Langmuir 2000, 16, 10463-10470
10463
Characterization of Porous Polylactide Foams by Image Analysis and Impedance Spectroscopy V. Maquet,† S. Blacher,*,‡ R. Pirard,‡ J.-P. Pirard,‡ and R. Je´roˆme† Center for Education and Research on Macromolecules and Department of Chemical Engineering, University of Lie` ge, Sart-Tilman, B6, 4000 Lie` ge, Belgium Received May 8, 2000. In Final Form: September 12, 2000 The texture of highly porous polylactide foams prepared by freeze-drying of poly-L-lactide (L-PLA) and poly-DL-lactide (DL-PLA) solutions of various concentrations was investigated by two novel methodologies, image analysis and impedance spectroscopy. Image analysis of scanning electron micrographs of transverse cross-sections at two different magnifications gave information on both the macroporosity (1 µm < width < 10 µm) and ultramacroporosity (width > 10 µm). Impedance spectroscopy was used to investigate the transport properties of the three-dimensional porous matrices by measurement of ionic conduction. Image analysis showed that: (a) the macroporosity, which mainly contributes to porosity, is independent of the sample composition; (b) when the concentration of the polymer solution is increased, the density of the ultramacropores decreases and their average diameter increases; (c) the distribution of the ultramacropores is more homogeneous in the semicrystalline L-PLA foams than in the amorphous DL-PLA counterparts, in which the ultramacropores tend to make clusters. The dielectric properties changed at low frequency, in relation to modifications in the ultramacroporosity. Ultramacropores of the L-PLA foams were found to be more open and more sensitive to the concentration of the polymer solutions compared with DL-PLA. Expectedly, the mechanical properties of the PLA foams changed with the structure of the ultramacroporous network. These results encourage further investigations on the texture of porous supports, to collect pertinent information on the physical macro- and ultramacroenvironment in which cells will reside.
Introduction Porous biodegradable synthetic and natural polymers are currently tested as implants for the regeneration of damaged and diseased tissues.1-3 They can be used either as an analogue of the extracellular matrix (ECM) or as substrate for cell transplantation. In this case, autologous cells collected from the patient tissues are seeded and expanded in vitro before being reimplanted in vivo for promoting tissue regeneration. This strategy is very promising for repairing a great number of organs and tissues, including liver, pancreas, intestine, cartilage, bone, skin, and neural tissues.4 One of the key features of tissue engineering is the scaffold porosity. The porous structure can indeed contribute to a new ECM and support angiogenesis and cell proliferation. The scaffold should ideally consist of a large population of pores. Indeed, pores larger than cell dimension (ca. 10 µm) are essential for sustaining cell infiltration, whereas pores smaller than 10 µm may contribute to cell attachment and create a large surface area for the growth of the tissue layer. According to the International Union of Pure and Applied Chemistry, porous materials with w > 50 nm are referred to as macroporous. In this work, pores with a width larger than the cell dimension (w > 10 µm) will be designated as ultramacropores and macropores will referred to pores smaller than 10 µm. The pore morphology is another key characteristic when the regeneration of highly organized * Author for correspondence. † Center for Education and Research on Macromolecues. ‡ Department of Chemical Engineering. (1) Frontiers in Tissue Engineering; Patrick, C. W., Mikos, A. G., McIntire, L. V., Eds.; Pergamon Press: Elsmford, NY, 1998. (2) Principles of Tissue Engineering; Lanza, R. P., Langer, R., Chick, W. L., Eds.; R G Landes Co. and Academic Press: Austin, 1997. (3) Polymers for Tissue Engineering; Shoichet, M. S., Hubbel, J. A., Eds.; VSP: Utrecht, The Netherlands, 1998. (4) Maquet, V.; Je´roˆme, R. In Porous Materials for Tissue Engineering; Trans Tech Publications: Uetikon-Zuerich; 1997; Vol. 250, pp. 1542.
