Mechanism of calcium disilicide-induced calcification of crystalline silicon surfaces in simulated body fluid under zero bias

Mechanism of calcium disilicide-induced calcificationof crystalline silicon surfaces in simulated bodyfluid under zero bias

Vladimir V. Seregin, Jeffery L. CofferDepartment of Chemistry, Texas Christian University, TCU Box 298860, Fort Worth, Texas 76129

Received 22 August 2006; revised 20 June 2007; accepted 20 July 2007Published online 13 December 2007 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/jbm.a.31691

Abstract: A dry-etch spark ablation method was used toproduce calcium disilicide (CaSi2/Si) layers on silicon sur-faces, and their biomineralization under zero bias was fol-lowed by means of scanning electron microscopy, X-rayenergy dispersive analysis, and Raman spectroscopy.CaSi2/Si wafers are bioinert at 258C and bioactive at 378C.Mechanistic insights regarding biomineralization werederived from an analysis of film growth morphology andchemical composition after various soaking periods instandard simulated body fluid (SBF). Changes in CaSi2 cal-cification behavior as a function of reaction temperatureand pH, SBF concentration, and various surface modifica-tion processes were also employed for this purpose. Dur-

ing CaSi2/Si calcification under zero bias, calcium phos-phate (CaP) growth is strongly dependent on the structuraldegradation of CaSi2 grains. Surface silanol groups, ini-tially present on the as-prepared material, cannot induceCaP nucleation, which begins only upon delamination ofCaSi2 layers. The calcium phosphate phases, which arepresent during various growth stages, possibly include acombination of Mg-substituted whitlockite, monetite, andtricalcium phosphate. � 2007 Wiley Periodicals, Inc.J Biomed Mater Res 87A: 15–24, 2008

Key words: calcium disilicide; spark ablation; calcification;bioactive silicon

INTRODUCTION

Many developed tissue engineering constructs,drug delivery systems, and biosensors integratewithin living organisms by employing tailored phys-icochemical properties. The bioactivity of such devi-ces is strongly dependent on complex chemical reac-tions. An alternative comes from the emerging fieldof bioelectronics,1 which targets the development ofmaterials whose biologically relevant actions can besignificantly altered by a simple application of elec-tric bias. As the search for new biomaterials contin-ues, one focus point is the development of bioactivesilicon-based platforms for constructing integratedcircuits which can operate in vivo.

Recently, our group demonstrated that CaSi2 grains,which were incorporated into Si wafers via a sparkablation technique, act as active growth layers forthe bias-assisted deposition of calcium phosphates(CaPs) in vitro.2 This approach, in contrast to thewell-studied solution phase routes to nanostructured

CaP,3 allows one to form patterned synthetic bonefilms on silicon substrates by taking advantage ofthe difference in conductivity between silicon andCaSi2. Calcium disilicide is a semimetal, and it is thestructure of the polymorphic CaSi2 bulk that accountsfor its metallic character, according to numerous theo-retical4–7 and experimental5,8 studies. In particular,the overlap between p and d orbitals of Si and Ca,respectively, results in a reduced energy gap betweenconduction and valance bands of the bulk crystallinematerial. Under electric bias, a bioactive CaP filmselectively forms on the more conductive CaSi2growth layers, which can be deposited in the form ofa pattern on semiconducting silicon platforms.2

In the last three decades, research pertaining tocalcium silicides was primarily focused on investigat-ing their fundamental properties with regard topotential microelectronic applications.9 At elevatedtemperatures, many metals react with silicon to formsilicides,9 and studies in this area can help to solvetechnological problems associated with metal–semi-conductor interfaces. Calcium silicides can be consid-ered structural analogs of silicides formed by otherelements, including rare-earths,5 and therefore mayserve as model systems for various heterojunctionsbetween silicon device components and metal con-

Correspondence to: J. L. Coffer; e-mail: [email protected] grant sponsor: Robert A. Welch Foundation

� 2007 Wiley Periodicals, Inc.

tacts. Unfortunately, methods for fabricating epitaxialfilms of calcium silicides on silicon are limited to theevaporation of calcium metal onto Si wafers followedby an anneal to produce CaSi2.

