber reaction, becausethe water content influences the rate of TEOS hydrolysis, which inturn affects both the hydrolysis and the condensation reactions in thesol–gel process. For this reason an excess of water is required andshould be maintained higher than 2.5 [15]. Because we have workedwith higher excess of water (ratio of water/TEOS is in the range from25 to 100) the low value is enough to ensure the recommended excesswater. To enhance the silica deposition on BC membranes our resultssuggest using a higher temperature, but in order to obtain silica parti-cles with decreasing size we do not need a high temperature. For thisreason, in order to optimize the synthesis conditions the response sur-face methodology will be used in a future work.

3.6. Antimicrobial effect of BC–silica composites

Due to the increasing issues provoked by bacteria adherence andbiofilm formation on different medical and industrial surfaces, alterna-tive methods for controlling this behavior are needed [8]. Nanostruc-tured surfaces inhibiting microbial colonization have been developedand optimized in the past 3 years [42]. Even though these modifiedsurfaces proved a great anti-microbial effect, many times they havethe disadvantage of being toxic or have a low biocompatibility [43,44].

eter of silica deposed on BCand b) silica content of BC–silica composites.

Table 4Analysis of variance (ANOVA) for the regressionmodel of silica content using the Plackett–Burman design.

Term Sum of squares DF Mean square F P

Model 1225.357 7 175.0510 14.00128 0.011335a

X1 939.87 1 939.87 52.7745 0.001907a

X2 0.653 1 0.653 0.0523 0.830394X3 8.003 1 8.003 0.6401 0.468482X4 10.830 1 10.830 0.8662 0.404687X5 40.333 1 40.333 3.2260 0.146901X6 162.803 1 162.803 13.0217 0.022586a

X7 23.520 1 23.520 1.8812 0.242099Error 50.010 4 12.503Total SS 1275.36 11R2 = 0.960 Adj R2 = 0.892

a P b 0.05 is considered significant.

Fig. 10. Graphic representation of log (CFU/mL) values obtained after the growth ofS. aureus in the presence of BC–silica–UA composites.

287A.S. Guzun et al. / Materials Science and Engineering C 42 (2014) 280–288

The use of natural, non-toxic surfaces is the most desired approach inthe development of improved surfaces to fight microbial colonization[45]. Our results using BC–silica composites demonstrate that most ofthe tested surfaces exhibit an anti-adherence and anti-biofilm activity,impairing the biofilm development of S. aureus. The most significantinhibitory effect was observed when using S1, S2, S3 and S12 compositesamples. These materials proved both anti-adherent and anti-biofilmeffects, impairing staphylococcal biofilm development for at least 72 hincubation. On the other hand, the composites S4, S5, S6, S7, S8, S9,S10 and S11 exhibited good anti-adherence profiles, but their inhibitoryactivitywas notmaintained over time (Fig. 10). The enhanced efficiencyof S1, S2, S3 and S12 composite samples may be due to their betteradsorption of the therapeutic agent (usnic acid) and better delivery ofit. Yet, additional tests may be needed for revealing the specific actionpathway in the bacteria cells since the adhesion of bacteria to mineralsurfaces can be affected by a variety of factors including pH, ionic strengthand cell surface properties. The first phase of bacterial adhesion andretention ismostly dominated by surface characteristics such as chemicalstructure, surface hydrophobicity, surface charge, surface free energy,roughness, surface area, pore structure and size [46,47]. Therefore, tobetter understand the interfacial mechanism, additional knowledgeregarding the adhesion forces and mechanisms between bacteria andmineral surfaces is required.

4. Conclusions

BC–silica hybrid materials were prepared using Stöber reaction. APlackett–Burman design was applied in order to evaluate the signifi-cance of process variables upon silica particles size and silica contentof BC–silica composites. The analysis of variance (ANOVA) applied totest the availability of PB design has revealed the significant processparameters which influence the silica particles' size and silica contentof BC–silica composites. Silica particles' size is significantly influencedby ammonia concentration, time and reaction temperature. Silica con-tent is also influenced by reaction temperature and water and TEOSratio. The BC structure also seems to control the size of silica particles,even if this influence is not very well understood. The SEM picturesreveal that the particles are well dispersed within the BC matrix andfor some reaction conditions the silica particles embedded the BC fibrils.Strong interaction between silica and BC was proved by FTIR spectra.From the transport properties, only water vapor transmission rate wasmeasured for the composites. The silica deposed on BC membranesinfluences this property of the hybrid material, all the BC–silica com-posites having a lower WVTR value that of pure BC. Our results havedemonstrated that it is possible to depose silica particles on BC wetmembranes in the range from 34.5 to 500 nm using different combina-tion between process parameters. This aspect is important in thoseapplications for which the silica particle sizes are playing a key role.