tissues such as nerves is concerned. We previous reported on a thermally induced phase separation (TIPS) technique for the preparation of porous resorbable polylactide (PLA) matrices.5,6 Briefly, the TIPS technique consists of the freeze-drying of a PLA solution in a suitable solvent. Phase separation takes place on cooling, and removal of the solvent results in a porous morphology. Depending on the critical solution temperature (Tc) and the freezing point of the solvent (Tf), either a liquid-liquid phase separation takes place (Tc > Tf)5 or the solvent crystallizes (Tc < Tf), and this solid-liquid phase separation is responsible for the final morphology.6 In this case, the progress of the crystallization front line of the solvent dictates the main orientation of the pores, the long axis of which being parallel to the cooling direction.6 This paper will deal with foams prepared by solidliquid phase separation of PLA solutions in 1,4-dioxane. Nucleation and growth of the solvent-rich phase results in an oriented network composed of ultramacropores of ca. 100 µm diameter dispersed in a 10 times less porous matrix (i.e., median pore size of 10 µm).5 The orientation of the ultramacropores is parallel to the cooling direction and depends on both the polymer concentration and polymer crystallinity. These highly oriented porous scaffolds have a great potential as temporary implants for supporting axonal regeneration in the case of lesions of rat sciatic nerve7 and spinal cord.8 A deeper characterization of the macro- and ultramacrostructure of this class of materials is desirable to understand better the interdependence of the TIPS process, the final properties of (5) Schugens, C.; Maquet, V.; Grandfils, C.; Je´roˆme, R.; Teyssie´., P. J. Biomed. Mater. Res. 1996, 30, 449. (6) Schugens, C.; Maquet, V.; Grandfils, C.; Je´roˆme, R.; Teyssie´., P. Polymer 1996, 37, 1027. (7) Maquet, V.; Martin, D.; Malgrange, B.; Franzen, R.; Schoenen, J.; Moonen, G.; Je´roˆme, R. J. Biomed. Mater. Res. 2000, 52, 639. (8) Maquet, V.; Martin, D.; Scholtes, F.; Franzen, R.; Schoenen, J.; Moonen, G.; Je´roˆme, R. Biomaterials, in press.
10.1021/la000654l CCC: $19.00 © 2000 American Chemical Society Published on Web 11/28/2000
10464
Langmuir, Vol. 16, No. 26, 2000
Maquet et al.
Table 1. Composition of Polymer Foams foam DL-PLA
1 5 DL-PLA 10 L-PLA 1 L-PLA 5 L-PLA 10 DL-PLA
resomer
polymer concentration
R206 R206 R206 L206 L206 L206
1 w:v % 5 w:v % 10 w:v % 1 w:v % 5 w:v % 10 w:v %
the freeze-dried matrices, and their ability to support tissue repair. Traditionally, mercury porosimetry is used to measure the porous volume and the pore size distribution of porous matrices. Nevertheless, there is experimental evidence that these matrices are shrunk, that is, compacted and not intruded by mercury. This behavior has been already observed in other systems presenting a high porosity.9,10 In this study, image analysis and impedance spectroscopy are discussed as nondestructive methods for quantification of PLA porosity. For this purpose, image analysis of two-dimensional matrix transverse cross-sections will be carried out. The transport properties of the three-dimensional structures will be assessed by the ionic conduction measured by the application of an alternative current to water-saturated polymer matrices. The mechanical properties of these foams will also be reported. Basic relations between structure and porosity of the freeze-dried foams and their mechanical properties will be searched for in the case of amorphous DL-PLA and semicrystalline L-PLA matrices. Materials and Methods DL-PLA, or Resomer R206 (Mn: 50 000), and L-PLA, or Resomer L206 (Mn: 60 000, 70% of crystallinity), were supplied by Boehringer Ingelheim (Germany). Polymer matrices were prepared by freeze-drying, as previously reported,5,6 from 1, 5, and 10 w:v % polymer solution in 1,4 dioxane, leading to the six foams listed in Table 1. Image Analysis. The foams were cut parallel to the surface and the transverse cross-sections were sputtered with platinum for 120 s under argon atmosphere. Samples