Fabrication of polymer scaffolds for tissue engineering using surface selective laser sintering

774

ISSN 1054-660X, Laser Physics, 2006, Vol. 16, No. 5, pp. 774–787.

©

MAIK “Nauka /Interperiodica”

(Russia), 2006.Original Text © Astro, Ltd., 2006.

1. INTRODUCTION

The development of artificial bone and cartilage tis-sues is one of the most rapidly developing branches ofmodern biomedicine. It involves the creation of perma-nent or temporary 3D matrix structures containingorganic bioactive agents or native cell cultures ofpatients [1–4]. The development of materials and tech-nologies for the formation of matrix structures is one ofthe key branches in tissue engineering [4–6]. Matrixstructures can be fabricated based on various materials:metals, ceramics, polymers, or composite materials.Polymers (including bioresorbed polymers that can beresolved inside the organism without yielding toxiccomponents) are the most promising materials. Themorphology and chemical composition of such materi-als should provide a certain rate of polymer reabsorp-tion and an effective growth of cells and, hence, tissue.

Various methods for the fabrication of highly porouspolymer-based materials have been developed duringthe last decade. However, the fundamental or technicallimitations of the majority of conventional methodsnecessitate the application of rapid prototyping [6, 7].

The modern methods for rapid 3D prototyping(materialization of computer images) involve computerimaging of the desired material object, image layering,and a layer-by-layer fabrication of a solid 3D object ofthe needed size, shape, and inner structure using vari-ous physicochemical methods. Laser stereolithographybased on the local laser photopolymerization (photo-

hardening) of a liquid monomer [8, 9] is one of the mostwidely spread methods for rapid prototyping. Alterna-tive approaches are based on the selective laser sinter-ing of polymer powders [10], fused deposition model-ing [11], and 3D printing [12]. However, the existingmethods of the rapid 3D prototyping meet serious prob-lems in the creation of porous bioresorbed scaffolds fortissue engineering. In particular, laser stereolithogra-phy is restricted to a class of photopolymerized (pre-dominantly, acrylic) systems. A few parameters ofthese systems (e.g., toxicity) do not satisfy the require-ments for the biomedical materials. In the selectivelaser sintering (SLS) of powders based on CO

2

lasers,the heating and melting of the entire volume of a parti-cle takes place. This leads to undesired degradation ofpolymers and thermally labile bioactive compounds.Fused deposition modeling has a narrow domain ofapplication, since it is developed for poly-

ε

-caprolac-tone and polyethylene. In addition, this method doesnot allow for the application of the functional fillers andbioactive compounds. 3D printing involves the applica-tion of organic solvents (e.g., chloroform) for the bind-ing of aliphatic polyesters. In addition, the two lastmethods may yield cytotoxic hydroxy acids. Therefore,the development of the new methods for the rapid pro-totyping of scaffolds for the tissue engineering is a top-ical problem of modern biomedical materials science.

Recently, we have realized a new version of the SLSof polymer microparticles—surface selective laser sin-

Fabrication of Polymer Scaffolds for Tissue Engineering Using Surface Selective Laser Sintering

E. N. Antonov

a

, V. N. Bagratashvili

a

, S. M. Howdle

b

, A. N. Konovalov

a

, V. K. Popov

a

, and V. Ya. Panchenko

a

a

Institute on Laser and Information Technologies, Russian Academy of Sciences, Pionerskaya ul. 2, Troitsk, Moscow oblast, 142190 Russia

b

School of Chemistry, University of Nottingham, Nottingham, NG7 2RD, United Kingdom

e-mail: [email protected] December 26, 2005

Abstract

—A new approach to the fabrication of individual implants and scaffolds for tissue engineering—sur-face selective laser sintering (SSLS)—is proposed and realized. In contrast to the conventional selective lasersintering, the SSLS method makes it possible to sinter polymer microparticles and melt the near-surface layerrather than the microparticle as a whole. The effect of the laser radiation parameters and the structure and com-position of the raw products on the structure and properties of the biomaterials sintered by the laser radiation isanalyzed. This approach makes possible both the application of thermally unstable polymers (e.g., polylactidesor polylactoglycolides) and the fabrication of scaffolds with incorporated bioactive proteins. The resultsobtained yield a regular physical basis for a new technology of the fabrication of various polymer scaffolds thatrepresent important materials and elements of modern tissue engineering. The flexibility of the SSLS methodis especially important at the stage of investigation of the cell and tissue responses needed for the optimizationof the material for a specific application in tissue engineering.

PACS numbers: 42.62.–h

DOI:

10.1134/S1054660X06050070

LASER METHODS IN BIOLOGYAND MEDICINE

LASER PHYSICS

Vol. 16

No. 5

2006

FABRICATION OF POLYMER SCAFFOLDS 775

tering (SSLS) [13–15]. In this method, the laser sinter-ing of polymer particles results from the absorption oflaser radiation by the carbon particles added to thepolymer powder at a relatively low concentration. Anypolymer particles can be used to synthesize scaffoldswith predetermined shapes and inner structures. Notethat, in this case, the chemical compositions of the orig-inal polymers and bioactive compounds remainunchanged.

This work is devoted to the SSLS optimization. Westudy the effect of the laser radiation parameters and thestructure and composition of the original componentson the structure and properties of the biomaterials sin-tered by the laser radiation. The results obtained serveas a regular physical basis for a new technology of thefabrication of various scaffolds that are used in moderntissue engineering.

2. SSLS PRINCIPLE

The fabrication of scaffolds using the laser sinteringof microparticles is based on layer-by-layer sintering inaccordance with the given program. A single step ofthis process is the binding (sintering or welding) of two

free polymer particles that results from the laser-induced melting of the interface layers.

