Available online at www.sciencedirect.com

Journal of the European Ceramic Society 30 (2010) 25632567

3D printing of bone substituteoaen

, Hoa al Aa

ersity

arch010

Abstract

Customized el mato significan positsimilar to th ia 3Dthe bioresorp d becbioresorbabl niquewas 14.9 MPa after sintering. XRD analysis revealed the occurrence of two other phases, CaNaPO4 and CaSiO3, both biocompatible and with thepotential of biodegradation. We conclude that it is possible to print tailored bone substitute implants using a bioactive TCP/glass composite. Theglass is not involved as reactive substance in the printing process. This offers the opportunity to alter the glass composition and therefore to varythe composition of the implant. 2010 Elsevier Ltd. All rights reserved.

Keywords: 3D

1. Introdu

The goautologousexcellent bbody.1 Howway, becautissue canwing fractuof synthetistruct the bare widely-tricalciumsorption prbone cemeand dicalcisorbable ca

CorresponE-mail ad

0955-2219/$doi:10.1016/jprinting; -Tricalcium phosphate; Bone substitute; Biodegradation; Bioactive glass

ction

lden standard for treating small bone defects isbone (bone derived from the same patient). It has

ioresorption capabilities and is never rejected by theever, large-scale defects are difficult to restore thisse the extraction of large amounts of autologouscause donor side problems, like morbidity or iliacres.24 For those defects solid implants consisting

c materials have to be manufactured, which recon-one and support it mechanically. Calcium phosphates

used as bone substitute materials.57 Especiallyphosphate (-TCP) is known for its good biore-

operties.810 -TCP can react with phosphoric acid asnt, forming dicalcium hydrogen phosphate (DCPD)um pyrophosphate (DCPP), which are also biore-lcium phosphates.1113 Biocompatible glasses, on

ding author. Tel.: +49 241 8080936; fax: +49 241 8082027.dress: chbergmann@ukaachen.de (C. Bergmann).

the other side, can be adjusted from bioinert to bioresorbableaccording to their composition.14,15 Both materials, in form ofgranules or pastes, are already in clinical application.

The hypothesis of this study was that a combination of-TCPand biocompatible glass is advantageous for the manufacturingof bone implants with a 3D-printing technique. 3D printing is arapid manufacturing method, were a binding solution is printedinto layers of powder, accordingly to a sliced virtual model.This method is already in the focus of scientific research forcustomized scaffolds.16,17 In this study the focus was to inves-tigate, if the different properties of -TCP and biocompatibleglass could prove advantage for the 3D printing of customized,bioresorbable, large-scale bone implants.

2. Experimental

2.1. Preparation of granulate

-TCP was synthesized, by heating calcium deficienthydroxyapatite (Ca3(PO4)2 art. no. 1.02143.9026 Merck Darm-stadt, FRG) for 1 h at 1000 C. Beforehand exact calcium to

see front matter 2010 Elsevier Ltd. All rights reserved..jeurceramsoc.2010.04.037calcium phosphate and biChristian Bergmann a,, Markus Lindner a, W

Armin Kirsten a, Rainer Telle bDepartment of Dental Materials and Biomaterials Research, University Hospit

b Department of Ceramics and Refractory Materials, RWTH Aachen UnivReceived 3 November 2009; received in revised form 24 M

Available online 23 May 2

implants for bone replacement are a great help for a surgeon to remodtly reduce operation times. The hypothesis of this study was that a come 45S5 Henchglass is suitable to manufacture customized implants vtion properties of the -TCP, its capability to react as bone cement, ane. Customized implants were manufactured using the 3D-printing techimplants usingctive glassesZhang b, Karolina Koczur a,rst Fischer a

chen, Pauwelsstrasse 30, D-52074 Aachen, Germany, Mauerstrasse 5, D-52064 Aachen, Germany2010; accepted 26 April 2010

xillofacial or craniofacial defects in an esthetical way, ande of -tricalcium phosphate (-TCP) and a bioactive glass-printing process. The composite was chosen because of

ause of the adjustability of the bioactive glass from inert to. The four point bending strength of the printed specimens

2564 C. Bergmann et al. / Journal of the European Ceramic Society 30 (2010) 25632567

phosphorus ratio of 1.5 was adjusted by adding 5 wt.% of cal-cium hydrogen phosphate dihydrate (CaHPO42H2O art. no.1.02144.90ture was an

Bioactivto as BGHSiO2 (SIKof sodiumMerck, Daart. no. 1.of calcium1.02144.90blended incible. Thepowder wawas poured60 C for 2X-ray diffr