were observed with a JEOL JSM-840 A at an accelerating voltage of 20 kV. Scanning electron microscopy (SEM) micrographs magnified ×10 and ×100, respectively, allowed the ultramacropores (ca. 100 µm width) and the macropores (ca. 10 µm width) to be discriminated properly, and they were analyzed by image analysis. Image treatment and statistic analysis were performed with a WorkStation Sun SPARC30, and the software Visilog5.0 from Noesis. Images were digitized on a matrix of 1024 × 1024 pixels with 256 Gy levels. Five images of different areas of the same sample were analyzed. Impedance Spectroscopy. A 1-cm3 specimen of each foam was degassed in distilled water and placed in a polystyrene UV type cell between two stainless steel electrodes. The cell was filled with distilled water and ionic AC electrical conductivity was measured by the frequency analyzer (Schlumberger) SI 1255 analyzer, supplied with the Chelsea dielectric interface (model CDI 4t) probing the 10-3-106 Hz frequency range. Mechanical Properties. Foams were cut with a razor blade into 1 × 1 × 1 cm pieces and analyzed with a texturometer (Stevens Farnell QTS-25, Essex, U.K.). Load versus time plots were recorded for two successive compression cycles. During these cycles, a spherical probe penetrated the polymer sample over a depth of 5 mm at a constant crosshead speed (forth and back) of 30 mm min-1. The time required for the probe to reach the sample surface after the first compression was proportional to the “elasticity” of the material. The equipment provided hardness, cohesiveness, and modulus values. Hardness is defined as the maximum load (g) during the first compression. Cohesiveness,
(9) Pirard, R.; Blacher, S.; Brouers, F.; Pirard, J.-P. J. Mater. Res. 1995, 10, 2114. (10) Pirard, R.; Blacher, S.; Sahouli, B.; Pirard, J.-P. J. Colloid Interface Sci. 1999, 217, 216.
Figure 1. SEM transverse cross-sections of L-PLA (a-d) and DL-PLA (e-h) foams at different magnifications: (a,e) ×10, (b,f) ×30, (c,g) ×100, (d,h) ×200. is the ratio of positive work during the second compression to the positive work during the first compression. A ratio of 1 indicates a completely reversible deformation. The modulus is defined as the gradient of positive work between 20 and 80% of the first peak load (g s-1). Texture analysis was performed for both the transverse and longitudinal cross-sections, except for foams prepared from 1 w:v % polymer concentration in which no pore orientation was observed.
Results The PLA matrices used in this study show two levels of porosity, that are, ultramacropores of ca. 100 µm width distributed in a macroporous matrix with pore size in the 10 µm range. This morphology is observed for both the L-PLA and DL-PLA foams prepared from 5 and 10 w:v % solutions. Conversely, no ultramacroporosity is observed for the foams prepared from dilute polymer solutions in dioxane (i.e., 1%). In this case, the depth resolution of the transverse cross-sections is not good enough for reliable two-dimensional image analysis to be performed. Image Analysis Characterization. Figure 1 shows typical transverse cross-sections of foams at various magnifications. The quantitative analysis of this type of morphology requires knowledge of both the density, size distribution, and dispersion of the ultramacropores and the density and size distribution of the macropores. Gray-
Porous Polylactide Foams
Figure 2. Extraction of the ultramacropore structure from the L-PLA foam (10%).
Figure 3. Extraction of the macropore structure from the L-PLA foam (10%).
level image transformations and binary image processing are the preliminary steps of the procedure.11,12 Figures 2 and 3 illustrate how the ultramacropores and the macropores have been extracted from the images. Processing Image Binarization. Figure 2 illustrates the main steps for the extraction of the ultramacroporosity. The original image (Figure 2a) is first treated by a (11) Serra, J. Image Analysis and Mathematical Morphology; New York, 1982. (12) Coster, M.; Chermant, J. L. Pre´ cis d’Analyse d’Images; CNRS: Paris, 1985.
