Membrane Formation via Thermally Induced PhaseSeparation in Polypropylene/Polybutene/Diluent System

HIDETO MATSUYAMA,1 HAJIME OKAFUJI,1 TAISUKE MAKI,1 MASAAKI TERAMOTO,1 NORIO TSUJIOKA2

1 Department of Chemistry and Materials Technology, Kyoto Institute of Technology, Matsugasaki, Sakyo-ku,Kyoto 606-8585, Japan

2 Asahi Chemical Industry Co., Ltd., 515 Kojima-cyo, Moriyama 524--0002, Japan

Received 22 December 2000; accepted 19 September 2001Published online 6 March 2002

ABSTRACT: Porous membranes were prepared from a polymer blend system by thethermally induced phase separation (TIPS) process. The polymer blend system wasisotactic polypropylene (iPP)/polybutene (PB) and the diluent was diphenyl ether(DPE). Two types of porous membranes were prepared by the extractions of DPE aloneand both DPE and PB after the phase separation. The effect of the addition of PB to theiPP solution on the phase diagram was investigated and the phase separation kineticswas measured by the light scattering method. The addition of PB resulted in the highersolute rejection property and lower water permeance. By the further extraction of PBfrom the porous iPP/PB membrane prepared by the extraction of DPE, the waterpermeance was approximately doubled, maintaining almost the same rejectionproperty. © 2002 Wiley Periodicals, Inc. J Appl Polym Sci 84: 1701–1708, 2002; DOI 10.1002/app.10550

Key words: thermally induced phase separation; polymer blend; polypropylene; po-lybutene; light scattering

INTRODUCTION

Thermally induced phase separation (TIPS) is amethod of making microporous membranes.1–11

TIPS is applied to a wide range of polymers thatcould not be used in the traditional phase inver-sion membrane formation due to the solubilityproblems. A variety of thermally stable, chemi-cally resistant membranes were produced by theTIPS process on a commercial scale.

In the TIPS process, only one polymer compo-nent and diluent, that is, two-component system,have been mainly used. Few studies have re-

ported on the formation of porous membranesfrom a two-polymer blend and diluent system.Castro produced microporous membranes byTIPS process from several two polymer blend/diluent systems such as polypropylene/chlori-nated polyethylene/diluent, polyethylene/chlori-nated polyethylene/diluent, and others.1 Themembrane formation from hydroxylate polypro-pylene/polypropylene/diluent system was re-ported by Chung and Lee.12

In contrast to the small number of works on theTIPS process, polymer blend was widely used toimprove thermal, rheological, and mechanicalproperties.13 Also in the field of membrane, ultra-filtration 14,15 and gas separation membranes16–18 were prepared from polymer blend systems.The advantage of the use of polymer blend in theTIPS process is that pore size can be controlled by

Correspondence to: H. Matsuyama (matuyama@chem.kit.ac.jp).Journal of Applied Polymer Science, Vol. 84, 1701–1708 (2002)© 2002 Wiley Periodicals, Inc.

1701

the addition of second polymer due to the changeof thermodynamics and kinetic properties. Fur-thermore, the properties of membrane materialssuch as hydrophilicity can be changed in theblend system, as shown in the study of Chung andLee.12 If only the one polymer component can bedissolved in a certain diluent, the polymer can berinsed away and high porosity will be achieved inthe porous membrane.

In this work, polypropylene/polybutene blendmembrane was produced by the TIPS process.The phase diagram and the pore growth kineticswere investigated. Because only polybutene (PB)could be rinsed away by toluene, two types ofporous membranes were prepared by the extrac-tions of diluent alone and both diluent and poly-butene after the phase separation. The mem-brane properties such as the solute rejection andwater permeability were compared for the twomembranes.

EXPERIMENTAL

Materials

Polymers used were isotactic polypropylene (iPP,Mw � 250,000; Aldrich Chemical Co.) and PB(�90% polyisobutene, Mn � 1290; Aldrich Chem-ical Co.). Diphenyl ether (DPE; Nakalai TesqueCo., Kyoto, Japan) was used as the diluent with-out further purification. Extractants for DPE andfor PB were methanol and toluene, respectively.

