Langmuir 2006, 22, 3235-3242
3235
Nondestructive Technique for the Characterization of the Pore Size Distribution of Soft Porous Constructs for Tissue Engineering Laleh Safinia, Athanasios Mantalaris, and Alexander Bismarck*,† Department of Chemical Engineering, Polymer & Composite Engineering (PaCE) Group, Imperial College London, South Kensington Campus, London, SW7 2AZ United Kingdom ReceiVed June 30, 2005. In Final Form: NoVember 9, 2005 Polymer scaffolds tailored for tissue engineering applications possessing the desired pore structure require reproducible fabrication techniques. Nondestructive, quantitative methods for pore characterization are required to determine the pore size and its distribution. In this study, a promising alternative to traditional pore size characterization techniques is presented. We introduce a quantitative, nondestructive and inexpensive method to determine the pore size distribution of large soft porous solids based on the on the displacement of a liquid, that spreads without limits though a porous medium, by nitrogen. The capillary pressure is measured and related to the pore sizes as well as the pore size distribution of the narrowest bottlenecks of the largest interconnected pores in a porous medium. The measured pore diameters correspond to the narrowest bottleneck of the largest pores connecting the bottom with the top surface of a given porous solid. The applicability and reproducibility of the breakthrough technique is demonstrated on two polyurethane foams, manufactured using the thermally induced phase separation (TIPS) process, with almost identical overall porosity (60-70%) but very different pore morphology. By selecting different quenching temperatures to induce polymer phase separation, the pore structure could be regulated while maintaining the overall porosity. Depending on the quenching temperature, the foams exhibited either longitudinally oriented tubular macropores interconnected with micropores or independent macropores connected to adjacent pores via openings in the pore walls. The pore size and its distribution obtained by the breakthrough test were in excellent agreement to conventional characterization techniques, such as scanning electron microscopy combined with image analysis, BET technique, and mercury intrusion porosimetry. This technique is suitable for the characterization of the micro- and macropore structure of soft porous solids intended for tissue engineering applications. The method is sensitive for the smallest bottlenecks of the largest continuous pores throughout the scaffold that contributes to fluid flow.
1. Introduction Porous polymeric materials play key roles in tissue engineering serving as three-dimensional (3D) scaffolds for cellular attachment and tissue development1 and membrane technology2 as filters for the separation of biological and other materials as well as chromatography and ion-exchange columns.3 Surface chemistry, pore size, and porosity are key characteristics of scaffolds that are critical in cell seeding and attachment and subsequent, if desired, transplantation into the host.4 A scaffold should possess high porosity with a complex network of channels and interconnected pores of appropriate size and shape to guarantee the ingress of large numbers of cells and allow the formation of cellular associations.5 The porous network influences cell growth, cellular organization, proliferation, and migration. In addition, interconnectivity of the pores affects the rate and depth of cellular in-growth6 and controls the circulation of culture medium and thus the exchange of nutrients and metabolites.7 Accurate control over the pore size, interconnectivity, and the overall porosity of scaffolds are therefore essential. * To whom all correspondence should be addressed. Phone: +44 20 7594 5578. E-mail:
[email protected]. † Polymer & Composite Engineering (PaCE) Group. (1) O’Brien, F. J.; Harley, B. A.; Yannas, I. V.; Gibson, L. J. Biomaterials 2005, 26, 433-441. (2) Lee, Y.; Jeong, J.; Youn, I. J.; Lee, W. H. J. Membrane Sci. 1997, 130, 149-156. (3) Zeman, L.; Zydney, L. A. Microfiltration and ultrafiltration: Principles and applications; Marcel Dekker: New York, 1996. (4) Blacher, S.; Maquet, V.; Pirard, R.; Pirard, J.-P.; Je´roˆme, R. Colloids Surf. A 2001, 187-188, 375-383. (5) Elma, H.; de Groot, J. H.; Nijenhuis, A. J.; Pennings, A. J.; Veth, R. P. H.; Klompmaker, J.; Jansen, H. W. B. Colloid. Polym. Sci. 1990, 268, 10821088. (6) van Tienen, T. G.; Heijkants, R. G. J. C.; Buma, P.; de Groot, J. H.; Pennings, A. J.; Veth, R. P. H. Biomaterials 2002, 23, 1731-1738.
