Influence of TMAO as co‐solvent on the gelation of silica‐PNIPAm core‐shell nanogels at intermediate volume fractions

Abstract We study the structure and dynamics of poly(N‐isopropylacrylamide) (PNIPAm) core‐shell nanogels dispersed in aqueous trimethylamine N‐oxide (TMAO) solutions by means of small‐angle X‐ray scattering and X‐ray photon correlation spectroscopy (XPCS). Upon increasing the temperature above the lower critical solution temperature of PNIPAm at 33 °C, a colloidal gel is formed as identified by an increase of I(q) at small q as well as a slowing down of sample dynamics by various orders of magnitude. With increasing TMAO concentration the gelation transition shifts linearly to lower temperatures. Above a TMAO concentration of approximately 0.40 mol/L corresponding to a 1 : 1 ratio of TMAO and NIPAm groups, collapsed PNIPAm states are found for all temperatures without any gelation transition. This suggests that reduction of PNIPAm‐water hydrogen bonds due to the presence of TMAO results in a stabilisation of the collapsed PNIPAm state and suppresses gelation of the nanogel.


Introduction
Stimuli-responsive polymer micro-and nanogels are crosslinked polymeric particles that are swollen with solvent molecules. Upon ap articular stimulus, e. g., pressure or temperature changes, the particles undergo av olume phase transition changing their configuration by releasing solvent molecules. [1] Within this particle class, microgels consisting of poly(Nisopropylacrylamide) (PNIPAm) are widely studied, as seen in a variety of recent reviews. [2][3][4][5] PNIPAm particles dispersed in water show ar eversible volume phase transition at al ower critical solution temperature (LCST) of 33°C. Below this LCST, the particles are swollen with water. They become hydrophobic above this temperature, releasing the water and forming a collapsed state. This process has been widely studied, predominately by light scattering determining the particle deswelling as a function of temperature. [6][7][8] Although the majority of these studies focuses on single particle properties, there has been a growing interest on higher particle packings. In particular, the use of PNIPAm micro-and nanogel particles for colloidal studies where the volume fraction is varied in-situ by changing the temperature and thus the particle size, has been questioned because the interaction potential changes as well. [9] Furthermore, interpenetration of PNIPAm particles may become possible which allows reaching very high volume fractions, even above 100 vol.%. [10][11][12] While below the LCST PNIPAm micro-and nanogels behave like as oft colloidal fluid, the increase of attractive interactions in the collapsed state above the LCST results in the formation of ac olloidal gel. [13][14][15][16][17][18] Recently, we studied the structure and dynamics of denselypacked core-shell nanoparticles consisting of as ilica core and a PNIPAm shell by means of coherent X-ray scattering. [19,20] At low volume fractions no gelation was observed, while gelation shows at emperature dependence at intermediate concentrations. Nevertheless, the gelation temperature was found to be around T gel � 37°Cf or high volume fraction, well above the LCST for these states. [20] The proximity of the LCST to biological-relevant temperatures makes PNIPAm ap romising material for applications in medicine and technology. [1,2,5] Therefore, the impact of cosolvents on PNIPAm properties, e. g., tuning the LCST becomes important. As PNIPAm can be dissolved as well in many alcohols, many experimental and theoretical investigations focus on water-methanol mixtures. They report ac ononsolvency, [21][22][23][24][25][26] i. e., ar eduction of the LCST below values found for the pure solvents. Ar eduction of the LCST has been reported for many other co-solvents, such as urea, [27,28] acetone, [29] and ethanol. [30] The reason for the reductions is unknown. For alcohol, al ocalization of alcohol molecules close to the polymer interface has been reported. Further, the reduction of the LCST is suggested to be ac onsequence of the release of water molecules from the PNIPAm hydrophobic moiety favoring the aggregation of ethanol molecules in the vicinity of the polymer. [30] Another co-solvent that was found to have an impact on the swelling of PNIPAm is the osmolyte trimethylamine N-oxide (TMAO). TMAO is known to stabilize proteins, [31,32] e. g., against high pressure. [33,34] Despite many studies in the last decades, the exact stabilization mechanism is still unclear, potentially TMAO stabilizes water hydrogen bonds [35,36] and hence the protein indirectly. Likewise, the addition of TMAO as co-solvent to aqueous PNIPAm micro-and nanogels leads to astabilisation of globular PNIPAm states, and consequently to an effective reduction of the LCST in the presence of TMAO. [37][38][39][40] These observations have been rationalized by af ormation of hydrogen bonds between TMAO and water molecules that are bound to PNIPAm. [37] Other studies concluded an increase in the magnitude of solvent-excluded volume effects because PNIPAm interacts preferentially with water. [40] Furthermore, studies on polystyrene reported a increased binding affinity of TMAO with collapsed conformation of polystyrene compared to the extended one, [41] which has been found as well for PNIPAm. [38] However, these investigations focused so far on low PNIPAm concentrations, i. e., the impact of co-solvents on the volume phase transition of PNIPAm micro-and nanogels in the single-particle limit. It is unclear how TMAO affects high volume fractions, in particular where gelation has been found.