10–12 In addition, CaSi2and CaSi melts, prepared by heating of Ca and Si instoichiometric proportion, exhibit superconductingbehavior at high pressures,13,14 and show potentialfor reversible hydrogen storage,15 respectively.

This manuscript further explores the potentialbiomaterial applications of calcium disilicide bydemonstrating its bioactivity in vitro in the absenceof electric bias, and contrasts these results with pre-viously reported electrically induced effects.2 Here,the biomineralization of CaSi2 growth layers onspark-processed silicon in simulated body fluid(SBF) was followed by means of scanning electronmicroscopy (SEM), energy dispersive X-ray (EDX)analysis, and Raman spectroscopy. It is shown thatgrains of CaSi2, comprised of silicon sheets interca-lated with calcium,4 decompose through a partialdelamination of silicide layers and grow CaP phasesvia a mechanism similar to that for bioactive glass/ceramics.16,17 Mechanistic insights regarding biomi-neralization are derived from an analysis of filmgrowth morphology and chemical composition aftervarious soaking periods in standard SBF. Changes inCaSi2 calcification behavior as a function of reactiontemperature and pH, SBF concentration, and varioussurface modification processes were also employedfor this purpose.

MATERIALS AND METHODS

Spark-processing of Si

Pieces (8 mm 3 14 mm) of p-type, <100>, boron-dopedCZ Si (9–18 X cm) were pre-cut by a diamond saw fromtwo Si wafers held together through their polished sidesby vacuum grease or parafilm. To obtain a hydrophilicsurface, any organic residue was removed prior to spark-ing through a series of ultrasound-assisted sequentialimmersions in hexane, chloroform, acetone, and de-ionized(DI) water. Thereafter, wafers were soaked in an

(NH4)2S2O8/H2SO4 (conc.) cleaning solution for 12–24 huntil a moment prior to spark ablation, at which pointthey were removed from the cleaning solution, thoroughlywashed with DI-water, and dried under a stream of N2.

Spark ablation was performed in a manner similar to apreviously published procedure.2 In this process, an Niwire cathode was inserted inside a glass capillary tube(1.2 mm inner diameter), which is partially filled withCaSi2 powder (Johnson Matthey) and pressed against a sil-icon wafer (anode) at the other end. Immediately prior tosparking, the capillary tube was dipped into the CaSi2powder up to a fixed 6-mm mark and mounted on top ofthe wafer. A conical steel plate (diameter � 1.2 mm) wasthen inserted inside the tube on top of the CaSi2 powder,and the tip of the cathode positioned at a fixed distance(3 mm) above the plate. The excess powder on the outsidewas removed by a gentle stream of N2. The contact to Siwas made with an alligator clip. All samples were exposedto the Tesla coil discharge for 900 s. Upon completion ofthe sparking events, samples were rinsed under a strongfocused stream of DI-water to remove any loose particlesand dried under a stream of N2. In some cases, waferswere cleaved to remove those parts of the wafers wherethe alligator clip was attached. Such samples possessslightly smaller dimensions (8 mm 3 12 mm) and arereferred to as ‘‘cleaved’’ in the discussion that follows.

Preparation of SBF and its modified solutions

The SBF solutions were prepared according to a litera-ture procedure.18 The pH values and the ion concentra-tions of the SBF and its modified solutions are shown inTable I. Each solution was prepared by dissolving reagentgrade CaCl2 2H2O, Na2SO4, NaCl, KCl, MgCl2�6H2O,NaHCO3, and K2HPO4 in a buffer of trishydroxymethyl-aminomethane and HCl.

Calcification at 258C

As-prepared samples (cleaved and uncleaved) wereimmersed in 20- and 80-mL aliquots of standard andmodified SBFs for set periods of 3, 4, and 5 weeks at roomtemperature. The samples were then removed form agiven solution, rinsed with DI-water, dried under a streamof N2 and analyzed by SEM/EDX. Surface analysis wasdone using a JEOL JSM-6100 scanning electron microscope

TABLE IIon Concentrations of Human Blood Plasma, SBF,

and its Modified Solutions18

Concentration (mM)

pH Ca2þ HPO422 Kþ Naþ Mg2þ Cl2 CO3

22 SO422

Blood plasma 7.40 2.50 1.00 5.00 142 1.50 103.0 27.0 0.50SBF 7.40 2.50 1.00 5.00 142 1.50 147.8 4.20 0.50pH 7.60 SBF 7.60 2.50 1.01 4.99 142 1.50 147.8 4.19 0.51Ca 1.50 SBF 7.40 3.74 1.01 5.01 142 1.50 150.3 4.20 0.50P 1.50 SBF 7.40 2.49 1.50 5.98 142 1.50 147.8 4.19 0.50

16 SEREGIN AND COFFER

Journal of Biomedical Materials Research Part A

coupled with an EDX analysis detector under 20-kV accel-erating voltage.

