Colloid Polym Sci (2009) 287:1369–1376 DOI 10.1007/s00396-009-2103-3

ORIGINAL CONTRIBUTION

Do additives shift the LCST of poly (N-isopropylacrylamide) by solvent quality changes or by direct interactions? Christian Hofmann & Monika Schönhoff

Received: 2 July 2009 / Revised: 20 August 2009 / Accepted: 21 August 2009 / Published online: 17 September 2009 # Springer-Verlag 2009

Abstract The phase transition of thermoresponsive poly (N-isopropylacrylamide) is studied under the influence of additives considered as model substances for drugs. A series of aromatic compounds with similar structures, mainly benzaldehydes, is chosen. The lower critical solution temperature (LCST) is determined by differential scanning calorimetry and 1H-NMR. All additives cause a down shift of the LCST, which depends on additive molecular structure and concentration. Since the LCST shifts are not correlated to hydrophobicity or solubility of the additive, the detailed substitution pattern is discussed as the controlling factor. The question whether LCST shifts can be explained by either the additives affecting the solvent quality or by specific interactions of additives with the polymer is addressed by LCST determination in dependence on polymer concentration. Though both factors are relevant, specific additive-polymer interactions are shown to play a major role in controlling the LCST. Keywords Poly(N-isopropylacrylamide) . LCST . Thermoreversible polymer . Microgel . NMR . DSC . Additive . PNiPAm

Electronic supplementary material The online version of this article (doi:10.1007/s00396-009-2103-3) contains supplementary material, which is available to authorized users. C. Hofmann : M. Schönhoff (*) Westfälische Wilhelms-Universität, Institut für Physikalische Chemie, Corrensstrasse 28/30, 48149 Münster, Germany e-mail: [email protected]

Introduction Polymers like poly(N-isopropylacrylamide) (PNiPAm) [1] or poly(vinyl methyl ether) (PVME) that show a so-called lower critical solution temperature (LCST) behavior are of great interest. Such polymers are water soluble, but will be insoluble when a certain temperature, the LCST, is exceeded. In the case of dilute solutions, where single chains collapse, this transition is called the coil-to-globule transition. For solutions of pure PNiPAm homopolymer in H2O, this temperature is about 32°C, and it increases by about 2°C when heavy water is chosen as solvent. Thus, PNiPAm is a promising compound for the build-up of socalled smart materials, like thermoresponsive films [2–4], particles [5], capsules [3], microgels [6], and selfassembling micelles [7, 8]. In particular, for controlled release applications, microgels, capsules, or micelles of thermoreversible polymers are suitable materials as drug carriers. Here, the carrier particle is loaded with a drug below or around the transition temperature, i.e., when the gel or capsule is in the swollen state. Above the transition temperature, the release of the drug with time is studied by appropriate methods [8–10]. Since it is known that even small amounts of additional solutes can have a distinct effect on the transition temperature of PNiPAm chains in solution [11], as well as in gels [12], profound knowledge of this influence is important for the development of drug release systems of PNiPAm. It has been found that release kinetics clearly differ for different model drugs [13]. Coughlan et al. have investigated both the release kinetics of parabens from PNiPAm hydrogels and the effect of those compounds on the LCST of linear PNiPAm homopolymer [14]. From their results, they concluded that from studying the influence of an additive on the LCST of linear PNiPAm, predictions on

