Porous biodegradable polymeric scaffolds prepared bythermally induced phase separation

Yoon Sung Nam, Tae Gwan ParkDepartment of Biological Sciences, Korea Advanced Institute of Science and Technology, 373-1 Kusong-dongYusong-gu, Taejon 305-701, South Korea

Received 3 December 1998; accepted 25 February 1999

Abstract: Poly(L-lactic acid) and its copolymers with D-lactic and glycolic acid were used to fabricate various porousbiodegradable scaffolds suitable for tissue engineering anddrug delivery based on a thermally induced phase separa-tion (TIPS) technique. A variety of parameters involved inTIPS process, such as types of polymers, polymer concen-tration, solvent/nonsolvent ratio, and quenching tempera-ture, were examined in detail to produce a wide array ofmicro- and macroporous structures. A mixture of dioxaneand water was used for a binary composition of solvent andnonsolvent, respectively. In particular, the coarsening effectof pore enlargement affected by controlling the quenchingtemperature was used for the generation of a macroporous

open cellular structure with pore diameters above 100 mm.The use of amorphous polymers with a slow cooling rateresulted in a macroporous open cellular structure, whereasthat of semicrystalline polymers with a fast cooling rate gen-erated a microporous closed cellular structure. The fabri-cated porous devices loaded with recombinant humangrowth hormone (rhGH) were tested for the controlled de-livery of rhGH, as a potential additional means to cell de-livery. © 1999 John Wiley & Sons, Inc. J Biomed Mater Res,47, 8–17, 1999.

Key words: biodegradable polymer; porous scaffolds; ther-mally induced phase separation

INTRODUCTION

Poly(lactic acid) (PLA) and poly(D,L-lactic-co-glycolic acid) (PLGA) have been widely used as three-dimensional (3D) polymer scaffolds for cell transplan-tation, since they are biodegradable and biocompat-ible.1 It is important that the scaffolds be highlyporous enough to allow a high density of cells to beseeded, and when the cell-device construct is im-planted in the body, they should permit the facile in-vasion of blood vessels for the supply of nutrients tothe transplanted cells.2 To fulfill these requirements,the polymer scaffolds should have several physicalcharacteristics such as high porosity, a large surfacearea, a large pore size, and a uniformly distributedand highly interconnected pore structure throughoutthe matrix.3,4 Until now, nonwoven polyglycolic acid(PGA) fiber mesh reinforced by physical bonding ofPLA and PLGA between the fibers has been the mostpopularly used scaffold for soft-tissue cell transplan-tation because of its highly porous and interconnected

structure.3,5,6 However, the PGA nonwoven mesh isnot suitable for the purpose of hard-tissue regenera-tion such as in bone, since it is mechanically fragile.

Several methods have been reported to fabricate po-rous scaffolds by using biodegradable PLA or PLGA.They are porogen leaching,5–7 emulsion freeze-drying,10 expansion in high pressure gas,11,12 3D print-ing,13 and phase separation techniques.14–16 The poro-gen-leaching method has been the best knownmethod, and involves the casting of a polymer/porogen composite membrane followed by aqueouswashing out of the incorporated porogen. Various po-rogens such as salts, carbohydrates, and polymers canbe used to produce porous materials. The salt-leachingtechnique was suitable for controlling pore sizes bychanging the size of salt particulates, but residual saltsremaining in the scaffolds, irregularly shaped pores,and their poor interconnected structure seem to beproblematic for cell seeding and culture. The emulsionfreeze-drying method is another approach for the pro-duction of porous scaffolds, but this method often re-sults in a closed cellular structure in the matrix. Theexpansion technique using a high pressured CO2 gasalso resulted in a closed pore structure inadequate forefficient cell seeding, and a combination technique ofgas-induced foaming and particulate leaching was re-

Correspondence to: T. G. ParkContract grant sponsor: Ministry of Science and Technol-

ogy, South Korea; Contract grant number: 97-NI-02-05-A-04

© 1999 John Wiley & Sons, Inc. CCC 0021-9304/99/010008-10

cently proposed to improve the pore structure.12 The3D printing technique using the principle of ink-jetprinting appears useful, while its application for tissueengineering is still in infancy.13