Figure 1 demonstrates the possible schemes of thelaser sintering of polymer particles. The first scheme(Fig. 1a) corresponds to the case when the absorptioncoefficient of the particle material

µ

a

is relatively high(1/

µ

a

is much less than the size

d

p

of the polymer parti-cle). In this case (1/

µ

a

�

d

p

), the heating and meltingtake place only at the front (with respect to the directionof the laser radiation) surface of the particle. The sinter-ing of the adjacent layers in the bulk results from theheat transfer. However, the particle is heated as a whole,and the front surface is significantly overheated.

This scheme is realized in the conventional SLS,when the polymer material is heated with mid-IR lasers(in particular, a CO

2

laser with a wavelength of10.6

µ

m). The sintering of the particles is accompaniedby a substantial (above the melting-softening point)heating and even overheating of the polymer. Evidently,this scenario is unacceptable, since the polymer mate-rial and the incorporated biologically active compoundexhibit thermal lability.

Figure 1b corresponds to the SSLS method consid-ered in this work. The laser radiation is weakly

Laser radiation

Polymer particles Overheat zone

Overheat zone

Laser radiation

Polymer particles

Carbon particles

(a)

(b)

Fig. 1.

Possible schemes for SLS of polymer particles: (a) conventional SLS (significant overheating of the polymer is possible) and(b) SSLS (the high-temperature zone is decreased and localized in a small volume in the vicinity of the absorbing microparticles).

776

LASER PHYSICS

Vol. 16

No. 5

2006

ANTONOV et

al.

absorbed by the polymer particles (1/

µ

a

�

d

p

) and theenergy is deposited to the system owing to the sensi-tizer added to the polymer powder. Carbon particlesthat strongly absorb laser radiation serve as the sensi-tizer. The size of these particles is much smaller thanthe size of the polymer particles (

d

p

�

d

c

). The numberdensity

N

c

of the carbon particles is much higher thanthe number density

N

p

of the polymer particles, so thateach polymer particle may be covered with several car-bon particles. The concentration of carbon particles ismuch less than the concentration of polymer particles(

V

c

�

V

p

). Black carbon particles, which stronglyabsorb laser radiation, heat white (nonabsorbing) poly-mer particles due to thermal contacts. This heatingleads to the local melting of the near-surface polymerlayer. The sintering of white particles can be realized inthe case when a single black particle contacts two whiteparticles. In this case, the high-temperature region islocalized near the microparticle surface (Fig. 1b). Thismakes it possible to realize the sintering of the polymerparticles in the absence of a temperature increase in thebulk. The SSLS optimization lies in the optimization ofthe size and concentration ratios of the white and blackparticles (and, if necessary, fillers), the optimization ofthe preparation of raw materials, and choice of the opti-mal parameters of the laser radiation (laser wavelength,laser beam spot, scanning rate, and scanning mode).

3. MATERIALS AND METHODS

The effect of the laser radiation parameters and themorphology and composition of the raw materials onthe characteristics of the sintered structures is studiedon the setup shown in Fig. 2. An IRE-Polyus LS-0.97diode laser with an output optical fiber, a wavelength of0.97

µ

m, and a power of up to 20 W serves as a cw radi-ation source. The laser radiation is focused on the pow-

der surface and is scanned along a predetermined tra-jectory. The laser spot size ranges from 200 to 600

µ

mand the scanning rate ranges from 0.1 to 3.0 mm/s. Thethickness of the irradiated powder layer is 0.1–0.5 mm.

The laser sintering is controlled in

situ both visuallyand using a Sony TVR17 video camera interfaced witha microscope. The dynamics of the melting process isdetermined with the frame-by-frame analysis of thevideo records. The temperature fields in the heatedmaterial are measured with an IRE IRTIS 2000M IRimager. Technological test measurements were per-formed and the scaffolds were fabricated on the proto-type of an SLS-80 setup at the Institute on Laser andInformation Technologies, Russian Academy of Sci-ences.

For the sintering, we employ D,L-polylactide (PLA)particles (Medisorb® 100 DL HIGH IV (Alkermes,Boston, United States), molecular mass

M

w

= 90 kD,polydispersity

M

n

= 1.4). Figure 3 shows microphoto-grams of the polymer particles used in the SSLS exper-iments. PLA powders are produced with various meth-ods and exhibit different particle sizes and shapes(Table 1). The ball-milled particles (Fig. 3a) are flaky,whereas the rotary-milled particles (Fig. 3b) representirregularly shaped granules. To homogenize the particlesize, we sift the powders through sieves with mesh sizesof 250 and 100

µ

m. In addition, we employ polymermicroparticles with a size of about 5

µ

m obtained usingthe supercritical media (SCM) (Fig. 3c). Carbon parti-cles with a typical surface area of 50 m

2

/g added to thepolymer powders serve as the sensitizer (blackparticles).

Original particles, sintered particles, and structuralfragments are studied with the scanning electronmicroscopy (SEM), gel permeation chromatography(GPC), and micro-Raman spectroscopy.

1

2

3

4

5

6

78

9

Fig. 2.

Scheme of the experimental setup for the SSLS investigation: (

1

) laser, (

2

) optical fiber, (

3

) microscope, (

4

) objective, (

5

) and(

8

) mirrors, (

6

) video camera, (

7

) IR imager, and (

9

) IR lens.

LASER PHYSICS

Vol. 16

No. 5

2006


4. RESULTS AND DISCUSSION

The radiation of CO

2

lasers employed in the conven-tional SLS is absorbed in a thin polymer layer with athickness of tens of microns. The energy transfer in thebulk of the powder is realized owing to the thermal pro-cesses. In this case, high temperature gradients mayemerge, and both the base polymer and the incorpo-rated bioactive agents can be thermally overheated anddestroyed.