The grathesized vi40 wt.% ofmilled in a230 C usinDenmark) aanalyzed b

2.2. 3D-pr

The gra(Designma3D-printingsaturation,tigated. Difart. no. 304ric acid (H7varying betwere testeddissolved iessary. The3D-printerof solutionand 0.94 mers were bwas variedtaken out opressed airsamples weical strengtthe forming

2.3. Mecha

The merectangularacteristic stest. The incross-head

Table 1Phase composition of the synthesized BGH glass in comparison to the theoreticalcomposition o

ent

, Zweat t

ener

irtuatomo(sur

nterfwit

ults

XRd a p/P ra

RD%. Tnd icenclized= 35. Those t1). T

1meneotinghanith

r bedere

ol. Wcess became instable. Therefore the best solution for theg process was the combination of 1 mol/l of orthophos-acid and 1 mol/l of pyrophosphoric acid with 20 wt.%ropanol. The printed structures could be easily taken outpowder bed and cleaned with compressive air. 0.414 mling solution per mm3 granulate was ideal for a detailedof the printed structures in combination with sufficient

nical strength for cleaning the structures with compres-r. Higher amounts of solution per mm3 granulate degradedfile accuracy.25 Merck Darmstadt, FRG). After heating, the mix-alyzed using XRD.e glass similar to 45S5 Henchglass, further referred

glass, was synthesized using 33.67 g of silicaRONSH 200, Quarzwerke, Frechen FRG), 31.34 gcarbonate anhydrous (Na2CO3 art. no. 6392.5000,rmstadt, FRG), 26.38 g calcium carbonate (CaCO302063.5000, Merck, Darmstadt, FRG) and 8.61 ghydrogen phosphate anhydrous (CaHPO4, art. no.25, Merck, Darmstadt, FRG). The powders werea dry mixer for 30 min and filled in a platinum cru-

crucible was placed into a preheated furnace and thes melted at 1300 C for 1 h and 45 min. The hot melt

into water. The produced glass frit was dried at4 h and analyzed by X-ray fluorescence yield and

action technique.nules consisting of -TCP and BGH glass were syn-a spray drying. First a water based suspension with-TCP and 60 wt.% of BGH glass was prepared andball mill. The milled suspension was granulated atg a spray dryer (Mobile minor 2000, Niro, Soeborg,nd the size distribution of the obtained granulate was

y a laser granulometer.

inting tests

nulate was used as powder in a 3D-printing devicete CX, Z-Corporation, Burlington, USA). For the

tests the binding solution, the layer thickness, theand the geometry in the building chamber was inves-ferent combinations of orthophosphoric acid (H3PO417, SigmaAldrich, Seelze, FRG) and pyrophospho-P2O7 art. no. 60352, SigmaAldrich, Seelze, FRG),ween 0 and 1.5 mol, in combination with isopropanol

as printing solutions. The phosphoric acids weren distilled water and mixed with isopropanol as nec-

solutions were filled into the printing heads of thedevice and printed into the powder bed. The amountprinted per mm3 granulate was varied between 0.376l. Structures with different thicknesses of the lay-

uilt up. The thickness of the applied powder layersbetween 25 and 100m. The printed structures weref the powder bed and cleaned using oil free com-supplied in-house from the university hospital. There post processed at 1000 C to increase the mechan-h. The temperature of 1000 C was chosen to avoidof -TCP, which starts at 1125 C.

nical tests

chanical tests were performed on specimens withcross section (6 mm 8 mm 45 mm). The char-

trength 0 was determined in a four point bendingner and outer roller spans were 20 and 40 mm. Thespeed during loading in the universal testing machine

Compon

SiO2Al2O3TiO2CaONa2OP2O5

(Z030were h

2.4. G

A vputerstl-fileware iprinted

3. Res

Theshowethe Caof the Xthan 1XRD afluorescrystaland 2phasestion cl(Table40 wt.%d50 = 4homog

Prinlower ttures wpowdemol/l wpropanthe proprintinphoricof isopof theof bindprofilemechasive aithe prof 45S5 Henchglass.

45S5 Henchglass theoretical BGH glass

45.00 wt.% 45.35 0.01 wt.%0.00 wt.% 0.14 0.01 wt.%0.00 wt.% 0.02 0.01 wt.%

24.50 wt.% 24.92 0.01 wt.%24.50 wt.% 23.10 0.01 wt.%6.00 wt.% 6.21 0.01 wt.%

ick, Ulm, FRG) was 5 mm/min. Ten specimens thatreated at 1000 C were tested.

ation of a 3D implant model

l 3D model of an implant was derived from a com-graphy data set. The model was converted into aface tesselation language) which is a standard soft-ace for rapid manufacturing devices. The model wash the 3D printer and heat treated at 1000 C.