The phase separation technique is based on thermo-dynamic demixing of a homogeneous polymer-solvent solution into a polymer-rich phase and a poly-mer-poor phase, usually by either exposure of the so-lution to another immiscible solvent or cooling thesolution below a binodal solubility curve. Thermallyinduced phase separation (TIPS) uses thermal energyas a latent solvent to induce phase separation.17,18,27

The quenched polymer solution below the freezingpoint of the solvent is subsequently freeze-dried toproduce porous structure. One of the advantages ofthis technique is that various porous structures can beeasily obtained by adjusting various thermodynamicand kinetic parameters. The TIPS technique has beenused commercially to produce microporous mem-branes for filtration, but the pore size of the resultantmembrane is too small to be applied for cell seeding,which requires a pore diameter at least above 100 mmwith an open cellular morphology. The TIPS techniquehas been previously reported for the preparation ofporous PLA scaffolds intended to use as tissue scaf-folds.16 Although a wide array of microporous (1–10mm) isotropic morphologies were reported in re-sponse to a slight change in the TIPS parameters,whether they have macroporous and open cellularstructure throughout the matrix was not discussed.Large pore size and an open cellular structure are criti-cal parameters for cell seeding and neovascularizationwhen implanted in vivo.

In this study, we investigated TIPS technique to fab-ricate 3D porous biodegradable polymer scaffoldswith the aim of attaining macroporous structure withan open cellular morphology. To achieve this goal, wecarefully studied the late stage of the phase separa-tion.19–22,26 After the early-stage development of themicrostructure in a phase-separating polymer solu-tion, the system continues to change in response to itstendency to reduce the interfacial free energy betweenthe two phases already separated in the earlier stage.This process, called a coarsening effect, results in anincrease in the size of phase-separated droplets with aconcomitant reduction in their numbers. The main fo-cus of this study was to use the coarsening process ofthe phase separation for the preparation of macro-porous scaffolds still having an open cellular structureideal for cell seeding. For this purpose, several experi-mental parameters such as polymer type and concen-tration, solvent/nonsolvent ratio, and quenching tem-perature were varied to optimize the morphology ofbiodegradable polymer scaffolds suitable for tissue re-generation. In addition, recombinant human growthhormone (rhGH) was loaded into the scaffolds by animmersion-soaking method to evaluate the pore inter-

connectivity, as well as to demonstrate the possiblecapability of controlled drug delivery.

MATERIALS AND METHODS

Materials

Poly(L-lactic acid) (PLLA), poly(D,L-lactic acid) (PDLLA),and poly(D,L-lactic-co-glycolic acid, lactide:glycolide ratio 85:15 (PLGA) were purchased from Medisorb (Cincinnati, OH).Weight average molecular weights and polydispersity (PI)were determined by gel-permeation chromatography; Mw =59,000 and PI = 2.07 for PLLA, Mw = 74,500 and PI = 2.07 forPDLLA, and Mw = 74,700 and PI = 1.62 for PLGA. A phe-nogel 00H-0445-K0 column (300 × 7.8 mm) was used as asize exclusion column. Tetrahydrofuran (THF) was eluted asa mobile phase and the sample was detected by using anultraviolet (UV) detector at 220 nm. rhGH was obtained as agift from Dong-A Pharm. Co., Ltd. (South Korea). The mi-crobicinchoninic acid (BCA) assay kit was obtained fromPierce. 1,4-Dioxane, methanol, acetic acid, and other re-agents were of analytical quality.

Cloud point and sol/gel transitioncurve determination

The polymer solution was prepared at various concentra-tions in an 87/13 (v/v) mixture of 1,4-dioxane and water,and warmed up to at least 10°C above the expected cloudpoint. The cloud point and sol/gel transition temperatureswere then determined visually by slowly cooling the poly-mer solution by 1°C. The temperature was maintained con-stant for 1 h for each visual observation.