Langmuir, Vol. 16, No. 26, 2000 10465
morphologic low-pass filter (closing transformation) to erase the macropores (Figure 2b). Then a threshold transformation leads to a binary image (Figure 2c). Finally, the ultramacropores are converted into measurable white pixels on a black pixel background, followed by opening transformation to eliminate artifacts, and image reconstruction from opening transformation image markers (Figure 2d). Concerning the macropores, the original image (Figure 3a) is first treated by a median filter to smooth small artifacts, followed by a gray-level thinning transformation to delimit the pore edges (Figure 3b). Threshold transformation into a binary image, image inversion, and opening transformation to eliminate small artifacts are then carried out as before (Figure 3c), followed by subdivision of the image into zones surrounding each pore (Figure 3d). The edges are then delimited by thinning transformation, and the image is inverted, which converts pores into measurable white pixels on a black pixel background (Figure 3e). Finally, each pore is identified by its gray level (Figure 3f). It must be noted that most of these transformations are parameter dependent. As a rule, if all the images are digitized under the same conditions, parameters are fixed once and for all, and the image binary processing is then completely automatic. In this study, the processing parameters have been changed for each set of images, because brightness and contrast were not the same. Measurements. From the images with a ×10 magnification, three characteristic features have been calculated: (1) Ultramacropore density; δum ) number of pixels characteristic of the ultramacropores/number of pixels of the whole image. (2) Equivalent circular diameter of the ultramacropores; D ) x4A/π, where A ) number of pixels for each pore multiplied by a calibration constant square. (3) The heterogeneous spatial distribution of the ultramacropores (named phase X) has been underscored by the covariance function C(x,λ).11 The covariance C(x,λ) of a stationary random function f(X) is defined as the probability for two points x and x + λ to belong to the same phase X. C(x,λ) has the following properties: C(x,λ) ) p, where p ) Prob{x∈X}, C(x,0) > C(x,λ), C(x,λ) f p2 when λ f ∞. The variation of C(x,λ) from p to p2 as a function of the modulus and the λ direction allow the dispersion of the phases to be studied. At constant phase concentration, a larger slope of C(x,λ) near the origin indicates a finer structure. Oscillations in the C(x,λ) curve indicate the existence of a structure with characteristic length (sizes, periodicity), and the directional character of C(x,λ) allows detection of anisotropy in the phase distribution. The covariance function has been calculated from the binary images in the two directions x and y, according to the algorithm12
C(λ) )
A(Mb∩Mb′) A(Z∩Z′)
where Mb is the binary image studied, Z the digital mask in which the covariance of the digital set Mb is measured, and Mb′ and Z′ are the values of Mb and Z after translation by a distance λ. The average distance between ultramacropores (λh) has been determined. From images recorded with a ×100 magnification, the macropore density (δm) has been calculated, where δm ) number of pixels characteristic of the macropores/number of pixels of the whole image. Because of the irregular shape of the macropores, calculation of an equivalent circular
10466
Langmuir, Vol. 16, No. 26, 2000
Maquet et al.
Table 2. Image Analysis for foam name
λh (mm)
δm
A (µm2) sk, k
porositya
0.108 ( 0.023 1.30, 2.33 0.135 ( 0.027 1.01, 2.00 0.097 ( 0.021 1.91, 8.32 0.116 ( 0.028 0.58, -0.17
0.32 ( 0.12
0.67 ( 0.08
0.89
0.80 ( 0.25
0.68 ( 0.07
0.26 ( 0.21
0.70 ( 0.07
0.52 ( 0.13
0.70 ( 0.05
0.092 ( 0.081 1.41, 1.48 0.080 ( 0.068 1.74, 3.04 0.066 ( 0.067 2.27, 6.31 0.060 ( 0.055 2.06, 5.43
1 5
0.087 ( 0.004
DL-PLA
10
0.062 ( 0.006
L-PLA
1 5
0.108 ( 0.003
L-PLA
10
0.064 ( 0.006
L-PLA
a
and L-PLA Foams
D (mm) sk, k
DL-PLA
DL-PLA
DL-PLA
δum
0.84 0.91 0.77
As measured by mercury porosimetry.1
Figure 4. Diameter size distribution of the ultramacropores for (a) the DL-PLA foams and (b) the L-PLA foams. Area size distribution of the macropores for (c) the DL-PLA foams and (d) the L-PLA foams. All the histograms have been constructed with 500 porosity data grouped in 20 classes for the diameter size distributions and in 30 classes for the area size distributions.