In contrast to the commercially available SLS sys-tems, the system under study employs near-IR laserradiation with a wavelength of 0.97

µ

m. At this wave-length, the polymer particles are virtually transparent(

µ

a

< 10

–1

cm

–1

). The heating takes place due to theabsorption of radiation by the black sensitizer particlesadded to the powder and located at the surface of thepolymer particles. Melting starts at the entire surface ofthe polymer particles (at many local points) and propa-gates into the depth. The temperature decreases withincreasing depth, and the maximum temperature isattained at the surface. The temperature gradientsemerge on a particle-size scale rather than the thicknessof the powder layer. For the appropriate irradiationparameters, the sintering can be realized with meltingof only the surface layers of particles when the innerregions remain unchanged (Fig. 4b).

4.1. SSLS Dynamics

Figure 4 shows an example of the sintering of PLAparticles that contain 0.1 wt % of carbon particles. Wedemonstrate the frames of a video record that corre-spond to various moments of the laser irradiation. Fig-ure 4a shows the initial stage of the integration of twoPLA particles over the surface fragment that containsdark carbon particles. Figures 4b and 4c correspond tothe subsequent stages. In Fig. 4d, the particles are com-pletely sintered and form a spherical particle that, inturn, is sintered with the low-lying layer of particles.

Choosing the laser energy and the scanning rate, wemay realize the sintering conditions that correspond toFigs. 4b–4d. The thermal actions and the degrees ofintegration are different, so that the resulting materialsexhibit different porosities and strengths. Evidently, thesintering dynamics, the thermal processes, and themechanical strength of the sintered structures shoulddepend on the shape of the polymer particles. Thephase of the process shown in Fig. 4a corresponds tothe beginning of melting of the surface. The sintering isrealized, but the major part of the particle volumeremains in the original state. Thus, the biochemicalproperties of the material are not violated. For the phaseshown in Fig. 4d, the material is completely melted.Note that a possible variation in the biochemical prop-erties depends on the duration of the thermal action andthe energy deposition. The scenario corresponding to

(‡) (b) (c)

12.5 kV 221

¥

200

m

m 1.31 kx 20

m

m 8.80 kx 5

m

m

Fig. 3.

Microphotograms (electron microscopy images) of the PLA particles fabricated with various methods: (a) granulated(rotary-milled) particles, (b) flaky (ball-milled) particles, and (c) spherical particles (SCM process).

Table 1.

Types of PLA particles used for SSLS

Sample type Particle size (

µ

m) Fabrication method Particle shape

I 1 Sifted through a sieve with a mesh size of 250

µ

m Rotary-milled Granules

2 Sifted through a sieve with a mesh size of 100

µ

m

3 Sifted through a sieve with a mesh size of 50

µ

m

II Sifted through a sieve with a mesh sizes of 100

µ

m(a width of 20–50

µ

m and a thickness of about 1

µ

m)Ball-milled Flakes

III 5

µ

m SCM process Spheres

778

LASER PHYSICS

Vol. 16

No. 5

2006

ANTONOV et

al.

Fig. 4a should be preferred in the biomedical applica-tions. However, in this case, the material may haveinsufficient strength. Therefore, based on the specificrequirements, one should choose a compromise thatcorresponds to the scenarios shown in Figs. 4b and 4c.

4.2. Effect of the Shape of Polymer Particles

A track of the sintered particles starts forming in theirradiation region when the temperature of the irradi-ated polymer increases to the melting point. The trackshape depends on the powder parameters (shape andsize of polymer particles and concentration of black

particles) and the laser radiation parameters. One of theproblems in the fabrication of scaffolds with predeter-mined inner structures and shapes lies in the sinteringshrinkage [10]. Figure 5 demonstrates the sinteredtracks obtained for various polymer powders. In thecase of the flaky polymer particles, the sintered trackmay exhibit discontinuities (Fig. 5b). The flaky parti-cles are attracted towards the center of the laser beamupon heating. This leads to the emptying of the neigh-boring regions and the formation of irregular sintereddrops not bound to each other.

Surface laser sintering of globular and granulatedpolymer particles leads to a significantly weaker

(‡)

(c) (d)

(b)

Fig. 4.

Microphotograms of the SSLS stages for two PLA particles: (a) beginning of the laser heating, (b) and (c) intermediatestages, and (d) end of the process.

(‡) (b) (c)

1 mm

Fig. 5.

Filaments resulting from the sintering of various PLA particles: (a) granules sifted through a sieve with a mesh size of250

µ

m, (b) flakes sifted through a sieve with a mesh size of 250

µ

m, and (c) spheres sifted through a sieve with a mesh size of 5

µ

m.

LASER PHYSICS

Vol. 16

No. 5

2006


shrinkage. The sintered tracks obtained upon the sinter-ing of spherical particles (Fig. 5c) have a width of about150

µ

m. At energies close to the melting threshold, thefilament width is close to the laser beam diameter. Thesmaller the polymer particles, the higher the spatial res-olution of the method. However, a decrease in the par-ticle size causes problems in the formation of powderlayers. Electrostatic interaction may impede the forma-tion of flat and smooth surfaces upon the mechanicalleveling of the layers. Note that the best spatial resolu-tion obtained with an alternative approach (fused depo-sition modeling) is 200–500

µ

m.

Thus, the sintering of particles with different origi-nal structures yields different structures and propertiesof the resulting materials. For the flaky original parti-cles, a real flake size is about 10

µ

m, and the sifted par-ticles with a size of 250

µ

m represent clustered con-glomerates.

The application of the filler-containing granulatedpowder excellently satisfies the requirements for highhomogeneity, reproducibility, size accuracy of the sin-tered elements, and spatial resolution. The applicationof flaky particles leads to the formation of loose,mechanically weak structures with high heterogeneityand porosity. Such a material may serve as an effectiveimplant provided that strict requirements for themechanical strength are missing or in the case when thebioresorption rate must be high or when high-molecu-lar compounds (e.g., growth factors and drugs) and cel-lular materials must be impregnated.