Scaffolds preparation

Processing condition 1

To observe the structure resulting from the nucleation andgrowth process of the polymer-rich phase, 1% (w/v) poly-mer concentration in a mixture of dioxane and deionizedwater was chosen. Water was added as a nonsolvent to en-courage phase separation. The volume ratio of dioxane andwater was set at 87:13. The clear polymer solution dissolvedabove the cloud point was poured into an aluminum foilmold (1 × 1 cm) which was fast-frozen in liquid nitrogentank or freezer (−15°C) and then incubated at those tempera-tures overnight. Solvent was removed by freeze-drying for 3days.

Processing condition 2

To compare the structures resulting from the spinodal de-composition and coarsening process, 500 mg of PLLA,PDLLA, or PLGA polymer dissolved in 5 mL of a mixture ofdioxane and water was warmed up above the cloud pointuntil the clear solution was formed. The volume ratios ofdioxane and water were varied at 84:16, 87:13, and 90:10. Thehomogeneously clear solutions were quenched under the

9POROUS BIODEGRADABLE POLYMERIC SCAFFOLDS

following conditions: liquid nitrogen bath, deep-freezer(−40°C), and freezer (−15°C), and then incubated overnightat the quenching temperature. Solvent was removed byfreeze-drying for 3 days.

Scanning electron microscopy (SEM) and mercuryintrusion porosimetry

The internal pore structures of the freeze-dried scaffoldswere observed using SEM (Philips 535M). The samples werefractured by a surgical blade and mounted on an aluminum

stub covered with a carbon adhesive and then coated withgold particles. Pore size distribution, total pore volume, sur-face area, and density of the scaffolds were determined bymercury intrusion porosimetry (Porous Materials, Ithaca,NY). The technique is based on the principle that the exter-nal pressure required to force a nonwetting liquid (mercury)into a pore, against the opposing force of the liquid surfacetension, depends on the pore size (the following Washburnequation)23:

P = −4s cos u

d

Figure 1. Schematic temperature-composition phase dia-gram of polymer solution. Figure 2. Cloud point curves and sol/gel transition curve

of the PLLA, PDLLA, and PLGA polymer solutions in amixture of dioxane and water with an 87/13 volume ratio.

Figure 3. SEM images of foam structures fabricated by using 1% (w/v) PLLA, PDLLA, and PLGA polymer solutions in amixture of 87/13 (v/v) dioxane/water. Polymer types: PLLA (a,b); PDLLA (c,d); PLGA (e,f). Quenching conditions: liquidnitrogen (a,c,e); −15°C (b,d,f).

10 NAM AND PARK

which describes the relationship between the applied pres-sure, P, and the pore size, to calculate the pore diameter, d,under the following assumptions: (a) the shape of the poresis a cylinder; (b) the contact angle of mercury onto pore wall,u, is 140°; and (c) the surface tension of mercury, s, is 480dyne/cm.

Protein loading, distribution analysis, andrelease experiment

Polymer scaffolds were cut into a dimension 5 × 5 × 4 mm.They were prewetted in pure ethanol and were stirred at 100rpm for 2 h using an orbital shaker; ethanol was replaced byexcess water with continuous shaking.24 rhGH was thenloaded into the prewetted scaffolds by dipping the scaffoldsinto the protein solution with a concentration of 4 mg/mL atroom temperature for 38 h. Loading quantities were deter-mined by measuring the residual amount of protein in the

solution. Protein-loaded scaffolds were then rinsed by dis-tilled water and put into the Coomassie blue solution, whichconsisted of 0.1% (w/v) Coomassie blue R-250, 45% (v/v)methanol, 45% (v/v) water, and 10% (v/v) glacial aceticacid. After incubation for several hours, Coomassie bluedestaining was accomplished by dipping the scaffolds intothe destaining solution consisting of 10% (v/v) methanol,10% (v/v) glacial acetic acid, and 80% (v/v) water. The invitro release experiment was accomplished in 33 mM phos-phate-buffered saline (PBS, pH 7.4, 0.1M NaCl) under staticconditions at 37°C. The incubation medium was replaced byfresh buffer at predetermined time intervals to maintain aconstant pH value. The released protein amount was deter-mined by the micro-BCA protein assay method.