Phase Diagram

Homogeneous two-polymer (iPP � PB) diluentsamples were prepared by a method previouslydescribed.7 The sample was placed between a pairof microscope cover slips. To prevent diluent lossby evaporation, a Teflon film of 100 �m thicknesswith a square opening was inserted between thecoverslips. The sample was heated on a hot stage(Linkam, LK-600PH) at 453 K and cooled to 298 Kat a controlled rate of 10 K/min. The temperatureof the stage was manipulated by a LinkamL-600A controller. Cloud points were determinedvisually by noting the appearance of turbidityunder an optical microscope (Olympus BX50,Tokyo, Japan).

To obtain the spinodal points, the sample onthe hot stage was quenched and maintained atvarious desired temperatures below the cloudpoint at the maximum cooling rate of 130 K/min.

If the isolated droplets started to form graduallyin the scattered position, the temperature wasconsidered in the metastable region between thebinodal and the spinodal.19 On the other hand,when the interconnected structure was immedi-ately formed all over the position, the tempera-ture was considered in the unstable region belowthe spinodal. Thus, the spinodal point, which isthe border between the metastable region andunstable region, was determined from the resultsat various different temperatures.

A DSC (Perkin–Elmer, DSC-7) was used to de-termine the dynamic crystallization temperature.The solid sample was sealed in an aluminum DSCpan, melted at 453 K for 3 min, and then cooled ata 10 K/min to 298 K. The onset of the exothermicpeak during the cooling was taken as the crystal-lization temperature.

Kinetic Study

The light scattering measurement was carriedout to obtain the structure growth data with apolymer dynamics analyzer (Otsuka ElectronicsCo., Hiraka, DYNA-3000).20 The hot stage waslocated between a He-Ne laser (5 mW) and adetector. The sample sealed with two coverslipswas placed on the stage and heated at 453 K.Then it was quenched to the desired temperatureat a cooling rate of 130 K/min. It was confirmedthat the light scattering occurred after the tem-perature reached the set value. The time intervalfor the measurement was 0.22 s.

Droplet growth kinetics was measured by theoptical microscope in the relatively later stage ofthe phase separation. The sample sealed with twocoverslips was heated at 453 K and cooled to 298K at a cooling rate of 10 K/min. After the temper-ature reached the cloud point, the droplet sizewas observed under a microscope. The image fromthe microscope was converted to a video signal.The video signal was passed through a videotimer and into a videocassette recorder. To obtainthe average droplet size of the polymer-leanphase, an image analysis (Mitani Co., Fukui,Japan, Win ROOF) was used.

SEM Observation

The sample sealed with two coverslips was heatedat 453 K and quenched in ice water. Then diphe-nyl ether was extracted from the membrane withmethanol and the methanol was evaporated. Insome cases, PB in the membrane was further

1702 MATSUYAMA ET AL.

extracted with toluene. The microporous mem-brane was fractured in liquid nitrogen and coatedwith Au/Pd. A SEM (Hitachi Co., Tokyo, Japan,S-800) with an accelerating voltage set to 15 kVwas used to examine the membrane cross sec-tions.

Solute Rejection Experiment

Membranes used for the filtration experimentwere prepared as follows. The homogeneous poly-mer-diluent sample was placed between a pair ofcopper plates (length: 150 mm, width: 150 mm,thickness: 0.5 mm). For adjusting membranethickness, the Teflon film of 150 �m thicknesswith a square opening in the center was insertedbetween the copper plates. The copper plateswere heated at 453 K in an oven for 15 min tocause melt-blending. Then the copper plates werequenched in ice water. After cooling, diphenylether was extracted from the membrane withmethanol and the methanol was evaporated. Insome cases, PB in the membrane was furtherextracted with toluene.

The apparatus and procedure for the filtrationexperiment were the same as those described pre-viously.21 The filtration experiment was con-ducted by using a stirred cell (Advantec Co.,UHP-25K). The feed solution was pressurized bynitrogen gas at 0.1 atm for iPP membrane and at0.5 atm for iPP/PB blend membrane. Solutes usedwere lysozyme from egg white (Seikagaku Co., 6�crystallized, Mw � 14,600, Stokes radius: 1.69nm22), ovalbumin (Sigma Chemical Co., Grade V,98% purity, Mw � 45,000, Stokes radius: 2.53nm22), ferritin from horse spleen (Nacalai TesqueCo., Mw � 440,000, Stokes radius: 6.77 nm22) andpolystyrene latex particle (Duke Scientific Co.,radius: 50 nm). The feed solutions were preparedby dissolving the proteins in a 0.05 mol/dm3 phos-phate-buffered solution (disodium hydrogen phos-phate � potassium dihydrogen phosphate, pH7.0). The protein concentrations were 0.1 g/dm3

