Aggregation behaviour of hydrophobically modified polyacrylate – Variation of alkyl chain length

Polymer 70 (2015) 194e206

Contents lists available at ScienceDirect

Polymer journal homepage: www.elsevier.com/locate/polymer

Aggregation behaviour of hydrophobically modified polyacrylate e Variation of alkyl chain length
vost a, b, Michaela Dzionara a, Marie-Sousai Appavou c, Sven Riemer a, Sylvain Pre Ralf Schweins d, Michael Gradzielski a, * €t Berlin, D-10623 Berlin, Germany Stranski-Laboratorium für Physikalische und Theoretische Chemie, Institut für Chemie, Technische Universita Helmholtz-Zentrum Berlin für Materialien und Energie GmbH, D-14109 Berlin, Germany c Jülich Center of Neutron Science (JCNS), Forschungszentrum Jülich GmbH Outstation at MLZ, Lichtenbergerstrasse 1, D-85747 Garching, Germany d Institut Laue-Langevin, DS/LSS Group, 71 Avenue des Martyrs, CS 20 156, F-38042 Grenoble, Cedex 9, France a

b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 17 March 2015 Received in revised form 6 June 2015 Accepted 10 June 2015 Available online 21 June 2015

The aggregation behaviour in aqueous solution of hydrophobically modified polyacrylates, synthesized by Atomic Transfer Radical Copolymerisation (ATRP) of mixtures of alkyl acrylate and t-butyl acrylate and subsequent hydrolysis of the t-butyl acrylate, was investigated by a combination of static and dynamic light scattering with small-angle neutron scattering (SANS). The degree of amphiphilicity was varied by the percentage of alkyl chains and the length of the alkyl chain (butyl to dodecyl), and, in addition, depends strongly on pH via the ionization of the polyacrylate backbone. SANS shows the formation of hydrophobic domains whose size scales with the length of the alkyl chain. The tendency for domain formation increases with the length of the alkyl chains and is much more pronounced for lower pH, while at high pH the electrostatic charging suppresses the formation of hydrophobic domains for chains shorter than octyl. Then only relatively large and loosely connected aggregates are formed. These hydrophobically modified copolymers show a pronouncedly pH dependent aggregation behaviour that is controlled by the length and percentage of hydrophobic modification and this widely tuneable aggregation behaviour could be interesting for the transport and controlled release of hydrophobic cargo molecules. © 2015 Elsevier Ltd. All rights reserved.

Keywords: Self-assembly Amphiphilic copolymer Polymer micelle pH-control Tuneable aggregation

1. Introduction Hydrophobically modified polyelectrolytes are an interesting class of water-soluble polymers, that in principle combine the properties of polyelectrolytes with the ability to self-assemble spontaneously, due to the hydrophobic interactions that arise from this modification [1e6]. Such systems have also been described thoroughly by theory [7] which has predicted the formation of necklace structures, of hairy, crew-cut, or braided micelles or the self-assembly into wormlike cylindrical micelles, depending on the ratio of block sizes and the ratio of the surface energies of the blocks. For instance, water soluble polymers with hydrophobic modifications will have a tendency to form segregated hydrophobic domains. Often the cross-linking of such domains, through polymer chains that have hydrophobic stickers in different

* Corresponding author. E-mail address: [email protected] (M. Gradzielski). http://dx.doi.org/10.1016/j.polymer.2015.06.010 0032-3861/© 2015 Elsevier Ltd. All rights reserved.

domains, leads to a substantially enhanced viscosity [8e11], i. e. they are rheological modifiers. A simple and classical case of hydrophobic cross-linking has been achieved for the case of doubly end-capped polymers where interconnected networks are frequently formed either by pure self-assembly or by connecting microemulsion droplets [12e15] with the corresponding ability to rheological modification, as it is frequently employed in commercial applications. More recent work then has also shown that the number of hydrophobic end-caps per polymer is a viable tool for controlling the rheological properties of amphiphilic networks formed by the polymers themselves [16] or when they cross-link microemulsion droplets [17,18]. It might be noted that one may also have polyelectrolyte gels with a hydrophobic modification, where microphase separation will be observed and such gels may show pH-dependent swelling [19,20]. pH-sensitive polymeric systems have been studied widely for the case of block copolymers with one ore my polyelectrolyte blocks [21,22]. For weakly hydrophobic copolymers it has for

S. Riemer et al. / Polymer 70 (2015) 194e206

instance been observed that for the case of P(nBA-stat-AA)-b-PAA di- and triblock copolymers (nBA: n-butyl acrylate, AA: acrylic acid) one can switch from having micelles to complete disintegration of the aggregates by raising the degree of ionisation [23]. For more hydrophobic block copolymers, such as ones made of poly(isobutylene)-block-poly(methacrylic acid) (PIB-b-PMAA) it has been observed that the formed micelles become smaller with increasing degree of ionisation, in agreements with simple packing parameter arguments [24]. And even for their interpolyelectrolyte complexes a reversible pH response has been confirmed [25]. A similar tendency for substantially reduced tendency for aggregation had been reported before for micelles of poly(styrene)-blockPoly((sulfamatecarboxylate) isoprene) copolymer micelles [26], i. e. even for systems where the hydrophobic block is below its glass transition point. Furthermore such pH dependent assembly of block copolymers is not only restricted to the formation of micelles but it has also been reported that vesicle formation can be switched on by raising the pH for poly[2-(methacryloyloxy)ethyl phosphoryl choline-block-2-(diisopropylamino)ethyl methacrylate copoly mers [27]. Of course, such pH response of copolymer micelles is also a very important aspect in designing drug delivery systems [28], a topic that will not dwelled on in more detail here. Of course, end-capped water-soluble polymers and block copolymers with polyelectrolyte chains are just a limiting case of hydrophobic modification and electrostatic stabilization. For copolymers one has easily the option for a fully continuous variation of the extent of hydrophobisation along the backbone of an otherwise hydrophilic polymer. For instance for statistical copolymers, one can tune the amphiphilicity of this polymeric system by the content of a hydrophobic monomer mixed with a hydrophilic monomer. Particularly interesting in that context is using a polyelectrolyte as water-soluble polymer with a multiple number of hydrophobic stickers, where the extent of charging can be controlled by pH, which is the case for polycarboxylates such as polyacrylate or polymethacrylate in a convenient pH-range. A hydrophobic modification may easily be introduced by having alkyl acrylate esters in the copolymer. The hydrophobicity is then controlled by the length of the alkyl chain of the ester and the percentage of hydrophobic substitution. In addition, it depends on pH, as at low pH the extent of ionisation of the polyacrylate backbone is low and the formed polyacrylic acid is itself rather hydrophobic and in pure form not even water-soluble. It should be mentioned that such copolymers are also in commercial use, as for instance in the case of Kollicoat MAE, which is a copolymer of methacrylic acid and ethyl acrylate, which is widely used as a film former, for coatings, and for drug delivery [29]. Such compounds are also often referred to as polysoaps and for instance the case of an alternating copolymer of maleic acid and alkyl vinyl ether with variation of the alkyl chain from butyl to hexadecyl has been studied intensely by means of fluorescence measurements. These experiments [30] showed that the extent of microdomain formation depends largely on the length of the alkyl chain and hardly occurs for alkyl chains shorter than octyl. Similarly hydrophobically modified polyacrylates (PAA), which contained statistically dodecyl or octadecyl modified acryl amide units, have been investigated by means of pyrene fluorescence and 13 C NMR. Here association of the alkyl units was only observed above a certain total concentration and the formed aggregates had lower aggregation numbers and higher polydispersity than the corresponding pure ionic micelles [31]. In contrast, the molecular weight of the polymers was found to have rather little effect on the association behaviour. Finally also PAA with perfluoroalkyl side chains have been studied with respect to their associative behaviour, where the perfluorinated chains lead to stronger aggregation than the corresponding hydrocarbon modification and at sufficient