RESULTS AND DISCUSSION

Figure 1 shows a schematic temperature-compo-sition phase diagram for a binary polymer/solvent

Figure 4. SEM images of the cross section of PLLA scaffolds as a function of coarsening time and solvent/nonsolventvolume ratio. The scaffolds were prepared by quenching a 10% (w/v) polymer solution under the following conditions: liquidnitrogen (a,d,g); −40°C (b,e,h); −15°C (c,f,i). The volume ratios of dioxane and water were 84/16 (a–c); 87/13 (d–f); 90/10 (g–i).

11POROUS BIODEGRADABLE POLYMERIC SCAFFOLDS

system. Above the binodal curve, a single polymersolution phase is formed, and if cooling below thecurve, polymer-rich and polymer-poor phases areseparated in a thermodynamic equilibrium state. Thespinodal curve is defined as the line at which the sec-ond-derivative Gibbs free energy of mixing is equal tozero, and it divides the two-phase region into an un-stable and a metastable region. If the system isquenched into the metastable region, phase separationoccurs in a nucleation and growth mechanism, leadingto a beadlike or isolated cellular structure. On theother hand, if the system temperature is quenched intothe unstable region, the phase separation takes placein a spinodal decomposition mechanism, resulting in amicroporous interconnected structure as reported.19,20

These events occur in the early stage of phase separa-tion, but in the later stage, the coalescence of phase-separated droplets continuously proceeds to minimize

the interfacial free energy associated with the interfa-cial area, which is called the coarsening process. Thiseffect has been scrutinized in the fabrication of syn-thetic membranes, because it is vitally important indetermining the final membrane morphology.21,22 Toattain macroporous scaffolds, it is highly desirable touse the coarsening effect, which induces the enlarge-ment of pores. The coarsening process, however, con-comitantly tends to generate more closed pores; thus,it is important to optimize various TIPS parameters toachieve a macroporous open cellular structure.

Various biodegradable polymers, PLLA, PDLLA,and PLGA (85/15), dissolved in a solvent/nonsolventmixture of dioxane/water were used for the genera-tion of liquid–liquid phase separation as reported.25

The reason for selecting dioxane as a solvent for bio-degradable polymers is that it has a low boiling point,101.1°C, with a relatively high melting temperature,

Figure 5. SEM images of the cross section of PLGA scaffolds. The scaffolds were prepared by quenching the 10% (w/v)polymer solution under the following conditions: liquid nitrogen (a,d,g); −40°C (b,e,h); −15°C (c,f,i). The volume ratios ofdioxane and water were: 84/16 (a–c); 87/13 (d–f); 90/10 (g–i).

12 NAM AND PARK

11.8°C, allowing easy removal by a freeze-dryingmethod. Figure 2 shows cloud points of various poly-mer solutions as a function of polymer concentration.Since the molecular weights of the polymers used inthis experiment were polydisperse, cloud point curveswere constructed instead of the binodal curve, whichis only applied to a polymer with monodisperse mo-lecular weight.26,27 Up to 25% (w/v) polymer concen-tration, the cloud point increases with increasing con-centration, but does not exhibit a critical polymer con-centration (fc) corresponding to an upper criticalsolution temperature (UCST). This means that thepolymer concentration range in this study was locatedin the left region of fc. The cloud point curves of thePDLLA and PLGA polymer solution were almostsimilar, but PLLA cloud points appeared about 20°Chigher than those of PDLLA or PLGA. This differenceis supposed to originate from the difference in thepolymer amorphous/crystallinity, since cloud pointswere independent of the polymer molecular weight.21

In addition, semicrystalline PLLA polymer solutionexhibited a gelation temperature curve appearing be-low the cloud temperature. This gelation behavior waspresumed to be derived from small crystallites ofPLLA, which were formed upon further slow coolingbelow the cloud point, and might play a role as physi-cal crosslinks in the phase-separated polymer-rich do-mains.26–28 From these different phase separation be-haviors, it is conceivable that PLLA would result indifferent morphologies from PDLLA or PLGA scaf-

folds under the same experimental conditions, whichwill be shown later.

