polyelectrolyte complexes for textile strengthening applied to painting canvas restoration

Colloids and Surfaces A xxx (xxxx) xxx–xxx

Contents lists available at ScienceDirect

Colloids and Surfaces A journal homepage: www.elsevier.com/locate/colsurfa

Preparation of silica/polyelectrolyte complexes for textile strengthening applied to painting canvas restoration ⁎

Krzysztof Kolmana, , Oleksandr Nechyporchuka,1, Michael Perssona,b, Krister Holmberga, ⁎ Romain Bordesa, a b

Department of Chemistry and Chemical Engineering, Chalmers University of Technology, Göteborg, Sweden AkzoNobel Pulp and Performance Chemicals, Bohus, Sweden

A R T I C L E I N F O

A B S T R A C T

Keywords: Silica Polyelectrolyte Adsorption Complexes Conservation Strengthening

We here report three different approaches to prepare silica-polyelectrolyte complexes for mechanical strengthening of cotton fibers. In the first approach, polyvinylpyrrolidone (PVP) was used as a stabilizing polymer to delay the adsorption of a poly(quaternary ammonium) species, PQA (a copolymer of dimethylamine and epichlorohydrin), on the surface of silica. In the second approach cationic starch (CS), which is a branched polyelectrolyte, was used and the adsorption of CS resulted in formulations with good colloidal stability. The third approach was based on reduction of the charge density of silica to prevent PQA adsorption. Lowering the pH reduced the surface charge of the silica and enabled control of the adsorption. As a result, the aggregation was prevented and only a thin layer of polymer adsorbed. For all formulations a second polyelectrolyte, carboxymethyl cellulose (CMC) was subsequently adsorbed on the cationic polyelectrolyte layer. The silica/ polyelectrolyte formulations were evaluated by dynamic light scattering (DLS). The obtained formulations were applied on model surfaces of degraded painting canvas. The performance of the silica particles coated either with one cationic polyelectrolyte and or with a layer of cationic polyelectrolyte followed by a layer of anionic polyelectrolyte were assessed by tensile testing and the morphology of the treated samples was investigated with SEM. The particles coated with a single cationic layer increased the maximum load at break by 29% at the cost of a reduction in strain. The particles coated with a double layer increased the maximum load to a lesser extent; however, higher values of strain were recorded. For all systems the mass uptake was limited to around 5 wt%.

1. Introduction Ageing of natural cellulose fibers originates from polymer chain degradation due to acid hydrolysis and oxidation [1]. It is manifested by color change, accompanied by brittleness and loss of mechanical resistance. For many applications, and especially for valuable artefacts of cultural heritage, e.g. painting canvases, it is important to prevent or limit the problems related to ageing [2,3]. Since cellulosic materials are sensitive to acidity, not only a mechanical reinforcement of the fibers has to be achieved [4], an alkaline reserve also needs to be applied to prevent further degradation [5,6]. Silica is suitable for the purpose because apart from being alkaline in nature it has been proven to provide enhanced mechanical properties to textiles when applied as a colloid [7]. Furthermore, the growing trend in art restoration towards minimum intervention promotes the utilization of nanoparticles, which due to small size and a high surface area work efficiently, even at low

⁎

1

concentration. Nowadays, nanoparticles are commonly used in art restoration for different purposes [8]. Calcium and magnesium hydroxide/carbonate nanoparticles are known to be excellent deacidification agents [6,9]. Titanium dioxide and zinc oxide nanoparticles have been tested for protection against soiling, fungi growth and UV-induced degradation [10,11]. Wood stabilization and strengthening, as well as protection against fire, UV light, microorganisms and insects have been achieved by treatment with functionalized silica sols [12]. Good colloidal stability is essential and constitutes a prerequisite for a surface treatment based on nanoparticles. This is especially true when the nanoparticles are used for mechanical reinforcement of fibrous material. Aggregation of the nanoparticles, triggered for instance by changes in pH or ionic strength or by adsorption at a solid surface [13–15], would lead to reduced penetration into the material to be treated and would also give a less homogeneous treatment. Another aspect to consider is the nanomaterial’s compatibility with the surface

Corresponding authors. E-mail addresses: [email protected], [email protected] (K. Kolman), [email protected] (R. Bordes). Present address: Swerea IVF, SE-431 22 Mölndal, Sweden.

http://dx.doi.org/10.1016/j.colsurfa.2017.04.051 Received 15 February 2017; Received in revised form 7 April 2017; Accepted 21 April 2017 0927-7757/ © 2017 Elsevier B.V. All rights reserved.

Please cite this article as: Kolman, K.P., Colloids and Surfaces A (2017), http://dx.doi.org/10.1016/j.colsurfa.2017.04.051

Colloids and Surfaces A xxx (xxxx) xxx–xxx

K. Kolman et al.

Fig. 1. Schematic illustration of different strategies to adsorb polyelectrolytes on silica particles.

were used. All these chemical were purchased from Sigma-Aldrich. The Dowex Marathon C was washed thoroughly with ethanol and dried before use, the other chemicals were used as received. Solutions of 0.001 N poly(diallyldimethylammonium chloride) (polyDADMAC) and 0.001 N sodium poly(ethylene sulfonate) (PESNa) were used as titrants for particle charge density measurements. Plain-weave cotton canvas was purchased from BarnaArts (Barcelona, Spain). Before treatment, the canvases were first washed in a washing machine at 60 °C for 40 min (no detergent added) and then washed at 85 °C for 1 h in a glass reactor with mechanical stirring. The chemicals used for the accelerated canvas degradation, hydrogen peroxide (35 wt%) and sulfuric acid (95–97 wt%), were purchased from Fisher Scientific and Merck Chemicals, respectively.