Several methods have been proposed to tailor-make porous scaffolds to obtain the desired pore structure.8 The major challenge for scaffold fabrication and characterization is the “accurate” and “nondestructive” characterization of the pore structure.9 Currently, there is no standard or even well accepted method that can be used for the accurate structural characterization of scaffolds. Commonly, the apparent density, overall porosity, specific pore volume, surface area, and the mean pore size of the porous solid are parameters that are measured in order to evaluate its structure. However, these values do not truly characterize the nature of the porous structure.10 Consequently, there is the need for the establishment of methods that will characterize the scaffold pore structure, in terms of pore diameter and its distribution, pore density, pore shape, and length. There are several techniques that have been developed for determining the pore size distribution of porous media [see the review in ref 11] including microscopic tools, such as scanning electron (SEM) and atomic force microscopy (AFM), X-ray microtomography (µCT), mercury intrusion porosimetry (MIP), gas adsorption, as well as various gas and liquid displacement methods, such as capillary flow porometry.12 Microscopy is most commonly employed to characterize the pore structure of membrane surfaces.13 The micrographs obtained provide qualitative information about the pore shape and size of (7) Klompmaker, J.; Jansen, H. W. B.; Veth, R. P. H.; de Groot, J. H.; Pennings, A. J.; Kuijer, R. Biomaterials 1992, 12, 810-816. (8) Guan, J.; Fujimoto, K. L.; Sacks, M. S.; Wanger, W. R. Biomaterials 2005, 26, 3961-3971. (9) Maquet, V.; Blacher, S.; Pirard, R.; Pirard, J.-P.; Vyakarnam, M. N.; Je´roˆme, R. J. Biomed. Mater. Res. 2003, 66A, 199-213. (10) Li, S. H.; de Wijn, J. R.; Layrolle, P.; de Groot, K. Key Eng. Mater. 2003, 240-2, 541-545. (11) Nakao, S.-I. J. Membrane Sci. 1994, 96, 131-165. (12) Stillwell, C. R.; Gupta, K. M. IEEE 11th Annual Battery Conference on Applications and Advances 1996, 267-271.
10.1021/la051762g CCC: $33.50 © 2006 American Chemical Society Published on Web 03/01/2006
3236 Langmuir, Vol. 22, No. 7, 2006
Safinia et al.
the porous solid. Image analysis of micrographs9 gives quantitative information with respect to the pore sizes. MIP has been widely used to measure the pore size distribution in the range of 7.5 nm to 75 µm.14 However, soft samples suffer from compression due to Hg penetration under high pressures, which can lead to incorrect pore size distribution.4 µCT is considered to be the most superior technique for the structural noninvasive characterization of 3D constructs. Using synchrotron radiation, it is possible to extent the spatial resolution down to the submicrometer range. The use of 2D analysis and 3D reconstruction software enables the determination of porosity, pore size, pore interconnectivity, as well as the internal layout of the porous solid.15 The structural characterization of fully hydrated samples is also possible since µCT does not require high vacuum.16,17 There are relatively simple methods for the nondestructive characterization of porous media, which are the liquid vapor pressure methods, capillary rise methods, and methods based on the application of Poiseuille’s law.18 Several liquid and gas displacement methods for the characterization of the largest pore size of porous solids have been described in the literature.19-23 These methods are based on measuring the pressure required to force liquid out of a pre-wetted membrane or to replace the wetting phase by another one, i.e., a gas/liquid displacing a liquid/ gas or a second immiscible liquid phase, and to form bubbles or droplet on the surface of the porous solid. A gas is forced through the pores of a solid filled with a liquid that fully wets (i.e., spreads without limits; contact angle θ between the solid and liquid equals 0° or cos θ ) 1) the surfaces of the pores when the applied pressure exceeds the capillary pressure, which is a function of the pore size and the liquid surface tension. The maximum bubble point test to determine the diameter of the largest pores or its limiting bottleneck is based on measuring the pressure required to force a gas through a liquid saturated porous medium.24-26 The pressure p at which the first bubbles appear on the downstream side of the porous medium is called the bubble point pressure. Using this pressure, the maximum pore diameter of a continuous imaginary cylindrical pore connecting the bottom with the top surface of the porous medium can be determined on the basis of eq 1