Here, we report results on the structure and dynamics of silica-PNIPAm core-shell nanogels at av olume fraction of � eff;20 ¼ 0:14 at 20°Cd ispersed in aqueous TMAO solutions. We employ coherent X-ray scattering in small-angle scattering geometry to cover the relevant length scales of the colloidal system. The dynamics are extracted by means of X-ray Photon Correlation Spectroscopy (XPCS). Upon increasing the temperature above the LCST, we find the gelation transition which has been reported recently for this system. [19,20] With increasing TMAO concentration the gelation transition shifts to lower temperatures. For the highest TMAO concentration studied in this work of 0.52 mol/L the system does not show any indication of gelation and rather behaves diffusive over the studied temperature range. This suggests that a1 :1 ratio of TMAO and NIPAm groups reduces the number of PNIPAm-water hydrogen bonds and results in as tabilisation of the collapsed PNIPAm state and suppression of gelation at intermediate volume fractions.

Results and Discussion
Illuminating ad isordered soft-matter sample with a( partially) coherent X-ray beam generates ag rainy diffraction pattern, the so-called speckle pattern. Analogous to dynamic light scattering (DLS) the dynamics of as ample can be studied by tracking the time evolution of the speckles via the intensity-intensity correlation function [42][43][44][45] probed at the modulus of the wave vector transfer ðÞ with wave length λ and scattering angle θ.
The average in Eq. 1i sp erformed over the experimental time t and detector pixels with the same q. This correlation function is connected to the intermediate scattering function that includes all information of the time evolution of the sample via The speckle contrast β is defined by experimental parameters, such as the degree of coherence, the beam size, and the detector pixel size. For soft matter systems as studied here, the correlation function can be described by aKohlrausch-Williams-Watts (KWW) function [42][43][44][45] Here, the shape of the g 2 -function is defined by the KWW exponent γ. The relaxation rate G q ðÞrelates to the relaxation time τ c via G q ðÞ¼1 = t c q ðÞand typically follows ap ower law as function of q as G q ðÞ/q p .T he type of dynamics can be qualified by both exponents, e. g., diffusion is characterized by 6phR ,B oltzmann's constant k B ,t emperature T,s olvent viscosity η and particle radius R.
In order to investigate the influence of TMAO on the temperature-dependence of the silica-PNIPAm nanogel, we compare the average sample structure obtained by SAXS with the dynamics revealed simultaneously by XPCS.