1370

the swelling and release behavior of hydrogels can be made. Anyhow, no parameter was identified by which the amount of LCST shift caused by an additive could be predicted. The influence of different cosolutes on the LCST has been subject to many studies in the past. When simple salts are added, the result is a decrease of the transition temperature, which depends linearly on the salt concentration [11]. An increase in the LCST is reported for some organic quaternary ammonium salts. In general, in conclusion from results for biological macromolecules, as well as PVME and poly(ethylene oxide), it is stated that the extent of decrease can be described by the series of Hofmeister [15]. Although PNiPAm is completely soluble in organic solvents like methanol, ethanol, or dimethylsulfoxide, adding these to an aqueous solution leads to a decrease of the LCST at low cosolvent content; the LCST is increasing only at high cosolvent fractions [16]. On the other hand, an increase of the transition temperature can be achieved by adding surfactants. Sufficiently high concentrations of sodium dodecylsulfate even completely prevent the polymer from precipitation [17]. For other surfactants, a similar behavior is reported [18]. In spite of the tremendous relevance of the LCST shift in the presence of drugs or other additives, systematic studies of the additive properties, which are controlling LCST shifts, are rare, and there are many different possible reasons for the effect of additives on the LCST in discussion [19]. Dhara et al. have studied the influence of different substituted benzoic acids and hydroxybenzenes on the LCST by turbidimetry. They observed a sharp transition and shifted LCSTs, but no systematic explanation for the differently pronounced influence of different additives [19]. However, they detected LCSTs in presence of additives that differed only by about 1 to 2 K from the transition temperature of gels loaded with the respective additive. Thus, conclusions from the LCST shifts of the homopolymer appear to be relevant to predict the transitions in delivery systems. 1 H-NMR and DSC, as methods to investigate LCST behavior, have the advantage over turbidimetry [18, 20, 21] that they give information about the width or distribution of the transition. The information obtained from either of the two methods is complementary since DSC detects the heat flow of the transition [18, 22], while in liquid state 1HNMR spectra, mobile polymer segments are selectively detected [23–28]. In addition, spin relaxation time measurements yield information about local segmental mobility [24, 29–31]. Here, we systematically study the influence of different substituted benzaldehydes (see Scheme 1) on the LCST of linear PNiPAM by 1H-NMR and DSC. We focus on the question which parameters govern the ability of an additive

Colloid Polym Sci (2009) 287:1369–1376

to influence the LCST. Hydrophobicity, solubility, or the detailed substitution pattern is analyzed as potential parameters controlling the phase transition temperature. Finally, we discuss the mechanism of the LCST shift, in particular, whether the influence of an additive on the LCST is promoted via a change of the solvent structure or via direct interactions of additive with the chain.

Materials and methods Materials Poly(N-isopropylacrylamide) with an average molecular weight of 76,700 g/mol and a polydispersity of 1.28 was purchased from Polymer Source, Montréal, Quebec, Canada. The polymer was dried at T=110°C for at least 2 hours. Salicylaldehyde (SA, 98%), m-hydroxybenzaldehyde (mHBA, 97%), ethylvanillin (EV, 99%), and 3,4-dimethox ybenzaldehyde (DMBA, 99%) were purchased from Aldrich. Benzoic acid (BA, 99.5%), methyl-p-hydroxybenzoate (MHB, 99 %), and p-hydroxybenzaldehyde (pHBA, 95 %) were purchased from Fluka (Scheme 1). The solutions were either prepared from ultrapure water (H2O, three stage purification system (Millipore); resistivity ≥ 18 MΩ cm) or

Scheme 1 Additives used in this study

Colloid Polym Sci (2009) 287:1369–1376

1371

from heavy water (D2O; isotope purity, 99.9 atom %, Aldrich).

1.0

Integral (a. u.)

0.8

DSC Differential scanning calorimetry was performed using a Setaram Micro DSC III heat conduction scanning microcalorimeter (Setaram, France). In all cases, the sample volume was 250µl. For all measurements, 250µl of water (H2O) were taken as reference. Both heating and cooling cycles were recorded over a temperature range from 5 to 50°C at a scanning rate of 0.5 K/min, since the thermograms were independent of the scan rate at scan rates below 0.6 K/min.

Results and discussion The polymer phase transition characterized by NMR and DSC For all samples studied by temperature-dependent onepulse NMR measurements, the phase transition can be clearly detected by the integrated signal intensity; an example which demonstrates the general behavior is shown in Fig. 1. For high and low temperatures, far apart from the LCST region, plateaus occur. At low temperatures, the largest intensity values are found, whereas, at high temperatures the signal vanishes completely. Between these plateaus, a continuous decrease occurs, which can be fitted using a phenomenological function [23] Imax
Imin
þ Imin I¼ 1 þ exp T
LCST k

ð1Þ

Here Imin, Imax, LCST, and k are fit parameters; T is the temperature. If the intensity is normalized, it is Imax =1; and if the transition of the polymer is complete, it is Imin =0, which is the case for all samples within a precision of a few percentage. LCST is the temperature, where the signal has

0.4 0.2 0.0 16

20

24

28

32

36

40

T [˚C]