195

concentration such systems show pronounced viscosity increase [32]. Structurally similar systems of amphiphilic statistical copolymer samples of sodium 2-(acrylamido)-2-methylpropanesulfonate and n-hexyl methacrylate with different degrees of polymerization and compositions have been investigated where it was found that they form aggregates with 2e7 polymer chains and possessing 1e5 hydrophobic microdomains, where the structures formed depend on the degree of polymerization and composition [33]. Such systems were also investigated with respect to their aggregation behaviour in water/methanol solvent mixtures [34]. For the case of statistical copolymers of 2-(acrylamido)-dodecanesulfonic acid (AMC12S), with 2-(acrylamido)-2-methylpropanesulfonic acid (AMPS) the formation of multimolecular aggregates has been observed [35]. Similar copolymers with a hydrophobic dodecyl chain and based on different amino acids were studied by light scattering and fluorescence. These investigations showed that the interchain aggregation increases with content of the hydrophobic modification, while it decreases with the hydrophobicity of the amino acid residue in the copolymer. Here then flower micelles with minimum loop size are formed [36]. It might be noted that also the formation of multi-chain aggregates has been observed for the case of chitosan hydrophobically modified with 4 mol% of ndodecyl chains [37]. For the case of amphiphilically modified chitosan rather large aggregates with diameters of 100e300 nm have been observed, that shrink substantially with increasing pH as the chitosan backbone then loses its hydrophilicity [38]. Accordingly, so far quite a bit of work has been done on the properties and aggregation behaviour of hydrophobically modified polyelectrolytes. However, systematic studies on the dependence of the aggregation behaviour as a function of the chain length of the hydrophobic modification for well-defined polymers are still largely missing. Therefore in our work we were addressing this issue for the case of rather short copolymers of acrylic acid and alkyl acrylate (degree of polymerization ~100), as they can be generated in a statistical fashion and with rather low polydispersity by means of atomic transfer radical copolymerisation (ATRP). In that context it has to be noted that previous work on the kinetics of radical polymerization of hydrophobically modified acrylates has shown that their rate increases with increasing length of the hydrophobic modification [39]. This means that our statistical copolymers should also not have a fully random distribution of the monomer units, but are expected to have a tendency for having the hydrophobic monomer polymerized initially, and this even more so the more hydrophobic the monomer. This then shall result in a copolymer with a gradient of monomeric units along its backbone, but this effect should also not be too pronounced as only an increase of the propagation rate coefficient by 3e4% per CH2 has been reported [39]. In our work we then explored to what extent such compounds do self-assemble and how the aggregation properties depend on the percentage of hydrophobic modification (controlled by the amount of hydrophobic monomer), the length of the hydrophobic side-chain, and the pH. This is interesting as such systems can be expected to be rather versatile with respect to their aggregation behaviour and should have a largely enhanced tendency for aggregation at low pH, as schematically depicted in Fig. 1. Accordingly we expect tunable self-assembly, as it is of relevance for the solubilisation and release properties. All this is then of prominent relevance for their potential use in cosmetic or pharmaceutical applications, but also in general, for their use as amphiphilic building blocks in more complex systems. In the following, we will discuss the synthesis and characterisation of such copolymers with a focus on their aggregation behaviour in aqueous solutions as a function of the molecular composition of the copolymers. They were prepared with alkyl side

196

S. Riemer et al. / Polymer 70 (2015) 194e206

Fig. 1. Scheme of the expected pH dependent structural behaviour of hydrophobically modified polycarboxylates. With decreasing pH the backbone becomes more protonated (neutral) and hydrophobic which leads to an increased tendency for association.

chains ranging from butyl to dodecyl and containing 10 or 20 mol% of hydrophobically modified side-chains. The backbone of acrylic acid will be fully charged at a pH above 9, but for lower pH the acrylic acid becomes increasingly protonated, thereby reducing the charge density along the polymer chain continuously. The aim then is to derive systematic correlations between the molecular build-up of the copolymers and their self-assembly properties as a function of the amount and length of hydrophobic side chains and the effect of pH on the aggregation behaviour. 2. Materials and methods 2.1. Materials & synthesis 2.1.1. Materials Toluene (>99.5%) from Fluka, methyl-2-bromopropionate (MBP, 98%) and hexane (p.a.) from Aldrich, 1,1,4,7,10,10hexamethyltriethylene-tetramine (HMTETA, 97%), copper(I) chloride (>99%), hexyl acrylate (98%), isooctyl acrylate (>90%), dodecyl acrylate (technical grade 90%) from SigmaeAldrich and diethylether (>99,5%) from Carl-Roth were used as supplied. t-Butylacrylate and n-butylacrylate were also used as supplied, dichloromethane was distilled before usage and all three were gifts from BASF. Milli-Q water was used for the sample preparation as produced by a Millipore filtering system. D2O was from Eurisotop (99.5% isotopic purity, Gif-sur-Yvette, France), sodium hydroxide (99%), sodium chloride (>99%) and Sudan III (technical grade, dye content >90%) were obtained from SigmaeAldrich. Silica (0.04e0.063 mm) and Aluminium oxide (90% active, 0.063e0.2 mm) for the columns were from Merck. 2.1.2. Synthesis The synthesis procedure was the same for all samples: MBP (1 eq.), copper(I) chloride (1.3 equivalents (eq.), relative to MBP) and the monomers (100 eq.) were dissolved in toluene (0.5 g per ml) at ambient temperature. Nitrogen gas was bubbled through the mixture for 30 min while stirring, in order to remove oxygen. The HMTETA was added drop-wise and the nitrogen gas was bubbled for another 20 min. Finally the flask was sealed and heated to 80
C. The reaction mixture was stirred for 20 h and the conversion was checked by 1H NMR. Once the conversion was complete the reaction was stopped by cooling down and opening the flask, otherwise the reaction time was extended to 40 h. The mixture was purified by column chromatography through a densely packed column of aluminium oxide (lower part) and silica (upper part) using dichloromethane as eluent. After evaporation of the solvents the products were obtained as yellowish gels with yields from 60 to 80% depending on the length of the column. Hydrolysis of the tert-butyl group was then done for 6 h at room temperature with 5 eq. trifluoroacetic acid in dichloromethane and yielded the final hydrophobically modified poly acrylic acid. To extract the hydrolysed products the solvent and the excess acid

were evaporated, the remaining product redissolved in dichloromethane, and precipitated with diethyl ether. The precipitant was washed several times with diethyl ether and hexane and during the washing continuously being grinded down, as during this process the product becomes brittle. After drying under vacuum, the product was pestled to a fine white powder. The yield was almost quantitative. 1 H NMR was used to confirm that the reaction was complete and to calculate the relative block lengths before hydrolysis from the integrals of the specific proton signals. To control the reaction progress, the vanishing of the double-bound proton-signals (6.33e5.80 ppm) was followed in comparison to the signal from the methyl protons from the tert-butyl group (1.50 ppm). In the hydrolysed product this signal is gone. The relative hydrophobic modification was calculated using the oxygen-neighbouring protons (CH2, 4.00 ppm) from the alkyl-chain (or the ending methyl group at 0.89 ppm) and the protons from the tert-butyl group. A more detailed description is given in the supplementary part (S7 and eq. S2a-f). In order to study the pH dependence of the copolymer a pH titration of the different produced copolymers was done. For that a certain amount of polymer was dissolved in water containing NaOH to deprotonate it completely. In the titration the excess of base was neutralized first, and beyond the first equivalence point the polymer then became protonated. The titration curves (see Figs. S3 e S5) look generically similar, showing a rather broad transitional range between pH 8 and pH 5, where the polymer becomes rather linearly decharged with lowering pH. The neutralization point cannot be precisely determined from such titrations as at very low pH the copolymer becomes insoluble and phase separates from the aqueous solution. However, taking this problem the best possible into account we could still determine the effective pKa, given as the point where 50% of the ionisable groups are deprotonated (a ¼ 0.5), and they are always in the range of 5.6e6.3, increasing with the chain length of the hydrophobic side chain, and being higher for the more highly substituted polymers (Table 1). This increase of pKaeff may be attributed to a compaction due to the presence of hydrophobic domains and accordingly a higher value for pKaeff might be expected with increasing tendency for hydrophobic domain formation. The degree of deprotonation a (ionized AA groups) can be defined as (where for n(OH) the initially added amount of NaOH was taken into consideration):

a¼


 n OH nðAAÞ

(1)

where n(OH) are the total moles of added NaOH and n(AA) the total moles of acrylic acid units (not considering whether they are actually protonated or not). In our further work we used the theoretical value for the chargeable groups contained to characterize our samples.

S. Riemer et al. / Polymer 70 (2015) 194e206

197

Table 1 Summary of the characteristic parameters degree of polymerization (DP) and molecular weight (Mn, Mw) of the polymers employed. The theoretical values were calculated from the amounts used for the synthesis, GPC was performed with THF as eluent with a polystyrene standard (both protected polymer). pH-titration was used to determine the mass per charged unit (unprotected). Mw,th, is the theoretical molecular weight of a charged unit under the assumption of full deprotonation. 1H NMR was used to determine the relative amount of hydrophobic modification (HM). In addition the effective pKaeff is given as defined by the point where 50% of the acid groups are ionised. 10 mol%

DPth

Mn,th [g/mol]

Mn (GPC) [g/mol]

PDI (GPC)

Mw,th [g/mol] (charged unit)

NMR-mol%

Mn,hyd [g/mol]

pKaeff

C4 C6 C8 C12

100 89 96 98

13040 11876 12978 14084

13140 15020 30730 14360

1.24 1.18 1.16 1.16

86.2 89.3 92.4 98.7

9.5% 9.1% 7.4% 9.7%

7979 7373 8153 9220

5.58 5.91 5.96 6.07

20 mol%

DPth

Mn,th [g/mol]

Mn (GPC) [g/mol]

PDI (GPC)

Mw,th [g/mol] (charged unit)

NMR-mol%

Mn,hyd [g/mol]

pKaeff

C4 C6 C8 C12

97 96 97 101

12593 13066 13726 15422

13650 11980 14560 17480

1.28 1.18 1.25 1.25

104 111 118 132

18.6% 20.2% 18.3% 17.8%

8255 8755 9371 10881

5.68 6.23 6.33 6.34

By gel permeation chromatography (GPC) we determined the apparent molecular weight and weight distribution of the nonhydrolysed polymers using THF as eluent with a flow rate of 1 ml per minute. The column was a SDV-type (styrene/divinylbenzene, 100 nm porosity and 5 mm particle size, 10000 nm (5 mm) and 1000000 nm (10 mm)) column from PSS GmbH. Calibration of the GPC column had been done by means of polystyrene (by PSS GmbH, 0.27e2570 kD) standards (for the measured GPC curves see Fig. S8). As our polymer deviates quite substantially from the calibration polymer the obtained molar masses have to be taken with some precaution, but show systematically the same trends as the theoretical values and, except for the 10 mol% isooctyl (C8) sample, are in very good quantitative agreement with the theoretical value. The polydispersity is always in the range of 1.15e1.25 (Table 1) as to be expected for polymers synthesized by ATRP.