to be treated. A lack of compatibility will lead to a less efficient treatment and will also increase the risks of release of the nanomaterial into the environment. Improved stability and chemical compatibility can often be achieved by adsorption of an oppositely charged polyelectrolyte [16]. However, this may lead to flocculation due to the strong electrostatic interactions between the particles and the polymer chains [17,18]. After the adsorption, the particle is neutralized and lacks electrostatic repulsion which leads to aggregation [19]. In this work, we present the preparation of particle/polyelectrolyte formulations consisting of colloidal silica, copolymer of dimethylamine and epichlorohydrin, cationic starch and sodium carboxymethyl cellulose. The three different strategies proposed to reduce or arrest aggregation in the preparation process of the silica/polyelectrolyte complexes are displayed in Fig. 1: (i) stabilization by neutral polymer, (ii) adsorption of polyelectrolyte with branched structure and (iii) silica surface charge reduction by pH change. In the second part of the work we performed preliminary tests regarding the application of the formulations on model surface of degraded cotton painting canvas. The treated samples were characterized by scanning electron microscopy and by assessment of the tensile strength.

2.2. Formulation preparation Schematic illustrations of the strategies to prevent silica aggregation by polyelectrolyte adsorption are presented in Fig. 1. As a reference PQA was added to silica, which led to immediate irreversible aggregation followed by sedimentation within a few minutes. For each strategy, two formulations were prepared − with and without additional adsorption of the anionic polyelectrolyte CMC. In the first approach, a stabilizing polymer, PVP, was used. The formulation stabilized with PVP was prepared as follows: 1 ml of a 50 wt% silica dispersion was mixed with 10.9 ml of ultra-pure water (Milli-Q system) and 1.75 ml of 40 wt% PVP solution. In the next step polyelectrolytes were adsorbed. First, 0.185 ml of a 6 wt% PQA solution and then 0.6 ml of a 3 wt% CMC solution were added. The second approach was based on adsorption of CS which is a branched polyelectrolyte with low charge density. In order to prepare this formulation, 1 ml of a 50 wt% silica dispersion was mixed with 5.5 ml of ultra-pure water and 5.5 ml of a 8.75 wt% solution of CS. The adsorption of the second polyelectrolyte was performed by adding 0.41 ml of a 3 wt% CMC solution. In the last strategy, the interactions between polyelectrolyte and silica were reduced by changing the pH close to that of the isoelectric point of silica. First, 1 ml of a 50 wt% silica sol was ion exchanged (IE silica) by Dowex Marathon C resin until pH 3 was reached. 0.36 ml of a 6 wt% solution of PQA at pH 3 and 5 ml of ultra-pure water were then added. The pH of the formulation was changed to 8 using Marathon A resin and 0.1 M sodium hydroxide solution. The anionic polyelectrolyte was adsorbed by addition of 1.12 ml of a 3 wt% CMC solution. All formulations were mixed using a vortex mixer and their final pH was set to 8 using 0.1 M sodium hydroxide solution and Dowex Marathon A resin. The dry mass content and the composition of each formulation are shown in Table 1.

2. Materials and methods 2.1. Materials Dispersion of silica nanoparticles Levasil CS50-28 (former Bindzil 50/80) was supplied by AkzoNobel Pulp and Performance Chemicals, Bohus, Sweden. The surface area of these particles was 80 m2/g which corresponds to a spherical diameter of 35 nm [20]. The average particle size obtained with DLS was close to 110 nm. The suspension had a pH of 9 (sodium ion stabilized), a concentration of 50 wt% silica, and a density of 1.4 g/ml. The polyelectrolytes, sodium carboxymethyl cellulose (CMC), Akucell AF0305, and a copolymer of dimethylamine and epichlorohydrin (poly(quaternary ammonium) − PQA), EkaFix 41, were supplied by AkzoNobel Pulp and Performance Chemicals. The degree of substitution of CMC was 0.77 and the viscosity of 1 wt% solution was 12 mPa·s. The weight average molecular weight of CMC was 650 000 g/ mol (determined by size exclusion chromatography). PQA was provided as a 50 wt% solution with a density of 1.15 g/ml and a pH of 5.5. The weight average molecular weight of PQA was 50 000 g/mol. Potato cationic starch (CS), Avecat 10, was supplied by Avebe Paper, the Netherlands. The degree of substitution of the CS was 0.02. The viscosity of a 15 wt% solution of the CS was 300 mPa·s. Polyvinylpyrrolidone (PVP) with weight average molecular weight of 10 000 g/mol was purchased from Sigma-Aldrich. For ion exchange and pH adjustments, resin Dowex Marathon C (H+ form), resin Marathon A (OH− form) and reagent grade sodium hydroxide and hydrochloric acid 2

Colloids and Surfaces A xxx (xxxx) xxx–xxx

K. Kolman et al.

2.7. Scanning electron microscopy

Table 1 Dry mass content and composition (without water) of the prepared formulations. Formulation

Dry mass content [wt%]

Silica Silica Silica + PVP + PQA Silica + PVP + PQA + CMC Silica + CS Silica + CS + CMC IE silica + PQA IE silica + PQA + CMC

The scanning electron microscopy (SEM) images were recorded on a LEO Ultra 55 microscope (Carl Zeiss) operating at an acceleration voltage of 3 kV. The samples of degraded warp thread were glued to the SEM holder using conductive tape. Directly before the measurement, a 10 nm gold coating was sputtered on the sample surface.