for lysozyme, 0.2 g/dm3 for ovalbumin, and 0.002g/dm3 for ferritin. The latex particle was dis-persed in an aqueous nonionic surfactant (0.01%Triton X-100) at a concentration of 1.03 � 1011

particles/dm3. The solute concentrations in fil-trate were measured by using a UV spectropho-tometer (Hitachi Co., U-2000) at the wavelengthsof 280 nm for lysozyme and ovalbumin, 275 nmfor ferritin, and 385 nm for latex particle. Beforethe filtration experiments, aqueous propanol so-lution (50 wt %) was passed through the mem-

brane to fill pores of the membrane with the so-lution.

RESULTS AND DISCUSSION

Extraction Behavior

Figure 1 shows the membrane weight after theextraction. The ordinate is the dimensionlessweight, which is defined as the membrane weightafter the extraction divided by the initial mem-brane weight. The axis denotes the weight per-centage of PB added to the iPP solution, in whichthe iPP weight percentage was fixed to be 10 wt%. First, DPE was extracted by methanol andthen PB was extracted by toluene. The data inboth cases are shown in Figure 1. The solid anddashed lines denote the predicted values when allDPE and DPE � PB are extracted. The experi-mental results in DPE extraction (E) and in DPE� PB extraction (F) agreed with the predictedsolid line and dashed line, respectively. This in-dicates that almost all DPE and PB could beeffectively extracted. Further, the agreement be-tween the experimental data and the predictedlines suggests that the concentration of iPP andPB in the polymer-lean phase is quite low. Whenthe polymer concentration in the polymer-leanphase is high, in which the polymers can flow outin the extraction process, the experimental data isexpected to be lower than the predicted line.Therefore, almost all iPP and PB exist in the

Figure 1 Membrane weight after the extraction. (E)DPE extraction; (F) DPE � PB extraction. The solidand dashed lines denote the predicted values when allDPE and DPE � PB are extracted, respectively. Mem-brane was prepared as follows. The homogeneous poly-mer–diluent sample was placed between a pair of glassplates (thickness: 2.8 mm) and after melt-blending, theglass plates were cooled at room temperature.

PP/PB/DILUENT THERMAL PHASE SEPARATION 1703

polymer-rich phase and form the membrane ma-trix.

The shrinkage results after the evaporation ofextractants are summarized in Table I. In bothcases of iPP alone and iPP/PB blend, the shrink-age degree of the membranes after the extractionof DPE were low. On the other hand, after theextraction of PB, the iPP/PB blend membraneshrunk well and the resultant volume was 69% ofthe initial volume. Even after the immersion intoluene, the iPP membrane did not shrink somuch.

Phase Diagram

Figure 2 shows the phase diagrams. The cloudpoint curve and the dynamic crystallization tem-perature are shown in Figure 2(a) when iPP con-centration was fixed at 10 wt % and additional PBconcentration was changed. The cloud points de-creased with the increase of PB concentrationbecause the total polymer concentration in-creased. In the polymer concentration regionhigher than the critical point, the increase of the

polymer concentration usually leads to the de-crease of the binodal line in the upper criticalsolution temperature-type phase diagram.23 Thecrystallization temperature slightly decreased asthe PB concentration increased. However, suchdegree of decrease was very small and the crys-tallization temperature was not influenced somuch by the addition of PB. Because PB is theamorphous polymer, only iPP can crystallize. Asshown in Table II, the heat of crystallizationslightly decreased with the increase of PB con-tent. This means that the crystallinity of iPP de-creased by the addition of PB.