Figure 3 shows a series of SEM pictures for variouspolymeric scaffolds fabricated based on a 1% (w/v)polymer concentration in a mixture of 87/13 (v/v)dioxane/water system. They were obtained by warm-ing the polymer solution above the cloud point tem-perature, subsequent rapid quenching at liquid nitro-gen temperature or −15°C, annealing overnight, andthen freeze-drying for 3 days. The scaffolds obtainedfrom the liquid nitrogen quenching resulted in astringy and threadlike structure, while those from−15°C quenching yielded a beady structure regardlessof the polymer type. The resultant microporous struc-tures, which were mechanically fragile, suggested thatat 1% (w/v) polymer concentration, both quenchingtemperatures permitted the polymer solution to be lo-cated within a metastable region. This is becausebeady/stringy structures are typically produced bythe nucleation and growth mechanism of the polymer-rich phase.

TABLE ICharacterization of Porous PDLLA Scaffolds by Mercury

Intrusion Porosimetry

Liquid Nitrogen −15°C

Porosity (%) 80.4 90.3Surface area (m2/g) 4.35 0.79Pore volume (cm3/g) 4.46 7.89

Figure 6. SEM images of the cross-section of PDLLA scaffolds. The scaffolds were prepared by quenching the 10% (w/v)polymer solution under the following conditions: liquid nitrogen (a,d); −40°C (b,e); −15°C (c,f). The volume ratios of dioxaneand water were: 87/13 (a–c); 90/10 (d–f).

13POROUS BIODEGRADABLE POLYMERIC SCAFFOLDS

Figure 4 shows SEM pictures of different morpholo-gies of 10% (w/v) PLLA dissolved in varying volumeratios of dioxane and water at different quenchingtemperatures. When the polymer solution wasquenched by liquid nitrogen, a microcellular porestructure was formed, suggesting that the quenchedstate of the polymer solution was located within theunstable region. The cell sizes in the scaffolds dramati-cally increased with increasing quenching tempera-ture, while they tended to become a more closed cel-lular structure. This can be explained by the aforemen-tioned coarsening effect, in which phase-separatedpolymer-poor (solvent-rich) droplets coalesce to formlarger droplet domains in the later stage of phaseseparation. Since the coarsening effect is a kinetic be-havior directed toward reaching the thermodynamicminimization of interfacial energy, one can manipu-late the cellular structure by arresting the phase sepa-ration process. In this study, coarsening time scaleswere expected to vary by adopting different quench-ing temperatures. The coarsening time decreased withdecreasing quenching temperature.

It should be mentioned that even though the liquidnitrogen–quenched scaffolds exhibited a microcellular

morphology, their closed cell structures were some-what different from the typical bicontinuous morphol-ogy produced by the spinodal decomposition mecha-nism, as shown in Figure 1.20–22,25 As the clear poly-mer solution was cast into a brass mold equilibratedwith liquid nitrogen temperature, it is conceivable thatan instantaneous heat gradient, although of presum-ably very short duration, developed in the directionfrom the contacting mold surface to the core region ofthe quenched polymer sample. This would create dif-ferent phase-separated morphologies from the centerto the surface in the resultant scaffolds by exertinglocally different phase-arresting time scales on the spi-nodal decomposition. Thus, the coarsening effect, un-til the final phase arresting occurred, was likely toplay a critical role in the central region from the sur-face in those time scales. This effect was expected to bemore pronounced in a large dimensional sample.From this reason, the prepared scaffolds had a more orless microporous surface skin structure, whereas thecentral region had a more macroporous structure be-cause of the locally occurring coarsening effect. Ac-cordingly, it should be noted that the term “quenchingtemperature” used here does not mean the tempera-

Figure 7. Pore size distribution of the PDLLA scaffolds. SEM pictures were shown to compare directly with the resultsdetermined by mercury intrusion porosimetry. (a) and (b) are the same as (d) and (f) of Figure 5, respectively. (b8) is alower-magnification (original ×70) picture of (b).