Dry mass composition [wt%] PVP

PQA

CMC

CS

10 9.6 9.5

100 48.9 48.3

– 50.3 49.7

– 0.8 0.8

– – 1.2

– – –

9.5 9.3 10.7 9.6

59.3 58.7 97.0 92.7

– – – –

– – 3.0 2.9

– 1.0 – 4.4

40.7 40.3 – –

2.8. Tensile testing The mechanical tensile tests were performed using an Instron 5565A equipped with a 100 N load cell and pneumatic clamps working at a pressure of 5 bars. The measurements were carried out using a gauge length of 15 mm and an extension rate of 9 mm/min (adopted from D3822 ASTM standard). Before the tensile tests, the specimens were conditioned at 60% relative humidity and 23 °C for 24 h. Each measurement was performed on 7 specimens.

2.3. Accelerated degradation of painting canvas The accelerated degradation of painting canvas was performed using a chemical treatment described elsewhere [21]. In brief, a 70 × 80 mm piece of new cotton canvas was immersed in 200 ml hydrogen peroxide solution mixed with 10 ml of sulfuric acid. The degradation was performed at 40 °C for 72 h with mild stirring. The degraded canvas was then thoroughly washed and dried under biaxial tension to resist shrinkage. After the canvas was dried, the warp threads were separated from it and used as samples for further tests. The degradation resulted in a decrease of the degree of polymerization from 6250 to 450 and a reduction of the tensile load from 12.9 N to 2.2 N for a warp thread.

3. Results and discussion 3.1. Preparation of silica-polyelectrolyte complexes As mentioned above, three different strategies were tested to reduce the flocculation (see Fig. 1). In the first a neutral (not charged) stabilizing polymer was used to kinetically hinder adsorption of the polyelectrolyte with high charge density on the silica particles. In the second approach cationic potato starch (CS) with low charge density and a branched structure was used. In the third approach, the charge density of silica was reduced by lowering the pH. All three approaches gave a controlled adsorption of the polymer, thus keeping flocculation at a minimum. Additionally, for the three different approaches, after adsorption of the cationic polyelectrolyte, an anionic polyelectrolyte, carboxymethyl cellulose (CMC), was adsorbed. Since CMC is chemically related to cellulose this treatment can be expected to improve the compatibility of the complexes with the cellulosic cotton fiber. The dry mass content and the composition of the formulations are presented in Table 1. The amounts of polyelectrolytes used were calculated based on charge density values of each component in the formulation (see Table 2) with an excess to ensure charge reversal after adsorption (charge overcompensation). In the first strategy, PVP was chosen due to its relatively weak affinity for silica surfaces [27,28] and almost equal amounts of silica and PVP were used. PVP was meant to prevent the adsorption of the strong cationic polyelectrolyte PQA, which interacts strongly with silica particles leading to kinetically driven flocculation prior to charge reversal [29]. The amount of PQA added to the formulations was 2.6 times higher than needed for silica neutralization, leading to pronounced charge overcompensation. As a result, the net charge of the particles turned positive which should result in strong interaction with the negatively charged surface of the degraded cellulosic material. It has been reported previously that excess of polyelectrolyte adsorbing on oppositely charged particles leads to smaller complexes and better colloidal stability [30]. Additionally, a layer of CMC was adsorbed on the silica/PQA complexes. The amount of CMC was such that it equals the amount of charges introduced by PQA. This means that the net charge of the complexes was reversed to be negative again, and close to

2.4. Particle charge density Particle charge density measurements of formulation components were performed using a PCD 02 particle charge detector (Mütek) connected to a Mettler DL21 titrator (Mettler Toledo), which allows monitoring the streaming potential that directly relates to the electrokinetic surface charge of the sample during the addition of an oppositely charged titrant [22,23]. The anionic and cationic titrants were PES-Na and polyDADMAC, respectively.

2.5. Dynamic light scattering The dynamic light scattering measurements were performed on a N4 Plus submicron particle analyser (Beckman Coulter). The formulations were diluted with ultra-pure water to a concentration of 0.05 wt%, sonicated for 5 min and filtered through 1.2 μm hydrophilic syringe filter. The measurements were performed at 90° for 300 s. The obtained autocorrelation functions were analysed using the Kohlrausch–Williams–Watts (KWW) fitting function and the CONTIN inverse Laplace fitting routine [24,25]. As a result of the KWW function fitting, an average size value is obtained, whereas CONTIN analysis yields a size distribution function. The CONTIN and the KWW fittings were performed using Matlab R2014b [26].

2.6. Application of the formulations The formulations were applied on the surface of degraded cotton threads (extracted from the warp direction of artificially aged canvases in order to have the same linear textile density) by an airbrush at a pressure of 2 bars. The dry mass content of the formulations that were sprayed is shown in Table 1. The spraying distance was around 20 cm and the spraying time was optimized to increase the mass of the samples by around 5 wt%. The uptake was monitored gravimetrically. The mass increase of the samples due to spraying of the formulations is presented in Table 3.