Figure 2(b) shows the phase diagrams whenthe ratio of PP to PB was fixed. The cloud pointsat the same total polymer concentration de-creased with the increase of the PB fraction. Thiscannot be explained by the increase of the totalpolymer concentration, which was the main causefor the result shown in Figure 2(a). First, we mustconsider the change of the compatibility betweenthe polymer and the diluent. The solubility pa-rameters for iPP, polyisobutene, and DPE were

Table I Shrinkage Results After the Extraction of DPE and PB

Membrane Typea ExtractantRelative

Membrane AreabRelative Membrane

ThicknessbRelative Membrane

Volumeb

iPP alone Methanol 94.0% 97.9% 92.0%iPP alone Methanol � toluene 88.7% 95.7% 84.9%iPP/PB blendc Methanol 97.4% 100% 97.4%iPP/PB blendc Methanol � toluene 77.6% 88.9% 68.9%

aMembrane preparation condition was the same as that used in the membranes for the solute rejection experiment.bValues relative to the initial values of membranes before the extraction.ciPP : PB ratio � 1 : 1.

Figure 2 Phase diagrams for polymer blend/diluent system. (a) Constant iPP concen-tration (10 wt %). (F) cloud point; (E) crystallization temperature. (b) Constant ratio ofPP to PB. (F) cloud point (iPP alone), (Œ) cloud point (iPP : PB � 2 : 1), (f) cloud point(iPP : PB � 1 : 1), (E) crystallization temperature (iPP alone), (‚) crystallization tem-perature (iPP : PB � 2 : 1), (�) crystallization temperature (iPP : PB � 1 : 1)

1704 MATSUYAMA ET AL.

reported as 18.8,24 16.5,24 and 20.7 MPa1/2,25 re-spectively. When the PB fraction in the polymerblend increases, the solubility parameter differ-ence between the polymer blend and the diluentbecomes large and thus the compatibility de-creases, which suggests that the cloud pointshould be shifted to the high-temperature posi-tion.6 Because this expectation is opposite to theexperimental result, the result cannot be ex-plained by the change of the compatibility. Theaddition of PB brings about the decrease of theaverage molecular weight because the molecularweight of PB is much lower than that of iPP.When the molecular weight is lower, the cloudpoint is shifted to the lower temperature positiondue to the entropy effect.23 This may be the rea-son for the decrease of the cloud point broughtabout by the increase of PB fraction.

The spinodal points were 0.7 and 0.5°C lowerthan the binodal points in the cases of iPP alone(15 wt %) and iPP (10wt %) � PB (5 wt %) system,respectively.

Kinetic Study

An example of the light scattering measurementis shown in Figure 3. The time shown in thisfigure corresponds to the elapse of time after thephase separation occurs. Clear maxima of the

scattered light intensity Is were observed eachtime. This indicates that the phase separationmechanism is not the nucleation and growthmechanism (NG mechanism) but the spinodal de-composition (SD).26 The peaks of Is shifted to asmaller angle with time, which means that thephase-separated structure was growing in thistime scale.20

The scattered angle is related to the wave num-ber as:

q � �4�n/�0�sin��/2� (1)

where n is the solution refraction index and �0 isthe wavelength of light in vacuo. The relationbetween qm, which is the wave number wheremaximum scattered light intensity appears, andthe average interphase periodic distance �m isgiven by

Figure 4 Time course of �m in both iPP alone (15 wt%) and iPP/PB systems (iPP 10 wt % � PB 5 wt %). (a)Quench depth: 1.2 K for iPP alone and 1.3 K for iPP/PBblend, (b) quench depth: 2.2 K for iPP alone and 2.3 Kfor iPP/PB blend.

Figure 5 Time course of average droplet diameter.Cooling rate was 10 K/min.

Table II Heat of Crystallization

PB Weight Percent 0 2 5 10

Heat of crystallization[J/g-iPP] 107 101 87.9 82.7

iPP concentration was fixed to be 10 wt%.

Figure 3 An example of light scattering measure-ment. IPP 10 wt % � PB 5 wt %, quench depth � 1.3 K.

PP/PB/DILUENT THERMAL PHASE SEPARATION 1705

qm � 2�/�m (2)

By using eqs. (1) and (2), the time course of �mcan be obtained from the light scattering experi-ment result. Figure 4 shows the time course of �min both iPP alone and iPP/PB blend systems (iPP :PB � 2 : 1). At almost the same quench depth,which is defined as the difference between thespinodal temperature and the experimental tem-perature, the larger �m was obtained in the iPPsystem than in the iPP/PB blend system. In bothtwo-quench depth cases, the difference of �mhardly changed as time passed. This means thatat the early stage of SD, �m is lower in the blendsystem and, however, the growth rate of �m in thelater stage was almost the same in the two sys-tems. �m in the early stage of SD can be related asradius of gyration Rg as 27:

�m � Rg�Ts � T�/T�2/3 (3)

where Ts and T are the spinodal temperature andexperimental temperature, respectively. In theiPP/PB blend system, the average molecularweight is smaller than in the iPP system, as de-scribed above. Thus, the smaller Rg value in the

blend system led to the smaller �m in the earlystage of SD.