2.2. Methods Densities were measured using a capillary Density Metre DMA 4500 by Anton Paar at 25
C. For pH titrations a Titrando System with the Software tiamo™ by Metrohm was used. The titrations were carried out at room temperature (22e24
C) and no effects of temperature were seen in a range of ±5
C. Light scattering was measured with a CGS-3 (compact goniometer system) with a HeNe-Laser at 632.8 nm wavelength and using a hardware correlator ALV-5000 from ALV GmbH (Langen, Germany). The scattered light was detected using two avalanche photo diodes (APD) for a pseudo cross-correlation at various angles between 30 and 150
. For dynamic light scattering (DLS) the majority of the measurements was performed at 90
. All samples were centrifuged at 5000 RPM for 10 min before the measurement in order to remove dust. The obtained field correlation function g(1) (t) was fitted using a double exponential decay with two relaxation modes:

  gð1Þ ðtÞ ¼ af *exp Gf *t þ as *expðGs *tÞ þ b

(2)

where af and as are the amplitude factors for the fast and slow mode (af þ as ¼ 1) and Gf and Gs the corresponding relaxations rates (Gi ¼ Dq2), respectively, while b accounts for some remaining small and constant background. The fast mode can be associated to the diffusive movement of the aggregated polymers, the slower one is typically seen in polyelectrolyte solutions due to charge interactions in solution, i. e. the so-called “slow mode” [40,41]. In static light scattering (SLS) the average scattered intensity for various angles (30
e150
) was recorded. The Rayleigh ratio Rq was

obtained by comparison with the scattering of toluene and correcting for the scattering of the cell. SANS measurements were performed at KWS1 from Jülich Center for Neutron Science (JCNS) at MLZ, Munich, Germany [42]. Three configurations were used (1.22, 7.72, 19.72 m sample-todetector distances) with a wavelength of 0.45 nm (wavelength spread, FWHM: Dl/l ¼ 0.1) to cover a q-range from 0.026 to 4.6 nm1, where q is the magnitude of the scattering vector that is defined as:

q¼

  4$p q $sin l 2

(3)

The 6Li glass detector has a size of 60  60 cm2 with 128 x 128 channels and was calibrated with a 1 mm water sample. Additional SANS experiments were performed at D11 at the ILL, Grenoble, France. Three configurations were used (1.4, 8.0, 34.0 m sample-todetector distances) with a wavelength of 0.6 nm (wavelength spread, FWHM: Dl/l ¼ 0.09) to cover a q-range from 0.0193 to 5.161 nm1 The 3He gas detector (CERCA) has a total size of 96  96 cm2 with a pixel size of 7.5  7.5 mm2 (128 x 128 channels) and was calibrated with a 1 mm water sample. The data reduction was done with BerSANS [43]. The scattering of the empty cell and the solvent (D2O) were subtracted as background before radial averaging and taking into account the transmissions the differential cross-sections were calculated. Subsequently the data sets obtained for the three different configurations were merged. Finally a constant background from the incoherent scattering was subtracted by extrapolating the intensity at large q by Porod's law. 3. Results and discussion In our investigation we studied polymers with a degree of polymerisation (DP) of ~100, which had a degree of substitution of 10 or 20 mol% (see Table 1) of the acrylic acid units by n-butyl (C4), n-hexyl (C6), isooctyl (C8), or n-dodecyl (C12) chains, to have a systematic variation of the hydrophobicity of the polymers. Of course, it might be noted that due to the rather low degree of polymerization in the individual copolymer molecules one has a rather wide distribution of the number k of hydrophobic moieties in the copolymers which may be described by a Poisson statistics (see Fig. S1 for an example at a given DP). Accordingly the individual hydrophobicity of the polymers can vary rather broadly. In our experiments we elucidated the effect of the degree of ionization a by employing a ¼ 0.15 and 1.0 (here we worked in reality with an excess of 15% of NaOH; to be safely on the side of full deprotonation). As mentioned before, these a values correspond to the theoretical charge per mass.

198

S. Riemer et al. / Polymer 70 (2015) 194e206

3.1. Phase behaviour All solutions were homogeneous, transparent and showed basically water-like viscosity. However, the polymers are poorly soluble in pure water without adding a small amount of sodium hydroxide. Decreasing the pH below 3 leads to turbidity and precipitation of the polymers, due to the then rather complete protonation of the polyacrylate (see also Figs. S4 and S5), which renders the polymer too hydrophobic to remain water-soluble, as similarly seen for pure PAA. Upon mixing, for a short time foam is formed which vanishes within a few minutes afterwards, indicating the surface activity of these copolymers. In order to find out whether these solutions are just molecular solutions or contain hydrophobic domains we probed this with a water insoluble dye. For that purpose we used „Sudan III“ as indicator dye and admixed it to solutions of the polyelectrolytes at different a of 0.15 and 1.0. In Fig. 2 pictures for 20 mol% n-butyl- and n-hexyl acrylate containing polyelectrolytes are shown. It is evident that the dye becomes well solubilized at low pH (a ¼ 0.15, pH ¼ 3e4) but much less for high pH (a ¼ 1.0, pH ¼ 12), while it is not at all solubilized into the pure water solution. This already indicates that the presence of hydrophobic domains in the polymer solutions depends strongly on pH, and to a certain extent also on the length of the hydrophobic alkyl modification. The aggregation behaviour was also studied by measuring the critical aggregation concentration (cac) of the copolymers with isooctyl and dodecyl side chains by means of the pyrene fluorescence method [44]. In Fig. S11 the intensity ratio I3/I1 is given as a function of the copolymer concentration for the different degrees of hydrophobic modification (10 or 20 mol%) and for a ¼ 0.15 or 1.0. One clearly observes the typical sudden decrease of the I3/I1 ratio that indicates the formation of hydrophobic domains, i. e. the cac. The cac values are summarised in Table S2 and show very low values of ~1 mg/L for a ¼ 0.15 for both degrees of substitution and for C8 and C12. At complete dissociation of the copolymers (a ¼ 1.0) the situation is different as here the cac increases by about two orders of magnitude for the C8 copolymer but for the C12 copolymer almost no change is seen for 20 mol% modification while a rise by a factor 40 occurs for the 10 mol% modification, i. e. for the C12 copolymer the pH dependence of the cac depends largely on the extent of hydrophobic modification. The observed large decrease of cac with the extent of hydrophobic modification seen for C12 is similar as seen before for octyl modified carboxymethylpullulan (CMP) [45], i. e. apparently here the fully charged PAA is significantly more hydrophilic than the CMP. This indicates that for the C12 copolymer the dodecyl chain is controlling the

aggregation behaviour irrespective of the charging conditions of the PAA backbone and does not depend much on the degree of hydrophobic modification. For the isooctyl modification the situation is quite different, as here apparently aggregate formation depends strongly on the charging of the PAA backbone and also on the extent of hydrophobic modification. Apparently for the C8 chain its hydrophobic effect is comparable to the desire of the charged NaPA chain to be in solution and accordingly here one has a rather wide tunability of the aggregation behaviour by pH. Of course, always an important point regarding the state of polymer solutions is their concentration, and in particular, whether one works above or below the overlap concentration cov. For the case of our copolymers one may estimate that concentration for the two extreme cases of either having a fully stretched chain or a random coil. In both cases cov may be estimated from cov ¼ Mw/ (NAv$L3eff), where Mw is the molecular weight of the polymer, NAv the Avogadro constant, and Leff the effective extension of the polymer. For a degree of polymerisation N ¼ 100 one has for the stretched case Leff ¼ 25 nm and Mw is ~10000 g/mol (Table 1). With these numbers on arrives at a value for cov of 1.1 g/l. In contrast, assuming a coil configuration and taking for Leff the end-to-end distance Ree of the polymer, which is Ree ¼ L$(C∞$N)0.5 ¼ 7.2 nm with C∞ ¼ 8.3 being the characteristic Flory ratio, and with L ¼ 0.25 nm as length of one monomeric unit, this model then yields an overlap concentration of 45 g/l. This means that effectively in all our experiments that cover the range of 1e10 g/l we work in the vicinity of the overlap concentration, and accordingly may expect the formation of volume filling polymer network or at least bridging of aggregates due to the extension of the individual polymer chains. 3.2. Structural characterisation 3.2.1. Light scattering In order to gain a first insight into the aggregation behaviour of these copolymers in aqueous solution we performed static and dynamic light scattering (SLS, DLS) experiments. This was done with little (a ¼ 0.15) and fully (a ¼ 1.0) deprotonated samples, for two concentrations (1 and 5 g/L) with a 20 mol% content of dodecyl and isooctyl moiety and for 5 g/L with a 20 mol% of n-butyl and nhexyl moiety. For a comparison 5 g/L of pure polyacrylic acid, prepared as the modified acrylates, was measured as well.