Fig. 1 Integrated intensity of the signal of PNiPAm methyl protons at 0.9 ppm in 1 wt% PNiPAm solution with 20 mM mHBA in D2O. The solid line is a fit according to Eq. 1

decreased by 50%. As in most of the references [25, 32], this definition is used for the transition temperature. The parameter k is a measure for the width of the transition as the slope of the inflection point tangent is m¼

Imax
Imin 4k

ð2Þ

Therefore, a small value of k is indicating a narrow transition. For all additives investigated, the solutions show a qualitatively similar behavior of the polymer methyl proton intensities, which is well-described by the fit with Eq. (1): major differences are found in the values of the LCST and the width parameter k. Generally, all additives investigated cause a decrease of the LCST, which is mostly accompanied by an increase of the width parameter, k. Figure 2 relates these two parameters for the different additives and shows a clear correlation: additives that cause a large down shift of the LCST, also show a large increase of the transition width. 1.4

Width parameter k [˚C]

NMR Nuclear magnetic resonance (NMR) spectra were recorded on an Avance 400 spectrometer (Bruker) at constant field strength of 9.4 T using a liquid state probe head (Bruker Diff-30). Acquisition of the spectra was performed with a relaxation delay of 20 s and a single 90° pulse. The number of scans was at least 32. In all cases the methyl group signal of the polymer was evaluated by integrating it after baseline correction. The sample temperature was controlled with an accuracy of 0.15 K by a heated airflow. The variation of the temperature was always performed starting at low temperatures and increasing it stepwise. At each temperature, the sample was equilibrated for at least 12 min before the measurement was started.

0.6

1.2 1.0 0.8 0.6 0.4 24

26

28

30

32

34

T [˚C]

Fig. 2 Width parameter k in dependence of LCST for 20 mM additive in 1% PNiPAm in D2O (square, SA; circle, DMBA; diamond, EV; up triangle, mHBA; down triangle, pHBA; and star, no additive)

1372

Colloid Polym Sci (2009) 287:1369–1376

A possible explanation could be interactions between the additive and the polymer chain. If the additive is replacing hydration water bound to the chains, this will result in a more hydrophobic nature of the polymer chain. Since this effect would lead to a lower transition temperature, the LCST shift caused by the additive should be larger when the strength of the interaction—and thus, the fraction of water molecules replaced by additive—is larger. Simultaneously, a larger fraction of additive molecules attached to the chains might cause a larger heterogeneity along the chain concerning the local hydrophobicity. Then, the transition is spread over a larger range of local transition temperatures, which explains the increase of the width of the transition and its correlation with the LCST shift. Such a broadening of the LCST was also found for PNIPAm adsorbed to surfaces, and there, it was explained by heterogeneity of the loop length of the adsorbed chains [22]. In addition to NMR, the phase transition is characterized by DSC. Generally, repeated heating and cooling cycles yield reproducible peak shapes and temperatures; only in a few cases, the first cycle slightly deviates and was not considered further. A hysteresis between heating and cooling is observed, which is typical [33]. Here, only the heating cycles are considered. From these, the onset and peak transition temperatures are determined. In order to compare DSC and NMR transition temperatures, Fig. 3 shows the DSC and NMR curves of a typical sample in D2O. In spite of a different mechanism of detection, the peak observed in DSC has about the same width as the transition region in the NMR signal and is found in the same temperature range. This agreement is found for all additives, see the comparison of DSC and NMR results in Fig. 4. For most additives, the LCST derived from NMR (cf. Eq. 1) is between the onset and the peak temperature observed in DSC. In addition, there is a correlation between

the width determined as difference of onset and peak temperature in DSC (see Fig. 4) and the width parameter from NMR experiments, see Fig. 2. Both decrease with increasing transition temperature. That means both methods give a similar picture of the process of precipitation in spite of detecting different quantities. The loss of segmental mobility observed in the NMR experiments seems to be strongly correlated to the heat of transition observed in DSC. Both methods have in common that they monitor the transition of all segments as a superposition, and thus, they deliver quantitative information about the width of the transition. In contrast to this, in turbidimetry the signal is not quantitatively related to the fraction of polymer precipitated such that a transition width cannot be extracted. The latter fact could be an explanation why Dhara et al. stated a sharp transition for benzoic acids, although, they have a stronger decreasing effect on the LCST as the benzaldehydes for which quite broad transitions are found here (cf. Fig. 4) [19]. Influence of different additives on the phase transition The transition temperatures for all additives are investigated by DSC in dependence of the additive concentration, see Fig. 5. Since all the solutions represented there are prepared from light water, the values for 20 mM additive concentration differ slightly from those shown in Fig. 4. This is due to the fact that the LCST is increased by about 2 K when D2O is used as the solvent instead of 1H2O. For all additives employed here, an increase in concentration leads to a decrease of the LCST. The functional dependence of the LCST on additive concentration, however, seems to follow different laws and clear differences of the influence of different additives on the LCST are observed. It is now interesting to try to identify the parameters that control the amount of LCST shift for a given additive.