Average droplet diameter determined from theoptical micrograph is plotted with time in Figure5. In these experiments, temperature was notfixed and gradually decreased at a constant cool-ing rate of 10 K/min. The droplet sizes increased

Figure 6 Cross sections of the membranes. (a) iPP alone (10 wt %, DPE extraction),(b) iPP 10 wt % � PB 5 wt % (DPE extraction), (c) iPP 10 wt % � PB 10 wt % (DPEextraction), (d) iPP 10 wt % � PB 10 wt % (DPE � PB extraction).

Figure 7 Relation between apparent solute rejectioncoefficient and solute Stoke radius. (‚) iPP alone (10 wt%, DPE extraction), (�) iPP 10 wt % � PB 10 wt %(DPE extraction), (F) iPP 10 wt % � PB 10 wt % (DPE� PB extraction).

1706 MATSUYAMA ET AL.

initially with time and reached constant values inall polymer cases. When temperature crossed thecloud point in the cooling process, phase separa-tion occurred and the droplet growth continueduntil the temperature reached the crystallizationtemperature. By the addition of PB, the constantdroplet size clearly decreased. As shown in Figure2(a), the region between the cloud point and thecrystallization temperature decreased with theincrease in amount of PB. Thus, in the polymerblend with the higher PB content, the dropletgrowth soon stopped and the resultant dropletsize was smaller.

SEM Observation

Figure 6 shows the cross sections of the mem-branes. As the PB content in the blend membraneincreased [Fig. 6(a)–(c)], the pore size decreasedbecause of the increase in the total polymer con-centration. This result agreed with the result inFigure 5, although the cooling conditions in thetwo cases were different. It is expected that afterthe PB extraction, the thickness of the polymerfibrils becomes thinner due to the effusion of PBfrom the matrix phase, which probably leads tothe larger pores. However, between the mem-brane structures after the PB extraction [Fig.6(d)] and before the PB extraction [Fig. 6(c)], aclear difference cannot be recognized at this mag-nification.

Solute Rejection

Figure 7 shows the relation between the apparentsolute rejection coefficient Ra and the soluteStokes radius. The apparent rejection coefficientis defined as (1 � Cs/C0), where C0 and Cs are thesolute concentrations in the feed and filtratephases. In Table III, the pure water permeanceare summarized in three cases. The water per-meance is defined as the volumetric flow ratedivided by the membrane area and the pressuredifference. From Figure 7 and Table III, the ad-dition of PB increased the solute rejection coeffi-cient remarkably and lowered the water per-

meance due to the formation of the smaller pores,as shown in Figure 6. By the extraction of PB, thewater permeance was approximately doubled,maintaining the same solute rejection property.This indicates that the use of the polymer blendsystem in the TIPS process has the advantage ofobtaining the higher permeability by the extrac-tion of one polymer component. The extraction ofPB brings about the enlargement of pores,whereas the membrane shrinkage shown in TableI brings about the pore size reduction. If theformer effect compensates for the latter, almostthe same solute rejection properties are obtainedin the cases after and before the PB extraction, asshown in Figure 7. The higher water permeancein the extraction of PB may be due to the forma-tion of small pores, through which only watermolecules can permeate and the larger solute can-not permeate. Further detailed investigation isnecessary to confirm this.

CONCLUSION

Porous membranes were prepared from iPP/PBblend system by the TIPS process. The addition ofPB to the iPP solution lowered the cloud point,whereas the dynamic crystallization temperaturehardly changed. The kinetic studies by the lightscattering showed that the structure formed atthe early stage of the spinodal decomposition wassmaller in the blend system than in the iPP sys-tem. However, the structure growth rate in thelatter stage was almost the same in both systems.When the polymer solution was cooled at the con-stant cooling rate of 10 K/min, the droplet sizesincreased first and reached constant value. Theconstant droplet sizes decreased with the increaseof the PB content because of the smaller regionbetween the cloud point and the crystallizationtemperature.