Surface treatments and calcification at 378C

Three types of surface-treated CaSi2-containing waferswere studied: (a) phosphate (P–CaSi2), (b) magnesium ion(Mg–CaSi2), and hydrofluoric acid (HF–CaSi2). In the caseof P–CaSi2 samples, the as-prepared CaSi2/Si wafers wereimmersed in 2 mL aliquots of K2HPO4 (0.5M) at 378C for2 h. Some of these samples were then immediately placedin SBF, while others were first pre-washed with DI-water/dried under a stream of N2 prior to SBF immersion. As-prepared Mg–CaSi2 samples were immersed in 2-mL ali-quots of MgCl2 (0.5M) at 378C for 2 h. These samples werethen pre-washed with DI-water, dried under a stream ofN2 and placed in SBF. In case of HF–CaSi2 samples, as-pre-pared CaSi2/Si wafers were dipped into a 48% HF solutionfor 30 s, then thoroughly washed with ethanol/DI-water,and dried under a stream of N2 prior to SBF exposure.

Surface-modified samples (alongside an as-preparedcontrol) were immersed in 3-mL aliquots of standard SBFfor 4 weeks. SBF was replaced with a fresh solution afterthe first 2-week period. The samples were then removedfrom the SBF, rinsed with DI-water, dried under a streamof N2 and analyzed by SEM/EDX. Next, as-prepared sam-ples were immersed in 25-mL aliquots of standard SBF forvarious set periods of time up to 4 weeks. SBF wasreplaced with a fresh solution after the first 2-week period.The samples were then removed from the SBF, rinsed withDI-water, dried at 378C (�24 h), and analyzed by SEM/EDX and Raman. Raman measurements were obtained ona custom-built system employing an Arþ laser source inconjunction with a Roper Scientific intensified CCD and aHolo Spec VPT System from Kaiser Optical Systems.

RESULTS

Calcification at 258C

Immersion of CaSi2/Si wafers in standard SBF atroom temperature results in no CaP formation on CaSi2growth layers, as assessed by FTIR. Similar resultswere obtained when SBF solutions modified withrespect to pH, Ca2þ, or HPO4

22 were used.SEM analysis of these samples indicates that the physi-cal state of CaSi2 grains has not significantly changed,while the corresponding EDX spectra show �2:1 Si/Capeak ratios, characteristic of original CaSi2 (Fig. 1). Aslight decrease in calcium peak intensities reflects par-tial ion-leaching, but this process is inefficient at 258Ccompared with the results obtained at 378C (vide supra).The only deposition products that were found on CaSi2growth layers during these experiments were isolatedspherulites such as the one shown in Figure 2. Compo-sitional analysis of these spherulites by EDX typicallyreflects low Ca levels, often at the detection limit of ourinstrument. Nonetheless, occasionally Si/Ca ratios onthe order of 3:1 are observed. Aside from Si and Ca, noother elements were detected, which indicates that cal-cium within the spherulites is most likely ion-pairedwith oxygen, carbonates, or silicates. It must also benoted that the routinely used EDX instrument isequipped with a beryllium window, and the presenceof oxygen and carbon is invisible to this detector.

Interestingly, CaP deposits were consistently foundon the uncleaved wafers in the areas where the Teslacoil electrode had been connected to Si. Minor siliconwafer surface erosion at these locations (as a conse-

Figure 1. Plan view SEM and EDX spectra of CaSi2 grains on spark-processed Si after 4 weeks in modified SBFs: (A) Ca1.50 SBF, (B) P 1.50 SBF, (C) pH 7.60 SBF, and (D) the EDX spectra associated with (A)–(C). All spectra are normalizedwith respect to Si peak intensity.