14 NAM AND PARK

ture at which the structure of phase-separated poly-mer solution was actually quenched, but representsonly the temperature to which the samples were ex-posed.

The increasing amount of water in the solvent mix-ture tended to generate large cellular pore sizes. Thiswas likely to be caused by the fact that as the nonsol-vent volume fraction increases, a weaker polymer–diluent interaction might induce the formation ofpolymer poor phase with greater droplet domains.29,30

In addition, it was reported that the gradual additionof water in a PLLA–dioxane–water system raised thecloud point curve and decreased intrinsic viscosity.31

In our system, the cloud point curve of 87/13 (v/v)dioxane/water system was located about 15°C higher

than that of 90/10 (v/v) dioxane/water system as re-ported.16 Thus, the enlarged pore sizes with increasingwater content were most likely due to the combinedeffects of weaker polymer–diluent interaction, largerquenching depth, and lowered viscosity in the cosol-vent system. The PLLA scaffolds prepared using 90/10 dioxane/water did not exhibit a well-defined lacystructure different from the other two scaffolds pre-pared by the 84/16 and 87/13 solvent/nonsolventcompositions. This trend can be understood by thepreferred occurrence of solid–liquid demixing involv-ing crystallization, gelation, and vitrification ratherthan liquid–liquid demixing in the spinodal decompo-sition region because of enhanced interaction betweenpolymer and diluent system.29,30

Figure 8. Cross-sectional visualization of protein distribution in the porous scaffolds. Protein was stained with a Coomassieblue dye.

15POROUS BIODEGRADABLE POLYMERIC SCAFFOLDS

Figures 5 and 6 show SEM pictures of various mor-phologies of PLGA and PDLLA scaffolds. Quenchingat a liquid nitrogen temperature produced micro-porous open cellular structures as a result of the spi-nodal decomposition mechanism. However, increas-ing the quenching temperature dramatically inducedthe coarsening effect of pore enlargement with closingcellular structures. This effect was particularly evidentfor PLGA and PDLLA at 87/13 and 84/16 solvent/nonsolvent volume ratios, respectively. As the greateramount of water incorporated into the solvent mix-ture, closed cellular macropores up to 1000 mm in di-ameter could easily be seen. The PLGA and PDLLAsamples at the 90/10 solvent/nonsolvent volume ratiochanged their morphologies from a microporous (<10mm) to a macroporous (∼100 mm) structure whilekeeping an open cellular architecture. The resultantmacroporous morphology seemed to be related to thecoarsening effect in the later stage of liquid–liquidphase separation in a relatively viscous state plus thecrystallization effect of solvent/nonsolvent mixtureused. The crystallization effect, typical of solid–liquidphase separation, occurs when dioxane is used as amajor solvent in the phase separation process.16,25 Thedata of porosity, surface area, and pore volume for thetwo PDLLA scaffolds prepared by quenching at liquidnitrogen temperature and −15°C are shown in Table I.Figure 7 shows the pore size distribution of the twoscaffolds as determined by mercury intrusion poro-simetry. The scaffold fabricated by quenching at liquidnitrogen temperature showed a unimodal pore sizedistribution centered about ∼3 mm, whereas the scaf-fold prepared at −15°C exhibits trimodal distributioncentered at 20, 45, and 170 mm. The results are in goodagreement with those observed in the SEM pictures.The multidistribution of pore sizes clearly suggests