4.3. Effect of the Black Particle Concentration and Laser Power

The preparation of the SSLS powders involves thor-ough mixing of the original components, which leads toa homogeneous distribution of the carbon particles overthe polymer surface (Fig. 6). In the course of sintering,

PLA + C (1%) 600× 50 mm

(‡)

(b)

PLA + C (1%) 4.02 kx 10 mm

Fig. 6. Microphotograms of the submicron carbon particles(white inclusions) on the surface of the granulated PLA par-ticles with scales of (a) 50 and (b) 10 µm.

1.0% C –4.0W 12.5 kV 50 mm573¥

(b)

(‡)

1.0% C –4.0W 12.5 kV 20 mm2.30 kx

Fig. 7. Microphotograms of the sintered PLA particles.After sintering, the carbon microparticles are located in thenear-surface layer inside PLA and in the zones of the parti-cle integration. Scales are (a) 50 and (b) 20 µm.

780

LASER PHYSICS Vol. 16 No. 5 2006

ANTONOV et al.

these particles are partially or completely envelopedwith the polymer (Fig. 7). A typical weight concentra-tion of the black particles is 0.5%. When the black par-

ticles are added to the pure powder at such a concentra-tion, the melting threshold and the absorption coeffi-cient at a wavelength of 0.97 µm decrease by twoorders of magnitude.

We employ PLA powders with carbon concentra-tions of 0.5, 0.4, 0.2, 0.1, and 0.04 wt %. When theblack particle concentration decreases, the penetrationdepth of the laser radiation, the melting energy thresh-old, and the thickness of the sintered layer increase. Inthis case, an increase in the scattering leads to the wid-ening of the melting region (i.e., worsening of the spa-tial resolution).

The melting threshold is estimated with the visualmicroscopic examination of the filaments at variouslaser intensities. Table 2 compares the results obtained.Even at a constant concentration of the black particles(0.5%), the melting thresholds differ for the polymerparticles of different types and sizes. An increase in theblack particle concentration leads to a decrease in themelting threshold.

It is seen that the minimum threshold corresponds tothe 100-µm granulated PLA. For the granulated PLA-250, the threshold is significantly higher. For flaky andfine PLAs, the thresholds are also higher than the PLA-100 threshold. The reason for this is the differencebetween the geometrical characteristics of the powderparticles. For larger particles, a higher heat capacityplays the dominating role in melting. For smaller parti-cles, the melting threshold is determined by the density.The thermal conductivity of the flaky particles is rela-tively low owing to a low density. Therefore, the initialstage of melting involves local changes that cannotimmediately be seen on the macroscopic scale. Forlarger particles, the surface concentration of carbon isrelatively high. For the flaky particles, carbon may pen-etrate deep inside the pores. In this case, the surfaceabsorption decreases, and the energy is more uniformlyabsorbed inside the layer. The same holds true for afiner PLA powder, in which carbon is distributed moreuniformly with respect to the powder depth. The effectlying in a decrease in the surface absorption is mani-fested upon visual observation. The granulated powderwith carbon is the darkest one (virtually black),whereas the flaky and fine PLAs are gray.

For a structure formed from the particles of onetype, the melting width and depth depend on the laserenergy parameters and the concentration of absorbingcenters. To reveal the dependence of the melting widthand depth on the laser radiation parameters, we performexperiments with PLA powders that consist of 100-µmgranulated particles with various carbon concentrations(0.4, 0.2, 0.1, and 0.04 wt %) at various laser intensitiesand scanning rates. The beam diameter at the powdersurface is 0.4 mm.

Figure 8 demonstrates the PLA filaments sintered atvarious laser energies. The filament width and the pow-der shrinkage in the adjacent regions increase with theenergy. The widths and thicknesses of the sintered fila-

Table 2. Melting thresholds for PLA powders of varioustypes in mixtures and 0.5% of black particles. Laser beamdiameter is 0.25 mm and the scanning rate is 3 mm/s

Type of PLA particles Melting threshold (J/cm2)

I, 250 µm 8.5 ± 1.3

I, 100 µm 3.8 ± 0.4

II, Flaky 6.0 ± 0.8

III, 5 µm SCM process 5.5 ± 0.8

Fig. 8. Laser-track filaments resulting from the sintering ofthe type-II PLA powder (Table 1) with a particle size of100 µm (0.2% of carbon) at various laser energies: sinteredstructures (left-hand side) prior to and (right-hand side)after the removal of the powder layer. The laser beam diam-eter is 0.4 mm. The scale division is 1 mm. The lower fila-ment is formed at an energy that is close to the meltingthreshold. Each next filament is formed upon adding theenergy that corresponds to the threshold. The upper filamentis obtained at an energy that exceeds the threshold level bya factor of seven. The shrinkage of the powder material isseen at high energies.



ments increase with both an increase in the laser energyat a constant scanning rate and an increase in the radia-tion dose (a decrease in the scanning rate at a constantenergy). In the vicinity of the melting thresholds, thefilament widths are close to the width of the laser beam,and the ratio of the melting depth to the width rangesfrom 0.65 to 0.75. However, at a minimum carbon con-centration of 0.04%, this ratio is 0.5 at a width that is afactor of 1.5 greater than the laser beam width. Whenthe energy is twice that of the threshold, the widths ofthe sintered filaments increase by a factor of 1.5. Notethat, in this case, the ratio of the melting depth to thewidth remains virtually unchanged. At higher energiesand carbon concentrations (0.2 and 0.4 wt %), the ratioof the melting depth to the width is close to unity. Forrelatively low concentrations, the depth-to-width ratiois virtually the same as in the case of the near-thresholdenergies.