dp )
4‚γlv cos θ p
(1)
where d is the diameter corresponding to the narrowest point of an imaginary pore connecting the bottom with the top surface, γl is the surface tension of the liquid, and θ is the contact angle of the liquid on the solid. In case of the liquid that fully wets the solid (cos θ), eq 1 becomes Cantor’s equation 24 (13) Ho, C.-C.; Zydney, A. L. J. Membrane Sci. 2000, 170, 101-112. (14) Shum, A. W. T.; Li, J.; Mark, A. F. T. Polym. Degrad. Stab. 2005, 87, 487-493. (15) Darling, A. L.; Sun, W. J. Biomed. Mater. Res. B 2004, 70B, 311-317. (16) Mu¨ller, B.; Beckmann, F.; Huser, M.; Maspero, F.; Sze´kely, G.; Ruffieux, K.; Thurner, P.; Wintermantel, E. Biomol. Eng. 2002, 19, 73-78. (17) Thurner, P.; Mu¨ller, B.; Beckmann, F.; Weitkamp, T.; Rau, C.; Mu¨ller, R.; Hubbell, J. A.; Sennhauser, U. Nucl. Instrum. Methods B 2003, 200, 397-405. (18) Bartell, F. E.; Osterhof, H. J. J. Phys. Chem. 1928, 32, 1553-1571. (19) Reichelt, G. J. Membrane Sci. 1991, 60, 253-259. (20) Zeman, L. J. Membrane Sci. 1996, 120, 169-185. (21) Gadam, S.; Phillips, M.; Orlando, S.; Kuriyel, R.; Pearl, S.; Zydney, A. J. Membrane Sci. 1997, 133, 111-125. (22) Piatkiewicz, W.; Rosinski, S.; Lewinska, D.; Bukowski, J.; Judycki, W. J. Membrane Sci. 1999, 153, 91-102. (23) Calvo, J. I.; Bottino, A.; Capannelli, G.; Hernandez, A. J. Membrane Sci. 2004, 239, 189-197. (24) Bechhold, H. Z. Phys. Chem. 1908, 64, 328-342. (25) Bigelow, S. L.; Bartell, F. E. J. Am. Chem. Soc. 1909, 31, 1194-1199. (26) Bartell, F. E.; Whitney, C. E. J. Phys. Chem. 1932, 36, 3115-3126.
dp )
4‚γlv p
(2)
Owing to the simplicity of the technique, the bubble point method has been used to estimate membrane integrity of many commercial membranes.27 Bechhold et al.28 were the first who developed a method based on liquid/liquid displacement. However, this method has severe limitations for the characterization of porous media with extremely small pores. The gas pressure required to intrude nanometer-size pores would be in the range of 30-300 bar leading to membrane compaction and rupture this limiting the applicability of the maximum bubble point test.29 For this reason liquid/liquid systems are used to determine the larger pore diameters. In this study, we adapted the nondestructive bubble point technique to characterize the pore sizes as well as pore size distribution of PU foams fabricated using the temperature induced phase separation (TIPS) technique with varying quenching temperatures. The results obtained were comparable to the SEM/ image analysis and MIP data, making the breakthrough technique a suitable characterization technique for determining the porous architecture of 3D polymer scaffolds. 2. Materials and Methods 2.1. Materials. Estane 58201 NAT 021, a polyether-based thermoplastic polyurethane (PU) compound with a density of 1.11 g/cm3, was kindly supplied by Noveon Europe BVBA, Brussels, Belgium. Dioxane (99% pure) purchased from Sigma-Aldrich was used for the fabrication of PU foams. The polymer and solvent were used without further purification. The test liquid, n-dodecane (99% pure), used to determine the maximum interconnected pore diameter of the polymer foams, was purchased from Riedel-de Hae¨n, Acroˆs Organics. Dodecane was purified by passing the solvent three times over a silica (Hopkins & Williams Ltd.) and basic-activated aluminum oxide (Acroˆs Organics) packed chromatography column. Deionized water from a NANOpure system (Barnstead, Dubuque, IA, conductivity 1.8 MΩ/cm) was used. 2.2. Preparation of PU Foams. The PU was vacuum-dried overnight (0.013 mbar) at 80 °C to remove all moisture. The preparation of porous PU foams (of diameter 100 mm and height 5 mm) involved a thermally induced phase separation (TIPS) process with a subsequent freeze-drying step, which has been presented elsewhere.30 PU polymer was dissolved in dioxane at 80 °C to give a polymer weight to solvent volume ratio of 5%. The mixture was stirred overnight to obtain a homogeneous polymer solution. The polymer solution was transferred to a lyophilisation flask and immersed in either liquid nitrogen or placed in a freezer (-25 °C) for 2 h. The frozen mixture was then transferred into an ethylene glycol bath maintained at -15 °C and connected to a vacuum pump (0.013 mbar) for 3 days to remove the solvent. The resulting porous scaffold was completely dried at room temperature in a vacuum oven for 24 h.