that the coarsening process induced by adjusting thequenching temperature played a critical role in form-ing a macroporous open cellular structure sufficientfor cell seeding and tissue regeneration. To elucidatewhether the 10% PDLLA scaffold fabricated by using90/10 dioxane/water volume ratio had an open cel-lular structure, rhGH was loaded into the scaffold byan immersion-soaking method and then stained witha Coomassie blue dye for cross-sectional visualization,as shown in Figure 8. The macroporous scaffold washomogeneously stained, indicating that it truly had anopen cellular structure. On the other hand, the PLLAscaffold with an apparent macroporous closed cellularstructure and the PDLLA scaffold appearing to have amicroporous open cellular structure, as observed intheir SEM pictures, did not show evidence of proteinstaining in the central region of their devices. Thisexperiment clearly indicates that they do not havemacroporous open cellular structures enough for thepenetration of rhGH through the pores. The loadingcapacity of rhGH with the open cellular scaffolds wasmuch higher than that of the closed cellular foams, butthe release profiles of rhGH from the three scaffoldswere similar: an initial rapid release up to 36–52% andsubsequent very slow release, as shown in Figure 9.The very slow release was likely due to rhGH adsorp-tion onto the surface of the scaffolds with a large sur-face area.32,33

CONCLUSION

Macroporous open cellular scaffolds with pore di-ameters above 100 mm can be readily fabricated byjudiciously using the coarsening process in the laterstage of thermally induced phase separation. Thesebiodegradable scaffolds could be potentially appliedfor tissue regeneration that requires effective cell seed-ing through the macropores in the matrix.

References

1. Athanasiou KA, Niederauer GG, Agrawal CM. Sterilization,toxicity, biocompatibility and clinical applications of polylacticacid/polyglycolic acid copolymers. Biomaterials 1996;17:93–102.

2. Mikos AG, Sarakinos G, Leite SM, Vacanti JP, Langer R. Lam-inated three-dimensional biodegradable foams for use in tissueengineering. Biomaterials 1993;14:323–330.

3. Mikos AG, Bao Y, Cima LG, Ingber DE, Vacanti JP, Langer R.Preparation of poly(glycolic acid) bonded fiber structures forcell attachment and transplantation. J Biomed Mater Res1993;27;183–189.

4. Cima LG, Vacanti JP, Vacanti C, Ingber DE, Mooney DJ, LangerR. Tissue Engineering by cell transplantation using biodegrad-able polymer substrates. J Biomech Eng 1991;113:143–151.

5. Freed L, Marquis JC, Nohria A, Emmanual J, Mikos AG. Neo-

Figure 9. In vitro release profiles of rhGH from the porousscaffolds. (Inset) Replot of rhGH release in cumulative re-lease amount.

16 NAM AND PARK

cartilage formation in vitro and in vivo using cells cultured onsynthetic biodegradable polymers. J Biomed Mater Res 1993;27:11–23.

6. Freed L, Vunjak-Novokovic G, Biron RJ, Eagles DB, LesnoyDC, Barlow SK, Langer R. Biodegradable polymer scaffolds fortissue engineering. Biotechnology 1994;12:689–693.

7. Mikos AG, Sarakinos G, Leite SM, Vacanti JP, Langer R. Lam-inated three-dimensional biodegradable foams for use in tissueengineering. Biomaterials 1993;14:323–330.

8. Mikos AG, Thorsen AJ, Czerwonka LA, Bao Y, Langer R, Win-slow DN, Vacanti JP. Preparation and characterization ofpoly(L-lactic acid) foams. Polymer 1994;35:1068–1077.

9. Mooney DJ, Sano K, Kaufmann PM, Majahod K, Schloo B,Vacanti JP, Langer R. Long-term engraftment of hepatocytestransplanted on biodegradable polymer sponges. J BiomedMater Res 1997;37:413–420.

10. Whang K, Thomas CH, Healy KE, Nuber G. A novel method tofabricate bioabsorbable scaffolds. Polymer 1995;36:837–842.

11. Mooney DJ, Baldwin DF, Suh NP, Vacanti JP, Langer R. Novelapproach to fabricate porous sponges of poly(D,L-lactic-to-glycolic acid) without the use of organic solvents. Biomaterials1996;17:1417–1422.

12. Harris LD, Kim BS, Mooney DJ. Open pore biodegradable ma-trices formed with gas foaming. J Biomed Mater Res 1998;42:396–402.

13. Park A, Wu B, Griffith LG. Integration of surface modificationand 3D fabrication techniques to prepare patterned poly(L-lactide) substrates allowing regionally selective cell adhesion. JBiomater Sci Polym Ed 1998;9:89–110.