diameter is irrelevant. Then, the macropore area distribution (A) has been measured. Mean values and standard deviations for δum, D, λh, δm and for the area of the macropores, A, are reported in Table 2. The third (sk) and the fourth (k) moments of the D and A distributions are also reported. For the sake of comparison, the total porosity as measured by mercury porosimetry is reported in the last column of Table 2. Comparison of δum and δm shows that essentially the macropores contribute to the foam porosity. The density of the macropores (δm) is largely independent of the sample composition, whereas the density of the ultramacropores (δum) decreases with the solution concentration (from 5 to at 10 w:v %), whatever the PLA type. Finally, on the
assumption that the two-dimensional porosity can be extrapolated to the three-dimension space, a combination of δum and δm systematically leads to a total porosity smaller than that one measured by mercury porosimetry, although the general trend is the same. Distributions of D are shown in Figure 4a and b. For amorphous DL-PLA samples, D distribution moves toward larger values upon increasing concentration, although the general shape is maintained (small variation of sk and k, see Table 2). In contrast, D distribution for the semicrystalline L-PLA samples changes from a rather broad distribution (sk = 1.9, k = 8.3) to a quasi-Gaussian one (sk = 0.6, k = -0.17), with a minor modification of the mean value. The A distribution of the macropores (Figure
Porous Polylactide Foams
Langmuir, Vol. 16, No. 26, 2000 10467
Figure 5. Covariance function of the ultramacropore structure for the L-PLA foams (a) L-PLA 5, (b) L-PLA 10 and DL-PLA foams (c) DL-PLA 5, (d) DL-PLA 10.
4c and d) is very broad and asymmetric for the two types of foams (Table 2), particularly for the L-PLA foams (higher values of sk and k). Upon increasing concentration, the A size distribution for the DL-PLA foams become more asymmetric and broader, and the mean value of A decreases slightly. The mean value of A for L-PLA foams does not change, whereas the broadness and asymmetry of the distribution tend to decrease. Finally, the mean interdistance of the ultramacropores increases with the polymer concentration. The covariance function for the ultramacropores (Figure 5), measured for the transverse cross-sections at magnification ×10, indicates that: (a) The covariance functions calculated for L-PLA in the two directions x, y do not show any significant difference. At long distances, the curve shows very damped oscillations, in line with an isotrope and almost homogeneous distribution of the ultramacropores. (b) Some anisotropy is observed for DL-PLA mainly at 10 v:w % concentration (DL-PLA 10). At long distances, mainly at this concentration, an oscillating behavior is observed, and, at short distances, a crossover is observed for the DL-PLA 5 and DL-PLA 10 foams, which is an indication of pore clustering. Dielectric Properties. It is well-known that the conducting properties of a liquid in a porous medium
provide information on the pore geometry and the pore surface area. Indeed, both the motion of free carriers and the polarization of the pore interfaces contribute to the total conductivity. The main study in this field deals with the frequency dependence of conductivity for water or brine-saturated rocks13 and porous glasses.14-16 Let ′ and σ′ be the dielectric constant and the conductivity, respectively. Dielectric properties are usually expressed by the frequency-dependent real and imaginary components of the complex dielectric permittivity:
*(ω) ) ′(ω) - i′′(ω) ) ′(ω)0 -
iσ′(w) ω
where i ) x-1, 0 is the dielectric permittivity of the vacuum, f is the frequency in Hz, and ω ) 2πf. Experimental data are usually shown as a log-log plot of ′ and ′′ against frequency. Alternatively, ′ can be plotted against ′′, which is referred to as the Cole-Cole plot. (13) Hilfer, R. Phys. Rev. B 1991, 44, 60. (14) Pissis, P.; Anagnostopolou-Konsta, A.; Apekis, L.; DaoukakiDiamanti, D.; Christodoulides, C. J. Noncryst. Solids 1991, 131, 1174. (15) Pissis, P.; Laudat, J.; Daoukaki, D.; Kyritsis. A. J. Non-Cryst. Solids 1994, 171, 201. (16) Gutina, A.; Axelrod, E.; Puzenko, A.; Rysiakiewitz-Pasek, A.; Kozlovich, N.; Feldman. Y. J. Non-Cryst. Solids 1998, 235-237, 302.