3. Structural/Morphological Analysis 3.1. Breakthrough Test. Pore size distribution of the PU foams was examined using the maximum bubble point test. A home-built cell described in ref 30 together with a Macroscope (Wild M420, Wild, Heerbrugg, CH), an external light source, and a digital camera connected to a computer (for details of the set up refer to Figure 1) was used. The scaffold was fitted into (27) Herna´ndez, A.; Calvo, J. I.; Pra´danos, P.; Tejerina, F. J. Membrane Sci. 1996, 112, 1-12. (28) Bechhold, H.; Schlesinger, M.; Silbereisen, K. in collaboration with Maier, L.; Nu¨rnberger, W. Kolloid Z 1931, 55, 172-198. (29) Gadam, S.; Phillips, M.; Orlando, S.; Kuriyel, R.; Pearl, S.; Zydney, A. J. Membrane Sci. 1997, 133, 111-125. (30) Safinia, L.; Blaker, J. J.; Maquet, V.; Boccaccini, A. R.; Mantalaris, A.; Bismarck, A. e-polymers 2005, 010.
NondestructiVe Technique for Tissue Engineering
Figure 1. Schematic diagram of the maximum bubble point test set up.
the displacement cell, which was connected on one side to a nitrogen cylinder and the other side to a water column manometer. n-Dodecane, a liquid that fully wets the polymer surface, was used to completely fill the pores of the scaffold. The cell was pressurized using nitrogen by successively increasing the flow rate. The pressure was measured. Eventually, nitrogen forces the liquid out of the pores resulting in rapid continuous bubbling, “bubble point” or the “first breakthrough” (Figure 2a,b). By continuing to increase the pressure in the cell, the number of bubble streams increases. Knowing the pressure at each subsequent breakthrough, the corresponding pore diameters can be determined using eq 2. To obtain statistically valid data, at least six samples from different parts of the PU foams were characterized. 3.2. Scanning Electron Microscopy (SEM)/Image Analysis. Scanning electron microscopy (JEOL JSM-840A, JEOL Ltd., Welwyn Garden City, U.K.) was used to observe the internal pore structure of the PU foams. The specimens were cut with a razor blade in the direction parallel and perpendicular to the surface. The resulting longitudinal and transverse sections were sputtered with gold in argon atmosphere prior to observation. Image and statistical analysis were performed on SEM micrographs of three horizontal cross-sections (magnification ×100) taken from three different areas of the same sample using AnalySIS software package. For the purpose of the image analysis, the pores were classified as particles, and the maximum feret diameter, the maximum distance of parallel tangents at opposite particle borders, of the pores was determined. To obtain statistically relevant data, about 30 for PU-25 and 50 pores for PU-196 were analyzed on each horizontal cross-section of the foam. 3.3. Mercury Intrusion Porosimetry (MIP). A mercury intrusion porosimeter (PoreMaster 33, Quantachrome) was used to determine the pore size distribution of the PU scaffolds. The Washburn eq 1 was used to calculate the pore diameter dp in relation to the applied external pressure p needed to force mercury into the pores. The mercury surface tension γlv is (480 mN/m), and the contact angle θ between mercury and polymer is assumed to be 140°. Details concerning the experimental and theoretical procedures for the measurements performed are reported elsewhere.31
4. Material Characterization of PU Foams 4.1. Physical Properties. Parameters such as specific pore volume, apparent density and porosity of the scaffolds were obtained using pycnometry (GeoPyc 1360 V1.03, Micromeritics) (31) Ritter, H.; Drake, L. Ind. Eng. Chem. 1945, 17, 782-786.