MECHANISM OF CALCIUM DISILICIDE-INDUCED CALCIFICATION 17


quence of the plasma discharge) is expected to pro-duce porous silica layers with a relatively high den-sity of silanols.2 Successful calcification of porousSiO2 layers alongside inactive CaSi2, both of whichhave been introduced onto the wafer surface viaspark ablation, is a positive control, which rules outpotential system contamination as a reason for theobserved lack of CaP formation on CaSi2.

Overall, CaSi2 appears to be bioinert at room tem-perature under the conditions of zero bias. In con-trast to these observations, similar CaSi2/Si samplescalcified readily in standard and all of the modifiedSBF solutions when an electric current (�0.7 mA)was applied, and produced CaP films with the samemorphology and composition, as had been shown inour previous work.2 Hence, the reference to the elec-trodeposition experiments is made only to underlinethe increased role of CaSi2 surface chemistry in zero-bias biomineralization in vitro.

Calcification of CaSi2/Si wafers at 378C

When the temperature is raised to 378C, as-pre-pared and surface-modified CaSi2/Si wafers calcifyreadily within a 4-week exposure period to SBF. Fur-thermore, the deposited CaP films, which cover theentire wafer surfaces, can be observed with theunaided eye when samples are first removed fromtheir corresponding SBF solutions. In the case ofmost samples, a subsequent gentle rinse with DI-water and/or the stream of N2 (which was appliedduring the drying step) removed overgrowth layersfrom the polished wafer surfaces around the sparkedareas, leaving the deposited CaP films anchored only

to CaSi2 growth layers. This delamination appearedto occur with varying degrees of difficulty, reflectingthe structural integrity of the deposited CaP films.For example, parts of the overgrowth layer on Mg–CaSi2 wafers separated from the surface on first con-tact with DI-water stream, while CaP films remainedattached to P–CaSi2 surfaces even after 3 min of sus-tained irrigation and exposure to a focused N2

stream (see Fig. 3). Overall, a SEM/EDX analysis

Figure 2. Plan view SEM of an isolated spherulite onCaSi2 growth layer, which was exposed to ‘‘P 1.50 SBF’’ so-lution for 4 weeks at 258C.

Figure 3. A CaP overgrowth layer formed on the surfaceof a CaSi2/Si wafer after surface treatment with K2HPO4

followed by a 4-week soak in SBF at 378C: (A) plan viewSEM, (B) tilted view showing the wafer edge covered byCaP, and (C) magnified view of (B).



revealed rather subtle changes in morphology andcomposition of the final CaP films formed on CaSi2growth layers of all wafers used in these experi-ments. Nonetheless, some mechanistic insightsregarding biomineralization can be inferred from thedifferences in coverage and chemical composition ofCaP films deposited from SBF after various surfacetreatments had been applied to CaSi2/Si wafers.

The interfacially controlled nucleation of CaP isexpected to begin once CaSi2 growth layers havedeveloped an appropriate negative surface potentialto gather positive calcium ions in SBF. Calcium disi-licide is a particularly interesting material in thisregard, since upon immersion in SBF, it may developthe necessary negative surface potential by exposingSi-OH (silanol) groups and/or by forming a complexbetween the silicide-bound calcium and the phos-phates from the fluid. In the latter case, biominerali-zation can occur in a manner similar to synthetic hy-droxyapatite.19 This is in addition to the surface ero-sion of CaSi2 grains in SBF and potential similaritieswith the known biomineralization mechanism of Bio-glassTM (vide supra).16

Figure 4 shows a typical morphology and composi-tion associated with as-prepared CaSi2 growth layerson spark-processed Si, which were calcified during 4weeks in SBF at 378C. Spherulites with diametersranging from 1 to 3 lm form a thin plate-like film atthe base of larger uniquely shaped CaP structures,which emerge above the wafer surface due to thelayer overgrowth. Such CaP films crack under rapiddrying conditions (or evacuation of the SEM cham-ber) and often appear in SEM images as isolated pla-telets. In addition, continuous overgrowth layers(similar to Fig. 3) cover the entire polished wafer sur-

faces prior to the DI-water rinse, which indicates thatthe system is under diffusion-controlled growth bythe end of the 4-week soak in SBF, and the interfa-cially controlled CaP growth is complete.

