Introduction

In a series of experimental studies recently made of aqueous poly(N-isopropylacrylamide) (PNIPA) solutions,1, 2, 3, 4 it was shown that the behavior of their cloud-point curves is considerably affected by the kind of end group of the PNIPA sample used1, 3 and also by primary structure, that is, the degree of branching of the sample.2 Furthermore, it was pointed out that the cloud-point curve does not seem to correspond directly to a liquid–liquid binodal.4

The PNIPA samples used previously1, 2, 3, 4 were synthesized by radical polymerization using azobis(isobutyronitrile) (AIBN) as an initiator and by living anionic polymerization using diphenylmethylpotassium as an initiator, so that they had hydrophobic groups at their chain ends, that is, the isobutyronitrile group for the former and the diphenylmethyl group for the latter. The above-mentioned aqueous solution behavior of PNIPA may arise from the effect of the hydrophobic chain-end groups, which is enhanced with decreasing molecular weight. To verify more clearly the effect of the chain-end groups on the behavior of the cloud-point curve, it is desirable to pursue further investigations of aqueous solution behavior of PNIPA samples with a hydrophilic chain-end groups.

In this study, therefore, we prepared such PNIPA samples by aqueous redox polymerization using a redox catalyst consisting of ammonium persulfate and sodium metabisulfite, whose chain-end groups are hydrophilic and/or ionic ones. We first characterized the PNIPA samples so prepared by analyzing dilute solution properties such as the mean-square radius of gyration 〈S2〉, second virial coefficient A2 and intrinsic viscosity [η], in methanol at 25.0 °C, then determined the cloud point in the aqueous solutions, and finally made a comparison of present results with the previous ones for the PNIPA samples having the hydrophobic chain-end groups.1

Experimental procedure

Materials

Original PNIPA samples were synthesized by aqueous redox polymerization using a redox catalyst consisting of ammonium persulfate (NH4)2S2O8 (Wako Pure Chemical Industries, Ltd, Osaka, Japan) as an oxidative reagent and sodium metabisulfite Na2S2O5 (Wako Pure Chemical Industries, Ltd) as a reductive reagent, following the procedure reported by Wooten et al.5 The monomer N-isopropylacrylamide (Tokyo Chemical Industry Co., Ltd, Tokyo, Japan; ∼20 g), which had been recrystallized three times from a 9:1 mixture of n-hexane and benzene and then dried in a vacuum for 12 h, was dissolved in 250 ml of phosphate-buffered saline (PBS) with pH=7.3–7.5 (Lonza Walkersville, Inc., Walkersville, MD, USA). Subsequently, 0.09 g of Na2S2O5 in 1 ml of PBS was added to the monomer solution, followed by 0.03 g of (NH4)2S2O8 in 1 ml of PBS. The reaction mixture was then stirred under dry nitrogen at 25 °C for 24 h. Then, the polymerization mixture was poured into 250 ml of methanol, to precipitate an original polymer sample.6 The resulting polymer was then dissolved in pure water and dialyzed seven times against pure water for 24 h using a cellulose tube.

The original samples so prepared were separated into fractions of narrow molecular weight distribution by fractional precipitation using acetone as a solvent and n-hexane as a precipitant, or by a column elution method with a 6:4 mixture of n-hexane and acetone as an eluent. We obtained nine test samples named Rx (x=10, 19, 21, 48, 66, 96, 130, 220 and 270), which are generally called R samples. Each of the samples was dissolved in 1,4-dioxane, then filtered through a Teflon membrane Fluoropore (Sumitomo Electric Industries, Ltd, Osaka, Japan) with a pore size of 1.0 μm, and finally freeze-dried from their 1,4-dioxane solutions. They were dried in a vacuum at ∼80 °C for 12 h just before use. The ratio Mw/Mn of the weight-average molecular weight Mw to the number-average molecular weight Mn was determined for the 9 samples from analytical gel permeation chromatography in the same manner as before1, 3 using tetrahydrofuran as an eluent and 12 standard polystyrene (PS) samples as reference standards.