10468
Langmuir, Vol. 16, No. 26, 2000
Maquet et al.
Several models have been proposed to interpret the frequency dependence of the permittivity data. The first one, proposed by Debye,17 considers an almost ideal situation with a single relaxation time (τ). The permittivity is then expressed as
* ) ∞ +
(0 - ∞) (1 + iωτ)
where 0 and ∞ are, respectively, the dc and the infinite frequency dielectric constant. In a Debye process, a maximum in ′′ is observed at f ) (2πτ)-1, in a log-log plot of ′′ versus f. In a Cole-Cole plot, a semicircle centered on the real axis is the Debye response. In an attempt to approach real systems, Cole and Cole18 introduced an empirical parameter (R) in the Debye equation, according to the equation
* ) ∞ +
(0 - ∞) [1 + (iωτ)1-R]
This equation accounts for a distribution of the relaxation times and results in a depressed semicircle in the ColeCole plot. Alternative empirical fittings19,20 and additional theoretical models21,22 are also available. In a different approach, Jonscher23 suggested, on the basis of experimental evidence, that the dielectric response of any dielectric solid agrees with a fractional power law * ∝ (iω)n-1, with 0 < n < 1. Dielectric dispersion in heterogeneous conducting materials is traditionally described by the Maxwell-Wagner theory.24,25 More recently, different models have been proposed to predict the geometrical features of a porous medium from the dielectric response.26 Freeze-dried matrices are three-dimensional solids consisting of an oriented (or not) ultramacropore network through which ionic species can migrate depending on the network structure. On the basis of previous works on water-saturated rocks and glasses, we have tried to extract information about the three-dimensional structure of the freeze-dried matrices from the dielectric response. Figure 6 shows the log-log plots for the real (′) and imaginary (′′) parts of the complex permittivity as a function of frequency of the dry foam. ′′ is negligible, whereas ′ is constant (′ ) 210-12F). This observation indicates that the main contribution to the total conductivity originates from the water in the saturated matrices. Figure 7 shows plots of ′′ and ′ as a function of the frequency for foams prepared from solutions of different polymer concentrations (i.e., 1, 5, and 10 w:v %), the electrodes being placed in the longitudinal direction. At frequencies higher than 1 Hz, the data agree with the Cole-Cole behavior, that is, ′′ ∝ ω-1 and ′ ∝ ω-q with q ≈ 2 in all cases. In the low-frequency region, the response changes with the PLA foam. ′ and ′′ tend to a plateau for the DL-PLA foams, which suggests a restriction to the free motion of the charge carriers; that is, confinement to spatially constricted regions. A continuous change in the curves is observed for L-PLA. Indeed, the L-PLA 1 sample (17) Debye, P. Polar molecules; Chemical Catalog Co.: 1929. (18) Cole, K. S.; Cole. R. H. J. Chem. Phys. 1941, 9, 341. (19) Davidson, D. W.; Cole, R. H. J. Chem. Phys. 1951, 18, 3417. (20) Havriliak, S.; Negami, J. S. J. Polym. Sci. Part C 1966, 14, 99. (21) Williams, G.; Watts, D. C. Trans. Faraday Soc. 1970, 66, 80. (22) Dissado, L. A.; Hill, R. M. Proc. R. Soc. London 1983, A390, 131. (23) Jonscher, A. K. Nature 1975, 253. (24) Maxwell, J. C. A Treatise on Electricity and Magnetism; Dover: New York, 1983. (25) Wagner, K. W. Arch. Elektrotech. (Berlin) 1914, 2, 371. (26) Hilfer, R. Adv. Chem. Phys. 1996, 92, 299.