Effects of surface treatments

In principle, the ability of the CaSi2 growth layerto induce nucleation of CaP may be strongly influ-enced by the surface Si-OH groups, much like lowporosity silicon.20 To evaluate the extent of surfacesilanol contribution to the CaP nucleation process,CaSi2/Si wafers were hydride-terminated via a sim-ple immersion in HF solution. Both the morphologyand composition associated with HF–CaSi2 samplesappear strikingly similar to the untreated CaSi2/Siwafers discussed above. Also similar to the control,continuous overgrowth layers initially covered theentire wafer surfaces, and upon their delaminationduring the DI-water rinse, one could observe multi-ple holes present within the exposed polished siliconareas, which probably were formed prior to SBF ex-posure during the HF-etch of p-type Si.

Exposure of CaSi2 growth layers to HPO422 and

Mg2þ containing solutions were originally expectedto increase and decrease the rate of CaP nucleation,respectively. Magnesium is known to inhibit biomi-neralization processes in vitro and in vivo,21,22 whilerates of calcification of bioactive glass/ceramics areraised with higher phosphate concentration in theirmatrix.16,21 Both Mg– and P–CaSi2 samples formedcontinuous CaP films throughout their surfaces uponSBF exposure. However, the former appeared underSEM imaging to have the thinnest overgrowth layers

Figure 4. A calcified CaSi2 growth layer on spark-processed Si after 4 weeks in SBF at 378C: (A) plan view SEM, (B) mag-nified view of (A), and (C) the EDX spectrum associated with (B).



and showed the largest degree of delamination ( seeSection ‘‘Calcification of CaSi2/Si wafers at 378C’’)during the postgrowth washing steps. On the otherhand, samples which had been exposed to the phos-phate solution possessed the thickest CaP, whichappeared barely cracked during the final SEM analy-sis. Figures 5 and 6 show the local microstructuresand associated compositions of calcified CaSi2growth layers on P– and Mg–CaSi2 wafers, respec-tively. For a given sample type, no significant differ-ence was noted between the samples which werepre-washed with DI-water prior to SBF immersionsand those wafers which were placed in SBF immedi-ately after the surface treatment.

Table II presents the Ca/P peak ratios which wereobtained from the EDX analysis of surface-modi-fied/calcified CaSi2 growth layers on Si. Variation inthese values reflects changes in the stoichiometry ofthe final CaP films depending on the type of surfacetreatment, and suggests relative biomineralizationrates for a constant SBF exposure time. The similar-ity between untreated samples and those whichwere hydride-terminated in HF indicates that the

rate of CaP nucleation remains unaffected despitethe reduced number of Si-OH groups on the surfaceof CaSi2. Therefore, CaSi2 growth layers do not initi-ate CaP nucleation exclusively through surface sila-nols. The initial presence of phosphates on thesurface of CaSi2, on the other hand, results in Ca-de-ficient phases of the final CaP films, as reflectedby the lower Ca/P peak ratio obtained from theP–CaSi2 samples. Substitution of Ca2þ in an apatiteframework by other cations (e.g. Naþ, Mg2þ) is anindication of an accelerated CaP growth rate. Surfacereactions with phosphate ions from the K2HPO4 so-lution can affect the nucleation rate of CaP in twoways: (1) phosphates can chemically attach to the sil-icide-bound calcium ions, thereby contributing to thenegative surface potential, and (2) physisorbed phos-phates can detach from the surface once the sampleis in SBF and increase the local saturation of SBFwith respect to CaP, thereby accelerating the onset ofnucleation. Contribution of the latter can be excluded,since similar results were obtained with those sam-ples which had been pre-washed in DI-water prior toSBF exposure. Finally, exposure to MgCl2 results in a

Figure 5. A calcified CaSi2 growth layer on spark-processed Si after surface treatment with K2HPO4 followed by a DI-water-rinse and a 4-week soak in SBF at 378C: (A) plan view SEM and (B) the EDX spectrum associated with (A).