Table 2 Charge density values of the formulation components at pH 8. Component Silica PQA CMC CS

3

Charge density at pH 8 [meq/g] −0.046 7.61 −4.69 0.12

Colloids and Surfaces A xxx (xxxx) xxx–xxx

K. Kolman et al.

dimethylamine and epichlorohydrin, and is considered as a strong polyelectrolyte, thus having a charge with a low pH dependence. However, the charge density determined for PQA remained close to 8 meq/g for a pH in the range 3–8 and decreased to around 5 meq/g at pH 11. This unexpected behavior can partly be explained by the nature of the product, which besides quaternary ammonium groups, may contain azetidinium, epoxy, and chlorohydrin groups [32]. These groups are pH sensitive, and thus a variation of the charge density on the alkaline side can be expected. It is also noteworthy that the charge density of PQA, even though lower at high pH, remains significantly larger than that of silica. Before PQA was adsorbed, the silica sol was ion exchanged in order to lower the pH of the dispersion to 3. At this pH silica particles have almost no charge and PQA does not trigger flocculation. In this region, silica sol is metastable, however. Its temporary colloidal stability is explained by two different phenomena. The electrostatic repulsion force between silica particles is too low to prevent collision; however, lack of hydroxyl ions at low pH hinders the formation of siloxane bonds and subsequent gelling [20]. Additionally, at low pH silica has a hairy layer of polysilicic acid chains on its surface, which impart some steric stabilization [33,34]. The best colloidal stability of the silica/PQA formulation was obtained by adding the PQA solution at pH 3 in a 5 fold charge excess (the charge values were calculated using the charge densities measured at pH 8), so that when the pH was raised to 8, the adsorbed layer of PQA overcompensates the silica charge, which means that the net charge of the complexes became positive. CMC was then adsorbed at this pH. The amount of added CMC was calculated to compensate the positive charge from PQA. By doing so the net charge of the formulation was reversed to become negative. In total, 6 different formulations were prepared using 3 different approaches. Three formulations were based on silica and a cationic polyelectrolyte, whereas the other three have an additional component, i.e. CMC. The dry content of each formulation was close to 10 wt%. The reference formulation was a 10 wt% silica dispersion without any additional components. The size of the different complexes was determined by DLS. Analyses of DLS autocorrelation functions are presented in Fig. 3. The top row shows the average (mean) size obtained by the KWW function, while the bottom row provides the size distribution function obtained by processing the raw data with the CONTIN algorithm. The average size of the silica particles obtained by DLS was 109 nm, which is much higher than the size values provided by the supplier − 35 nm. The difference can be attributed to different measurement methods and to the high polydispersity of the silica sol, which strongly influences the outcome of the fitting by the KWW function. The size of 35 nm is calculated from surface area measurements obtained by the Sears method, assuming a monodisperse system [35]. Moreover, DLS measurements is strongly influenced by the larger species in polydisperse samples, especially when average size values are reported [36]. The CONTIN algorithm is based on an inverse Laplace transform and yields a size distribution function of the measured sample; hence, it works relatively well also with polydisperse samples. The size distribution function obtained for silica particles shows broad distribution of particles between 30 and 120 nm with a maximum of intensity around 65 nm. In Fig. 3A size values at different stages of preparation of the PVP stabilized formulation are presented. After PVP was adsorbed, no significant change in average size or in size distribution function was observed. The mean size increased to around 115 nm and the size distribution functions obtained for silica and silica/PVP almost overlapped. The results indicate that PVP adsorbed on the surface of silica particles but did not trigger any aggregation. Similar findings regarding PVP adsorption on silica surface have been published earlier [37]. When the cationic polyelectrolyte PQA was adsorbed, the average size of the complexes increased to 310 nm, which indicates aggregation. This can be attributed to the system passing through the charge

Table 3 Mechanical properties of threads collected from artificially degraded cotton canvases before (first row) and after treatment with the formulations. The last column reports the mass increase of the threads due to the treatment. Formulation

Max. load [N]

Strain at max. load [%]

Mass increase [%]

No treatment Silica Silica + PVP + PQA Silica + PVP + PQA + CMC Silica + CS Silica + CS + CMC IE silica + PQA IE silica + PQA + CMC

2.23 2.73 2.87 2.48

± ± ± ±

0.27 0.33 0.32 0.40

17.66 10.26 12.93 15.72

± ± ± ±

4.10 1.38 2.48 3.08

– 7.6 ± 1.8 5.9 ± 0.4 5.3 ± 0.2

2.29 2.16 2.80 2.75

± ± ± ±

0.16 0.16 0.22 0.20

14.28 11.73 14.53 14.31

± ± ± ±

4.48 2.48 2.49 2.68

4.2 4.8 6.0 6.7

± ± ± ±

0.8 0.3 0.5 1.5

the surface charge of silica. Despite electrostatic repulsion between CMC and the surface of the degraded cellulosic material, CMC, as a cellulose derivative, is expected to interact strongly with the cotton via hydrogen bonding, after evaporation of the water. The formulations obtained using the second approach were based on cationic starch (CS). The CS that was employed in this study has a low charge density and branched polymer chains. It has been reported that branched polyelectrolytes create smaller flocs, i.e. weak aggregates, in the papermaking process [31]. Since the interaction with silica particles was expected to be strong enough to ensure full coverage, PQA was not used. The amount of added CS was in a 1.8 charge ratio to the negative silica; thus, a charge overcompensation occurred and the net charge of the particles became positive. CMC was also adsorbed on the silica/CS complexes in a quantity sufficient to fully neutralize the positive charge of the added CS. In the third approach, the charge density of the silica particles was varied by adjustment of the pH. In Fig. 2 the variations of charge density as a function of pH are presented for silica (Fig. 2A) and for PQA (Fig. 2B). The charge density of silica particles increased with pH, whereas the charge density of PQA decreased. PQA is a copolymer of

Fig. 2. Charge density of (A) silica particles and (B) PQA as a function of pH.