Langmuir, Vol. 22, No. 7, 2006 3237
in conjunction with helium displacement pycnometry (AccuPyc 1330 V3.00, Micromeritics). Six foam samples (12 mm × 5 mm) were analyzed to obtain statistically relevant data. 4.2. Permeability; Pore Interconnectivity. PU foams (22 mm in diameter and 5 mm wide) were placed in the cylindrical flow cell compartment of the Electrokinetic Analyzer (EKA, Anton Paar KG, Graz, Austria). With the foam in place and two silver/silver-chloride (Ag/AgCl) electrodes lodged into two small compartments on both ends of the cell, the cell was rinsed with 1 mM KCl solution at a constant temperature of 20 °C, applying a maximum pressure difference (∆p) of 250 mbar prior to measurements. Once the system was stabilized, the overflow of the electrolyte passing through the foam sample was collected over a fixed time keeping ∆p constant. By increasing ∆p from 50 to 250 mbar and from eq 3 the permeability κ (m2/Darcy) was deduced. To obtain statistically valid data 6 data points were collected for each pressure value
κ)
∆Q Lη ‚ ∆p A
(3)
where ∆Q is the flow rate (m3/s), L is the sample length (m), η is the fluid viscosity (1 mPas for water), ∆p is the pressure drop across the sample (Pa), and A is the cross-sectional area of the PU sample (m2). As the porosities of the two different PU scaffolds (PU-196 and PU-25) were comparable (in the range of 60-70%), the permeability was used to compare pore interconnectivity and pore diameter. Further details for calculating the permeability of porous scaffolds can be found elsewhere.32 4.3. BET Measurement (Nitrogen Adsorption Technique). The surface area of the PU foams was determined using nitrogen adsorption isotherms at 77 K utilizing a surface area analyzer (Micromeritics ASAP 2010). The sample holder was filled with PU foam pieces. Before performing the gas sorption experiments, surface contaminants were removed via a “degassing” step. The samples are heated inside glass sample cells under vacuum (heating temperature of 80 °C, Tg ) 110 °C) overnight. For the analysis part, nitrogen (the adsorbate) was admitted into the evacuated sample chamber. The detailed explanation of the experimental and theoretical procedures for the measurements performed are reported elsewhere.33 4.4. Mechanical Properties. Compression tests were carried out using a Lloyds universal testing machine (Lloyds EZ50, Lloyds Instruments Ltd., Fareham, U.K.). Foam samples (10 mm × 10 mm × 5 mm) were compressed at a loading rate of 20 mm/min by means of a compression plate with a load cell of 1 kN, until a compression of 75% of the initial test piece thickness is attained. Afterward, the unloading curve was measured at the same rate until the separation between the compression plate and the base plate is equal to the initial test piece thickness. Three different foam samples were analyzed to obtain statistically relevant data.
5. Results 5.1. Morphological and Structural Analysis: SEM and Image Analysis. The SEM micrographs (Figure 3a,b) show the axial cross sections of PU foams demonstrating the porous structure and the distinct macro- and microporous architecture. The longitudinal cross-sections of the PU foams (Figures 4 and 5a-d) highlight the pore interconnectivity forming the 3D pore (32) Sohier, J.; Haan, R. E.; de Groot, K.; Bezemer, J. M. J. Controlled Release 2003, 87, 57-68. (33) Li, J.; Favis, B. D. Polymer 2001, 42, 5047-5053.
3238 Langmuir, Vol. 22, No. 7, 2006
Safinia et al.
Figure 2. Images of a PU foam in dodecane (a) and after pressurizing the cell with nitrogen, showing the initial breakthrough of the first bubble stream (b). The foam diameter is 6 mm.
Figure 4. SEM micrographs of a longitudinal cross-section through PU-25 polymer scaffold demonstrating the pore throats linking the macroporous structure.
Figure 3. SEM micrographs of the axial cross-section through PU polymer scaffold, (a) PU-196 (×120) (b) PU-25 (×120) showing the macro- and micropore structure of the PU foams.
network. The larger pores are orientated almost perpendicular to the surface following the freezing direction. When the polymer solution was quenched in liquid nitrogen (PU-196), a microcellular pore structure was formed with pore diameters in the range of 20-40 µm (Figure 3a). Increasing the quenching temperature (PU-25) resulted in a dramatic increase in the scaffold pore size. PU-25 (Figure 3b) had macropores with a diameter exceeding 100 µm. It is assumed that at a rapid cooling rate (quenching at -196 °C) leads to the formation of smaller solvent crystals, which do not have enough time to grow. During the solvent extraction step, the solvent crystals are removed resulting in the small pores.34 Nam and Park35 suggested that increasing the quenching temperature of the polymer solution results in scaffolds with a closed cell pore structure as compared to scaffolds quenched in liquid nitrogen (Figure 4). The white pockets on the SEM (34) Li, H.-R.; Yen, Y.-J. J. Biomed. Mater. Res. B: Appl. Biomater. 2004, 71B, 52-65. (35) Nam, Y. S.; Park, T. G. J. Biomed Mater. Res. 1999, 47, 8-17.