.   IðqÞ ¼ Ið0Þ$exp  R2G $q2 3

(4)

Fig. 2. Photographs of the uptake of a hydrophobic dye (Sudan III). A: 20 mol% n-butyl; B: 20 mol% n-hexyl; 1 is a water blank sample, 2 a ¼ 0.15, 3 a ¼ 1.0 (all samples are at a polymer concentration of 2 g/L.).

S. Riemer et al. / Polymer 70 (2015) 194e206

Mw ¼

Ið0Þ c$K

199

(4a)  2

K¼4

p2 $n20 $

dn dc

(4b)

NA $l4

The static light scattering data was analysed by Guinier plots (see figs. S9) for these samples (eq. (4)), yielding I(0), from which we derived the molecular weight, Mw (eq. (4a)), employing an estimated value of 0.15 ml/g for the refractive index increment of the different copolymers (it might be noted here that in particular for the C8 case one sees in several cases two slopes for the Guinier plot, which may be attributed to the fact that at lower q the larger and more loose structure of the aggregates is seen, while at higher q then their more compact domains become visible; cf. for this also the SANS experiments in 3.2.2). The obtained values are summarised in Table 2. Only for samples with a dodecyl moiety the calculated molar masses are rather similar, independent of concentration and pH of the sample. For the shorter iso-octyl modification, one observes a pH dependent behaviour with the protonated samples having a substantially lower Mw. Depending on the degree of ionisation a the Mw values are either higher (a ¼ 1.0) or lower (a ¼ 0.15; more hydrophobic) than for the dodecyl substitution. For the still shorter hydrocarbon chains (butyl and hexyl), especially for the hexyl moiety a much larger molecular weight is observed, irrespective of a, while for n-butyl the Mw is lower again, with larger molar mass for low pH, i. e. the protonated case. Pure polyacrylic acid, prepared like the modified polyacrylates, shows a similar behaviour as the n-butyl modification but even with somewhat larger Mw. We have to point out that for the short hydrophobic chain (butyl and hexyl) as well as for the pure PA the data-quality was not very good, despite the fact that we carefully avoided having dust or other contaminations in the samples. It seems that in particular for the dodecyl moiety and to a much lesser extent for the protonated iso-octyl moiety the system is forced into a more well-defined structure due to the more pronounced hydrophobic interactions. In contrast for the shorter hydrophobic modifications the electrostatic repulsion leads to more extended polymer conformations and correspondingly bigger, loose aggregates are formed (see also Fig. 9). They are much larger than the polymer or simple aggregates that could be formed by combining enough hydrophobic units to have 50e80 together, as would be required for forming one hydrophobic micellar domain. The DLS intensity autocorrelation functions (Fig. 3, see also Fig. S2) show a decay that can be described with two exponential

Fig. 3. Intensity correlation functions for 20 mol% samples with 5 g/L at a scattering angle of 90
. The bad data quality for C4 and a ¼ 1.0 is easy to observe, for the other curves the biexponential decay is nicely seen, as well as the discussed pronouncing of the fast mode for longer chain lengths.

functions (eq. (2)). Comparing the relative amplitudes for the different samples one finds that the amplitude of the fast mode increases with increasing angle, as to be expected as one focuses more on the smaller structural scale with correspondingly faster relaxation modes. For the samples with dodecyl or iso-octyl moiety the amplitude of the fast mode increases linearly for angles from 30 to 100
and then stays almost constant for higher angles. In general, the fast mode becomes much more pronounced with increasing chain length of the hydrophobic modification (Fig. 4), which indicates that it is associated with the presence of hydrophobic domains. Similarly we observe a larger fast amplitude for lower degree of ionisation. The observation of a pronounced fast relaxation mode strongly indicates the formation of compact hydrophobic aggregates with radii of 2e4 nm (Table 2). For pure polyacrylic acid, which compared to the modified copolymer acrylates can not have the same type of hydrophobic interactions, one finds a similar light scattering behaviour as for the modified acrylates with a short hydrophobic modification. Accordingly we may assume that the slower mode, getting more pronounced with a higher charge density in the polymer, can be ascribed to the “slow mode” of polyelectrolytes [40,41,46e48]. For all samples for both modes the power law dependence of the relaxation rate G was determined: G ~ qn,f(n,sl) and the characteristic exponents for the fast (n,f) and the slow mode (n,sl) are given in Table 2. The fast mode scales with the square of the scattering

Table 2 Parameters obtained from the light scattering experiments. Intensity extrapolated to zero-angle, I(0), molecular weight Mw and the calculated polymer aggregation number Nagg,p as obtained from SLS by dividing the Mw (of the aggregates) by the Mn,hyd of the hydrolysed polymer (see Table 1). Power law exponent nf and nsl for the fast and slow relaxation mode, respectively, amplitude of the fast mode Af,90
and corresponding hydrodynamic radius RH,90 at 90
from DLS (*: not diffusive). SLS intensities and deduced Mw and Nagg,p are estimated to have an error of 10%. Moiety

c [g/L]

a

I(0) [1/cm]

Mw [kg/mol]

Nagg,p

Af,90


nf

nsl

RH,90 [nm]

C0; PAA C0, PAA C4 C4 C6 C6 C8 C8 C8 C8 C12 C12 C12 C12

5 5 5 5 5 5 1 1 5 5 1 1 5 5

0.15 1.0 0.15 1.0 0.15 1.0 0.15 1.0 0.15 1.0 0.15 1.0 0.15 1.0

8.60E-4 5.88E-4 4.15E-4 1.02E-4 1.46E-3 1.21E-3 1.32E-5 5.73E-5 3.94E-5 2.05E-4 2.16E-5 2.22E-5 8.73E-5 7.08E-5

1050 730 512 126 1800 1500 81 350 49 250 133 137 108 100

130.6 89.2 62.0 12.3 205.6 171.3 8.6 37.3 5.2 26.7 12.2 12.6 9.9 9.2

e e 0.142 0.286 0.112 0.078 0.570 0.265 0.609 0.203 0.808 0.657 0.766 0.527

e e 1.98 2.62 2.15 2.08 2.04 2.45 1.98 1.96 1.92 1.93 1.90 1.93

e e 2.53 2.73 2.57 2.46 2.29 2.30 2.27 2.04 2.1 2.3 2.16 2.26

e e 2.7 1.7* 4.5 4.4 2.7 3.9* 2.4 2.4 4.0 4.3 3.2 3.2

200

S. Riemer et al. / Polymer 70 (2015) 194e206

Fig. 4. Development of the relative amplitudes for the fast mode derived from biexponential fits over the scattering angle for pure PAA, and the butyl (C4), hexyl (C6), isooctyl (C8) and dodecyl (C12) modified copolymers (20 mol% hydrophobic modification; concentration of 5 g/l) (at 25
C with 10 mM of NaCl). Left side a ¼ 0.15, right side a ¼ 1.00.

vector q (Fig. S4), which indicates diffusive behaviour. Only for the iso-octyl copolymer at high degree of ionisation and at low concentration a pronounced deviation of the exponent to ~2.62 is observed. A reason for this deviating behaviour might be that for iso-octyl one is exactly at a point where the hydrophobic interaction is just about similar to the electrostatic repulsion of the backbone. The calculated hydrodynamic radii of the different samples (deduced from the fast, diffusive mode) are always similar in the range of a few nm for the same concentration (a summary of parameters can be found in Table 2), independent of the degree of deprotonation. The slow mode, being an order of magnitude slower than the fast mode, systematically has a q-dependency higher than two, i. e. in the range of 2.2e2.8 which indicates a non-diffusive character and in a range typically observed for polyelectrolyte or polymer networks [49,50]. In summary, the fast mode represents the smaller hydrophobic domains inside the larger polymer network, from which arises the slow mode. Both techniques, static and dynamic light scattering show complementary but similar results for largely protonated (a ¼ 0.15) and deprotonated (a ¼ 1.0) samples. For the dodecyl modified polymer in general and for the deprotonated iso-octyl polymer, DLS gives a pronounced diffusive mode which can be ascribed to compact aggregates (Fig. 4). For the same samples SLS gives smaller overall molar masses, which indicates that the samples with less hydrophobic modification form larger systems with interconnected polymer chains, facilitated by the presence of the hydrophobic alkyl chains. The obtained hydrodynamic radius RH (derived from the fast mode) then is not giving directly the size of the hydrophobic domains contained but their effective mobility within these much bigger agglomerates. Therefore these values (Table 2) are somewhat bigger than expected for the micellar domains themselves. The amplitude of this fast mode becomes smaller with the shortening of the hydrophobic modification, thereby reflecting the decreasing amount of material contained in them. 3.2.2. Small-angle neutron scattering (SANS) In order to obtain a more refined structural picture of the organisation of the copolymers in solution, SANS experiments were performed to gain information using sub-nm resolution. For each copolymer with 20 mol% butyl to dodecyl modification and for a of 0.15 and 1.0 we studied different concentrations of 1, 5 and 10 g/L. In general, the scattering curves show rather little concentration dependence when normalised for the copolymer concentration and, especially for the longer hydrophobic modifications and at lower degree of ionisation, are quite similar for q-values larger than