4

Colloids and Surfaces A xxx (xxxx) xxx–xxx

K. Kolman et al.

Fig. 3. Size values of the formulations obtained using (A) PVP as stabilizing polymer, (B) CS as branched polyelectrolyte and (C) a pH change as a way to reduce electrostatic interactions. The top pictures show average size values obtained by the KWW function fitting, and the bottom pictures present the size distribution functions obtained by the CONTIN algorithm.

was detrimental. Two main fractions with sizes 40–190 nm and 250–700 nm can be seen on the size distribution function obtained by the CONTIN algorithm. Again, it is likely that the aggregation upon addition of CMC occurs due to presence of free PQA in the formulation.

neutralization point, thus becoming unstable [38]. The size distribution function shows three populations with center size values of 115 nm, 340 nm and 600 nm. The most prominent fraction is the one with 115 nm size. Despite some aggregation, the silica/PVP/PQA formulation had good colloidal stability and no sedimentation occurred over a period of two months. In the next step CMC was adsorbed, which caused severe aggregation of the particles. The average size of the complexes increased to 1 μm along with a strong broadening with complex sizes between 300 nm and 900 nm. Adsorption of CMC resulted in deterioration of the colloidal stability and after a few hours the aggregates started to sediment. Most likely, the strong aggregation was caused by free polyelectrolytes in solution (an excess of PQA was used). Because of the poor colloidal stability, the silica/PVP/PQA/CMC system was used for further tests directly after preparation. The size values of the silica/CS/CMC at every stage of preparation are reported in Fig. 3B. After adsorption of CS, the silica/CS complexes with an average size of 270 nm were formed. Thanks to the branched structure and the low charge density of the cationic starch, it was possible to obtain complexes stabilized electrosterically [39]. The size distribution function shows that silica/CS complexes with sizes between 50 nm and 300 nm were obtained. The distribution has two maxima, at 85 nm and at 190 nm, which suggests that these are the most abundant complexes in the formulation. Addition of CMC caused an increase of the average size of the complexes to 370 nm. The increase is not as pronounced as for the silica/PQA/CMC system, however. The branched structure of the potato starch imparts steric stabilization to the complexes. It has been reported previously that starch transfers extraordinary steric stabilization to colloidal particles [40–43]. The size distribution function of the silica/CS/CMC formulation became broad with the most prominent population being located between 100 nm and 250 nm. The sizes of the complexes obtained by adsorbing PQA and CMC on ion-exchanged silica particles are shown in Fig. 3C. The adsorption of PQA on silica particles at pH 3 followed by a raise in pH to 8 resulted in the formation of complexes with an average size value of 113 nm, showing no aggregation, with an identical size distribution to bare silica. These results show that by simply reducing pH, it is possible to arrest flocculation, in line with previous results [44]. Addition of CMC (performed at pH 8), however, triggered an increase of the average size to 375 nm. Again the passage through the charge neutralization point

3.2. Application of the formulations to the mechanical reinforcement of degraded cotton threads The formulations were applied on the surface of artificially degraded painting cotton canvas using an airbrush right after preparation to prevent aggregate growth. After application the threads were dried at room temperature and conditioned under controlled temperature and humidity. The mass increase of the samples is listed in Table 3; the average lies around 6 wt%. The SEM pictures showing the surface of degraded cotton fibers before and after treatment are presented in Fig. 4. The surface of degraded cotton fibers without any treatment is visible in Fig. 4A. The cell wall structure of the fiber remained intact; however, some minor cracks could be noted. Silica sol with a solids content of 10 wt% was used as a reference and the surface of the treated fibers is presented in Fig. 4B. The silica particles tend to form agglomerates and are not distributed evenly on the surface of the fiber. It was also noticed that during handling/rubbing of the treated threads the adhesion of silica was not strong enough, which was probably due to poor compatibility between the silica particles and the fibers. Fig. 4C and 4D show the fibers after spraying with the silica/PVP/PQA and the silica/PVP/PQA/ CMC formulations, respectively. The distribution of the formulations on the surface of the fibers is improved compared to the treatment with the bare silica. Similar results regarding improvement of silica film-forming properties by PVP have been published previously [45]. The films of the PVP-stabilized formulations appear to be quite porous, which may be caused by the relatively high polymer content compared to the other formulations. It may also reflect the steric hindrance provided by PVP. The surfaces of fibers treated with the CS-based formulations are shown in Fig. 4E (silica/CS) and 4F (silica/CS/CMC). The nanoparticles are uniformly distributed over the surface and create a dense and continuous film. In Fig. 4G and 4H fibers treated with the IE silica/PQA and the IE silica/PQA/CMC formulations, respectively, are shown. As can be seen, the fiber surface coverage is good. An excess of material can be observed at the edges of fibers treated with the IE silica/PQA formula5

Colloids and Surfaces A xxx (xxxx) xxx–xxx

K. Kolman et al.

Fig. 4. SEM pictures of warp threads taken from degraded cotton canvas (A) before and after treatment with (B) silica, (C) silica + PVP +PQA, (D) silica + PVP + PQA + CMC, (E) silica + CS, (F) silica + CS + CMC, (G) ion exchanged silica + PQA and (H) ion exchanged silica + PQA + CMC. The images were taken at a magnification of 5 000 × , while the insets present the same sample at a magnification of 25 000 × .

based on ion-exchanged (IE) silica (Fig. 5C) but not for the CS-based formulation (Fig. 5B). The systems ranked in terms of maximum load improvement as follow: silica/PVP/PQA (+28.7%), IE silica/PQA (25.6%) and IE silica/PQA/CMC (23.3%). However this was at the expense of the plasticity of the material, with the silica/PVP/PQA system reducing the strain the most. Surprisingly, the formulation containing cationic starch performed relatively poorly. No change of the maximum load and no increase of the strain at brake were observed. The samples treated with the silica/ CS/CMC formulation became much stiffer than the samples treated with silica only, which can be seen in Fig. 5B (blue curve). It is likely that this behavior emanates from the branched structure of the cationic starch, which in turn limits the particle interaction with the fiber surface. Since the starch molecules are quite bulky, they perform well when it comes to stabilization of silica in formulations but at the same time they may act like a spacer after drying and reduce the strengthening effect of the silica particles.