-4.4

0.4 0.2

-5.2

0.0

-5.4 24

28

32

4

30 28

3

26 2

24 22

∆Peak-Onset / K

-5.0

32

T [˚C]

-4.8

0.6

5

DSC (Heating); Heat flux (mW)

-4.6 0.8

1

H-NMR; Integral (a. u.)

1.0

1 SA

DMBA

EV

mHBA

pHBA

36

T [˚C]

Fig. 3 1H-NMR integral (circles; the solid line is a fit according to Eq. 1) and DSC thermogram for 1 wt% PNiPAm/20 mM DMBA in D2O

Fig. 4 Transition temperatures extracted from NMR data (squares) and DSC data (diamonds, onset temperatures; and triangles, peak temperatures) for samples with 20 mM additive and 1 wt% PNiPAm in D2O. Full circles indicate the width of the transition in DSC as characterized by the difference of peak and onset temperatures

Colloid Polym Sci (2009) 287:1369–1376

1373

a)

30

30

TOnset [°C]

TOnset [°C]

28 25

26 24

20 0

5

10

15 cAdditive / mM

20

22

25

1.0

Fig. 5 Onset temperatures measured by DSC in dependence on additive concentration at 1 wt% polymer concentration in H2O. Hexagons indicate MHB; pentagons indicate BA; down triangles indicate pHBA; up triangles indicate mHBA; diamonds indicate EV; squares indicate SA; and circles indicate DMBA. Lines are guides to the eye

1.5 log pO/W

2.0

b)

30

Since at the LCST, where the chain collapse takes place, hydrophobic interactions between chains play a major role, and since, in addition, it is known that hydrophobic comonomers decrease the LCST, one can speculate about the role of the hydrophobicity of the additives. The hydrophobicity of a compound is commonly described by log POW, where POW is the octanolwater partition coefficient, i.e., the ratio of concentrations of a compound in octanol and water in equilibrium. Figure 6a and b correlate the LCST shift to the hydrophobicity as described by log POW and to the additive solubility in water, respectively. In neither case, correlation can be identified in Fig. 6. Therefore, a property of the additive other than hydrophobicity or solubility must control the LCST shift. Another possible factor explaining the LCST shifts could be found in the chemical structure, i.e., the substitution pattern of each additive, which can be assumed to control specific polymer-additive interactions. Dhara et al. have investigated the effect of differently substituted benzoic acids on the LCST of PNiPAm [19]. The benzoic acids studied there show a much more pronounced influence on the LCST as compared to the corresponding benzaldehydes (SA, mHBA, and pHBA) investigated here. Consistently, benzoic acid in our study (see pentagons in Fig. 5) has a stronger influence on the LCST as any of the benzaldehydes. Within the series of benzaldehydes, as well as within the series of benzoic acids, however similar dependencies on the substitution pattern occur, which suggest a relevance of the position of single substituent concerning the acids, salicylic acid has the least influence, while p-hydroxybenzoic acid and m-hydroxybenzoic acid show about the same influence, which is clearly larger than that of salicylic acid [19]. Similarly, for the benzaldehydes, salicylaldehyde shows the

TOnset [°C]