0.2 nm1 (see Fig. 5), i. e. in the range that should be associated with the structure of the hydrophobic domains (their form factor). The changes that can be seen in the range of 0.3e0.6 nm1 can be attributed to a structure factor that becomes more prominent with increasing concentration and indicates a rather well defined mean spacing between the hydrophobic domains. Mainly for the samples with C4 the scattering increases with increasing concentration; and the formation of hydrophobic domains, including a correlation peak, is only wellvisible for the highest concentration with the protonated sample (a ¼ 0.15), but the formation of small aggregates is already visible for the lower concentrations. Sticking to the picture that the contrast mainly comes from the hydrocarbon chain, the low intensities confirm the absence of hydrocarbon-rich domains for the C4 copolymer in the deprontonated state (a ¼ 1.0). Starting from C6 for the protonated samples (a ¼ 0.15) a well defined globular structure may be inferred from the scattering curve in the intermediate and high q-range. However, it is certainly interesting to note that even for a ¼ 1.0 already some formation of aggregates can be seen even at the lowest concentration and for the highest concentration a weak correlation peak appears that indicates a mean spacing of ~8e9 nm between these small domains. This behaviour basically vanishes largely for the deprotonated samples, which indicates that here only much fewer and/or smaller hydrophobic domains are present and apparently the strongly charged polymer is mostly present in the form of extended chains that do not easily allow the formation of hydrophobic aggregates. This explains the much lower scattering intensity at intermediate to high q. For the low q-regime a q2.6 dependence of the scattering intensity is observed that might be attributed to the formation of a polymeric network on a larger length scale and is similarly seen for C4, C6, and C8 (the exponent for mass fractals is between 2 and 3). The onset of this q2.6 increase is shifted to lower q with increasing concentration, thereby indicating the presence of larger network structures with increasing concentration. Only for the C12 samples within the observed q-range no such increase is seen. Instead a much more pronounced domain scattering is seen both for low and high degree of ionisation, i. e. here the hydrophobicity is strong enough to allow for formation of hydrophobic domains under all probed conditions. For a ¼ 0.15 and 10 g/l for all samples a well visible correlation peak is observed in the range of 0.3e0.5 nm1, which moves to lower q with increasing chain length of the hydrophobic modification. This peak indicates the rather high degree of ordering of these hydrophobic domains that must be facilitated by the presence of the somewhat charged polymer

S. Riemer et al. / Polymer 70 (2015) 194e206

201

Fig. 5. SANS intensity, normalised to the concentration, as a function of q for samples of 20 mol% hydrophobically modified polyacrylate; C4: butyl, C6: hexyl, C8: isooctyl, C12: dodecyl; for various concentrations and for degrees of deprotonation a ¼ 0.15 and 1.0. All samples contained in addition 10 mM NaCl; identical concentrations have the same colours, closed symbols stand for protonated (a ¼ 0.15), open for deprotonated (a ¼ 1.0) samples. The samples with 1 and 5 g/L for C4 and C12 were measured at D11, all other samples at KWS1 (for the error bars see Fig. S14). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

chains and leads to a corresponding average spacing between 20 and 12 nm. Effectively one has here micellar domains surrounded by a charged polymer corona. It is also interesting to compare directly the scattering curves for the variation of the length of the hydrophobic chain. In Fig. 6a this is done for the case of 10 g/L concentration and a ¼ 0.15 and one always observes a weak correlation peak, whose position shifts to lower q with increasing length of the hydrocarbon chain (Fig. 6a) in a rather linear way (Fig. 6b). This would be in agreement with a domain size that is directly proportional to the chain length (and thereby correspondingly the spacing between the domains scales identically), i. e, the size is simply given by the stretched length of the alkyl chain as conventionally observed in surfactant micelles [51]. The intensity of the peak is increasing strongly with increasing

alkyl chain length. For the case of a domain size directly proportional to the chain length as determined by the number of carbon chains nc one would expect that the scattering intensity of the domains is described by:

I ¼ B þ f$

n4c 1 þ 0:0389$nc

(5)

Such a dependence arises as one has a proportionality of ~n3c from the increase of the volume of the individual aggregates and another to ~nc for the increase of the volume fraction of domains present (as the mass concentration is constant), while the division by 1 þ 0.0389*nc accounts for the fact that for constant mass concentration of the polymer with increasing chain length of the hydrophobic modification the number of hydrophobic chains in the

Fig. 6. a) SANS scattering intensity as function of the magnitude q of the scattering vector for samples with a concentration of 10 g/L and a ¼ 0.15 and for different alkyl chain lengths (measured at KWS1) b) position qmax of the peak and peak intensity as a function of number of C atoms in the alkyl chain. The increase in intensity is described by eq. (5). Pure poly acrylic acid (prepared as the other polymers) is given as a reference for pure backbone scattering.

202

S. Riemer et al. / Polymer 70 (2015) 194e206

sample decreases. As shown in Fig. 6b for the peak intensity this relation is very well fulfilled using eq. (5) with a background B ¼ 0.257 cm1, which accounts for scattering of the polymer backbone, and a proportionality factor f of 1.46 104 cm1. Accordingly this picture of having domains whose size is simply determined by the length of the hydrophobic chain is confirmed. In that sense obviously the hydrophobically modified polymers behave identical to simple surfactants. The increase at low q may be interpreted as an interconnection of hydrophobic domains and a network formed by the polyelectrolyte chains as the stretched polymer has a length of about 25e27 nm (0.252 nm/monomer unit) and accordingly is able to bridge the average distance d between the hydrophobic domains of 11e20 nm (as seen from the peak at qmax ~ 0.3e0.55 nm1 and d ¼ 2p/qmax), thereby forming a sample-spanning network. Therefore we extrapolated an effective I0,eff for the hydrophobic domains (micelles) from the intermediate q-range (0.7e1.5 nm1) by using Guinier's approximation (eq. (4), see also Figs. S3 and S12). From I0,eff we then calculated an effective molar mass Mw,eff and an aggregation number NAgg,p for the hydrophobic domains (Table 3) for the samples with a concentration of 5 g/L at a ¼ 0.15 (for a ¼ 1.0 this does not work as there only for C12 a similar scattering pattern is observed). NAgg,p was simply obtained by dividing Mw,eff by the molecular weight of the respective alkyl chain (butyl to dodecyl). In a next step the SANS data was fitted for q > 0.1 nm1 (below which the scattering upturn occurs, for an example see Fig. S13) with a model of polydisperse spheres that are formed by the hydrophobic chains. This model is described by eqs. (6) and (7), where R is the radius of the aggregates, DSLD the contrast (difference in scattering length densities between the hydrocarbon chain and the average of the medium), and q the magnitude of the scattering vector. The polydispersity of the scattering domains, which is experimentally evidenced by the absence of oscillations at higher q in the scattering intensity curves, was described with a LogNormal distribution (eq. (8)), with the width parameter s, the mean radius Rm, and the number density parameter N(fp) expressed in terms of the volume fraction fp, that can described using eq. (8a) and eq. (8b) where < R3> is the third moment of the LogNorm distribution of the radii.