SAXS results
The SAXS intensity I(q)i ss hown in Figure 1f or four TMAO concentrations and temperatures between 20°Ca nd 45°C. The 2D speckle patterns were averaged for each state before obtaining I(q)v ia azimuthal integration. For 0mol/L TMAO, the data follow our previous results. [19,20] At low temperature the I(q) resembles the form factor of spherical core-shell particles, i. e., the intensity drops with increasing q and shows oscillations. With increasing temperature, the slope at low q changes resulting in as teeper increase for q ! 0. This behaviour is typically connected to the formation of large-scale structure, such as agglomerates, within the sample and was rationalized by ac hange from repulsive to attractive interaction in PNIPAm systems above the LCST. [13,19,20,46,47] For TMAO concentrations up to c = 0.26 mol/L similar results are obtained. However, the upturn of I(q)f or q ! 0s ets in at lower temperatures and appears slightly stronger at 45°C. In contrast, for c = 0.52 mol/L TMAO the I(q)d on ot change significantly with T. Compared to the lower concentrations, the I(q)d rop at low q is furthermore steeper at low temperatures. This indicates ad ifferent structure at this concentration that seems to be independent of temperature.
To gain more details on structural changes as af unction of TMAO concentration and temperature, we calculated effective structure factors S(q)b yn ormalization of I(q)b yt he form factor P(q)which was measured from diluted colloidal dispersions. The structure factors are shown in Figure 2f or 20°Ca nd 45°C.  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57 These temperatures represent both the swollen and collapsed state of the PNIPAm shell. At 20°C ( Figure 2a), structure factors with am aximum at q max � 0.048 nm À1 can be found for c � 0:26 mol/L. As they vary only slightly with c,w ec onclude that TMAO has only very weak impact on the structure of the dispersion. For c = 0.52 mol/L S(q)shows only weak modulations around 1. This suggests that the structure of the dispersion resembles basically the structure of the form factor sample with Sq ðÞ�1w hich is representative for dilute colloidal dispersions. Furthermore, the slight upturn for q ! 0nm À 1 and the weak modulations may suggest appearance of as mall number of agglomerates due to attractive polymer-polymer interactions. [7,47,48] At elevated temperatures where I(q)s hows the upturn at low q,w ef ound adistinct c-dependence of S(q) (Figure 2b). For q ! 0t he S(q)i ndeed grows, in particular for the low TMAO concentrations. The peak position shifts to q max � 0:077 nm À1 which indicates ar eduction of the average next-neighbour distance as well as the formation of agglomerates as mentioned above. Furthermore, the S(q)s hows different amplitudes for q < 0:15 nm À1 ,i .e., it decreases with increasing c. The structure factor at c ¼ 0:52 mol/L shows am ore complex shape with a peak at q max � 0:1nm À 1 and modulations for q ! 0. Compared to the lower TMAO concentrations this suggests ac loser packing with different types of agglomerates, however, the effects are too weak to draw afinal conclusion.
Altogether the evolution of S(q)with T indicates atransition from repulsive interactions -r epresented by the S(q)f or c � 0:26 mol/L in Figure 2a -t owards attractive interactions at high temperatures. This is in agreement with our recent studies on pure silica-PNIPAm dispersions. [19,20] In contrast, at high TMAO concentrations the interaction potential seems to be less affected by the temperature change, assuming an unchanged sample structure. Nevertheless, we would like to note that a more detailed analysis of S(q)w ill need an extended q-range towards lower q,e .g., using ultra-small-angle X-ray scattering geometries [49,50] to capture the formation and type of large-scale clusters in the sample.

XPCS results
The dynamics of the core-shell particles have been studied by means of XPCS. Correlations from series of speckle patterns have been analyzed following Eq. 1. Due to the short exposure and thus the limited count rate in the speckle patterns, the analysis was limited to typically q < 0:046 nm À1 .T he resulting g 2 -functions are shown in Figure 3f or different TMAO concentrations. The g 2 functions of the pure silica-PNIPAm sample suggest first aspeeding-up upon heating up to T = 40°C. This is connected to the reduction of solvent viscosity η at higher temperatures and consequently ar eduction of the relaxation time τ c . Upon further heating to T gel = 42°C, the sample slows down and becomes almost static over the studied time range at 45°C. In general, the correlation functions are modeled by a single, stretched exponential decay following Eq. 3. At T gel = 42°C, two relaxation processes are used to model the data [51] and to capture the appearance of the slow relaxation process. This behaviour was reported recently for different pure silica-PNIPAm systems and has been related to the formation of a colloidal gel upon heating. [19,20] Most interestingly, the gelation transition at T gel takes place well above the LCST -i nt his case at approximately 10°Ch igher temperatures. This matches the phase behaviour found for similar volume fractions. [20] For low TMAO concentrations (c � 0:26 M), the g 2 -functions show asimilar behaviour. However, the slow-down of dynamics appears to happen at lower gelation temperatures T gel compared to the pure sample, i. e., at approximately 38.5°Cf or 0.11 mol/L TMAO and 34°Cf or 0.26 mol/L TMAO. In contrast, the sample at c ¼ 0:52 mol/L TMAO only shows the speed-up with increasing temperature without indication of gelation. This is in line with the observation of the I(q), where as well no structural transition was found.