28 26 24 22 0

5

10 15 20 Solubility . g/L

25

30

Fig. 6 LCST values derived from DCS onset temperatures in solutions with 15 mM additive and 1 wt% PNiPAm in dependence on (a) log POW value and (b) solubility of the additive. Symbols as described in the caption of Fig. 5

least LCST shift, whereas, that of pHBA is similar to that of mHBA, see Fig. 5. Another evidence for a possible relevance of the 1,3- and the 1,4-substitution pattern is the LCST with added EV, which is quite similar to that with pHBA or mHBA. Ethylvanillin has a hydroxyl group in para position, which seems to be more relevant than the additional ether group. The DBMA having two substituents in para and in meta position might be deviating from this general trend, since these substituents are not hydroxy, but methoxy groups. They might have a reduced influence on the LCST due to their lower dipole moments as compared to the hydroxy groups. In summary, the number and steric accessibility of the OH group or the aldehyde group may be an important factor: in benzoic acids, these groups are well accessible and can interact. Considering the aldehyde series from pHBA to mHBA and finally to SA, the position of the OH group on the ring is getting closer to the aldehyde substituent, which might prevent interactions. Altogether,

1374

Influence of the polymer concentration: is the LCST shift controlled by direct additive-polymer interaction or by solvent quality? The decrease of the LCST with increasing additive concentration might indeed be explained by an interaction of additives with the polymer chains: If there is a fraction of the additive favorably interacting with the chain, and this interacting fraction is in equilibrium with freely dissolved additive, then increasing the total additive concentration will lead to a larger number of additive molecules replacing water along the polymer chain. Since the additives are rather hydrophobic, the result is a reduced solubility of the polymer-additive complex, and thus, a reduced LCST in presence of the additive. However, there is an alternative explanation for the concentration dependence of the LCST: The solvent quality will change when additive is added to water. Since the additives studied here are rather hydrophobic, the solvent becomes somewhat more hydrophobic and has a reduced ability to form hydrogen bonds. Thus, the solvent quality is reduced, and polymer precipitation occurs already at lower temperatures. Though in this case LCST shifts would be expected to scale with additive hydrophobicity, a relevance of the detailed substitution pattern might arise from detailed changes of the water hydrogen bond network being disturbed by the additive. Due to additive being contained in the water, the solvent quality is changed. It would be interesting to distinguish between these two mechanisms, which are possibly controlling the LCST shift occurring upon additive addition. Both mechanisms should scale with the additive concentration, thus, no conclusion can be drawn from the results of Fig. 5. However, only the mechanism of direct interactions will additionally scale with the polymer concentration, since in this case it is the ratio of polymer segments to additive that should determine the LCST. Therefore, in order to clarify whether the decrease in LCST is caused by solvent quality effects or by direct interactions between additive and PNiPAm, samples containing different amounts of PNiPAm were investigated. The results for three different series with different additive content are shown in Fig. 7. In the case of additive-free PNiPAm solution, no concentration dependence is found. This is consistent with the results reported in literature for polymer concentrations up to 10% [34]. For an additive content of 15 mM and above 2% polymer concentration, no concentration dependence of

32

T [˚C]

the ability of the substituents to interact with the polymer— in dependence on their relative position on the ring— controls the amount of shift of the LCST rather than macroscopic thermodynamic properties such as solubility or hydrophobicity.

Colloid Polym Sci (2009) 287:1369–1376

28

24

20 1

2

3

4

5

cPolymer [wt.%]

Fig. 7 Dependence of the DSC onset temperatures (open symbols) and peak temperatures (filled symbols) on the polymer content at different pHBA concentrations (circles, 0 mM; diamonds, 15 mM; and squares, 25 mM). Lines are guides to the eye

the peak temperatures is observed either, but the onset temperature does decrease when the polymer concentration is increased. At 5 wt% the onset temperature is about 4°C lower than at 2 wt%. Since the peak temperature does not change significantly in this region, this implies a broader transition for high polymer concentration. The comparison of NMR and DSC data (Fig. 3) showed that the peak temperature is always close to the LCST derived by NMR, and the latter was defined as the center of the transition. Thus, here, it is mainly the width of the transition that is affected by the polymer concentration, not the transition temperature. A different dependence is observed in the case of an additive concentration of 25 mM, see the squares in Fig. 7. There, an increase of both the onset and the peak temperature is detected with the polymer concentration increasing from 1% to 5%. An increase of the transition width is present as well, but it is less pronounced. Thus, in general, the transition width increases with increasing additive content and with increasing polymer concentration. This means that also in DSC, a broader transition is, generally, detected for lower LCSTs, consistent with the findings from NMR, see Fig. 3. The most interesting finding is certainly the pronounced increase of the peak temperature with polymer concentration for samples with 25 mM pHBA. This cannot be explained by solvent quality but only by specific interactions. When the additive content is kept constant and the polymer concentration is increased, the number of additive molecules per polymer segment is decreased. This number of additives per segment is, apparently, controlling the hydrophobicity of the chains and therefore, the decrease of the transition temperature. However, no evidence of such an interaction mechanism is detected for the lower additive concentration of 15 mM, at least not for the polymer concentrations above 2%. In this context, it is interesting to