Figure 6. Real (′, O) and imaginary (′′, 9) components of the complex permittivity for the dry L-PLA L-PLA 10 foam.
shows a loss peak, which tends, however, to disappear when the concentration of the polymer solution is increased. Slight increase in ′ is also observed with the solution concentration in the same frequency range. This dispersive behavior indicates that the higher the L-PLA solution concentration, the more the charge carriers move freely over extended regions of the foams. This result leads to the conclusion that the L-PLA structures are more “open” and more sensitive to the polymer concentration than the DL-PLA ones. Mechanical Properties. The mechanical properties of the PLA foams are reported in Table 3 for longitudinal and transverse cross-sections, except for the DL-PLA 1 and L-PLA 1 foams, which show no evidence of pore orientation. The effect of the polymer solution concentration on the mechanical properties of the freeze-dried foams is straightforward. The hardness and the modulus significantly increase with the polymer concentration for both the amorphous DL-PLA and the semicrystalline L-PLA foams. Conversely, the foams prepared from dilute solutions (L-PLA 1 and DL-PLA 1) are more cohesive than foams L-PLA 5,L-PLA 10 and DL-PLA 5,DL-PLA 10 prepared at higher concentrations. Comparison of the mechanical properties in the longitudinal and transverse directions for the DL-PLA 5, DL-PLA 10, L-PLA 5, and L-PLA 10 foams shows that they are harder and more deformable when the deformation is parallel to the orientation of the ultramacropores. The results reported in Table 3 also show that at constant polymer concentration and direction of the crosssection, semicrystalline L-PLA foams tend to be harder and of a higher modulus than the amorphous DL-PLA counterparts, except for the L-PLA 1 and DL-PLA 1 foams. In the transverse direction and at constant concentration, the DL-PLA foams are slightly more cohesive than L-PLA foams. It must be pointed out that the measurements by the texturometer provide only relative data that must be discussed on a comparative basis. Discussion Porous polymers are promising materials for tissue engineering because they offer a temporary environment for cell seeding and subsequent transplantation into the host. The porosity and structure of these scaffolds are key characteristics for the success of this strategy. Mercury porosimetry is currently used to measure pore volume
Porous Polylactide Foams
Langmuir, Vol. 16, No. 26, 2000 10469
Figure 7. Real (′) and imaginary (′′) components of the complex permittivity for the L-PLA and 1, 5, and 10% solutions. Table 3. Mechanical Properties for Foams foam DL-PLA
1 5 DL-PLA 5 DL-PLA 10 DL-PLA 10 L-PLA 1 L-PLA 5 L-PLA 5 L-PLA 10 L-PLA 10 DL-PLA
crosssection
hardness (g)
transverse longitudinal transverse longitudinal transverse longitudinal transverse longitudinal
81 ( 15 2331 ( 310 664 ( 124 8694 ( 424 3290 ( 430 35 ( 4 3059 ( 372 676 ( 33 8888 ( 697 6226 ( 216
DL-PLA
and L-PLA
cohesiveness
modulus (g/s)
0.53 ( 0.04 12 ( 3 0.23 ( 0.04 377 ( 37 0.41 ( 0.03 95 ( 15 0.26 ( 0.04 1201 ( 51 0.37 ( 0.015 469 ( 34 0.7 ( 0.02 4 ( 0.6 0.19 ( 0.02 494 ( 16 0.46 ( 0.03 101 ( 1 0.21 ( 0.008 1621 ( 150 0.33 ( 0.03 866 ( 10
and pore size distribution of porous biomaterials. However, this technique is limited to pore size in the 7.5 nm-75 µm range, and shrinkage of the sample by mercury penetration under high pressure can lead to incorrect pore size distribution. In this study, two novel methods were used to investigate the texture of porous PLA foams: image analysis and impedance spectroscopy. These two techniques have the advantage to be nondestructive for the
DL-PLA
foams prepared from