Z∞ Iðq; Rm Þ ¼ 1 NðfpÞ$

LogNormðR; NðfpÞ; s; Rm Þ$Fðq; RÞ$dR

1

NðfpÞ ¼

  D E 9 R3 ¼ R3m $exp $s2 2

Fðq; RÞ ¼

!2 4p$R3 $DSLD$ðsinðqRÞ  qR$cosðqRÞÞ

(7)

3

ðqRÞ

NðfpÞ lnðR=Rm Þ2 exp LogNormðR; NðfpÞ; s; Rm Þ ¼ R 2$s2

! (8)

Table 3 Parameters obtained from SANS for samples with c ¼ 5 g/l and a ¼ 0.15. Given are I0eff extrapolated from the intermediate q-range, calculated effective molar mass Mw,eff and aggregation number NAgg,p for the alkyl side chains, assuming that only they are forming the hydrophobic domains. The estimated error of the SANS intensity I0eff and the deduced quantities is estimated to be 10%. 20 mol%

I0eff [1/cm]

Mw,eff [g/mol]

NAgg,p

C4 C6 C8 C12

0.12 0.25 0.48 0.82

3640 9070 24800 37540

64 108 219 222

(8a)

(8b)

For the scattering length densities (SLD) of the core we assumed pure hydrocarbon chains and the volume fractions are given by their content in the samples. In Table 4 the results are given, and one finds that the radius of the aggregates increases step-wise and is directly proportional to the chain length of the hydrophobic alkyl chain (Fig. 7), where for the data shown in Fig. 7 we employed the Rm values in parentheses (where existing) from Table S4, as they should correspond to the value arising from the alkyl chains. In addition, a smaller systematic increase of the size with increasing concentration of the samples is seen, as it has similarly been observed for hydrophobically modified polyacrylamide [52]. Here it might be noted that for the C4 polymers the data quality (due to low scattering intensity) is rather poor and accordingly the deduced parameters associated with rather large errors. The ratio between fitted and calculated volume fraction fp/F gives an idea about the extent of alkyl chains contained in the spherical micelles and in addition, how large the tendency for hydrophobic aggregation is. This means, it also accounts for the incorporation of PAA into the hydrophobic domains, which may be expected at low pH when PAA becomes hydrophobic. For short chains at low concentration the observed scattering is much lower than expected for complete aggregation of all chains. The short hydrocarbon chains have a much lower tendency to form aggregates because of their lower hydrophobicity and a larger part remains dispersed in solution and is not contained in hydrophobic domains. In contrast, for long chains the ratio is above one and quantifies the amount of PAA that is effectively a part of the hydrophobic domains. Due to the statistical distribution of the hydrophobic modification this is expected to occur, in order for the system to minimize the hydrophobicehydrophilic interactions. With the radii, the density and the molar mass of the alkyl chain, it is possible to calculate the number of alkyl chains forming an aggregate (NAgg, see Table 4) and the number of involved polyelectrolyte molecules by means of:

0

(6)

fp$3  4$p$ R3

NAgg ¼

NAv $r$4$p$R3m M$3$ðfp=FÞ

(9)

where NAv is the Avogadro constant, r and M density and molar mass of the given alkyl chain. The factor fp/F accounts for the fact that not all alkyl chains are in the aggregates or for strong aggregation tendency also some PAA might be incorporated (a ¼ 0.15) and has to be considered in eq. (9) only when it is larger than 1 (otherwise is set to one). Similarly for the calculation of the effective sphere radius due to the alkyl chains fp/F was set to one for the cases it was exceeding this value and this effective Rm then is given in parentheses in Table 4. This radius is also the one considered in the calculation of the head group area ah per alkyl chain. The values for Rm and NAgg are systematically smaller than those obtained from the extrapolation to I0 in the Guinier approach (Table 3). This may be attributed to the fact that with the approximation applied before one has a systematically too high estimate of the low q intensity value as also some scattering of the polymer chains is contributing to it. However, the values derived from the full fit can be considered to be more realistic. The SANS curves for the fully deprotonated samples (a ¼ 1.0) show substantially lower scattering intensities than the corresponding samples at a ¼ 0.15 (Fig. 5). Especially the ones with the

S. Riemer et al. / Polymer 70 (2015) 194e206

203

Table 4 Parameters calculated from a spherical model for samples with a ¼ 0.15. Performed at: a) KWS1,b) D11; DSLD is the scattering contrast of the pure chains and D2O, F the calculated volume fraction (eq. S(1)) of the hydrophobic chains, fp the volume fraction from the fit, Rm the mean radius of the hydrophobic domains (eq. (8b)), and NAgg the aggregation number of the alkyl chains as calculated from eq. (9) using values for the density and molar mass of the pure hydrocarbons, respectively. The radii in brackets are reduced radii where the ratio fp/F was set to one if it exceed one. The errors for the fitted parameters Rm and fp were estimated from the fit routine to be about 2% each. Moiety b)

C4 C4b) C4a) C6a) C6a) C6a) C8a) C8a) C8a) C12b) C12b) C12a)

c [g/L]

DSLD [nm2]

F (calc.)

Fp/F

Rm [nm]

NAgg (alkyl)

ah [nm2]

1 5 10 1 5 10 1 5 10 1 5 10

6.79E-04 6.79E-04 6.79E-04 6.77E-04 6.77E-04 6.77E-04 6.85E-04 6.85E-04 6.85E-04 6.73E-04 6.73E-04 6.73E-04

0.000237 0.001183 0.002343 0.000291 0.001423 0.002876 0.000253 0.001333 0.002693 0.000427 0.002334 0.004225

0.382 0.143 0.825 1.326 1.200 1.187 1.619 1.559 1.459 0.988 0.985 1.264

1.00 1.14 1.34 1.37 1.52 1.59 1.72 1.78 1.84 2.22 2.31 2.48

25.6 38 61.7 37.9 57.2 66.2 61.7 71.1 83.9 122.5 138.0 135.1

0.49 0.43 0.37 0.52 0.45 0.43 0.44 0.42 0.39 0.51 0.49 0.49

butyl and hexyl moiety apparently have very little tendency for the formation of hydrophobic domains and accordingly were not analysed by means of the spherical model, as that makes no sense here. The information for I0 and Mweff as deduced from the Guinier approximation is summarised in Table S3. In contrast, the isooctyl and the dodecyl chains lead to some domain scattering, and in particular for the case of dodecyl one still has a rather pronounced formation of hydrophobic domains, despite the electrostatic repulsion. However, these domains are systemically smaller than for the low deprotonated ones. In Fig. 8 we compare for a ¼ 1.0 the 10 g/l samples and have also included for a comparison the scattering of a sample of pure poly acrylic acid, prepared by the same procedure as the copolymers, where the pure PAA accounts for the scattering arising from the backbone. Fits with the sphere model (eqs. (6) and (7)) give smaller fp values as the calculated volume fractions F and the ratio is getting closer to 1 for increasing concentrations (Table 5). However, it never exceeds one, which confirms our picture that the values larger than one seen for a ¼ 0.15 are due to the presence of unprotonated PAA, which is not the case for a ¼ 1.0. The fit parameters are summarized in Table 5 and show consistently smaller values for radius and aggregation numbers compared to the less charged (a ¼ 0.15) counterparts. This means that here the electrostatic repulsion of the backbone and its increased hydrophilicity does not only suppresses the formation of hydrophobic domains itself but if formed they are also smaller in size. For the longer alkyl chains, and here in particular for the isooctyl, one apparently has the situation that electrostatics (that works against aggregate formation) and the hydrophobic effect of the alkyl chain are just about

Fig. 7. Radius of the hydrophobic domains as a function of the number of carbons in the hydrophobic alkyl chain for samples with a ¼ 0.15, as deduced from the SANS curves (the error for the values of R is 2%).

(1.25) (1.43) (1.50) (1.46) (1.54) (1.62)

(2.29)

Fig. 8. SANS scattering intensity as function of the magnitude q of the scattering vector for samples that contained 20 mol% hydrophobic substitution (and for a comparison pure poly acrylic acid) with a concentration of 10 g/L and a ¼ 1.0 for different alkyl chain lengths (data from KWS1).

similarly strong, i. e. in this range one can tune very subtly the aggregation via pH and the length of the hydrophobic modification. In contrast, for the longer dodecyl chain one observes pronounced formation of hydrophobic domains in aqueous solution, which depends somewhat on pH but is seen also for high pH. However, in general more open (“fluffy”) aggregates are formed for higher degree of ionisation, as also evidenced by the viscosity of the samples. For instance, the 10 g/l sample with 20 mol% dodecyl modification has at a ¼ 0.15 a viscosity of 0.948 mPas and at a ¼ 1.0 one of 1.0804 mPas. By employing the viscosity formula for hard spheres of Thomas [53] this translates into a hard sphere volume fraction of 0.0234 for a ¼ 0.15 and one of 0.0661 for a ¼ 1.0, i. e. here one observed an effective swelling of the aggregates in solution by about a factor 3. From the geometry of the hydrophobic domains one can calculate the head group area (ah ¼ 4pR2m/Nagg) per alkyl chain, where here we employed the radius relating to the alkyl chains (in parentheses in Table 4), which means we focus on the interface between hydrophobic alkyl chains and the hydrophilic polymer. In general, it can be noted that the values obtained for ah are in the range of 0.4e0.7 nm2, i. e. in the range as it is typically observed for surfactants [54]. Here it can be noticed that for a ¼ 0.15 the ah values are 0.4e0.5 nm2 and for a ¼ 1.0 0.6e0.7 nm2, which can be understood such, that at a ¼ 1.0 the repulsive interaction due to the electrostatic charging acts in the same fashion as in the case of head group areas of ionic surfactants

204

S. Riemer et al. / Polymer 70 (2015) 194e206

Table 5 Parameters calculated from a spherical model for the samples with a ¼ 1.0. F is the calculated volume fraction (eq. S(1)) of the hydrophobic chains, fp the volume fraction from the fit, Rm the mean radius of the hydrophobic domains, Nagg the aggregation number of the alkyl chains (eq. (9); using the density and molar mass of the pure hydrocarbons), and ah the area per alkyl chain of these hydrophobic domains (ah ¼ 4pR2m/Nagg), respectively. (DSLD is the scattering contrast of the pure chains and D2O; data from: a) KWS1,b) D11) The errors for the fitted parameters Rm and fp were estimated from the fit routine to be about 2% each. Moiety a)

C8 C8a) C12b) C12b) C12a)

c [g/L]

DSLD [nm2]

F (calc.)