Colloid Polym Sci (2009) 287:1369–1376

compare the ratio of additive to polymer segments. At 15 mM, the number of additives per monomer ranges from 1.7 10−3 down to 0.3 10−3 for the polymer concentrations between 1% and 5%, while at 25 mM additive concentration, the ratios vary from 2.8 10−3 to 0.56 10−3. Assuming a low fraction of interacting additive, the actual ratio of interacting additive per monomer is much lower. This implies that only few monomers can interact with the additive, which might lead to the situation that this interaction becomes relevant only (a) above a certain additive concentration, thus, it is detectable at 25 mM; or (b) at very low polymer concentration, see the initial increase of peak and onset temperature for 15 mM additive.

Conclusions Both NMR and DSC measurements of the LCST of PNiPAm in the presence of small aromatic compounds show similar results. A clear decreasing effect on the LCST is found for all molecules studied. The actual strength of this effect is varying for the different molecules, however, no correlation to either solubility or hydrophobicity of the additive is found. Instead, the chemical structure, i.e., the detailed substitution pattern, plays a major role. Adding benzaldehydes to PNiPAm solutions not only leads to transition temperatures shifted towards lower values but even results in a broader transition. The mechanism by which the additive shifts the LCST cannot be explained by only regarding the solvent quality. Direct interactions of PNiPAm and the additive molecules play an important role controlling the LCST. This is consistent with the finding that instead of the solubility or hydrophobicity, the detailed substitution pattern controls the LCST shift.

References 1. Schild HG (1992) Poly (N-Isopropylacrylamide)—experiment, theory, and application. Prog Polym Sci 17:163–249 2. Kuckling D (2009) Responsive hydrogel layers-from synthesis to applications. Colloid Polym Sci 287:881–891 3. Sukhishvili SA (2005) Responsive polymer films and capsules via layer-by-layer assembly. Curr Opin Colloid Interface Sci 10:37–44 4. Glinel K, Dejugnat C, Prevot M, Schöler B, Schönhoff M, Klitzing RV (2007) Responsive polyelectrolyte multilayers. Colloids Surf, A Physicochem Eng Asp 303:3–13 5. Rusu M, Kuckling D, Möhwald H, Schönhoff M (2006) Adsorption of novel thermosensitive graftcopolymers: Core-shell particles prepared by polyelectrolyte multilayer self-assembly. J Colloid Interf Sci 298:124–131 6. Ballauff M, Lu Y (2007) "Smart" nanoparticles: Preparation, characterization, and application. Polym Bull 48:1815–1823 7. Liu RX, Fraylich M, Saunders BR (2009) Thermoresponsive copolymers: from fundamental studies to applications. Colloid Polym Sci 287:627–643