In the next step we take ac loser look on the relaxation times and the type of dynamics. In Figure 4a the relaxation times extracted from the KWW fits are shown for q ¼ 0:023 nm À1 for all studied TMAO concentrations. Here, results of the intermediate concentration of c ¼ 0:16 mol/L are added as well. For T � 32 � Ca nd c � 0:26 mol/L, the relaxation times match each other well. Increasing the temperature further, the relaxation times increase by several orders of magnitude. Note that the statistical accuracy is limited for t c � 5s due to the limited exposure time on the samples. Nevertheless, few values could be modelled from longer XPCS series measured at some temperatures. The onset of slow-down at T gel is clarified by the dashed lines. In contrast, τ c for c ¼ 0:52mol/L TMAO shows ac ontinuous decrease with T. This follows in general the expectations for diffusion of nanoparticles in al iquid. Therefore, we calculated the diffusive relaxation time for particles of R ¼ 75 nm, i. e., the collapsed state, shown as solid line in Figure 4a. The viscosity of water-TMAO as solvent was extrapolated from literature values [52] and was corrected for the volume fraction of hard spheres. [53] We relate the difference between τ c for the sample at 0.52 mol/L TMAO and the calculation for diffusion of af actor of approximately 1.3 deviating from the hard sphere case to an effective slow-down of dynamics because of non-zero particle-particle interaction. This suggests that the sample is in acollapsed state for all temperatures. Furthermore, this is in line with the S(q) data at low temperatures (see Figure 2a), where indications were found for few agglomerates due to polymer-polymer interactions typically reported for collapsed PNIPAm states. [13,46,47] The type of dynamics is further studied by the exponents p describing the q-dependency of τ c and γ which is am easure of the shape of the g 2 -function. Both exponents are given in Figure 4b for T = 20°C. Due to the limited statistical accuracy above T gel ,t he exponents cannot be extracted. For all concentrations we found p ¼ 2a nd g � 0:8, which is close to the expectations for diffusion of p ¼ 2and g ¼ 1. The slightly lower values of γ suggest sub-diffusive or heterogeneous dynamics [54][55][56] and have been reported before for PNIPAm systems. [19] The shift of T gel observed in Figure 4a is highlighted in Figure 5( top). The error bar for T gel was set to DT = 1°Cw hich equals to the temperature steps used. Remarkably, we found a linear relation between T gel and c with arate of dT gel dc ¼ À30:8 � CL/ mol. Compared to T gel ,t he reduction of the LCST with TMAO concentration is less pronounced (dashed line in Figure 5(top)). Here, alinear model based on data for TMAO concentrations up to 1mol/L reported in literature was used, [37,39] leading to ar ate of À7:0 � CL/mol. Both lines intercept at c LG ¼ 0:36 � 0:07 ðÞ mol/ Lw hich may suggest that no gelation takes place above this  value. For c > c LG the volume phase transition as prerequisite for gelation will not have taken place. In fact, our measurement at c ¼ 0:52 mol/L does not show any indication for gelation.