The optimization of the process should be aimed atthe minimization of the width of the sintered track atthe maximum depth of the melted layer and the mini-mum heating of single particles. Based on these crite-ria, under the above experimental conditions, the opti-mal carbon concentration is 0.1 wt % at the maximumscanning rate of the laser beam. As in the case of theanalysis of the sintering dynamics shown in Fig. 4, wemay state that the sintering in the vicinity of the meltingthreshold (lower filament in Fig. 8) should be preferredfor biomedical applications. With regard to the require-ments for the mechanical strength, the optimum modemay correspond to the formation of one of the filamentsadjacent to the lower filament. In the upper filament, thematerial is significantly overheated and partiallydestroyed, which leads to a variation in the biochemicalproperties.

4.4. Introducing Fillers in the Polymer Powders

The SLS method allows for a wide variation in rawmaterials. Various fillers that affect both the mechanicaland biological properties of the sintered composite maybe introduced into the polymer powders. We employhydroxylapatite (HA) as a filler at a concentration of upto 20 wt %. HA is an osteoconducting material that vir-tually always facilitates the bone tissue formation. Inaddition, in the presence of HA, the sintering shrinkageis absent, and the quality of sintered structures becomeshigher. Figure 9 shows several examples of HA-con-taining structures sintered at various laser intensities. Inthe case of sintering in the vicinity of the meltingthreshold, HA particles can be found both in the bulkand at the surface (Figs. 9a and 9b). The polymer com-pletely envelopes the HA particles when the energy issignificantly higher than the threshold (Fig. 9c).

4.5. SSLS Temperature Fields

Various thermal modes can be realized upon thelaser irradiation of polymer particles. The possibilities

for the control of the thermal mode are limited: it canbe varied with a variation in the laser intensity or parti-cle size. Note that the thermal mode may play a signif-icant role in the fabrication of biocompatible and biore-

04CHA4 12.5 kV 50 mm600¥

(‡)

(b)

(c)

04CHA4 12.5 kV 200 mm221¥

20HA10 12.5 kV 50 mm573¥

Fig. 9. Sintered structures consisting of PLA with a HAcontent of 20%: (a) and (b) in the vicinity of the sinteringthreshold and (c) far above the threshold.

782


ANTONOV et al.

sorbed 3D structures with the SLS method. First, thismode affects the mechanical strength of the sampleand, hence, its porosity. Second, a relatively high tem-perature in the laser irradiation region may lead to vari-ations in the properties of the biopolymer and the bio-active component introduced into the sintered powder.Therefore, in this work, we measure and calculate thetemperature fields for the SSLS of polymer particles.

4.5.1. Temperature measurements. An IR imager8 (Fig. 2) is used for the measurements of the tempera-ture field generated upon the powder sintering. Lens 7is used for image magnification. Thermal IR radiationrelated to the local heating of the powder is directed tothe IR imager with mirror 5. Using this system, westudy the laser-induced powder heating. The spatial res-olution is 100 µm per pixel and the time resolution (IR-imager exposure time) is 1 s.

Figure 10 shows typical temperature profilesobtained at various laser powers. Figure 11 demon-strates the dependence of the temperature at the centerof the laser beam on the incident power. Note that thedependence differs from the linear dependence. Asalient point corresponds to an incident power of 9 mW(melting threshold). Apparently, this behavior is relatedto the activation of an additional mechanism for thethermal loss: heat transfer due to the motion of themelted powder and the energy expended to melt thepolymer and, possibly, to evaporate it at a temperaturehigher than 100°C.

Note that the leading edge is predominantly deter-mined by the laser power and the absorption coefficientof the powder, whereas the trailing edge is determinedby the cooling process that depends on the size of theheated region and the thermal conductivity. The lead-ing-edge duration is given by

(1)dTdt------

P0µa

S----------- 1

ρCp

----------,≈

and the trailing-edge duration is represented as

(2)

where P is the incident laser power, S = πD2/4 is thecross section area of the laser beam, ρ is the powderdensity, Cp is the specific heat, χ is the thermal conduc-tivity, and µa is the absorption coefficient of the powder.

The shortest leading and trailing edges (about 0.8 s)are observed for the granulated PLA powder with car-bon. The longest edges (about 2.5 s) correspond to thesame powder in the absence of carbon. An increase inthe leading-edge duration is caused by a relatively lowabsorption coefficient of the carbon-free sample. Forthe carbon-containing powder, we have (dT/dt)1 =60 K/s, whereas, for the pure powder, we obtain (dT/dt) =10 K/s. Therefore, the ratio of the correspondingabsorption coefficients is α1/α = (dT/dt)1P/(dT/dt)P1 =60 K/s 6000 mW/(10 K/s 970 mW) = 37. Note that theratio of the melting thresholds is greater than one hun-dred. Knowing the cooling time (Fig. 11a), we estimatethe thermal conductivity with expression (2): χ ~ d2/τ ≈1 × 10–6 m2/s.

For the PLA powder, the reason for an increase inthe trailing-edge duration lies in the fact that, prior tocooling, the size of the heated area (D = 1.6 mm) isgreater than this size for the carbon-containing powder(D = 1.1 mm). Apparently, the temperature stabilizationobserved for the mixtures of the polymer powder withcarbon in a certain interval of the heating times is dueto the balance of the heating and cooling rates owing tothe processes of melting and heat transfer from theheated area to the cold area. Note also that substantiallylonger edges correspond to the flaky powder. Similarlyto the previous case, we find that the absorption of theflaky powder is lower than the absorption of the granu-lated powder: µa1/µa2 = 60 K/s 1570 mW/(28 K/s970 mW) = 4. The sizes of the heated areas of the flaky

τ D2/χ,∼

100

–1000 0 1000 x, µm

90

80

70

60

50

40

30

T, °C

7

6543

2

1

Fig. 10. Stationary temperature profiles resulting from theaction of the laser radiation on the PLA powder with a car-bon content of 0.5% (granules) at P = (1) 2.5, (2) 3.5, (3)4.3, (4) 5.4, (5) 6.3, (6) 7.2, and (7) 9.0 mW.