14. Lo H, Ponticiello MS, Leong KW. Fabrication of controlledrelease biodegradable foams by phase separation. Tissue Eng1995;1:15–28.

15. Lo H, Kadiyala S, Guggino E, Leong KW. Poly(L-lactic acid)foams with cell seeding and controlled-release capacity. JBiomed Mater Res 1996;30:475–484.

16. Schugens C, Maguet V, Grandfils C, Jerome R, Teyssie P. Poly-lactide macroporous biodegradable implants for cell transplan-tation. 2. Preparation of polylactide foams by liquid–liquidphase separation. J Biomed Mater Res 1996;30:449–461.

17. Mulder M. Basic principles of membrane technology. Dor-drecht, The Netherlands: Kluwer Academic; 1996. p 71–156.

18. Bansil R, Liao G. Kinetics of spinodal decomposition in homo-polymer solutions and gels. Trends Polym Sci 1997;5(5):146–154.

19. Siggia ED. Late stage of spinodal decomposition in binary mix-tures. Phys Rev A 1979;20:595.

20. Song SW, Torkelson JM. Coarsening effects on microstructureformation in isopycnic polymer solutions and membrane pro-duced via thermally induced phase separation. Macromol-ecules 1994;27:6389–6397.

21. Song SW, Torkelson J. Coarsening effects on the formation ofmicroporous membranes produced via thermally inducedphase separation of polystyrene-cyclohexanol solutions. JMembr Sci 1995;98:209–222.

22. Guo HF, Laxminarayan A, Cabeba GT, Solc K. Morphologicalstudies of late-stage spinodal decomposition in polystyrene-cyclohexanol system. J Appl Polym Sci 1995;55:753–759.

23. Ritter HL, Drake LC. Pore-size distribution in porous materi-als. 1. Pressure porosimeter and determination of completemacropore-size distribution. Ind Eng Chem 1945;17:782–786.

24. Mikos AG, Lyman MD, Freed LE, Langer R. Wetting of poly(L-lactic acid) and poly(D,L-lactic-co-glycolic acid) foams for tissueculture. Biomaterials 1994;15:55–58.

25. Nam YS, Park TG. Biodegradable polymeric microcellularfoams by modified thermally induced phase separationmethod. Biomaterials. In press.

26. Tsai FJ, Torkelson JM. Roles of phase separation mechanismand coarsening in the formation of poly(methyl methacrylate)asymmetric membranes. Macromolecules 1990;23:775–784.

27. van de Witte P, Dijkstra PJ, van den Berg JWA, Feijen J. Phaseseparation processes in polymer solutions in relation to mem-brane formation. J Membr Sci 1996;117:1–31.

28. Coombes AGA, Heckman JD. Gel casting of resorbable poly-mers. Biomaterials 1992;13:217–224.

29. Lloyd DR, Kinzer KE, Tseng HS. Microporous membrane for-mation via thermally induced phase separation. I. Solid–liquidphase separation. J Membr Sci 1990;52:239–261.

30. Lloyd DR, Kim SS, Kinzer KE. Microporous membrane forma-tion via thermally induced phase separation. II. Liquid–liquidphase separation. J Membr Sci 1991;64:1–11.

31. van de Witte P, Dijkstra PJ, van den Berg JWA, Feijen J. Phasebehavior of polylactides in solvent–nonsolvent mixtures. JPolym Sci Polym Phys 1996;34:2553–2568.

32. Crotts G, Sah H, Park TG. Adsorption determines in-vitro pro-tein release rate from biodegradable microspheres: Quantita-tive analysis of surface area during degradation. J Control Rel1997;47:101–111.

33. Schmitz JP, Zardeneta G, Agrawal CM, Lee T, Athanasiou K,Milam SB. Protein interactions with polylactide-polyglycolideparticles and scaffolds. Tissue Eng 1997;3:257–266.

17POROUS BIODEGRADABLE POLYMERIC SCAFFOLDS