pore structure and to extract information at different scales. The lower limit of image analysis is fixed by the quality of the SEM pictures. For instance, Figure 8 shows pores smaller than 1 µm in the walls of the matrix. This detail in the porous structure is not easily studied by image analysis, although it contributes to the whole porosity of the PLA foams. This limitation can explain the difference in porosity measured by image analysis and mercury porosimetry. Information on the macroporosity is important for understanding how cells interact mutually and with the external medium. For example, nutrients and oxygen supply are expected to depend on the macroporosity. This study shows that the macropores contribute to a large extent to the foam porosity and that the macropore size distribution is broad and asymmetric toward the higher values, independently of the PLA type. Nevertheless, the asymmetry and broadness of the distribution increase with the polymer concentration for the amorphous DL-PLA foams, in contrast to what happens for the semicrystalline L-PLA foams.
10470
Langmuir, Vol. 16, No. 26, 2000
Maquet et al.
validation of these two techniques. For example, the more homogeneous dispersion of the ultramacropores and the narrower porous size distribution in the L-PLA foams observed by image analysis expectedly results in longer distance motion of the charge carriers. This structural detail could lead to more open structure with better diffusion for biological fluids. In contrast, saturation observed in the conductivity of the DL-PLA foams can be related to the ultramacropore clustering and the broad porous size distribution. Interestingly enough, the mechanical properties of the freeze-dried foams are related to the ultramacropore size distribution. Actually, the more homogeneous L-PLA foams exhibit the better mechanical properties, that are, higher modulus and hardness, than the DL-PLA foams. Conclusions
Figure 8. SEM micrograph of the L-PLA foam prepared from 1 w/v % solution (high magnification: ×6500).
Ultramacropores are essential for cell infiltration and tissue organization. Image analysis shows that the size of the ultramacropores increases and their number decreases when the concentration of the polymer solution is increased, whatever the PLA type, which is basically an effect of the solution viscosity. However, ultramacropores in amorphous DL-PLA foams show a large size distribution at all concentrations and a tendency to clustering, particularly at 10 w:v %. In contrast, ultramacropores in semicrystallyne L-PLA foams are homogeneously distributed and show a size distribution that is narrower at higher concentrations. These differences are expected to affect cell distribution, cell interactions, and subsequent tissue organization and function. The internal consistency of the data supplied by image analysis and impedance spectroscopy is an indirect
Image analysis and impedance spectroscopy have been used as potential tools for the characterization of the texture of ultramacroporous PLA foams prepared by freeze-drying. These two techniques actually provide valuable information on the structure of a series of PLA foams. They are complementary to mercury porosimetry, which is a more global method and which does not allow morphological details to be distinguished. They are nondestructive in contrast to mercury porosimetry and they probe the structure at the cell size scale. To our knowledge, this combination of techniques has never been reported before and could be helpful for scientists active in the field of highly porous materials. Finally, differences in the mechanical properties of the L-PLA and DL-PLA foams can be, at least partly, explained by data collected from these two techniques. Acknowledgment. CERM is indebted to the Service Fe´de´raux des Affaires Scientifiques, Techniques et Culturelles (SSTC) in the frame of the “Poˆle d’Attraction Interuniversitaires: Chimie et Catalyze Supramole´culaire, and S.B. to the Conseil de la Recherche, ULG, Belgium, for financial support. LA000654L