tion. To evaluate the potential of the treatment for fiber strengthening, tensile tests of the degraded warp threads before and after treatment were performed. Representative load-strain curves are presented in Fig. 5 and values of mechanical properties for each sample are listed in Table 3 (average of 7 specimens). In these tests we focused on two criteria. The first is the maximum load at break, which reflects the capacity of the threads to adsorb the stress of the elongation. The second is the extension at break, which gives an indication on how the treatment affects the plasticity of the material. In practice, and especially for art, the material is never stressed to that extent as it would cause destruction of the object. For instance, in the case of paintings, it would mean cracking the paint layer. This is far beyond the stress induced under real conditions, by a wood frame for instance. Still, it gives an indication of the potential performance of the treatments. Initially, the degraded canvas had a maximum load at break of 2.2 N and an extension at break equal to 17% of the initial sample length. After treatment with 10 wt% dispersion of silica (the reference formulation), the maximum load increased to 2.7 N, whereas the strain decreased to 10%. For all the system used, when a polyelectrolyte was added to the silica, it was possible to reach similar maximum load values as the ones obtained when silica solely was applied; however, the strain at brake was on average higher. Such behavior was observed for the PVP-stabilized formulation (Fig. 5A) and for the formulation

4. Conclusion We describe three different strategies to reduce or overcome flocculation in silica/polyelectrolyte systems. In the first approach, a stabilizing polymer (PVP) was used to hinder the electrostatic interactions between silica particles and a high charge density cationic

Fig. 5. Tensile curves recorded for degraded warp threads treated with formulations obtained using (A) stabilizing polymer (strategy 1), (B) branched polyelectrolyte (strategy 2), and (C) pH induced adsorption (strategy 3).

6

Colloids and Surfaces A xxx (xxxx) xxx–xxx

K. Kolman et al.

polyelectrolyte − a copolymer of dimethylamine and epichlorohydrin (PQA). As a result, a formulation with good stability was obtained. In the second approach, a different cationic polyelectrolyte, cationic starch (CS), was used. The branched structure and the low charge density of CS gave reduced aggregation and imparted electrosteric stabilization. The third approach was based on control of the surface charge density. Adsorption of PQA on silica nanoparticles at pH 3 caused no aggregation and only a thin polyelectrolyte layer was adsorbed. A second polyelectrolyte, carboxymethyl cellulose (CMC), was subsequently adsorbed on the cationized particles for each approach. While adsorption of CMC caused major aggregation for the PVP-stabilized system, the other two formulations remained stable. The formulations were then applied on the surface of model degraded cotton threads taken from painting canvas. The amount of applied formulations was controlled so that the sample’s weight uptake remains close to 5%. The mechanical performance of the treated samples was measured using tensile testing, using maximum load and strain at break as criteria. Whereas the cationic systems tended to considerably improve the thread in terms of maximum load, they also significantly reduced the strain. The addition of a final negatively charged layer was beneficial and the loss of strain was reduced. Based on the preparation route, the stability of the system and the improvement of the mechanical performance, the most promising formulation for mechanical reinforcement of degraded canvases is that of ionexchanged silica coated with a cationic (PQA) and an anionic (CMC) layer. Having a cellulose derivative as the outermost layer gives the material a certain chemical resemblance with cotton.