These observation are now discussed in the framework of the TMAO-PNIPAm interaction. Therefore, we calculated the molar ratio r TP of TMAO molecules and NIPAm groups of the PNIPAm shells. The results are shown in Figure 5( bottom). We found r TP ¼ 1, i. e., equal amount of TMAO molecules and NIPAm groups, for c TP ¼ 0:40 � 0:05 ðÞ mol/L. Interestingly, this coincides with c LG ,f or larger TMAO concentrations PNIPAm is collapsed and consequently no gelation can be observed, as observed for c ¼ 0:52 mol/L. We interprete this observation by the TMAO-PNIPAm interaction. TMAO was reported to reduce the number of hydrogen bonds between PNIPAm and water [37] and in addition preferentially bind to the collapsed PNIPAm state. [38,41] Our results suggest that a1:1 molar ratio of TMAO and NIPAm groups fosters the collapse of PNIPAm and supresses gelation at intermediate volume fractions.

Conclusions
In summary, we measured coherent SAXS patterns of silica-PNIPAm core-shell nanogels with the presence of TMAO as cosolvent. At low TMAO concentrations, I(q)s hows an upturn for q ! 0a tt emperatures above the LCST. This suggests a transition from repulsive to attractive interactions as reported previously as af ingerprint for gel formation in this system. [19] The sample at the highest TMAO concentration studied (0.52 mol/L) shows the I(q)ofadilute colloidal fluid with weak S (q)c ontribution only. This is supported by the XPCS results, where the dynamics of this sample at 0.52 mol/L are diffusive over the whole temperature range. The combination of SAXS and XPCS results suggests, that the PNIPAm shell is collapsed for all temperatures, in particular enforcing av olume phase transition by the presence of TMAO. In contrast, the relaxation times obtained for c � 0:26 mol/L increase by several orders of magnitude at certain temperatures above the LCST. This is in agreement with the I(q)and our previous studies on pure silica-PNIPAm core-shell systems where this upturn was ac onsequence of ag elation transition. Here, the gelation temperature T gel depends on the TMAO concentration c. Above ap articular TMAO concentration, the particles stay in the collapsed state and do not form agel state.
These observations can be rationalized as follows: First, the gelation of the nanogel takes place at approximately 42°C which is in agreement with our previous observation. [20] Second, the gelation temperature is found to be reduced linearly with increasing TMAO concentration up to 0.26 mol/L, much stronger than the decrease of the LCST. Third, above at hreshold TMAO concentration that corresponds to a1:1 ratio of TMAO and NIPAm groups, we found collapsed PNIPAm states for all temperatures and no gelation transition. This indicates that one TMAO molecule per NIPAm unit is necessary to reduce the number of hydrogen bonds between PNIPAm and water below ac ritical threshold that leads to (1) as tabilisation of the collapsed state which consequently implies (2) the suppression of gelation.
Our results validate recent work on the role of TMAO on the phase behavior of PNIPAm nanogels [38,41] and indicate the proposed stabilisation of the collapsed state due to preferred interaction between water and TMAO. Nevertheless, amolecular level understanding of the interplay between TMAO-NIPAm interactions, in particular for different nanoparticle volume fractions, is still lacking which may motivate further theory and simulations studies on dense nanogel states.

Experimental Section Samples
We studied colloidal core-shell nanoparticles consisting of silica spheres coated with PNIPAm. The PNIPAm shell had been crosslinked using 4wt.% methylene bisacrylamide. Details of the synthesis are given in Refs. [20,57]. The stock solution with approximately 16.5 wt.% silica-PNIPAm particles was diluted 1:1involume with TMAO solutions, resulting in TMAO concentrations c in the solvent of 0.11 mol/L, 0.16 mol/L, 0.26 mol/L, 0.52 mol/L. The silica cores had ar adius of 50 nm and ad ispersity of 9%.T he size of the PNIPAm shell was measured by DLS (sample SP 1 in [20]) and found to have at hickness of 50 nm at 20°Ca nd 25 nm in the collapsed state at 45°C. This corresponds to an effective volume fraction of F eff;20 ¼ 0:141 in the swollen state at 20°Ca nd F eff;45 ¼ 0:059 at 45°C. For calculation of structure factors, particle form factors were measured from ad ilute dispersion F FF � 0:01 ðÞ .B efore the experi-