micrograph highlight the pore throats interconnecting the individual pores. Increasing the quenching temperature resulted in a slight increase in porosity (Table 1); however, the biggest change is observed in the degree of interconnectivity within the scaffold. Figure 5a,c reveals the interconnected pore structure of the scaffolds quenched in liquid nitrogen (PU-196), which possess a ladder-like structure throughout the thickness of the foam (Figure 5c). It appears that much smaller pores interconnect the macropores. In contrast, the PU-25 scaffolds have direct connections between the macropores through small openings in the pore walls (Figures 4 and 5d). The results of the image analysis (Figure 6a,b) show that the interconnected pore range of the PU-196 foams is in the range 30-40 µm, whereas for the PU-25 foams, the pores are in the range of 50-60 and 90-100 µm. The SEM results together with the image analysis show that the morphology of the scaffolds can be manipulated by simply altering the quenching temperature during the preparation. 5.2. Structural Analysis: Breakthrough Test. Figure 7 shows the pressure required to displace the fully wetting liquid n-dodecane (cos θ ) 1) from the largest pores of the PU foams as a function of increasing gas flow rate. The gas pressure required to displace a fully wetting liquid increases with increasing flow rate until the bubble pressure is reached, i.e., the point at which the first continuous stream of bubbles appears on the downstream surface of the sample (Figure 2a,b). At this point, the graph (∆p ) f(Q)) starts to level off, pointing toward a rather narrow pore size distribution for the scaffold. The minimum pressure required to overcome the capillary pressure and force dodecane through the PU-196 and PU-25
NondestructiVe Technique for Tissue Engineering
Langmuir, Vol. 22, No. 7, 2006 3239
Figure 5. SEM micrographs showing the vertical cross-section through PU polymer scaffold, (a), (c) PU-196 × 50 (near the surface), ×120, respectively (b) PU-25 × 100 (near the surface), ×120, respectively. Table 1. Average Envelop Density, Porosity, Specific Pore Volume, BET Surface Area, Permeability, and Compressive Modulus of the PU Scaffolds Prepared by TIPS Process
scaffold
envelope density, g/cm3
porosity, %
specific pore volume, cm3/g
BET surface area, m2/g
permeability κ, 10-12m2 (Darcy)
compression modulus E, kPa
PU-25 PU-196
0.4 ( 0.03 0.4 ( 0.04
70.6 ( 2.4 68.7 ( 2.5
2.0 ( 0.24 1.5 ( 0.20
0.25 1.03
2.52 (2.56) 1.25 (1.27)
28 ( 3 78 ( 8
scaffolds, resulting in the first bubble breakthrough was 2370 ( 30 and 1536 ( 40 Pa, respectively (Figure 8a,b). The microporous structure of the PU-196 scaffolds necessitated that higher pressure was required to force dodecane through the pores of the scaffold in comparison to the larger structure of the PU-25 scaffolds. Using Cantor’s equation the maximum diameter of an imaginary cylindrical pore running through from the bottom to the top surface of the PU-196 and PU-25 scaffolds was determined to be 43 ( 1 and 66 ( 3 µm, respectively. More precisely, these pore diameters reflect the narrowest parts or bottlenecks of the largest pores. These results are in agreement with the SEM micrographs and the image analysis results (Figure 3a,b and 5a,b). After reaching the first breakthrough (bubble) point, the pressure was further increased stepwise to obtain subsequent bubble breakthrough points. As soon as further bubble streams appeared, the pressure was recorded. From this pressure, the values for the corresponding pore diameters could be calculated.
The number of bubble streams relates to the number of pores going through the entire scaffold. Figure 8c,d shows the pore size distribution of the largest pores that are most likely to contribute to the fluid flow through the PU-196 and PU-25 scaffolds, respectively. The reproducibility of the test is excellent. The data presented in Figure 8 are averaged values of six independent measurements on fresh samples from various foams, and the error given is the standard deviation. As can be seen, the experimental scatter is rather small. An inherent limitation of the breakthrough test is the requirement of safely distinguishing between the individual bubble streams, which becomes harder as the number of streams appearing on the surface of the scaffold increases. 5.3. Structural Analysis: Mercury Intrusion Porosimetry (MIP). MIP was used to characterize the microstructure (pore size distribution) of the PU foams (Figure 9a,b). In PU-196 (Figure 9a), the pores are in the range of 150-250 µm, whereas for
3240 Langmuir, Vol. 22, No. 7, 2006
Figure 6. Pore size distributions of the PU foams (a) PU-196 and (b) PU-25 using the software package AnalySIS.