D analysis of the synthesized calcium phosphateure-TCP phase. Hydroxyapatite, which forms whentio is higher than 1.5, was not detected. The accuracyanalysis of-TCP in respect of the HA content is lesshe synthesized BGH glass was also analyzed usingts element composition was determined using X-raye yield. The XRD diagram did not reveal any peaks ofmaterial, only an intensity increase between 2 = 28, which is typical for the short range order of glass

e X-ray fluorescence yield showed a phase composi-o the theoretical composition of 45S5 Henchglasshe spray dried granules of 60 wt.% BGH glass and

-TCP had a grain size distribution of d10 = 18m,and d90 = 92m. The grains consisted out of a

us mixture of BGH glass and -TCP (Fig. 1).solutions with concentrations of phosphoric acid

2 mol/l were not suitable for binding the printed struc-high enough strength to get them safely out of the. Concentrations of phosphoric acid with 2 or moreonly printable with the addition of 20 wt.% of iso-ithout isopropanol the print heads overheated and

C. Bergmann et al. / Journal of the European Ceramic Society 30 (2010) 25632567 2565

Fig. 1. SEM micrograph of a 60 wt.% BGH glass and 40 wt.% -TCP granule.Both components are distributed homogeneously.

The layer thickness had also a high influence on the mechani-cal strengthor higher oners easily bsmall layerof the layerwere achie

The mecheat treatmof the specheat treatmstructures.before andthe only doand was rep

Fig. 2. XRDand at 1050 Cture. After th(NaCaPO4 an

4. Discussion

The calcium to phosphorus ratio of hydroxyapatite (HA)is 10% hiCa/P = 1.67second phamore calciuof -TCP.(CPP) (Ca/0.3% of CPing mixturemain peakdiagram inless than 1%

The BGlar, but notHenchglasssition rangresearchersbe bond to

The sizeulates were

durieneolasstheted ithechancoun

Highd carateayerers,

ntingctureereof the printed structures. With a thickness of 87.5mly very brittle structures were producible. Single lay-roke under shear stresses. On the other hand a toothickness of 35m or lower caused direct sheering

s during the printing process. The best printing resultsved using a layer thickness between 50 and 75m.hanical properties were significantly increased by theent of the printed structures. The bending strengthimens treated at 1000 C was 14.9 3.6 MPa. Theent changed the phase composition of the printedFig. 2 shows the XRD diagrams of the specimensafter the heat treatment. The -TCP phase which isminant phase in the non-treated sample disappearedlaced by a phase mixture of NaCaPO4 and CaSiO3.

layershomogBGH gduringfabrica

Forthe meinto aclayer.der bedeteriothick lder laythe priof struglass wdiagrams of an untreated (grey line marked with black squares)heat treated (black line marked with white circles) printed struc-

e heat treatment, the -TCP phase disappeared and new phasesd CaSiO3) occurred.

tration of t1 mol/l of p0.414 ml ofthese paramCT data wa

The heincreasedthan a poscalcium phapproximabone.24

Reactioglass durinated the phboth bioacon the phasRespectingphase of thThereforegher than that of -TCP (-TCP Ca/P = 1.5, HA). One may assume that 1% of HA will form as ase during the synthesis of -TCP, when there is 0.1%m in the mixture than needed for the exact Ca/P ratio

If the ratio is lower than 1.5 calcium pyrophosphateP = 1) will be formed as a second phase. In this caseP will form, if there is 0.1% less calcium in the start-than needed for the synthesis of-TCP. The missing

of HA at 2 = 31.73 (ICDD no. 24-33) in the XRDdicates a phase purity of the -TCP with an error of

, as Peters et al. already described elsewhere.18H glass was synthesized with a composition simi-exactly as the theoretical composition of the 45S5

. Yet the BGH glass is still within the compo-e of a bioactive glass, as described by different.14,19,20 As found by the researchers the glass canboth, hard and soft tissue.distribution and round shape of the spray dried gran-suitable for producing plain and flawless powder

ng the 3D-printing process. The granules had also aus distribution of the two components -TCP and, which is important for a reliable binding reactionprinting process and homogeneous properties of themplant.optimization of the 3D-printing parameters not onlyical strength of the build-up structure had to be takent, but also the profile and the bonding between eacher amounts of acid solution printed into the pow-used an intensification of the binding reaction, butd the profile of the 3D structures. A too thin or toothickness influenced the bonding between the pow-resulting in shearing of the layers during or afterprocess. The best parameters found for the printings using a granulate consisting of -TCP with BGHa layer thickness between 50 and 75m, a concen-he binding solution of 1 mol/l orthophosphoric acid,yrophosphoric acid and 20 wt.% of isopropanol, andsolution printed per 1 mm3 -TCP granulate. Witheters the reconstruction of an implant derived froms printed (Fig. 3).