To specify the chain-end group of the present PNIPA samples, we synthesized an extra PNIPA sample with a very small Mw by mixing 10 g of N-isopropylacrylamide in 250 ml of PBS with 1.50 g of Na2S2O5 in 25 ml of PBS and 0.50 g of (NH4)2S2O8 in 25 ml of PBS, in the same manner as above. The reaction mixture was stirred under dry nitrogen at 25 °C for 30 min. The resulting polymer was dialyzed and then dried in the above-mentioned manner. The extra sample was named R0.5, whose values of Mw, Mn and Mw/Mn were determined to be 4.77 × 103, 3.34 × 103 and 1.43, respectively, from analytical gel permeation chromatography with a column Shodex SB-804 HQ (Showa Denko KK, Tokyo, Japan) connected to a solvent delivery pump PU-980 (JASCO Corporation, Tokyo, Japan) and a refractive index detector RI-8000 (Tosoh Corporation, Tokyo, Japan); a 0.2 M aqueous NaNO3 solution was used as an eluent and nine standard poly(oxyethylene) samples as reference standards.

Tetrahydrofuran used in analytical gel permeation chromatography was of reagent grade with no stabilizer. Deuterated dimethyl sulfoxide used for 1H- and 13C-nuclear magnetic resonance (NMR) spectroscopy was of reagent grade. Methanol used for static light scattering (LS) and viscosity measurements was purified by distillation after refluxing over calcium hydride for ∼6 h. Water used for the determinations of the cloud point was highly purified through a water purification system Simpli Lab (Millipore Corporate, Billerica, MA, USA); its resistivity was 18.2 MΩ cm.

1H- and 13C-NMR

To determine the values of the fraction fr of racemo diads for all the nine samples, R10 through R270, 1H NMR spectra for the samples in deuterated dimethyl sulfoxide at 170 °C at ∼1 wt% were recorded on a spectrometer EX-400 (JEOL Ltd, Tokyo, Japan) at 399.8 MHz using an rf pulse angle of 90° and a pulse repetition time of 8 s. To specify the chain-end group of the present PNIPA samples, the 13C NMR spectrum of the extra sample R0.5 in deuterated dimethyl sulfoxide at 150 °C at ∼20 wt% was recorded on the same spectrometer, operated at 100.5 MHz with a pulse repetition time of 3 s. Tetramethylsilane was added to each solution as an internal standard.

Light scattering

LS measurements were carried out to determine Mw and A2 for all nine samples, R10 through R270, and 〈S2〉 for the eight samples, R19 through R270, in methanol at 25.0 °C. A light-scattering photometer Fica 50 (Fica, Saint Denis, France) was used for all measurements with vertically polarized incident light of wavelength λ0=436 nm. To calibrate the apparatus, the intensity of the light scattered from pure benzene was measured at 25.0 °C at a scattering angle of 90°, where the Rayleigh ratio RUu(90°) of pure benzene was taken as 46.5 × 10−6 cm−1.7 The depolarization ratio ρu of pure benzene at 25.0 °C was determined to be 0.41±0.01. The scattered intensity was measured at eight different concentrations and at scattering angles ranging from 22.5 to 142.5°, and then converted to the excess unpolarized components ΔRUv of the reduced scattered intensity using the scattered intensity from the solvent methanol. The data obtained were treated by using the Berry square-root plot.8 For all samples, corrections for the optical anisotropy were unnecessary as the degree of depolarization was negligibly small.

The most concentrated solution of each sample was prepared gravimetrically and made homogeneous by continuous stirring at room temperature for 1 or 2 days. It was optically purified by filtration through a Teflon membrane Fluoropore with a pore size of 0.45 or 0.10 μm. The solutions of lower concentration were obtained by successive dilution. The weight concentrations of the test solutions were converted to polymer mass concentrations c using the densities of the respective solutions calculated from the partial specific volumes v2 of the samples and the density ρ0 of the solvent methanol. The quantity v2 was measured using an oscillating U-tube density meter DMA5000 (Anton-Paar, Graz, Austria). The values of v2 so determined in methanol at 25.0 °C were 0.869, 0.876, 0.878 and 0.883 cm3 g−1 for the samples R10, R19, R21 and R48, respectively, and 0.884 cm3 g−1 for the samples with Mw≳6 × 105, independent of Mw. For ρ0 of methanol at 25.0 °C, we used the literature value 0.7866 g cm−3.9

The refractive index increment ∂n/∂c was measured at a wavelength of 436 nm using a differential refractometer DR-1 (Shimadzu Corporation, Kyoto, Japan). The values of ∂n/∂c in methanol at 25.0 °C were determined to be 0.1865, 0.1864, 0.1855 and 0.1846 cm3 g−1 for samples R10, R19, R21 and R48, respectively, and 0.1851 cm3 g−1 for samples with Mw≳6 × 105, independent of Mw. For the refractive index n0 of methanol at 25.0 °C at a wavelength of 436 nm, we used the literature value 1.3337.9