2 4 6 8 10 12P, mW

40

50

60

70

80

90

100

110

120T, °C

Fig. 11. Plot of the stationary temperature at the center ofthe laser beam vs. incident power for the PLA powder witha carbon content of 0.5% (granules).



and granulated powders are equal, but the trailing edgeis longer for the granulated powder. The reason for thisis a lower thermal conductivity related to a higherporosity of this sample.

For the PLA DE powder (Fig. 12c), the trailing edgeis approximately the same as in the case of the PLA/Cpowder. The diameter of the heated area is approxi-mately the same (about 1 mm). Hence, the thermal con-ductivities of these powders are virtually equal. For thePLA DE powder, the leading edge is significantlylonger. This is due to both a higher powder density(closer packing for fine powder) and a lower absorptioncoefficient.

Note also that, for different powders, the meltingthresholds with respect to temperature are also differ-ent. The reasons for this are, first, uncertainty of thethreshold for polymers and, second, the method for thethreshold measurement (detection of the moment corre-sponding to the beginning of motion). It is seen that thegranulated PLA powder exhibits the maximum temper-ature threshold, whereas the threshold of the flaky andfine PLA powders are substantially lower. In our opin-ion, this effect is related to the more homogeneous dis-

tribution of carbon for porous and fine particles. A het-erogeneous distribution of carbon leads to the localoverheating at the places with the highest concentrationof the absorbing particles.

4.5.2. Simulation of temperature inside a poly-mer particle. The results of the temperature measure-ments in the course of the SSLS of polymer particlesdemonstrate the possibility for the temperature controlof the polymer volume aimed at the optimization of thesintering conditions and the prevention of the undesiredoverheating of the polymer particles. The spatial reso-lution of the IR imager is no higher than 100 µm. Todetermine the temperature inside a single particle, weperform a simple thermal calculation. At a relativelyhigh concentration of carbon particles at the surface ofthe polymer particle, the heating can be considered as asurface process with a uniform energy input. Indeed,the leveling of the temperature of the surface layer issignificantly faster (the characteristic time is ~x2/χ,where x is the distance between the absorbing carbonparticles) than the heating of the polymer particle (thecharacteristic time is ~a2/χ, where a is the particle

40

20

60

80

100

120

T, °C

(b)

20 400 10t, s

205 15 25 3530

40

50

60

70

80

90

100

110

30

20 400 1030

5 15 25 3530

40

50

60

70

80

90

100

20

T, °C

405060708090

100110

30

(a)

(d)(c)

Fig. 12. Dynamics of the laser heating in the course of SSLS for (a) PLA with a carbon content of 0.5% (granules), a particle sizeof 50–250 µm, P = 10.9 mW, a threshold of P0 = 4.5 mW, T0 = 63°C, and D0 = 1100 µm; (b) PLA with a carbon content of 0.5%(flakes), a particle size of 50–250 µm, P = 18.1 mW, a threshold of P0 = 9 mW, T0 = 48°C, and D0 = 1100 µm; (c) PLA DE powderwith a carbon content of 0.5% (small spheres), a particle size of about 10 µm, P = 25 mW, a threshold of P0 = 7.5 mW, T0 = 41°C,and D0 = 950 µm; and (d) PLA (granules), a particle size of 50–250 µm, P = 1100 mW, a threshold of P0 = 600 mW, T0 = 80°C,and D0 = 1600 µm.

784


ANTONOV et al.

radius). In the case under study, this condition is satis-fied with a relatively high accuracy, since x � a (Fig. 6).

For the analysis of the main regularities of such aheating mode, we employ the following simplifiedmodel. A particle represents a sphere whose surfaceuniformly absorbs laser radiation, so that the absorptioncoefficient is constant over the surface. For the pur-poses of simplification, we assume that the absorptioncoefficient is relatively low, so that the energy depositedonto the back surface (lower part) is approximatelyequal to the energy absorbed by the upper part. Theabsorbed energy density is I0µacosϑ, where I0 is theincident laser intensity, µa is the absorption coefficient,and ϑ is the azimuthal angle. In the analysis, we assumethat the absorbed energy density is independent of ϑ.This assumption results in certain corrections to thecharacteristic times and temperatures of heating, sincethe thermal front represents an elliptical surface ratherthan a sphere. In this assumption, we solve a cen-trosymmetric heat-conduction problem. In sphericalcoordinates, the corresponding equation is written as

(3)

and the boundary condition at r = a is given by

(4)

where Q(t) = Pa(t)/πD2 is the energy density absorbedby the particle surface, Pa = σI0 is the absorbed energy,and σ is the absorption cross-section of the particle.

An averaged absorption coefficient of many powderlayers can be estimated as

(5)

∂T∂t------

χr---∂2

rT( )∂r

2----------------,=

∂T∂r------

Q t( )cpρχ------------,=

µ0 σN σ/D3.= =

We find σ from expression (5) and obtain

(6)

Let a particle be irradiated at the initial moment at aconstant intensity. In this case, a thermal wave startspropagating from the particle surface. It reaches thecenter of the particle at the moment t0 ~ a2/χ. For thetimes much longer than t0, the temperature can be rep-resented as

(7)

This expression represents an exact solution to Eq. (3)in the case Q(t) = Q0 = const. Note that, at any moment,

the derivative of temperature at the particle surface

is constant and depends on the energy input (see bound-ary condition (4)). Figure 13 schematically shows thedynamics of the temperature profile. Thus, in contrastto the case of the volume absorption of the laser radia-tion in the presence of absorbing centers, we alwaysobtain a strong temperature gradient that is determinedby the laser intensity and the absorption coefficient.Using expression (7), we can find the maximum tem-perature difference between the surface and the centerof the particle:

(8)

Here, P0 is the laser power and R0 is the laser beamradius.