The BC–silica composites were also tested for their antimicrobialeffect, the results obtained demonstrating that some of the samplesexhibit inhibitory effect on S. aureus biofilm development. These resultsare encouraging enough to recommend that these composites be usedin biomedical applications.

Acknowledgments

Financial support from UEFISCDI-Romania (Research Grant-SABIOM-No. 114/2012) is gratefully acknowledged.

References

[1] P. Judeinstein, C. Sanchez, J. Mater. Chem. 6 (1996) 511–525.[2] G. Kickelbick, Introduction to hybrid materials, in: G. Kickelbick (Ed.), Hybrid

Materials. Synthesis, Characterization, and Applications, Wiley-VCH Verlag GmbH& Co. KGaA, Weinheim, 2007, pp. 1–48.

[3] M.T. Nistor, A.P. Chiriac, L.E. Nita, C. Vasile, Int. J. Pharm. 452 (2013) 92–101.[4] P. Loyer, W. Bedhouche, Z.W. Huang, S. Cammas-Marion, Int. J. Pharm. 454 (2013)

727–737.[5] R. Coco, L. Plapied, V. Pourcelle, C. Jérôme, D.J. Brayden, Y.-J. Schneider, V. Préat, Int. J.

Pharm. 440 (2013) 3–12.[6] A.M. Grumezescu, B.S. Vasile, A.M. Holban, Lett. Appl. Nanobiosci. 2 (2013) 120–123.[7] I. Anghel, A.M. Holban, A.M. Grumezescu, E. Andronescu, A. Ficai, A.G. Anghel, M.

Maganu, V. Lazar, M.C. Chifiriuc, Nanoscale Res. Lett. 7 (2012) 690.[8] I. Anghel, A.M. Holban, E. Andronescu, A.M. Grumezescu, M.C. Chifiriuc,

Biointerphases 8 (2013) 12.[9] A.M. Grumezescu, C. Saviuc, C.M. Chifiriuc, R. Hristu, D.E. Mihaiescu, P. Balaure, G.

Stanciu, V. Lazar, IEEE Trans. Nanobiosci. 10 (2011) 269–274.[10] Y.A. Shchipunov, T.Y.U. Karpenko, A.V. Krekoten, Compos. Interface. 11 (2005)

587–607.[11] E. Effati, B. Pourabbas, Powder Technol. 219 (2012) 276–283.[12] Z. Luo, R.Y. Hong, H.D. Xie, W.G. Feng, Powder Technol. 218 (2012) 23–30.[13] T. Angelova, N. Rangelova, R. Yuryev, N. Georgieva, R. Müller, Mater. Sci. Eng. C 32

(2012) 1241–1246.[14] Y. Zhang, L. Ye, Compos. Part B 56 (2014) 749–755.[15] S. Sequeira, D.V. Evtuguin, I. Portugal, A.P. Esculcas, Mater. Sci. Eng. C 27 (2007)

172–179.[16] R.J.B. Pinto, P.A.A.P. Marques, A.M. Barros-Timmons, T. Trindade, C.P. Neto, Compos.

Sci. Technol. 68 (2008) 1088–1093.[17] K. Xie, Y. Yu, Y. Shi, Carbohydr. Polym. 78 (2009) 799–805.[18] R.S. Gill, M. Marquez, G. Larsen, Micropor. Mezopor. Mater. 85 (2005) 129–135.[19] J. Shi, L. Lu, W. Guo, J. Zhang, Y. Cao, Carbohydr. Polym. 98 (2013) 282–289.[20] W. Zhang, S.Y. Chen, W.L. Hu, B.H. Zhou, Z.H. Yang, N. Yin, Carbohydr. Polym. 86

(2011) 1760–1767.[21] J. Gutierrez, A. Tercjak, I. Algar, A. Retegi, I. Mondragon, J. Colloid Interface Sci. 377

(2012) 88–93.[22] N. Shah, M. Ul-Islam, W.A. Khattaka, J.K. Park, Carbohydr. Polym. 98 (2013)

1585–1598.[23] W. Hu, S. Chen, J. Yang, Z. Li, H. Wang, Carbohydr. Polym. 101 (2014) 1043–1060.[24] A. Ashori, S. Sheykhnazari, T. Tabarsa, A. Shakeri, M. Golalipour, Carbohydr. Polym.