at treatment of the -TCP, BGH glass mixturethe four point bending strength two times highert treatment of printed structures consisting of pureosphates.2123 Nevertheless the bending strength oftely 15 MPa is still 10 times lower than that of natural

ns between the -TCP, the DCPD and the bioactiveg the heat treatment of the printed structures, gener-ases CaNaPO4 and CaSiO3. Those two phases are

tive with the potential of biodegradation dependinge composition and the porosity of the implant.2530the printing process of the structures, the glassye granules has no effect on the cement reaction.

the glass content can be varied to generate tailored

2566 C. Bergmann et al. / Journal of the European Ceramic Society 30 (2010) 25632567

Fig. 3. Printeimplant was p-TCP and B

biodegradaprocess.

5. Conclu

The stuTCP and aprinted struresolutionStructuresing implanthe phase crhenanite (compatibilOnly the while the gbility of thbiodegrada

The disahigh produtially degrafor the patiweaken thether develoto be improity. The fothe additio

Acknowled

The authMineral Enwith the ch

Reference

1. Infante-CGarcia Afacial su2007;29

2. Vail TV, Urbaniak JR. Donor-site morbidity with use of vascularized auto-genous fibular. J Bone Joint Surg Am 1996;78:20411.

emers FW, Blokhuis TJ, Patka P, Bakker FC, Wippermann BW,armanin tr3;66

ringtomplic6;32yssinseointter Rbsteropharuganed naDoe

nthe Gh fouin Cs calc3;17

shinoercalataininmate

eiss Fcompmateelt D,hnerciumoverical m6;27

nch6;1:3

nch6;17tz H,pouros Partrz FCrpholls 200ers Fa-tric7.

derssacce

50.d structure of a model of an implant generated from CT data. Therinted with a commercial 3D printer using granulate consisting ofGH glass.

tion capabilities of the implant after the treatment

sion

dy confirmed that a composite material made of -bioactive glass is suitable for the build-up of 3D

ctures. The samples were printed with a maximumof 50m according to the smallest applied layers.generated from computer tomography data, display-ts for real applications, were producible. Althoughomposition changed to wollastonite (CaSiO3) and

CaNaPO4) after the post process at 1000 C the bio-ity should still be given as proved by others.2530-TCP phase was necessary for the printing process,lass phase had no influence. Therefore the adjusta-

e glass composition holds the possibility for tailoredtion kinetics in an in vivo application.dvantage of customized, 3D-printed implants is the

ction costs for a limited number of cases. Also a par-dation of synthetic implants is not a priori advantageent in every case, because remaining fragments canmechanical strength of the new grown bone. In fur-

pments the degradation behavior of the implants hasved by optimizing the phase composition and poros-

3. BloHaics200

4. ArCo199

5. FraOsMa

6. Wenan

7. Mubas

8. vonLewit

9. Kleiou198

10. Hointcon

Bio11. Th

BioBio

12. ApBocal

13. Grlog200

14. He198

15. He200

16. SeiofRe

17. Fiemo

ria18. Pet

bet203

19. Anthe145rmation of new tissue should also be stimulated by

n of growth factors or other pharmaceuticals.

gement

ors want to acknowledge Dr. Arno Kaiser, Institute ofgineering, RWTH Aachen University, for his supportaracterisation of BGH glass.

s

ossio P, Gutierrez-Perez JL, Torres-Lagares D, Garcia-Perla, Gonzalez-Padilla JD. Bone cavity augmentation in maxillo-

rgery using autologous material. Rev Esp Cir Oral Maxillofac:719.

20. Kokuboterials 1

21. GbureckResorbaprinting.

22. Gburecktricalciuing. J M

23. Mller Bal. Bio-m2009;74

24. Reilly DJoint Sur

25. Knabe CH. Therapidly33544.HJTM. Autologous bone versus calcium-phosphate ceram-eatment of experimental bone defects. J Biomed Mater Res BB:52631.n ED, Smith WJ, Chambers HG, Bucknell AL, Davino NA.ations of iliac crest bone graft harvesting. Clin Orthop Relat R9:3009.et P, Mathon D, Lerch A, Autefage A, Collard P, Rouquet N.egration of composite calcium phosphate bioceramics. J Biomedes 2000;50:12530.TJ, Siegel RW, Bizios R. Enhanced functions of osteoblasts onse ceramics. Biomaterials 2000;21:180310.