The temperature increases at a rate given by

(9)

Using the above estimate of the thermal conductivity(χ = 1 × 10–6 m2/s), we find the characteristic heatingtime of a particle with the radius a = 100 µm: t0 ~ a2/χ ≈0.01 s.

Let us make a few estimates. It follows from themeasurements of the temperature dynamics and profilethat, at a laser power of P0 = 10.9 mW and a beamdiameter of 2R0 = 1100 µm, the temperature increases

at a rate of = 60 K/s. It follows from expression (8)

that such a rate corresponds to the absorption coeffi-

cient µ0 = ≈ 22 cm–1 or a penetration depth

of about 0.45 mm. For the estimate, we employ P0 =10 mW, R0 = 0.4 mm, and the values of density and specificheat for paraffin (ρ = 870 kg/m3 and C = 1600 J/kg K).Using this result, we calculate the temperature differ-ence between the particle surface and its inner part that

Q t( ) Pa t( )/πD2 µ0DI0/π.= =

T t r,( )Q0

2χρCa----------------- 6χt r

2+( ) T0.+=

∂T∂r------

∆T T2 T1–Q0a

2χρC--------------

µ0P0

π2χρC----------------- a

R0-----⎝ ⎠

⎛ ⎞ 2

.= = =

∂T∂t------

3Q0

ρCa-----------

6µ0P0

π2ρCR02

--------------------.= =

∂T∂r------

∂T∂t------

π2ρCR02

6P0--------------------

T(t, r), K

T0

T1

T2

t � t0 (T ~ r2)

t � t0

0 a r, µm

Fig. 13. Temperature distributions inside a spherical parti-cle at various moments after the beginning of the laser irra-diation.



is realized in the experiments at an irradiation power ofP = 100 mW: ∆T ≈ 1 K. Here, we use the value χ = 1 ×10–6 m2/s. Note that the temperature difference can beincreased due to an increase in the irradiation power P0upon a simultaneous increase in the scanning rate. Forexample, a temperature gradient of tens of degrees canbe realized at a power of about one watt and a scanningrate of tens of millimeters per second.

4.6. SSLS Effect on the Molecular Compositionof the Material

The GPC measurements of the molecular mass andpolydispersity of the polymers yield a slight monotonicdecrease in Mw and Mn with an increase in the laserintensity (Table 3). If the energy is a factor of eighthigher than the threshold, an increase in the energyleads to an increase in the polymer molecular weightrather than to its decrease. When the energy is an orderof magnitude higher than the threshold, the samples arenot soluble in chloroform. This indicates the cross link-ing in the polymer structure and the possible degrada-tion. The simulation of sintering at a nonlaser heatingshows that the parameters obtained at the energy that istwice higher than the melting threshold are comparablewith the parameters obtained for a conventional sinter-ing at 70°C.

The GPC results correlate with the Raman data. TheRaman measurements of PLA are performed in thespectral ranges 800–1800 and 2900–3000 cm–1, whichcorrespond to the CH, CH3, COC, and COO vibrationalbands. The spectra of the sintered samples are identicalto the spectrum of the original polymer. Minor varia-tions in the Raman spectra of the irradiated PLA sam-ples in comparison with the Raman spectra of the unir-radiated samples are observed only for the energy thatexceeds the melting threshold by an order of magni-tude. However, in this case, the radiation intensity is an

order of magnitude higher than the intensity used forthe polymer sintering.

4.7. SSLS Effect on the Biological Properties of the Materials

We preliminary studied the SSLS effect on the activ-ity of ribonuclease incorporated in the PLA particles[15]. At certain conditions for the SSLS of PLA, thisenzyme retains the biological activity (the degradationis virtually absent). After the SLS process, the ribonu-clease activity decreases to a level of 70–80% relativeto the original value. This scenario is realized uponlaser irradiation when the sintering threshold isexceeded by a factor of 1.5–2. At higher intensities, theribonuclease activity decreases owing to its thermaldegradation.

Carbon nanophase used as the sensitizer is a bio-compatible material [16, 17]. However, the thermalaction on the polymer–carbon mixture causes destruc-

Table 3. PLA molecular mass and polydispersity

Material Mw Mn

Type-I PLA (250 µm) 89370 63982

Sintered in the vicinityof the threshold

88282 63093

Twofold excess over the threshold 83753 59186

Fourfold excess over the threshold 81841 58128

Eightfold excess over the threshold 102889 58848

Tenfold excess over the threshold Chloroform-insoluble

Conventional sintering at 70°C 83672 61044

Conventional sintering at 90°C 85794 62878

(‡) (b)

20 mm 1 mm25.0 kV 35.8×

Fig. 14. SEM images of the cell cultures grown on the surface of 3D porous PLA/C matrices fabricated with SSLS: (a) 3T3 fibro-blasts (five days after seeding) and (b) ovine meniscal chondrocytes (24 h after seeding).

786


ANTONOV et al.

tion and chemical reactions that may yield toxic com-pounds. In addition, the toxicity of the submicron car-bon particles is insufficiently studied. Therefore, oneshould test the toxicity of the sintered structures. Thetests that involve the growth of cell cultures on thePLA/C scaffolds show the absence of toxicity for thesamples sintered at various conditions (including sin-tering at a laser energy that is three to four times higherthan the polymer melting threshold [18]). Figure 14ademonstrated the 3T3 fibroblast cultures grown on thesurface of the PLA/C scaffolds fabricated with SSLS.Fibroblasts are well proliferated on the surface of the sin-tered samples, which indicates their biocompatibility.