Figure 6. A calcified CaSi2 growth layer on spark-processed Si after surface treatment with MgCl2 followed by a DI-water-rinse and a 4-week soak in SBF at 378C: (A) plan view SEM and (B) the EDX spectrum associated with (A).



slower nucleation of CaP, as indicated by theincreased Ca/P peak ratio. In SBF, hydrated magne-sium is expected to poison CaP nucleation sites onCaSi2 by complexing with surface silanols and shield-ing local surface areas, which may otherwise partici-pate in the nucleation process. Similar to the case ofP–CaSi2 samples, the contribution of physisorbedMgCl2 can be excluded. Overall, under zero-bias con-ditions, biomineralization of CaSi2 growth layers onspark-processed silicon is most affected by the reac-tion temperature, while the functional groups, whichare present on the surface of as-prepared silicide, donot significantly contribute to this process.

Evolution of CaSi2 morphology

As discussed previously, CaSi2 grains retain theirphysical state even after a month in various SBF sol-utions at 258C (see Fig. 1). When CaSi2/Si wafers areimmersed in SBF at 378C, CaSi2 undergoes gradualstructural decomposition. Figure 7 shows delamina-tion of CaSi2 grains during the first week of SBF ex-

posure, by the end of which the grains appear underthe SEM as piles of thin plates [see Fig. 7 (C)]. Nosimilar structures were found present on the surfaceof CaSi2/Si wafers, which had been exposed to SBFfor longer periods of time. Therefore, structuraldecomposition of CaSi2 is understood to come tocompletion within 10 days of a biomineralizationprocess at 378C. In addition, as the surface evolves,the concentration of calcium within these grainsreduces [Fig. 7 (D)].

Another change in calcification behavior of CaSi2growth layers which is induced by the temperatureelevation is the formation of isolated spherulites(data not shown). On average, these spherulitesappear 3–5 times smaller than those that form at258C (see Fig. 2). The frequency of occurrence ofsuch structures increases as biomineralization pro-gresses. Although it is difficult to establish a defi-nite trend in their compositional change using ourEDX instrument (aside from Si and Ca, no otherelements were detected), a typical starting calciumpeak intensity (1:2 ratio relative to Si) appears togradually decrease during the first two weeks andremain constant thereafter, until CaP overgrowthlayers cover wafer surfaces and such spherulites areno longer visible. Incorporation of phosphate ionswithin these spherulites, indicated by appearance ofthe phosphorous signature peak in their associatedEDX spectra, occurs in between 16 and 20 days inSBF.

A typical Raman spectrum associated with calci-fied CaSi2 layers, which had been exposed toSBF for 28 days at 378C, is presented in Figure 8

TABLE IICa/P Peak Ratios Obtained from the EDX Analysis of

Surface-Modified/Calcified CaSi2 Growth Layers

Surface Treatment

None HF K2HPO4 MgCl2

Ca/P 1.44 1.45 1.34 1.52

Figure 7. Plan view SEM and EDX spectra of time-dependent surface evolution of CaSi2 grains on spark-processed Si inSBF at 378C: (A) day 3, (B) day 5, (C) day 7, and (D) the EDX spectra associated with (A)–(C). All spectra are normalizedwith respect to Si peak intensity.



(an enlargement of the P–O stretching region bet-ween 800 and 1200 cm21).23 The peaks at 951 cm21

and 977 cm21 are due to the phosphate symmetricstretch modes, but the spectrum itself does not iden-tify with any of the known CaP phases. The mea-sured Raman spectra resemble those for tricalciumphosphate (peaks at 947 and 970 cm21), whitlockite(broad peak at 970 cm21), and monetite (slopingbroad peak at 986 cm21).23,24 The peaks appearslightly shifted toward lower wavenumbers comparedwith those samples that had been soaked in SBF forshorter periods of time. Overall, the data suggest thatCaP films, which deposit on CaSi2/Si within a month,may include a mix of several CaP phases such as Mg-substituted whitlockite, tricalcium phosphate, andmonetite.

DISCUSSION

The growth of biomimetic CaP and carbonatephases in vitro and in vivo is strongly dependent ona great variety of experimental parameters, and, as aresult, many researchers establish their own theoriesregarding the biomineralization mechanism of anyparticular substrate.16,17,21,23,25 For example, themechanism of carbonate incorporation into an apa-tite framework is still under debate;23 even less isknown about the role of magnesium in mineraliza-tion processes occurring in biological systems.22