90 (2012) 413–418.[25] H. Maeda, M. Nakajima, T. Hagiwara, T. Sawaguchi, S. Yano, J. Mater. Sci. 41 (2006)

5646–5656.

288 A.S. Guzun et al. / Materials Science and Engineering C 42 (2014) 280–288

[26] S. Yano, H. Maeda, M. Nakajima, T. Hagiwara, T. Sawaguchi, Cellulose 15 (2008)111–120.

[27] W. Stöber, A. Fink, E. Bohn, J. Colloid Interface Sci. 26 (1968) 62–69.[28] M.A. Dabbaghian, A.A. Babalou, P. Hadi, E. Jannatdoust, Int. J. Nanosci. Nanotechnol. 6

(2010) 104–113.[29] J.H. Zhang, P. Zhan, Z.L. Wang, W.Y. Zhang, N.B. Ming, J. Mater. Res. 18 (2003)

649–653.[30] K.-S. Chou, C.-C. Chen, Ceram. Int. 34 (2008) 1623–1627.[31] X.-D. Wang, Z.-X. Shen, T. Sang, X.-B. Cheng, M.-F. Li, L.-Y. Chen, Z.-S. Wang, J. Colloid

Interface Sci. 341 (2010) 23–29.[32] R.L. Plackett, J.P. Burman, Biometrica 33 (1944) 305–325.[33] D. Klemm, D. Schumann, U. Udhardt, S. Marsch, Prog. Polym. Sci. 26 (2001)

1561–1603.[34] H.S. Barud, R.M.N. Assunção, M.A.U. Martines, J. Dexpert-Ghys, R.F.C. Marques, Y.

Messaddeq, S.J.L. Ribeiro, J. Sol–Gel Sci. Technol. 46 (2008) 363–367.[35] N. Pijarn, A. Jaroenworaluck, W. Sunsaneeyametha, R. Stevens, Powder Technol. 203

(2010) 462–468.[36] N. Jia, S.-M. Li, M.-G. Ma, J.-F. Zhu, R.-C. Sun, Bioresources 6 (2011) 1186–1195.[37] Y. Hu, V. Topolkaraev, A. Hiltner, E. Baer, J. Appl. Polym. Sci. 81 (2001) 1624–

1633.

[38] K.L. Spence, R.A. Venditti, O.J. Rojas, J.J. Pawlak, M.A. Hubbe, Bioresources 6 (2011)4370–4388.

[39] L.E. Nielsen, J. Macromol. Sci. Chem. A1 (5) (1967) 929–942.[40] T.V. Duncan, J. Colloid Interface Sci. 363 (2011) 1–24.[41] T.M. Arantes, A.H. Pinto, E.R. Leite, E. Longo, E.R. Camargo, Colloids Surf. A

Physicochem. Eng. Asp. 415 (2012) 209–217.[42] A.M. Grumezescu, Curr. Org. Chem. 17 (2013) 90–96.[43] M.C. Chifiriuc, A.M. Grumezescu, C. Saviuc, R. Hristu, V. Grumezescu, C. Bleotu, G.

Stanciu, D.E. Mihaiescu, E. Andronescu, V. Lazar, R. Radulescu, Curr. Org. Chem. 17(2013) 1023–1028.

[44] A.I. Cotar, A.M. Grumezescu, K.S. Huang, M.C. Chifiriuc, R. Radulescu, BiointerfaceRes. Appl. Chem. 3 (2013) 559–565.

[45] A.M. Grumezescu, A.I. Cotar, E. Andronescu, A. Ficai, C.D. Ghitulica, V. Grumezescu,B.S. Vasile, M.C. Chifiriuc, J. Nanoparticle Res. 15 (2013) 1766.

[46] M. Diao, E. Taran, S. Mahler, T.A.H. Nguyen, A.V. Nguyen, Colloids Surf. B:Biointerfaces 115 (2014) 229–236.

[47] J. Esteban, R. Pérez-Tanoira, C. Pérez-Jorge-Peremarch, E. Gómez-Barrena, Bacterialadherence to biomaterials used in surgical procedures, in: K. Kon, M. Rai (Eds.),Microbiology for Surgical Infections: Diagnosis, Prognosis and Treatment, AcademicPress-Elsevier, 2014, pp. 41–57.