The cell adhesion and proliferation on the surface ofthe PLA/C scaffolds fabricated with SSLS is also dem-onstrated in Fig. 14b, where we show the cultures of theovine meniscal chondrocytes that occupy more than50% of the surface 24 h after seeding.

4.7. Synthesis of 3D Structures

Figures 15–17 demonstrate 3D multilayer structuressintered from PLA powders with particle sizes rangingfrom 250 to 50 µm. In all of the cases, thin filaments areformed. The filament cross section is nearly rectangu-lar, and the filament sizes are determined by the sizes ofthe laser beam and particles.

Thus, the SSLS technology employing near-IRdiode lasers makes it possible to fabricate 3D structureswith various porosities. The size and shapes ofmicropores can be predetermined and controlled duringthe layer-by-layer sintering. The microporosity is deter-mined by the size and shape of the polymer particlesand the sintering conditions. The SSLS method allowsa wide-range variation in the macro- and microporosityand fully complies with the requirements and problemsof the modern tissue engineering.

PLA250–01 25.0 kV 2 mm23.1¥

(‡)

(b)

PLA250–02 25.0 kV 500 mm57.5¥

Fig. 15. Fragments of a structure sintered from the PLApowder with a particle size of 250 µm and a carbon concen-tration of 0.4% at a laser beam diameter of 0.4 mm withscales of (a) 2 mm and (b) 500 µm.

(‡)

(b)

PLA100–01 25.0 kV 2 mm22.2¥

PLA100–02 25.0 kV 500 mm57.5¥

Fig. 16. Fragments of a structure sintered from the PLApowder with a particle size of 100 µm and a carbon concen-tration of 0.4% at a laser beam diameter of 0.4 mm withscales of (a) 2 mm and (b) 500 µm.



5. CONCLUSIONS

The proposed SSLS technology makes it possible tofabricate 3D porous polymer matrices with intercon-nected pores. The size, inner structure, chemical com-position, and shape can be predetermined. Note themulti-purpose character of this method, since it can beused for the polymer particles of any composition, size,and shape. In addition, the method allows for the appli-cation of various functional fillers, including nonpoly-mer fillers (e.g., calcium phosphates). The proposedSSLS scheme enables one to significantly decrease thetemperature gradients inside the material and to elimi-nate the limitations on the SLS applications for the fab-

rication of biocompatible polymer matrices for tissueengineering.

ACKNOWLEDGMENTSThis work was supported by the Russian Foundation

for Basic Research (project nos. 04-02-16933 and 05-02-08069-ofi_a), the Ministry of Science and Educa-tion of the Russian Federation (Program for the Supportof the Leading Scientific Schools), and The WellcomeTrust Foundation (Collaborative Research InitiativeGrant nos. 062760 and 073913).

REFERENCES1. J. P. Vacanti and R. Langer, Science 260, 920 (1993).2. M. Sittinger, J. Bujia, N. Rotter, et al., Biomaterials 17,

237 (1996).3. J. S. Temenoff and A. G. Mikos, Biomaterials 21, 431

(2000).4. W. Sun and P. Lal, Comput. Methods Programs Biomed.

67, 885 (2002).5. G. Ciardelli, V. Chiono, C. Cristallini, et al., J. Mat. Sci.:

Mat. Med. 15, 305 (2004).6. K. F. Leong, C. M. Cheah, and C. K. Chua, Biomaterials

24, 2363 (2003).7. D. W. Hutmacher, Biomaterials 21, 2529 (2000).8. E. N. Antonov, A. V. Evseev, M. A. Markov, et al., Opt.

Tekh. 13, 55 (1998).9. M. N. Cooke, A. N. Konovalov, et al., J. Biomed. Mater.

Res. B 64, 65 (2003).10. J. T. Rimell and P. M. Marquis, J. Biomed. Mater. Res.

53, 414 (2000).11. I. Zein, D. W. Hutmacher, K. C. Tan, and S. H. Teoh,

Biomaterials 23, 1169 (2002).12. S. Limpanuphap and B. Derby, J. Mat. Sci.: Mat. Med.

13, 1163 (2002).13. E. N. Antonov, V. N. Bagratashvilli, A. N. Konovalov,

et al., in Proceedings of XI International ConferenceNew Information Technology in Medicine, Pharmacol-ogy, Biology and Ecology, Gurzuf, Ukraine, 2003, p.209.

14. V. K. Popov, E. N. Antonov, V. N. Bagratashvilli, et al.,Materials Processing for Properties and Performance(2004), Vol. 3, p. 332.

15. E. N. Antonov, V. N. Bagratashvilli, M. J. Whitaker,et al., Prog. Mod. Nat. Sci. 1 (Suppl. 1), 97 (2005).

16. R. L. Price, M. C. Waid, K. M. Haberstroh, andT. J. Webster, Biomaterials 24, 1877 (2003).

17. M. Kaibara, H. Iwata, H. Wada, et al., J. Biomed. Mater.Res. 31 (3), 429 (1996).

18. E. N. Antonov, V. N. Bagratashvilli, A. N. Konovalov,et al., to be published.

PS50–03 25.0 kV 500 mm57.5¥

(‡)

(b)

PS50–04 25.0 kV 200 mm178¥

Fig. 17. Fragments of a structure sintered from the powderwith a particle size of 50 µm at a laser beam diameter of0.4 mm with scales of (a) 500 and (b) 200 µm.

Fabrication of polymer scaffolds for tissue engineering using surface selective laser sintering

Documents