R. Bioresorbable composite bone paste using polysaccharideno hydroxyapatite. Biomaterials 2004;25:382935.rnberg MC, von Rechenberg B, Bohner M, Grnenfelder S, vanH, Mller R, et al. In vivo behavior of calcium phosphate scaffolds

r different pore sizes. Biomaterials 2006;27:518698.PAT, Driessen AA, de Groot K. Biodegration behavior of var-ium phosphate materials in bone tissue. J Biomed Mater Res

:76984.M, Egi T, Terai H, Namikawa T, Takaoka K. Repair of long

ted rib defects using porous beta-tricalcium phosphate cylindersg recombinant human bone morphogenetic protein-2 in dogs.rials 2006;27:493440., Apelt D, Brand B, Kutter A, Zlinszky K, Bohner M, et al.atibility and resorption of a brushite calcium phosphate cement.rials 2005;26:438394.Theiss F, El-Warrak AO, Zlinszky K, Bettschart-Wolfisberger R,

M, et al. In vivo behavior of three different injectable hydraulicphosphate cements. Biomaterials 2004;25:143951.ML, Gbureck U, Wright AJ, Tremayne M, Barralet JE. Bio-

ediated resorption of brushite cement in vitro. Biomaterials:217885.LL. Ceramic implants for humans. Adv Ceram Mater0624.LL. The story of Bioglass. J Mater Sci: Mater Med:96778.Rieder W, Irsen S, Leukers B, Tille C. Three-dimensional printingus ceramic scaffolds for bone tissue engineering. J Biomed MaterB: Appl Biomater 2005;74B:7828., Beckmann F, Huser M, Irsen SH, Leukers B, Witte F, et al. Theogy of anisotropic 3D-printed hydroxyapatite scaffolds. Biomate-8;29:3799806., Reif D. Functional materials for bone regeneration fromalcium phosphate. Mat -wiss u Werkstofftech 2004;35:

on H, Liu G, Kangasniemi K, Juhanoja J. Evaluation ofptance of glass in bone. J Mater Sci: Mater Med 1992;3:

T. Bioactive glass ceramics: properties and applications. Bioma-991;12:15563.U, Hlzel T, Klammert U, Wrzler K, Mller FA, Barrale JE.

ble dicalcium phosphate bone substitutes prepared by 3D powderAdv Funct Mater 2007;17:39405.U, Hlzel T, Biermann I, Barralet JE, Grover LM. Preparation of

m phosphate/calcium pyrophosphate structures via rapid prototyp-ater Sci: Mater Med 2008;19:155963.

, Deyhle H, Fierz FC, Irsen SH, Yoon JY, Mushkolaj S, etimetic hollow scaffolds for long bone replacement. Proc SPIE

01:74010D1D.T, Burstein AH. The mechanical properties of cortical bone. J Boneg Am 1974;56:100122., Berger G, Gildenhaar R, Howlett CR, Markovic B, Zreiqatfunctional expression of human bone-derived cells grown onresorbable calcium phosphate ceramics. Biomaterials 2004;25:

C. Bergmann et al. / Journal of the European Ceramic Society 30 (2010) 25632567 2567

26. Ramselaar MMA, Van Mullem PJ, Kalk W, Driessens FCM, De WijnJR, Stols ALH. In vivo reactions to particulate rhenanite and particulatehydroxylapatite after implantation in tooth sockets. J Mater Sci: MaterMed 1993;4:3117.

27. Driessens FCM, Ramselaar MMA, Schaeken HG, Stols ALH, VanMullem PJ, De Wijn JR. Chemical reactions of calcium phosphateimplants after implantation in vivo. J Mater Sci: Mater Med 1992;3:4137.

28. Jalota S, Bhaduri SB, Tas AC. A new rhenanite (-NaCaPO4) and hydrox-yapatite biphasic biomaterial for skeletal repair. Mater Res Part B: ApplBiomater 2007;80B:30416.

29. Teramoto H, Kawai A, Sugihara S, Yoshida A, Inoue H. Resorption ofapatite-wollastonite containing glass-ceramic and beta-tricalcium phos-phate in vivo. Acta Med Okayama 2005;59:2017.

30. Wue W, Liu X, Zheng XB, Ding C. In vivo evaluation of plasma sprayedwollastonite coating. Biomaterials 2005;26:345560.

3D printing of bone substitute implants using calcium phosphate and bioactive glassesIntroductionExperimentalPreparation of granulate3D-printing testsMechanical testsGeneration of a 3D implant model

ResultsDiscussionConclusionAcknowledgementReferences