Viscosity

Viscosity measurements were carried out for all nine samples, R10 through R270, in methanol at 25.0 °C using a conventional capillary viscometer of the Ubbelohde type. The flow time was measured to a precision of 0.1 s, keeping the difference in the flow time between the solvent and solution longer than 20 s. The test solutions were maintained at 25.0 °C within ±0.005 °C, during the measurements.

The most concentrated solution of each sample was prepared in the same manner as in the case of the LS measurements. The solutions of lower concentration were obtained by successive dilution. The polymer mass concentrations c were calculated from the weight fractions and the densities of the solutions. Density corrections were also made in the calculations of the relative viscosity ηr from the flow times of the solution and solvent. The data obtained for the specific viscosity ηsp and ηr in the range of ηr<1.8 were treated, as usual, by the Huggins (ηsp/c vs c) and Fuoss–Mead (ln ηr/c vs c) plots, respectively, to determine [η] and the Huggins coefficient k′. (Note that the two plots have the same intercept.)

Transmittance of light

The intensity of light passing through aqueous solutions of the samples R10 and R19 at various weight fractions w was measured using a home-made apparatus with incident light of wavelength 650 nm from a laser diode module. A given test solution contained in a cylindrical cell of outer diameter 10 mm was immersed in a water bath and stirred continuously. The temperature of the water bath was made to increase at a rate of ∼1.5 °C h−1. During the continuous increase in temperature from 30 to 35 °C, the intensity of the light passing through the cell was monitored by a photodiode. The output of the photodiode and the solution temperature, measured simultaneously using a platinum resistance thermometer combined with a programmable digital multimeter 7555 (Yokogawa Electric Corporation, Tokyo, Japan), were recorded on a personal computer at intervals of 10 s. Then, the (relative) transmittance, as defined as the ratio of the intensity of light through the test solution at a given temperature to the intensity at a lower temperature at which the test solution is apparently transparent, was determined as a function of temperature.

The measurements were carried out for each sample at eight concentrations in the range of 0.5%≲w≲10%. The most concentrated solution of each sample was prepared gravimetrically and made homogeneous by continuous stirring for 2 days at ∼5 °C. The solutions of lower concentration were obtained by successive dilution.

Results and Discussion

Characterization

Stereochemical compositions and chain-end groups

As shown in the second column of Table 1, the values of the fraction fr of the racemo diads determined for all nine samples, R10 through R270, from 1H NMR spectra using the assignments of the 1H signals as proposed by Isobe et al.,10 are almost the same (0.53–0.54) and rather in good agreement with the fr values (0.51–0.52) determined previously for the samples synthesized by radical polymerization using AIBN as an initiator in tert-butanol and benzene,1 which are hereafter called T and B samples, respectively.

Table 1 Values of fr and Mw/Mn and results of LS and viscosity measurements for poly(N-isopropylacrylamide) in methanol at 25.0 °C
Full size table

Figure 1 shows the 13C NMR spectrum of the extra sample R0.5 in the range of the chemical shift δ from 10 to 80 p.p.m., where signals from the methylene carbon Ca (34≲δ/p.p.m.≲36)11 and the methine carbon Cb (41≲δ/p.p.m.≲42)11 in the main chain, the methine carbon Cc (38≲δ/p.p.m.≲40)11 and the methyl carbon Cd (20≲δ/p.p.m.≲22)11 in the side group, and the carbons in the solvent dimethyl sulfoxide (38≲δ/p.p.m.≲40) are observed. We note that a signal from the carbonyl carbon (172≲δ/p.p.m.≲174)11 is out of the range.

Figure 1
figure 1

13C nuclear magnetic resonance spectrum for the poly(N-isopropylacrylamide) sample R0.5 in the range of 10⩽δ/p.p.m.⩽80. The signal with the asterisk is from the methylene carbon adjacent to the initiating chain-end group. DMSO, dimethyl sulfoxide.