[12] B. Mahltig, C. Swaboda, A. Roessler, H. Böttcher, Functionalising wood by nanosol application, J. Mater. Chem. 18 (2008) 3180, http://dx.doi.org/10.1039/b718903f. [13] L. Borgnino, Experimental determination of the colloidal stability of Fe(III)montmorillonite: effects of organic matter, ionic strength and pH conditions, Colloids Surf. A: Physicochem. Eng. Asp. 423 (2013) 178–187, http://dx.doi.org/ 10.1016/j.colsurfa.2013.01.065. [14] M. Pavlovic, R. Huber, M. Adok-Sipiczki, C. Nardin, I. Szilagyi, Ion specific effects on the stability of layered double hydroxide colloids, Soft Matter 12 (2016) 4024–4033, http://dx.doi.org/10.1039/C5SM03023D. [15] T. Abe, S. Kobayashi, M. Kobayashi, Aggregation of colloidal silica particles in the presence of fulvic acid, humic acid, or alginate: effects of ionic composition, Colloids Surf. A: Physicochem. Eng. Asp. 379 (2011) 21–26, http://dx.doi.org/10. 1016/j.colsurfa.2010.11.052. [16] P. Rouster, M. Pavlovic, I. Szilagyi, Improving the stability of titania nanosheets by functionalization with polyelectrolytes, RSC Adv. 6 (2016) 97322–97330, http:// dx.doi.org/10.1039/C6RA23707J. [17] A.M. Howe, R.D. Wesley, M. Bertrand, M. Côte, J. Leroy, Controlled association in suspensions of charged nanoparticles with a weak polyelectrolyte, Langmuir 22 (2006) 4518–4525, http://dx.doi.org/10.1021/la053327s. [18] D. Bauer, E. Killmann, W. Jaeger, Flocculation and stabilization of colloidal silica by the adsorption of poly-diallyl-dimethyl-ammoniumchloride (PDADMAC) and of copolymers of DADMAC with N-methyl-N-vinyl-acetamide (NMVA), Colloid Polym. Sci. 276 (1998) 698–708, http://dx.doi.org/10.1007/s003960050299. [19] J. Gregory, S. Barany, Adsorption and flocculation by polymers and polymer mixtures, Adv. Colloid Interface Sci. 169 (2011) 1–12, http://dx.doi.org/10.1016/j. cis.2011.06.004. [20] R.K. Iler, The Chemistry of Silica: Solubility, Polymerization, Colloid and Surface Properties, and Biochemistry, Wiley, 1979. [21] O. Nechyporchuk, K. Kolman, M. Oriola, M. Persson, K. Holmberg, R. Bordes, 2017. Accelerated ageing of cotton canvas as a model for further consolidation practices. Submitted. [22] R. Wäsche, M. Naito, V.A. Hackley, Experimental study on zeta potential and streaming potential of advanced ceramic powders, Powder Technol. 123 (2002) 275–281, http://dx.doi.org/10.1016/S0032-5910(01)00462-4. [23] K. Böckenhoff, W. Fischer, Determination of electrokinetic charge with a particlecharge detector, and its relationship to the total charge, Fresenius, J. Anal. Chem. 371 (2001) 670–674, http://dx.doi.org/10.1007/s002160100897. [24] S.W. Provencher, CONTIN: A general purpose constrained regularization program for inverting noisy linear algebraic and integral equations, Comput. Phys. Commun. 27 (1982) 229–242, http://dx.doi.org/10.1016/0010-4655(82)90174-6. [25] S.W. Provencher, A constrained regularization method for inverting data represented by linear algebraic or integral equations, Comput. Phys. Commun. 27 (1982) 213–227, http://dx.doi.org/10.1016/0010-4655(82)90173-4. [26] I.-G. Marino, Regularized Inverse Laplace Transform (RILT) – MatLab Script, (2007). https://se.mathworks.com/matlabcentral/fileexchange/6523-rilt. [27] L. Gargallo, E. Cid, Adsorption characteristics of polyvinylpyrrolidone at solidliquid interface, Colloid Polym. Sci. 255 (1977) 556–559, http://dx.doi.org/10. 1007/BF01549742. [28] L.M. Al-Harbi, S.A. Kosa, M.K. Baloch, Q.A. Bhatti, E.-S.E.-B.H. El-Mossalamy, Adsorption of polyvinylpyrrolidone over the silica surface: As affected by pretreatment of adsorbent and molar mass of polymer adsorbate, Int. J. Polym. Sci. 2016 (2016) 1–9, http://dx.doi.org/10.1155/2016/2417292. [29] R.K. Iler, Relation of particle size of colloidal silica to the amount of a cationic polymer required for flocculation and surface coverage, J. Colloid Interface Sci. 37 (1971) 364–373, http://dx.doi.org/10.1016/0021-9797(71)90304-3. [30] G.M. Lindquist, R.A. Stratton, The role of polyelectrolyte charge density and molecular weight on the adsorption and flocculation of colloidal silica with polyethylenimine, J. Colloid Interface Sci. 55 (1976) 45–59, http://dx.doi.org/10. 1016/0021-9797(76)90007-2. [31] E. Antunes, F.A.P. Garcia, P. Ferreira, A. Blanco, C. Negro, M.G. Rasteiro, Use of new branched cationic polyacrylamides to improve retention and drainage in papermaking, Ind. Eng. Chem. Res. 47 (2008) 9370–9375, http://dx.doi.org/10. 1021/ie801216t. [32] Q.Y. Yue, B.Y. Gao, Y. Wang, H. Zhang, X. Sun, S.G. Wang, R.R. Gu, Synthesis of polyamine flocculants and their potential use in treating dye wastewater, J. Hazard. Mater. 152 (2008) 221–227, http://dx.doi.org/10.1016/j.jhazmat.2007.06.089. [33] J.J. Adler, Y.I. Rabinovich, B.M. Moudgil, Origins of the non-DLVO force between glass surfaces in aqueous solution, J. Colloid Interface Sci. 237 (2001) 249–258, http://dx.doi.org/10.1006/jcis.2001.7466. [34] M. Kobayashi, M. Skarba, P. Galletto, D. Cakara, M. Borkovec, Effects of heat treatment on the aggregation and charging of Stöber-type silica, J. Colloid Interface Sci. 292 (2005) 139–147, http://dx.doi.org/10.1016/j.jcis.2005.05.093. [35] G.W. Sears, Determination of specific surface area of colloidal silica by titration with sodium hydroxide, Anal. Chem. 28 (1956) 1981–1983, http://dx.doi.org/10. 1021/ac60120a048. [36] E. Tomaszewska, K. Soliwoda, K. Kadziola, B. Tkacz-Szczesna, G. Celichowski, M. Cichomski, W. Szmaja, J. Grobelny, Detection limits of DLS and UV–vis spectroscopy in characterization of polydisperse nanoparticles colloids, J. Nanomater. 2013 (2013) 1–10, http://dx.doi.org/10.1155/2013/313081. [37] S. Robinson, P.A. Williams, Inhibition of protein adsorption onto silica by polyvinylpyrrolidone, Langmuir 18 (2002) 8743–8748, http://dx.doi.org/10.1021/ la020376l. [38] J. Hierrezuelo, I. Szilagyi, A. Vaccaro, M. Borkovec, Probing nanometer-thick polyelectrolyte layers adsorbed on oppositely charged particles by dynamic light scattering, Macromolecules 43 (2010) 9108–9116, http://dx.doi.org/10.1021/ ma1014462.