1375 8. Zhang JX, Qiu LY, Jin Y, Zhu KJ (2006) Thermally responsive polymeric micelles self-assembled by amphiphilic polyphosphazene with poly(N-isopropylacrylamide) and ethyl glycinate as side groups: Poylmer synthesis, characterization and in vitro drug release study. J Biomed Mater Res Part A 76:773–780 9. Wang Q, Zhao Y, Yang Y, Xu H, Yang X (2007) Thermosensitive phase behavior and drug release of in situ gelable poly (N-isopropylacrylamide-co-acrylamide) microgels. Colloid Polym Sci 285:515–521 10. Lynch I, Dawson KA (2004) Release of model compounds from "Plum-Pudding"-type gels composed of microgel particles randomly dispersed in a gel matrix. J Phys Chem B 108:10893– 10898 11. Dhara D, Chatterji PR (2000) Phase transition in linear and crosslinked poly(N-isopropylacrylamide) in water: effect of various types of additives. J M S Rev Macromol Chem Phys 40:51–68 12. Kosik K, Wilk E, Geissler E, László K (2007) Distribution of phenols in thermoresponsive hydrogels. Macromolecules 40:2141–2147 13. Coughlan DC, Quilty FP, Corrigan OI (2004) Effect on drug physicochemical properties on swelling/deswelling kinetics and pulsatile drug release from thermoresponsive poly(Nisopropylacrylamide) hydrogels. J Control Release 98:97–114 14. Coughlan DC, Corrigan OI (2006) Drug-polymer interactions and their effect on thermoresponsive poly(N-isopropylacrylamide) drug delivery systems. Int J Pharm 313:163–174 15. Schild HG, Tirrell DA (1990) Microcalorimetric detection of lower critical solution temperatures in aqueous polymer solutions. J Phys Chem 94:4352–4356 16. Winnik FM, Ringsdorf H, Venzmer J (1990) Methanol- water as a co-nonsolvent system for poly(Nisopropylacrylamide). Macromolecules 23:2415–2416 17. Eliassaf J (1978) Aqueous solutions of poly(N-isopropylacrylamide). J Appl Polym Sci 22:973–874 18. Schild HG, Tirrell DA (1991) Interaction of poly(N-isopropylacrylymide) with sodium n-alkyl sulfates in aqueous solution. Langmuir 7:665–671 19. Dhara D, Chatterji PR (1999) Effect of hydrotropes on the volume phase transition in poly(Nisopropylacrylamide) hydrogel. Langmuir 15:930–935 20. Liu HY, Zhu XX (1999) Lower critical solution temperatures of N-substituted acrylamide copolymers in aqueous solutions. Polymer 40:6985–6990 21. Durme KV, Rahier H, Mele BV (2005) Influence of additives on the thermoresponsive behavior of polymers in aqueous solution. Macromolecules 38:10155–10163 22. Schönhoff M, Larsson A, Welzel PB, Kuckling D (2002) Thermoreversible polymers adsorbed to colloidal silica: a H-1 NMR and DSC study of the phase transition in confined geometry. J Phys Chem B 106:7800–7808 23. Larsson A, Kuckling D, Schönhoff M (2001) 1H NMR of thermoreversible polymers in solution and at interfaces: the influence of charged groups on the phase transition. Coll Surf A 190:185–192 24. Hanyková L, Labuta J, Spěvaček J (2006) NMR study of temperature-induced phase separation and polymer–solvent interactions in poly(vinyl methyl ether)/D2O/ethanol solutions. Polymer 47:6107–6116 25. Andersson M, Maunu SL (2006) Volume phase transition and structure of poly(N-isopropylacrylamide) microgels studied with 1H-NMR spectroscopy in D2O. Colloid Polym Sci 285:293– 303 26. Spěváček J (2005) Phase separation in aqueous polymer solutions as studied by NMR methods. Macromol Symp 222:1–13 27. Rusu M, Wohlrab S, Kuckling D, Möhwald H, Schönhoff M (2006) Coil-to-globule transition of PNIPAM graft copolymers

1376 with charged side chains: a H-1 and H-2 NMR and spin relaxation study. Macromolecules 39:7358–7363 28. Spevacek J (2009) NMR investigations of phase transition in aqueous polymer solutions and gels. Curr Opin Colloid Interface Sci 14:184–191 29. Yushmanov PV, Furó I, Iliopoulos I (2006) Kinetics of Demixing and Remixing Transitions in Aqueous Solutions of Poly(Nisopropylacrylamide): A Temperature-Jump 1H NMR Study. Macromolecular Chem Phys 2006:1972–1979 30. Zeng F, Tong Z, Feng HQ (1997) Nmr investigation of phase separation in poly(N-isopropyl acrylamide)/water solutions. Polymer 38:5539–5544

Colloid Polym Sci (2009) 287:1369–1376 31. Tokuhiro T, Amiya T, Mamada A, Tanaka T (1991) NMR study of poly(N-isopropylacrylymide) gels near phase transition. Macromolecules 24:2936–2943 32. Otake K, Inomata H, Konno M, Saito S (1990) Thermal analysis of the volume phase transition with nisopropylacrylamide gels. Macromolecules 23:283–289 33. Wang X, Qiu X, Wu C (1998) Comparison of the coil-to-globule and the globule-to-coil transitions of a single poly(N-isopropylacrylamide) homopolymer chain in water. Macromolecules 31:2972–2976 34. Winnik FM (1990) Fluorescence studies of aqueous solutions of poly(N-isopropylacrylamide) below and above their LCST. Macromolecules 23:233–242