Studies of biomineralization rates to determine thegrowth mechanisms are relatively scarce and usuallydeal with bulk solution data, when direct observa-tions in situ or postgrowth surface microtopogra-

phies are needed.25 It is generally accepted, however,that a substrate can initiate CaP nucleation in vitroonce it develops an appropriate negative surfacepotential to gather positive calcium ions in SBF. Forexample, nucleation of apatite on a Na2O–SiO2 glassis initiated by the negative charge due to the surfaceSi-OH;26 a similar effect can be achieved throughsurface hydroxyl, carboxyl, or phosphate groups. Forthose materials whose surface erodes during thecourse of biomineralization, such as CaSi2, the situa-tion is more complex. The five-stage mechanism forhydroxycarbonate apatite formation on the surface ofbioactive glass, formulated by Hench and Wilson16

dictates that upon proper surface erosion, calciumand phosphate ions within bioglass migrate to thesurface and form a CaO–P2O5-rich film on top of theSiO2-rich layer.16 This, in turn, triggers the growth ofamorphous CaP through electrostatic ion accumula-tion and its later crystallization into various apa-tites.15 Our work on CaSi2 draws many similaritieswith this well-established mechanism.

In theory, CaSi2 grains, where silicon sheets areintercalated with calcium,4 may initiate CaP nuclea-tion by utilizing surface Si-OH groups similar to bio-active glasses or as synthetic hydroxyapatite byforming a complex between the silicide-bound cal-cium and phosphate ions from the fluid.19 The calci-fication study conducted at 258C indicates that, inthe initial state, the surface of CaSi2 growth layersdoes not provide adequate structural sites withproper coordination for calcium binding. Calciumdisilicide is bioactive only at the physiological tem-perature of 378C and bioinert at 258C. Changes inCaSi2 calcification behavior as a function of reactiontemperature were studied for two reasons. First,there is a variety of potential applications in den-tistry and medicine for hydroxyapatite formed atlow or near-physiological temperature.27 For exam-ple, it can be used as a synthetic enamel for rapidtooth repair.28 Yet, the main purpose of the tempera-ture reduction is to further decrease the rate of CaPprecipitation and promote deposition of calcium car-bonates (as a negative control) on CaSi2 growthlayers over CaP. The mechanisms of crystallizationand transformation of calcium carbonates formulatedby Sawada29 indicate that during the nucleationstage, amorphous calcium carbonate is more likelyto transform into calcite if the reaction temperatureis kept at 258C. Vaterite, a less stable polymorph, ismore abundant at 378C than calcite at the early met-astable stage.29 Fast dissolution kinetics of vaterite(relative to calcite) assure that there is less competi-tion for the nucleation sites between carbonates andCaP. On the other hand, calcite is the preferredphase at 258C.29 In this case, heterogeneous nuclea-tion of CaP on CaSi2 is expected to be significantlyretarded, if not completely inhibited.

Figure 8. Typical Raman spectrum of a calcified CaSi2layer on spark-processed Si after 4 weeks in SBF at 378C.



Modified SBF solutions, such as the ones used inthis work, are known to accelerate growth of CaPs invitro,18 but did not have any effect on CaSi2 growthlayers; however, the porous silica areas, which werepresent on the same wafers, on the other hand, con-currently calcified. Surface treatments with HF,K2HPO4, and MgCl2 indicate that both Si-OH andsurface-bound phosphates participate in the nuclea-tion process, with less importance being placed onthe silanol groups. Finally, the time-dependent sur-face evolution study shows that the structural decom-position of CaSi2 grains (and the associated localrelease of calcium) is a prerequisite to CaP deposition.The mechanistic details regarding zero-bias biominer-alization of CaSi2 growth layers on spark-processedsilicon in SBF can thus be summarized as follows:

� Within 10 days, CaSi2 grains undergo decomposi-tion accompanied by the local release of calcium.

� CaSi2 growth layers acquire negative surfacecharge through surface hydrolysis and adsorp-tion of phosphates.

� Within 20 days, spherulites, which are initiallydeposited in the form of calcium carbonate/sili-cate, acquire phosphates through ion exchangewith the SBF solution.

� The CaP phases, which are present during vari-ous growth stages, possibly include a combina-tion of Mg-substituted whitlockite, monetite,and tricalcium phosphate.