Safinia et al.
and the other techniques has two causes: the high pressure required to force mercury into small nanopores (dp < 7 nm) and second that mercury penetrates larger pores (dp > 75 µm) on its own before the first measurements are done, i.e., at pressures below 0.2 bar. It is important to note that the upper detection limit of the PoreMaster 33 corresponds to pore diameters of 250 µm. Therefore, the pore size distribution represents the pore diameters that constricted the intruded mercury flow, rather than the largest pores themselves. During MIP mercury surrounds the scaffolds and penetrates the pores from all sides. MIP therefore provides a pore size distribution although it is unable to discriminate how the pores are interconnected and which pores contribute to the flow through a porous medium. 5.4. Material Characterization: Physical Characterization of the PU Foams. As shown in Table 1, virtually no or only very small differences in the average envelope density, overall porosity, and specific pore volumes were detected between the foams produced at different quenching temperatures. The PU foams (PU-196 and PU-25) have a total porosity in the range of 6070%. The BET surface area decreases for the foams prepared at -25 °C, which is a clear indication that pore sizes increase.36 Therefore, parameters such as average apparent density, porosity, specific pore volume and BET surface area do not provide any relevant information about the structural properties of the polymer scaffolds. Pore size as well as the degree of pore interconnectivity affects the fluid flow through porous media. Permeability measurements are a tool to characterize the degree of pore interconnectivity.32 Figure 10 shows the flow rate of a KCl solution through the porous polymer scaffolds as a function of applied pressure drop across the sample. The gradient obtained provides a value for fluid conductance (∆Q/∆p). The steeper gradient of the PU-25 foam (2 × 10-10 m3/Pas) compared to the PU-196 foam (1 × 10-10 m3/Pas) indicates a higher permeability of the PU-25 scaffolds. Since both scaffolds have a high degree of pore interconnectivity, the increased permeability (Table 1) can mainly be attributed to the larger pore diameters as demonstrated by the breakthrough test and, therefore, reduced resistance against flow. 5.5. Structural Characterization: Mechanical Properties. Figure 11 shows the stress-strain curve for the two PU foam samples prepared at different quenching temperatures. It is evident from Figure 11 that both scaffolds, PU-196 and PU-25, undergo elastic deformation during compression. The compression modulus (Table 1) decreases with increasing quenching temperature. The increase in the compression modulus can be attributed to the fact that the PU-25 foam possesses much larger open pores. The overall porosity of the foams seems to be unaffected by the processing temperature, which resulted in larger but fewer pores and therefore thicker pore walls.
6. Discussion
Figure 7. Maximum bubble point measurement: Pressure as function of nitrogen flow rate to determine the breakthrough pressure at which continuous bubbling from the foam surface first occurs. n - dodecane was used as test liquid.
PU-25 (Figure 9b), pores with a diameter of 280 µm were detected. Maquet et al.9 suggested that MIP underestimates pore volumes, which are systematically smaller than those measured by pycnometry. They conclude that the discrepancy between MIP
In an effort to develop an effective characterization technique to quantify pore size and the pore size distribution that affects the flow through porous media through soft porous solids, we propose an inexpensive nondestructive breakthrough method that provides a quantitative measure of the pore structure. We demonstrated the applicability of the technique by studying PU foams of similar overall porosity yet different pore architectures. The scaffolds were produced by TIPS by varying the quenching temperatures. Using the breakthrough test, the maximum pore diameter of PU-196 and PU-25 was calculated to be 43 ( 1 and 95 ( 3 µm, respectively. The breakthrough test enables the (36) Yuan, Z.; Favis, B. D. Biomaterials 2004, 25, 2161-2170.
NondestructiVe Technique for Tissue Engineering
Langmuir, Vol. 22, No. 7, 2006 3241
Figure 8. (a and b) Nitrogen flow rate through the wet PU samples vs applied pressure: (a) PU-196 and (b) PU-25. (c and d) number of pore vs pore diameter: (c) PU-196 and (d) PU-25.