Acknowledgements The authors would like to thank Anders Mårtensson from Chalmers University of Technology for performing the SEM measurements. Frida Iselau, also from Chalmers University of Technology, is acknowledged for valuable discussions and for help with the PCD and the size exclusion chromatography measurements. This work has received funding from the European Union’s Horizon 2020 research and innovation programme under the NanoRestArt project (Grant Agreement No. 646063). References [1] N. Ryder, Acidity in canvas painting supports: deacidification of two 20th century paintings, Conservator 10 (1986) 31–36, http://dx.doi.org/10.1080/01410096. 1986.9995015. [2] M. Oriola, A. Možir, P. Garside, G. Campo, A. Nualart-Torroja, I. Civil, M. Odlyha, M. Cassar, M. Strlič, Looking beneath Dalí’s paint: non-destructive canvas analysis, Anal. Methods 6 (2014) 86, http://dx.doi.org/10.1039/c3ay41094c. [3] P. Baglioni, E. Carretti, D. Chelazzi, Nanomaterials in art conservation, Nat. Nanotechnol. 10 (2015) 287–290, http://dx.doi.org/10.1038/nnano.2015.38. [4] P. Ackroyd, The structural conservation of canvas paintings: changes in attitude and practice since the early 1970, Stud. Conserv. 47 (2002) 3–14, http://dx.doi.org/10. 1179/sic.2002.47.Supplement-1.3. [5] J. Wouters, CHEMISTRY: coming soon to a library near you? Science (80-.) 322 (2008) 1196–1198, http://dx.doi.org/10.1126/science.1164991. [6] R. Giorgi, D. Chelazzi, P. Baglioni, Nanoparticles of calcium hydroxide for wood conservation. The deacidification of the Vasa warship, Langmuir 21 (2005) 10743–10748, http://dx.doi.org/10.1021/la0506731. [7] H. Mahfuz, F. Clements, V. Rangari, V. Dhanak, G. Beamson, Enhanced stab resistance of armor composites with functionalized silica nanoparticles, J. Appl. Phys. 105 (64307) (2009), http://dx.doi.org/10.1063/1.3086431. [8] P. Baglioni, D. Chelazzi, R. Giorgi, Nanotechnologies in the Conservation of Cultural Heritage, Springer, Netherlands, Dordrecht, 2015, http://dx.doi.org/10.1007/97894-017-9303-2. [9] R. Giorgi, C. Bozzi, L. Dei, C. Gabbiani, B.W. Ninham, P. Baglioni, Nanoparticles of Mg(OH) 2: synthesis and application to paper conservation, Langmuir 21 (2005) 8495–8501, http://dx.doi.org/10.1021/la050564m. [10] M. Afsharpour, M. Hadadi, Titanium dioxide thin film: environmental control for preservation of paper-art-works, J. Cult. Herit. 15 (2014) 569–574, http://dx.doi. org/10.1016/j.culher.2013.10.008. [11] O.M. El-Feky, E.A. Hassan, S.M. Fadel, M.L. Hassan, Use of ZnO nanoparticles for protecting oil paintings on paper support against dirt, fungal attack, and UV aging, J. Cult. Herit. 15 (2014) 165–172, http://dx.doi.org/10.1016/j.culher.2013.01. 012.

7

Colloids and Surfaces A xxx (xxxx) xxx–xxx

K. Kolman et al.

1039/C6SM00165C. [43] F. Iselau, T. Phan Xuan, G. Trefalt, A. Matic, K. Holmberg, R. Bordes, Formation and relaxation kinetics of starch–particle complexes, Soft Matter 12 (2016) 9509–9519, http://dx.doi.org/10.1039/C6SM01312 K. [44] J.F. Bringley, A. Wunder, A.M. Howe, R.D. Wesley, T.A. Qiao, N.B. Liebert, B. Kelley, J. Minter, B. Antalek, J.M. Hewitt, Controlled, simultaneous assembly of polyethylenimine onto nanoparticle silica colloids, Langmuir 22 (2006) 4198–4207, http://dx.doi.org/10.1021/la0534118. [45] H. Kozuka, A. Yamano, M. Fujita, H. Uchiyama, Aqueous dip-coating route to dense and porous silica thin films using silica nanocolloids with an aid of polyvinylpyrrolidone, J. Sol-Gel Sci. Technol. 61 (2012) 381–389, http://dx.doi.org/10.1007/ s10971-011-2638-x.

[39] B. Kronberg, K. Holmberg, B. Lindman, Surface Chemistry of Surfactants and Polymers, John Wiley & Sons, Ltd., Chichester, UK, 2014, http://dx.doi.org/10. 1002/9781118695968. [40] P. Raveendran, J. Fu, S.L. Wallen, Completely green synthesis and stabilization of metal nanoparticles, J. Am. Chem. Soc. 125 (2003) 13940–13941, http://dx.doi. org/10.1021/ja029267j. [41] H. De Bruyn, E. Sprong, M. Gaborieau, J.A. Roper, R.G. Gilbert, Starch-graft(synthetic copolymer) latexes initiated with Ce4+ and stabilized by amylopectin, J. Polym. Sci. A: Polym. Chem. 45 (2007) 4185–4192, http://dx.doi.org/10.1002/ pola.22189. [42] F. Iselau, T. Phan Xuan, A. Matic, M. Persson, K. Holmberg, R. Bordes, Competitive adsorption of amylopectin and amylose on cationic nanoparticles: a study on the aggregation mechanism, Soft Matter 12 (2016) 3388–3397, http://dx.doi.org/10.

8