Full size image

The small signal with the asterisk observed at δ≃54 p.p.m. may arise from methylene carbons adjacent to the initiating chain-end groups. Analogous to polymers synthesized by aqueous redox polymerization using (NH4)2S2O8 and Na2S2O5,12, 13 the initiating chain-end of sample R0.5 may probably be sulfonate, sulfate or hydroxy groups, as shown in Figure 2. According to the report by Ebdon et al.,12 which stated that the signals from the methylene carbons adjacent to the sulfonate, sulfate and hydroxy groups at the initiating end of a vinyl polymer appear at 51≲δ/p.p.m.≲52, 61≲δ/p.p.m.≲62 and 64≲δ/p.p.m.≲65, respectively, the signal with the asterisk in Figure 1 may be assigned to the methylene carbon adjacent to the sulfonate end-group. As signals from the methylene carbons adjacent to the sulfate and hydroxy end groups are inappreciable in Figure 1, it may be concluded that almost all of the initiating chain-ends in the sample R0.5 are sulfonate groups. Although we could not directly specify the type of the initiating chain-end group for the nine samples, R10 through R270, because of the weakness of the signal from the methylene carbon adjacent to the initiating chain-end group, the sulfonate group may also be predominant in the samples synthesized in the same manner as was sample R0.5. As for the terminating end group of the present PNIPA samples, we do not have any detailed information.

Figure 2
figure 2

Chemical structures of probable initiating chain-end groups: sulfonate, sulfate and hydroxy groups.

Full size image

Mean-square radius of gyration in methanol at 25 °C

The values of Mw and Mw/Mn determined from LS measurements in methanol at 25.0 °C and analytical gel permeation chromatography, respectively, for all nine samples, R10 through R270, are given in the third and fourth columns, respectively, of Table 1. The fifth column of the table gives the values of 〈S2〉 for the eight samples R19 through R270 determined simultaneously from the LS measurements in methanol at 25.0 °C. We note that the value of 〈S2〉 for sample R10 has been omitted because the slope of the Berry square-root plot against the square of the magnitude of the scattering vector was not sufficiently large for R10 to evaluate 〈S2〉 accurately.

Figure 3 shows double-logarithmic plots of 〈S2〉 (in Å2) against Mw; the unfilled circles represent the present values for the R samples, and the solid curve smoothly connects the data points. For comparison, plots for the T and B samples have been reproduced from Figure 1 of ref. 2. The filled triangles and squares represent the values for the T and B samples, respectively, and dashed curves smoothly connect the respective data points.

Figure 3
figure 3

Double-logarithmic plots of 〈S2〉 (in Å2) against Mw for poly(N-isopropylacrylamide) in methanol at 25.0 °C: (○), R samples; (▴), T samples;2 (▪), B samples.2 The curves smoothly connect the respective data points.

Full size image

The slope of the plot of the R samples in the range of Mw≳106 is 0.94 and intermediate between 1.12 and 0.90 for the T and B samples, respectively. The slope 0.94 is less than 1.2, which corresponds to linear flexible polymers with very large Mw in good solvents, and smaller than unity, which corresponds to unperturbed linear polymers. This implies that the primary structure of the R samples is not linear but branched, as in the cases of the T and B samples.2 The values of 〈S2〉 for the R samples are in rather good agreement with the values for the T samples, but definitely larger than the values for the B samples. For a given polymer chain, 〈S2〉 generally decreases with the increasing degree of branching.14 Considering the fact that the R, T and B samples have almost the same stereochemical composition, therefore, the degrees of branching of the R and T samples agree with each other and are smaller than that of the B samples. Strictly, however, the data points for the R samples deviate downward from the point for the T samples for Mw≳2 × 106. This implies that the degree of branching is somewhat larger for the R samples than for the T samples in the range of very large Mw.

Second virial coefficient in methanol at 25.0 °C

The values of A2 determined from LS measurements in methanol at 25.0 °C for all nine samples, R10 through R270, are given in the sixth column of Table 1. The A2 values are of order 10−4 cm3 mol g−2, indicating that methanol at 25.0 °C is a good solvent for the present R samples as well as for the previous T and B samples.

Figure 4 shows double-logarithmic plots of A2 (in cm3mol g−2) against Mw for PNIPA in methanol at 25.0 °C. For comparison, plots for the T and B samples have been reproduced from Figure 2 of ref. 2. The symbols and curves have the same meaning as those in Figure 3. The values of A2 for the R samples are in rather good agreement with the values for the T samples over the whole range of Mw examined and are definitely larger than the values for the B samples. The quantity A2 is proportional to an effective volume, excluded to a polymer chain by the presence of another, and decreases with an increasing degree of branching. Therefore, the above result indicates that the effective volume of a given R sample is almost the same as that of a T sample with the same Mw.