characterization of pore interconnectivity throughout the thickness of a porous medium. The comparison of the pore sizes determined using the breakthrough test with the results obtained from the image analysis of the SEM pictures shows that the size of the largest pore is only 50% of the visible largest pores in the SEM cross-section, which highlights the fact that the method is sensitive for the smallest bottleneck in the largest pore. By continuing the test after the first bubble breakthrough occurred and counting the number of subsequent breakthroughs at small pressure intervals, we determined the pore size distribution of the largest interconnected pores (Figure 8c,d). The pressure necessary to force the liquid out of smaller pores is significantly larger, and therefore, the method cannot give quantitative information about the percentage of large and small pores. Nevertheless, our results are complemented by the data obtained from SEM micrographs and provide a real, quantitative measure of pore interconnectivity. Analysis of scaffold pore size distribution remains a major challenge. Commonly used methods include a combination of image analysis of SEM micrographs and impedance spectroscopy providing information on 2D and 3D levels without affecting the structure (as for instance by MIP) or the risk of artifacts.9 SEM micrographs (Figures 3a,b and 5a-d) confirm that pore size distribution can be adjusted by selecting the appropriate freezing temperature (and cooling rate) during TIPS. Although image analysis of SEM micrographs (Figure 6a,b) provide a “quantitative” measure of pore size distribution, this is applicable only to the small section of the foam that can be viewed by SEM. Image analysis of SEM micrographs provides no measure of
pore interconnectivity (co-continuity) and it is highly dependent on the quality of the SEM micrographs.4 Furthermore, image analysis of SEM micrographs is an involved process that requires trained personnel and specialized equipment. In contrast, the breakthrough test is easy to perform and requires little equipment which can be easily made and assembled everywhere; essentially the only requirement being the identification of a nonsolvent for the solid that spreads (wets the solid fully) without limits on/ through the porous medium without dissolving or swelling the material. The breakthrough test provides both a measure of the pore size distribution and directly relates the maximum pore size that contributes to flow through the porous medium. Mercury porosimetry is currently the most commonly used technique to measure the pore volume and pore size distribution of porous biomaterials. The high operational pressures of the MIP can result in the creation of additional tiny pores in porous structures if thin films are used for the analysis.36 Shrinkage of the sample due to Hg penetration under high pressure can affect pore size distribution especially in soft materials.4 Furthermore, the samples cannot be reused after the characterization. In contrast, the breakthrough technique is a nondestructive technique making it an attractive tool for characterizing the pore size distribution of porous media. µCT is considered to be the most superior of all of characterization techniques for porous media. The use of 2D analysis and 3D reconstruction software allows not only the morphological examination of the internal architecture of a porous network, but provides also information about the interconnectivity
3242 Langmuir, Vol. 22, No. 7, 2006
Safinia et al.
Figure 11. Stress-strain behavior of PU foams under compressive loading.
7. Conclusion
Figure 9. Pore size distribution of PU foams determined using mercury intrusion porosimetry (MIP), (a) PU-196, (b) PU-25.
Figure 10. Permeability measurements: determining the fluid conductance from the induced flow ∆Q (m3/s) and the pressure drop across the sample (∆P) (Pa).
of porous solids and a quantitative measure of porosity. However, it still requires investment of significant resources.
We introduce a simple, inexpensive, quantitative and nondestructive technique to characterize the pore sizes as well as pore size distribution of the narrowest parts or bottlenecks of the largest continuous pores connecting the bottom with the top of a porous solid. The technique is especially useful for the characterization of soft porous solids for tissue engineering applications. The applicability and reproducibility of the breakthrough technique was demonstrated on two polyurethane foams with almost identical overall porosity yet distinctively different pore morphology. The samples were prepared using the TIPS process, which enables to produce porous foams with wellcontrolled pore size and morphology by varying the quenching temperature of the polymer solution. The results of the breakthrough test were compared to conventional characterization techniques, such as SEM combined with image analysis and the BET and MIP techniques. Pore sizes and pore size distribution data were in good agreement with the results obtained from the conventional techniques. The breakthrough technique is sensitive to the micro- and macropore structure of polymer foams and provides an accurate measure of the pore diameter of a continuous pore throughout the scaffold that contributes to fluid flow. However, for porous solid material of unknown structure, the presented method is sensitive only to the largest pores, whereas the pressure necessary to get a breakthrough in smaller pores is significantly larger. Thus, the method cannot really give quantitative information about the percentage of large and small pores. Acknowledgment. L.S. is grateful to the U.K. Engineering and Physical Science Research Council (EPSRC) for providing the financial support for her Doctoral Trainee Award. Furthermore, the authors thank Dr. Julian Jones (Department of Materials, Imperial College London) for performing the MIP measurements and Dr. Raquel Verdejo-Marquez (Department of Chemistry, Imperial College London) and Alexis Baltazar-y-Jimenez (PaCE group, Department of Chemical Engineering, Imperial College London) for taking the SEM micrographs shown. LA051762G