Fp/F

Rm [nm]

NAgg (alkyl)

ah [nm2]

5 10 1 5 10

6.85E-04 6.85E-04 6.73E-04 6.73E-04 6.73E-04

0.001260 0.002515 0.000427 0.002134 0.004010

0.798 0.854 0.553 0.806 1.000

0.96 1.06 1.88 1.61 1.79

17.4 23.4 74.4 46.7 64.2

0.67 0.60 0.60 0.70 0.63

where ah also increases with increasing effective degree of charging of the surfactant head group [54]. 4. Conclusions The aggregation behaviour of hydrophobically modified polyacrylates in aqueous solution has been investigated for different degrees of deprotonation (pH). The hydrophobicity was varied systematically by the percentage of hydrophobic modification and in particular by the length of the hydrocarbon moiety, for which we employed butyl, hexyl, isooctyl, and dodecyl, which are statistically distributed along the polyelectrolyte backbone. Here it should be noted, that due to the higher reactivity of monomers with increasing length of the hydrophobic modification [39] we do not expect to have copolymers with perfect random distribution of the monomer units, but with some tendency for gradient formation. For the aggregation properties percentage and hydrophobicity of this moiety and hydrophilicity and electrostatic repulsion of the polymer backbone, as controlled by the degree of ionisation, play a major role. Due to close distances of the potentially charged groups in the backbone, a high charge density not only renders the polyelectrolyte much better soluble but also increases the stiffness of the polyelectrolyte chain, which affects the ability of the attached side chains to arrange for the formation of hydrophobic, micellar domains, rendering it more difficult with increasing charge density. Light scattering experiments showed the formation of larger sized aggregates where different polymer chains are interconnected by hydrophobic contact points or for longer alkyl chains by the hydrophobic domains, which is also reflected by the low qincrease of the SANS scattering intensity (I ~ q2.6). These fluffy

aggregates of polymer chains are the larger the shorter the hydrophobic alkyl chain of the copolymers, as apparently the formation of hydrophobic domains restricts the growth of larger assemblies, since it leads to having more alkyl chains of one polymer contained in a given aggregate thereby reducing the effective extension of the polymer chain. The presence of the micellar domains is seen in thorough detail by means of SANS, which shows that their size grows directly proportional with the length of the hydrophobic alkyl chain. In addition, we found that hydrophobic domains are much more easily formed for low deprotonation of the PAA backbone and their formation depends also on the percentage of hydrophobic modification of a given copolymer. For full deprotonation no domains are formed for short alkyl chains, for isooctyl partial micelle formation is seen and only for dodecyl a rather complete formation is seen. Apparently only for dodecyl the hydrophobic effect is sufficiently strong to induce aggregation and overcome electrostatic repulsion and hydrophilicity of the PAA chain. At low pH part of the (undissociated) PAA becomes incorporated into the spherical hydrophobic domains. This shows that in summary the formation of aggregates (see Fig. 9) in these amphiphilic copolymers is controlled by a combination of the hydrophilic properties of the polymer backbone, as controlled by pH, and the hydrophobic driving force for aggregation, as controlled by the length, and to a lesser extent the percentage, of the alkyl chains of the hydrophobic modification. In summary this investigation shows how the amphiphilic and aggregation properties of hydrophobically modified polyacrylates can be tuned systematically by the length of the hydrophobic modification and the pH of the aqueous solution. This means that these copolymers are interesting amphiphiles, where the

Fig. 9. Scheme of the different aggregation behaviour for the differently long hydrophobic side chains and at different pH.

S. Riemer et al. / Polymer 70 (2015) 194e206

amphiphilic and aggregational properties can easily be controlled over a large range by the choice of molecular architecture and the extent of ionisation. Acknowledgements We are grateful to the Jülich Center for Neutron Scattering (JCNS) at the FRM-II (Munich, Germany) and the Institute LaueLangevin (ILL, Grenoble, France) for allocation of SANS beam time, and to Christoph Wieland from Fraunhofer IAP (Potsdam, Germany) for performing the GPC measurements. For the viscosity measurements we would like to thank Jana Lutzki. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.polymer.2015.06.010. References [1] K.T. Wang, I. Iliopoulos, R. Audebert, Viscometric behaviour of hydrophobically modified poly(sodium acrylate), Polym. Bull. 6 (1988) 577e582. [2] C.L. McCormick, J.C. Middleton, D.F. Cummins, Water-soluble copolymers. 37. Synthesis and characterization of responsive hydrophobically modified polyelectrolytes, Macromolecules 25 (1992) 1201e1206. [3] P. Guenoun, H.T. Davis, M. Tirrell, J.W. Mays, Aqueous micellar solutions of hydrophobically modified polyelectrolytes, Macromolecules 29 (1996) 3965e3969. [4] L.E. Bromberg, D.P. Barr, Aggregation phenomena in aqueous solutions of hydrophobically modified polyelectrolytes. A probe solubilization study, Macromolecules 32 (1999) 3649e3657. [5] F. Candau, J. Selb, Hydrophobically-modified polyacrylamides prepared by micellar polymerization, Adv. Coll. Interface Sci. 79 (1999) 149e172.
, E. Buhler, Structural and rheological properties [6] C. Esquenet, P. Terech, F. Boue of hydrophobically modified polysaccharide associative networks, Langmuir 20 (2004) 3583e3592. [7] A.V. Dobrynin, M. Rubinstein, Hydrophobically modified polyelectrolytes in dilute salt-free solutions, Macromolecules 33 (2000) 8097e8105. [8] C.L. McCormick, T. Nonaka, C.B. Johnson, Water-soluble copolymers: 27. Synthesis and aqueous solution behaviour of associative acrylamideNalkylacrylamide copolymers, Polymer 29 (1988) 731e739. [9] S.W. Shalaby, C.L. McCormick, G.B. Butler, Water soluble polymers, in: Synthesis, Solution Properties and Applications ACS Symposium Series, vol. 467, American Chemical Society, Washington, DC, 1991. [10] S. Biggs, A. Hill, J. Selb, F. Candau, Copolymerization of acrylamide and a hydrophobic monomer in an aqueous micellar medium: effect of the surfactant on the copolymer microstructure, J. Phys. Chem. 96 (1992) 1505e1511. [11] W.K. Ng, K.C. Tam, R.D. Jenkins, Evaluation of intrinsic viscosity measurements of hydrophobically modified polyelectrolyte solutions, Eur. Polym. J. 35 (1999) 1245e1252. [12] M. Gradzielski, A. Rauscher, H. Hoffmann, Hydrophobically cross-linked micellar solutions: microstructure and properties of the solutions, J. Phys. IV 3 (C1) (1993) 65e79. [13] E. Alami, M. Almgren, W. Brown, J. Francois, Aggregation of hydrophobically end-capped poly(ethylene oxide) in aqueous solutions. Fluorescence and light-scattering studies, Macromolecules 29 (1996) 2229e2243. [14] E. Michel, M. Filali, R. Aznar, G. Porte, J. Appell, Percolation in a model transient network: rheology and dynamic light scattering, Langmuir 16 (2000) 8702e8711. [15] P. Malo de Molina, M.-S. Appavou, M. Gradzielski, Oil-in-water microemulsion droplets of TDMAO/decane interconnected by the telechelic C18-EO150-C18: clustering and network formation, Soft Matter 10 (2014) 5072e5084. [16] C. Herfurth, P. Malo de Molina, C. Wieland, S. Rogers, M. Gradzielski, A. Laschewsky, One-step RAFT synthesis of well-defined amphiphilic star polymers and their self-assembly in aqueous solution, Polym. Chem. 3 (2012) 1606e1617. [17] P. Malo de Molina, C. Herfurth, A. Laschewsky, M. Gradzielski, Structure and dynamics of networks in mixtures of hydrophobically modified telechelic multiarm polymers and oil in water microemulsions, Langmuir 28 (2012) 15994e16006.
vost, C. Herfurth, M.-S. Appavou, [18] P. Malo de Molina, F. Ihlefeldt, S. Pre A. Laschewsky, M. Gradzielski, Phase behavior of nonionic microemulsions with multi-end-capped polymers and its relation to the mesoscopic structure, Langmuir 31 (2015) 5198e5209. [19] O.E. Philippova, D. Hourdet, R. Audebert, A.R. Khokhlov, pH-responsive gels of hydrophobically modified poly(acrylic acid), Macromolecules 30 (1997) 8278e8285.