Figure 4
figure 4

Double-logarithmic plots of A2 (in cm3 mol g−2) against Mw for poly(N-isopropylacrylamide) in methanol at 25.0 °C. The symbols and curves have the same meaning as those in Figure 3.

Full size image

The seventh column of Table 1 gives the values of the interpenetration function Ψ for the eight samples R19 through R270 in methanol at 25.0 °C calculated from

with the values of Mw, 〈S2〉 and A2 given in the third, fifth and sixth columns, respectively, of the table, where NA is the Avogadro constant. The quantity Ψ, which is nearly equal to 0.24 for linear flexible polymers with very large Mw in good solvents, is sensitive to the density profile of repeat units constituting a given polymer14, 15 and becomes larger than 0.24 if the segment density of the polymer is larger than that of the linear flexible polymers.

Figure 5 shows plots of Ψ against log Mw for the R samples in methanol at 25.0 °C along with plots for the T and B samples, which have been reproduced from Figure 4 of ref. 2. The unfilled circles, filled triangles, filled squares and solid and dashed curves have the same meaning as those in Figures 3 and 4. For comparison, literature data for linear PS (♦),16, 17 regular 4-arm star PS ()18 and regular 6-arm star PS (),19 all measured in benzene at 25.0 °C, are also shown in Figure 5; the dotted curves smoothly connect the data points for each kind of PS. The value of Ψ for PS increases with the increasing number of arms because of the above-mentioned nature of Ψ. Note that the linear PS corresponds to two-arm star PS.

Figure 5
figure 5

Plots of Ψ against logMw. The symbols ○, ▴, ▪ and solid and dashed curves have the same meaning as those in Figure 3. The other symbols represent the literature data for polystyrene (PS) in benzene at 25.0 °C: (♦) linear PS;16, 17 () regular 4-arm star PS;18 () regular 6-arm star PS.19 The dotted curve smoothly connects the data points for each kind of PS.

Full size image

As a natural consequence of the behavior of 〈S2〉 and A2, shown in Figures 3 and 4, respectively, Ψ for the R samples is close to that for the T samples in the Mw range of ≲2 × 106, but deviates upward as Mw increases from ∼2 × 106. As for the B samples, Ψ increases with increasing Mw and becomes larger than Ψ for the regular four-arm star PS, which indicates that the degree of branching in the B samples is definitely larger than that in the R and T samples.

Intrinsic viscosity in methanol at 25 °C

The values of [η] and k′ determined from viscosity measurements in methanol at 25.0 °C for all nine samples, R10 through R270, are given in the eighth and ninth columns of Table 1. In accordance with the fact that methanol at 25.0 °C is also a good solvent for the present R samples as in the cases of the previous T and B samples, deduced from the above-mentioned results of A2, the k′ values lie between 0.3 and 0.4.

Figure 6 shows double-logarithmic plots of [η] (in 100 ml per g) against Mw for PNIPA in methanol at 25.0 °C. For comparison, plots for the T and B samples have been reproduced from Figure 3 of ref. 2. The symbols and curves have the same meaning as those in Figures 3 and 4. As in the cases of 〈S2〉 and A2, shown in Figures 3 and 4, respectively, the values of [η] for the R samples are in rather good agreement with the values for the T samples and definitely larger than the values for the B samples. The quantity [η] is proportional to the hydrodynamic (molar) volume and decreases with the increasing degree of branching. Therefore, the above result indicates that the hydrodynamic volume of a given R sample is almost the same as that of a T sample with the same Mw. The slope of the plot for the R samples in the range of Mw≳106 is 0.58 and is smaller than 0.8, which corresponds to linear flexible polymers with very large Mw in good solvents, as in the cases of the T and B samples.2 As a minor point, we note that the values of [η] for the R samples with Mw≲105 are somewhat larger than the values for the T samples. This subtle discrepancy may arise from the difference in the effective hydrodynamic volume of the chain-end group between the R and T samples.

Figure 6
figure 6

Double-logarithmic plots of [η] (in 100 ml per g) against Mw for poly(N-isopropylacrylamide) in methanol at 25.0 °C. The symbols and curves have the same meaning as those in Figure 3.