The addition of PB increased the solute rejec-tion and lowered the water permeance owing tothe formation of smaller pores. The extraction of

Table III Water Permeance of Membranes Prepared in Various Conditions

Membrane Type Extracted Species Water Permeance [m3/(m2 s Pa)]

iPP alone (10 wt %) DPE 3.29 � 10�9

iPP/PB blend (iPP10 wt % � PB 10 wt %) DPE 2.35 � 10�10

iPP/PB blend (iPP10 wt % � PB 10 wt %) DPE � PB 4.51 � 10�10

PP/PB/DILUENT THERMAL PHASE SEPARATION 1707

PB from the membrane was effective to improvethe water permeance.

REFERENCES

1. Castro, A. J. U.S. Pat. 4,247,498, 1981.2. Caneba, G. T.; Soong, D. S. Macromolecules 1985,

18, 2538.3. Hiatt, W. C.; Vitzthum, G. H.; Wagener, K. B.;

Gerlach, K.; Josefiak, C. in Microporous Mem-branes via Upper Critical Temperature Phase Sep-aration; Lloyd, D. R., Ed.; ACS Symposium Series269; American Chemical Society: Washington, DC,1985; p 229.

4. Lloyd, D. R.; Kinzer, K. E.; Tseng, H. S. J MembrSci 1990, 52, 239.

5. Tsai, F.-J.; Torkelson, J. M. Macromolecules 1990,23, 775.

6. Lloyd, D. R.; Kim, S.-S.; Kinzer, K.E. J Membr Sci1991, 64, 1.

7. Kim, S.-S.; Lloyd, D. R. J Membr Sci 1991, 64, 13.8. Vadalia, H. C.; Lee, H. K.; Myerson, A. S.; Levon, K.

J Membr Sci 1994, 89, 37.9. Mehta, R. H.; Madsen, D. A.; Kalika, D. S. J Membr

Sci 1995, 107, 93.10. Matsuyama, H.; Berghmans, S.; Lloyd, D. R. J

Membr Sci 1998, 142, 213.11. Matsuyama, H.; Berghmans, S.; Lloyd, D. R. Poly-

mer 1999, 40, 2289.12. Chung, T. C.; Lee, S. H. J Appl Polym Sci 1997, 64,

567.

13. Utracki, L. A. Polymer Alloys and Blends; CarlHanser Verlag: Munich, 1989.

14. Nunes, S. P.; Peinemann, K. V. J Membr Sci 1992,73, 25.

15. Blicke, C.; Peinemann, K. V.; Nunes, S. P. J MembrSci 1993, 79, 83.

16. Chiou, J. S.; Paul, D. R. J Appl Polym Sci 1987, 34,1037.

17. Bikson, B.; Nelson, J. K.; Muruganandam, N. JMembr Sci 1994, 94, 313.

18. Park, C.; Jo, W. H.; Park, H. C.; Kang, Y. S. Poly-mer 2000, 41, 1765.

19. van de Witte, P.; Dijkstra, P. J.; van den Berg,J. W. A.; Feijen, J. J Membr Sci 1996, 117, 1.

20. Matsuyama, H.; Kudari, S.; Kiyofuji, H.; Kitamura,Y. J Appl Polym Sci 2000, 76, 1028.

21. Matsuyama, H.; Yuasa, M.; Kitamura, Y.; Ter-amoto, M.; Lloyd, D. R. J Membr Sci 2001, 179, 91.

22. Matsuyama, H.; Iwatani, T.; Kitamura, Y.; Ter-amoto, M.; Sugo, N. J Appl Polym Sci 2001, 79,2456.

23. Flory, P. J. Principles of. Polymer Chemistry; Cor-nell Univ. Press: Ithaca, NY, 1953.

24. Brandrup, J.; Immergut, E. H. Polymer Handbook,4th ed.; John Wiley and Sons: New York, 1999.

25. Barton, A. F. M. Handbook of Solubility Parame-ters and Other Cohesion Parameters, 2nd ed.; CRCPress: Boca Raton, FL, 1991.

26. Hashimoto, T.; Sasaki, K.; Kawai, H. Macromole-cules 1984, 17, 2812.

27. Lal, J.; Bansil, R. Macromolecules 1991, 24, 290.

1708 MATSUYAMA ET AL.