205

[20] M. Vamvakaki, C.S. Patrickios, M. Gradzielski, P. Lindner, Amphiphilic networks based on cross-linked star polymers: a small-angle neutron scattering study, Langmuir 23 (2007) 10433e10437. € rster, V. Abetz, A.H.E. Müller, Polyelectrolyte block copolymer micelles, [21] S. Fo Adv. Polym. Sci. 166 (2004) 173e210. [22] S. Dai, P. Ravi, K.C. Tam, pH-Responsive polymers: synthesis, properties and applications, Soft Matter 4 (2008) 435e449. [23] C. Charbonneau, M.M. De Souza Lima, C. Chassenieux, O. Colombani, T. Nicolai, Structure of pH sensitive self-assembled amphiphilic di- and triblock copolyelectrolytes: micelles, aggregates and transient networks, Phys. Chem. Chem. Phys. 15 (2013) 3955e3964. [24] D.V. Pergushov, E.V. Remizova, M. Gradzielski, P. Lindner, J. Feldthusen, A.B. Zezin, A.H.E. Müller, V.A. Kabanov, Micelles of polyisobutylene-blockpoly(methacrylic acid) diblock copolymers and their water-soluble interpolyelectrolyte complexes formed with quaternized poly(4-vinyl pyridine), Polymer 45 (2004) 367e378. [25] M. Burkhardt, M. Ruppel, S. Tea, M. Drechsler, R. Schweins, D.V. Pergushov, M. Gradzielski, A.B. Zezin, A.H.E. Müller, Water-soluble interpolyelectrolyte complexes of polyisobutylene-block-poly(methacrylic acid) micelles: formation and properties, Langmuir 24 (2008) 1769e1777. [26] M. Uchman, K. Prochazka, M. Stepanek, G. Mountrichas, S. Pispas, M. Spirkova, A. Walther, pH-dependent self-assembly of polystyrene-blockpoly((sulfamatecarboxylate) isoprene) copolymer in aqueous media, Langmuir 24 (2008) 12017e12025. [27] J. Du, Y. Tang, A.L. Lewis, S.P. Armes, pH-sensitive vesicles based on a biocompatible zwitterionic diblock copolymer, J. Am. Chem. Soc. 127 (2005) 17982e17983. [28] Y. Bae, S. Fukushima, A. Harada, K. Kataoka, Design of environment-sensitive supramolecular assemblies for intracellular drug delivery: polymeric micelles that are responsive to intracellular pH change, Angew. Chem. Int. Ed. 42 (2003) 4640e4643. [29] A. Dashevsky, K. Kolter, R. Bodmeier, pH-independent release of a basic drug from pellets coated with the extended release polymer dispersion Kollicoat® SR 30 D and the enteric polymer dispersion Kollicoat®MAE 30 DP, Eur. J. Pharm. Biopharm. 58 (2004) 45e49. [30] O. Anthony, R. Zana, Fluorescence investigation of the binding of pyrene to hydrophobic microdomains in aqueous solutions of polysoaps, Macromolecules 27 (1994) 3885e3891. [31] F. Petit-Agnely, I. Iliopoulos, R. Zana, Hydrophobically modified sodium polyacrylates in aqueous solutions: association mechanism and characterization of the aggregates by fluorescence probing, Langmuir 16 (2000) 9921e9927. €nyi, Associating polyelectrolytes with [32] F. Petit, I. Iliopoulos, R. Audebert, S. Szo perfluoroalkyl side chains: aggregation in aqueous solution, association with surfactants, and comparison with hydrogenated analogues, Langmuir 13 (1997) 4229e4233. [33] A. Hashidzume, A. Kawaguchi, A. Tagawa, K. Hyoda, T. Sato, Synthesis and structural analysis of self-associating amphiphilic statistical copolymers in aqueous media, Macromolecules 39 (2006) 1135e1143. [34] T. Mori, A. Hashidzume, T. Sato, Micellization behavior of an amphiphilic statistical copolymer in water-methanol mixtures, Polym. J. 41 (2009) 189e194. [35] X. Wu, Y. Qiao, J. Wang, Associative properties of amphiphilic statistical polymers in aqueous media, J. Chem. Eng. Data 55 (2010) 919e924. [36] T. Kawata, A. Hashidzume, T. Sato, Micellar structure of amphiphilic statistical copolymers bearing dodecyl hydrophobes in aqueous media, Macromolecules 40 (2007) 1174e1180. [37] E.V. Korchagina, O.E. Philippova, Multichain aggregates in dilute solutions of associating polyelectrolyte keeping a constant size at the increase in the chain length of individual macromolecules, Biomacromolecules 11 (2010) 3457e3466. [38] L. Liu, X. Xu, S. Guo, W. Han, Synthesis and self-assembly of chitosan-based copolymer with a pair of hydrophobic/hydrophilic grafts of polycaprolactone and poly(ethylene glycol), Carbohydr. Polym. 75 (2009) 401e407. [39] A.P. Haehnel, M. Schneider-Baumann, K.U. Hiltebrandt, A.M. Misske, C. Barner-Kowollik, Global trends for kp? Expanding the frontier of ester side chain topography in acrylates and methacrylates, Macromolecules 46 (2013) 15e28. [40] M. Sedl
ak, The ionic strength dependence of the structure and dynamics of polyelectrolyte solutions as seen by light scattering: the slow mode dilemma, J. Chem. Phys. 105 (1996) 10123e10133. [41] B.D. Ermi, E.J. Amis, Domain structures in low ionic strength polyelectrolyte solutions, Macromolecules 31 (1998) 7378e7384. [42] A. Feoktystov, H. Frielinghaus, Z. Di, S. Jaksch, V. Pipich, M.-S. Appavou, E. Babcock, R. Hanslik, R. Engels, G. Kemmerling, H. Kleines, A. Ioffe, D. Richter, T. Bruckel, KWS-1 high resolution SANS instrument at JCNS: current state, J. Appl. Cryst. 48 (2015) 61e70. [43] U. Keiderling, The new 'BerSANS-PC' software for reduction and treatment of small angle neutron scattering data, Appl. Phys. A-Materials Sci. Process. 74 (2002) s1455es1457. [44] C.-L. Zhao, M.A. Winnik, G. Riess, M.D. Croucher, Fluorescence probe techniques used to study micelle formation in water-soluble block copolymers, Langmuir 6 (1990) 514e516. [45] W. Henni, M. Deyme, M. Stchakovsky, D. LeCerf, L. Picton, V. Rosilio, Aggregation of hydrophobically modified polysaccharides in solution and at the airewater interface, J. Coll. Interface Sci. 281 (2005) 316e324.

206

S. Riemer et al. / Polymer 70 (2015) 194e206

€ rster, M. Schmidt, M. Antonietti, Static and dynamic light scattering by [46] S. Fo aqueous polyelectrolyte solutions: effect of molecular weight, charge density and added salt, Polymer 31 (1990) 781e792.
k, E.J. Amis, Dynamics of moderately concentrated saltfree poly[47] M. Sedla electrolyte solutions: molecular weight dependence, J. Chem. Phys. 96 (1992) 817. [48] A. Pohlmeier, S. Haber-Pohlmeier, Ionization of short polymethacrylic acid: titration, DLS, and model calculations, J. Colloid Interface Sci. 273 (2004) 369e380. €m, B. Lindman, Dynamic and viscoelastic properties during the [49] B. Nystro thermal gelation process of a nonionic cellulose ether dissolved in water in the presence of ionic surfactants, Macromolecules 28 (1995) 967e974.
vost, R. Schweins, N.J. Wagner, M. Gradzielski, [50] I. Hoffmann, P. Heunemann, S. Pre Self-aggregation of mixtures of oppositely charged polyelectrolytes and

[51] [52]

[53]

[54]

surfactants studied by rheology, dynamic light scattering and small-angle neutron scattering, Langmuir 27 (2011) 4386e4396. S.S. Berr, Solvent isotope effects on alkyltrimethylammonium bromide micelles as a function of alkyl chain length, J. Phys. Chem. 91 (1987) 4760e4765. Y.A. Shashkina, Y.D. Zaroslov, V.A. Smirnov, O.E. Philippova, A.R. Khokhlov, T.A. Pryakhina, N.A. Churochkina, Hydrophobic aggregation in aqueous solutions of hydrophobically modified polyacrylamide in the vicinity of overlap concentration, Polymer 44 (2003) 2289e2293. D.G. Thomas, Transport characteristics of suspension: VIII. A note on the viscosity of Newtonian suspensions of uniform spherical particles, J. Colloid Sci. 20 (1965) 267e277. M.J. Rosen, Surfactants and Interfacial Phenomena, third ed., Wiley-Interscience, Hoboken, USA, 2004.