Full size image

The last column of Table 1 gives the values of the Flory–Fox factor Φ for the eight samples, R19 through R270, in methanol at 25.0 °C calculated from

with the values of Mw, 〈S2〉 and [η] given in the third, fifth and eighth columns, respectively, of the table. The quantity Φ, which is nearly equal to 2 × 1023 mol−1 for linear flexible polymers with very large Mw in good solvents, is sensitive to the degree of branching of a given polymer and takes a larger value if the polymer has branches.

Figure 7 shows plots of Φ against logMw for the R samples in methanol at 25.0 °C along with plots for the T and B samples, which have been reproduced from Figure 5 of ref. 2 and plots for the linear, regular 4-arm and regular 6-arm PS.16, 17, 18, 19 All the symbols and curves have the same meaning as those in Figure 5. The difference in the behavior of Φ among the R, T and B samples is almost the same as in the case of Ψ shown in Figure 5.

Figure 7
figure 7

Plots of Φ (in mol−1) against logMw. The symbols ○, ▴, ▪ and solid and dashed curves have the same meaning as those in Figure 3. The other symbols ♦, , and the dotted curves have the same meaning as those in Figure 5.

Full size image

From the results of the 1H- and 13C-NMR spectra and the dilute solution properties given in this subsection, it may be concluded that the R samples have almost the same stereochemical compositions as those of the T and B samples, the same degree of branching as that of the T samples for Mw≲2 × 106, and the hydrophilic sulfonate group at the initiating end in place of the hydrophobic isobutyronitrile group for the T and B samples.

Cloud point in aqueous solutions

Figure 8 shows plots of the (relative) transmittance against temperature for the aqueous solutions of the PNIPA samples, R10 and R19, both at w=5.05%. It should be noted here that the shape of the transmittance curve for each solution is almost independent of the rate of increase in temperature if the rate is slower than 1.5 °C h−1. For comparison, the transmittance curves (dashed) previously determined for the aqueous solutions of the T samples, T5 (Mw=5.17 × 104) and T13 (Mw=1.31 × 105), and the B samples, B5 (Mw=4.65 × 104) and B14 (Mw=1.44 × 105), all at w≃5%, have been reproduced from Figure 2 of ref. 1. The curves for the R samples are located to the right (high temperature side) of the curves for the T and B samples. It is interesting to see that the curve moves to the right (high temperature side) with increasing Mw for the T and B samples, but moves to the left (low-temperature side) with increasing Mw for the R samples.

Figure 8
figure 8

Temperature dependence of the transmittance of light passing through the aqueous solutions of the poly(N-isopropylacrylamide (PNIPA) samples R10 and R19, both at w=5.05%. The dashed curves represent the previous results for the aqueous solutions of the PNIPA samples T5, T13, B5 and B14, all at w≃5%.1 The horizontal line indicates the threshold value 90% and the symbol on each curve indicates the cloud point.

Full size image

The cloud point in a given test solution is, in principle, determined to be the temperature at which the solution just starts to become turbid. As seen from Figure 8, however, it is somewhat difficult to determine such a temperature unambiguously because the transmittance for each solution starts to decrease from 100% rather gently. Following the previous studies,1, 3 therefore, the cloud point in each solution was determined as the temperature at which the transmittance became the threshold value 90%, for convenience. In Figure 8, the symbols and represent the cloud points determined for samples R10 and R19, respectively, and the symbols ▾, ▴, and represent the cloud points previously1 determined for samples T5, T13, B5 and B14, respectively. As a natural consequence of the behavior of the transmittance curves, the cloud points for the R samples are the highest among the three kinds of samples.

Finally, Figure 9 shows the cloud-point curves in aqueous solutions of the R samples R10 () and R19 () in the range of 0.5%≲w≲10%. For comparison, the cloud-point curves previously determined in aqueous solutions of the T samples T5 (▾) and T13 (▴) and of the B samples B5 () and B14 () have been reproduced from Figure 3 of ref. 1. The solid and dashed curves smoothly connect the cloud points for the R samples and the T and B samples, respectively. In the figure, the literature data reported by de Azevedo et al.20 are also shown by the symbols . Although their PNIPA sample with Mw=6.15 × 105 was prepared in the same manner as in the case of the present R samples, their experimental procedure to determine the cloud point was different from ours: the temperature at which the slope of the plot of the reciprocal of the intensity of light scattered from an aqueous PNIPA solution against temperature changes suddenly during the heating process was adopted as the cloud point. Despite the difference in Mw and the procedure to determine the cloud point, the present data for R10 and R19 qualitatively agree with the data by de Azevedo et al.20

Figure 9
figure 9

Cloud-point curves in aqueous solutions of poly(N-isopropylacrylamide): () R10; () R19; (▾) T5;1 (▴) T13;1 () B5;1 () B14;1 () literature data for the sample with Mw=6.15 × 105 by de Azevedo et al.20 The curves smoothly connect the respective data points.

Full size image

As expected from the results shown in Figure 8, the cloud-point curves for the R samples are located on the high temperature side among the three kinds of samples over the whole range of w examined. It is more important to note that the cloud-point curve for the R samples shifts to the low-temperature side with increasing Mw, while the curves for the T and B samples in the range of w≳2% shift to the opposite side. Otake et al.21 and Schild and Tirrell6 pointed out that the cloud point in an aqueous solution of PNIPA synthesized by aqueous redox polymerization was somewhat higher than that of PNIPA synthesized by radical polymerization in organic solvents using AIBN as an initiator, although the reductive reagent N,N,N′,N′,-tetramethylethylenediamine of a redox catalyst used by them was different from ours. Furthermore, Xia et al.22, 23 reported that the cloud point in an aqueous solution of PNIPA synthesized by atom transfer radical polymerization with a hydrophilic nonionic group at the initiating chain-end decreased with increasing molecular weight. These experimental results are consistent with ours, as shown in Figure 9. The dependence of the cloud point on Mw for the R samples is rather consistent with the prediction of conventional polymer solution thermodynamics.24 We note that such dependence could also result from the hydrophilicity of the chain-end group: the hydrophilic effect increases with decreasing Mw.

It was previously3 shown that the cloud-point curve in an aqueous solution of a PNIPA sample with a strongly hydrophobic chain-end group, which was synthesized by living anionic polymerization and had the diphenylmethyl group at the initiating chain-end, shifted remarkably to the low-temperature side with decreasing Mw. Furthermore, it was recently reported that amphiphilic A-B block copolymers with a PNIPA block as the hydrophilic part formed aggregates in their aqueous solutions, and their cloud points were lower than those in an aqueous PNIPA solution.25, 26, 27, 28 These facts imply that the above-mentioned noteworthy difference in the behavior of the cloud-point curve between the R sample and the T and B samples arises from the hydrophobicity of the chain-end groups, whose effect relatively increases with decreasing Mw.

We close our discussion with reference to the previous opinion about the difference in the behavior of the cloud points between the T and B samples:1, 2 the increase in the number of branch points results in the increase in the number of hydrophobic chain-end groups and, therefore, in the decrease in the cloud point.

Conclusion

The series of PNIPA samples having sulfonate groups at the initiating chain-ends were prepared by aqueous redox polymerization using a redox catalyst consisting of (NH4)2S2O8 and Na2S2O5, which were referred to as R samples. The R samples were shown to have the same stereochemical compositions as those of the PNIPA samples, previously1 synthesized by radical polymerization using AIBN as an initiator in tert-butanol and benzene, which were called T and B samples, respectively. From the analyses of 〈S2〉, A2 and [η] in methanol at 25.0 °C, the R samples were shown to have the same degree of branching as that of the T samples in the range of Mw≲2 × 106, but have a somewhat larger degree of branching for Mw≳2 × 106 (the degree of branching of the B samples is the largest of all.2).

For the aqueous solutions of the two R samples, R10 (Mw=1.04 × 105) and R19 (Mw=1.86 × 105), the cloud-point curves were determined in the range of 0.5%≲w≲10%. The curves for R10 and R19 were appreciably higher than the curves previously1 determined for the two T samples (Mw=5.17 × 104, Mw=1.31 × 105). (The curves for the B samples were lower than those for the T samples.1) It was found that the cloud-point curve for the R samples shifts to the low-temperature side with increasing Mw, while the curves for the T and B samples in the range of w≳2% shift to the opposite side. The difference in the behavior of the cloud-point curve between the R and T samples, both with almost the same degree of branching and stereochemical composition, arises from the difference in the chain-end group: the R samples have the hydrophilic sulfonate group and the T samples have the hydrophobic isobutyronitrile group.