CONCLUSION

The development of new techniques for fabricatingbioactive silicon platforms is an important area ofresearch in the field of bioelectronics. Incorporationof calcium disilicide into silicon wafers via spark-processing technique is a facile dry-etch method forproducing silicon surfaces capable of biomineralizingin the presence and in the absence of an electric biasin vitro. The process of CaSi2/Si calcification underzero bias appears to follow the same mechanism asbioactive glasses/ceramics, during which CaPgrowth is significantly dependent on the structuraldegradation of CaSi2 grains. Surface functionalgroups, initially present on the as-prepared material,cannot induce CaP nucleation which begins onlyupon delamination of CaSi2 layers. Our work withCaSi2/Si demonstrates that control over biominerali-zation rate and mechanism can be eventuallyachieved by tuning the reaction temperature and/orapplying electric current instead of tailoring thestarting chemical composition and structure of a bio-material, which is typically done in case of other bio-materials that CaSi2 mimics: bioglasses and poroussilicon.

The authors thank Professor Waldek Zerda and MonikaWieligor for their assistance with the Raman measure-ments.

References

1. Hill RT, Lyon JL, Allen R, Stevenson KJ, Shear JB. Microfabri-cation of three-dimensional bioelectronic architectures. J AmChem Soc 2005;127:10707–10711.

2. Seregin VV, Coffer JL. Bias-assisted in vitro calcification ofcalcium disilicide growth layers on spark-processed silicon.Biomaterials 2006;27:3726–3737.

3. Kumta P, Sfeir C, Lee D, Olton D, Choi D. Nanostructuredcalcium phosphates for biomedical applications: Novel syn-thesis and applications. Acta Biomater 2005;1:65–83.

4. Fahy S, Hamann DR. Electronic and structural properties ofCaSi2. Phys Rev B 1990;41:7587–7592.

5. Bisi O, Braicovich L, Carbone C, Lindau I, Iandelli A, OlceseGL, Palenzona A. Chemical bond and electronic states in cal-cium silicides: Theory and comparison with synchrotron-radi-ation photoemission. Phys Rev B 1989;40:10194–10209.

6. Satta G, Profeta G, Bernardini F, Continenza A, Massidda S. Elec-tronic and structural properties of superconducting MgB2, CaSi2,and related compounds. Phys Rev B 2001;64: 104507–104516.

7. Gemming S, Seifert G. Nanotube bundles from calcium disili-cide: A density functional theory study. Phys Rev B 2003;68:075416–075423.

8. Jia JJ, Callcott TA, Asfaw A, Carlisle JA, Terminello LJ,Ederer DL, Himpsel FJ, Perera RCC. Occupied electronicstates of CaSi2 and CaSi: Soft-x-ray-fluorescence spectroscopy.Phys Rev B 1995;52:4904–4913.

9. Maex K, Van Rossum M. Properties of Metal Silicides. Ste-venage: INSPEC; 1995.

10. Schopke A, Wurz R, Schmidt M. Epitaxial growth of CaSi2 onSi(111). Surf Interface Anal 2002;34:464–467.

11. Morar JF, Wittmer M. Metallic CaSi2 epitaxial films onSi(111). Phys Rev B 1988;37:2618–2621.

12. Vogg G, Brandt MS, Stutzmann M. Kinetics of the topotacticformation of siloxene. Chem Mater 2003;15:910–915.

13. Affronte M, Sanfilippo S, Nunez-Regueiro M, Laborde O,Lefloch S, Bordet P, Hanfland M, Levi D, Palenzona A,Olcese GL. New superconducting CaSi2 phase with Tc up to14 K under pressure. Phys B 2000;284:1117–1118.

14. Sanfilippo S, Elsinger H, Nunez-Regueiro M, Laborde O,LeFloch S, Affronte M, Olcese GL, Palenzona A. Supercon-ducting high pressure CaSi2 phase with Tc up to 14 K. PhysRev B 2000;61:R3800–R3803.

15. Aoki M, Ohba N, Noritake T, Towata S. Reversible hydridingand dehydriding properties of CaSi: Potential of metal sili-cides for hydrogen storage. Appl Phys Lett 2004;85:387–388.

16. Hench LL, Wilson J. An Introduction to Bioceramics. RiverEdge, NJ: World Scientific; 1993.

17. Kokubo T, Kim HM, Kawashita M. Novel bioactive materialswith different mechanical properties. Biomaterials 2003;24:2161–2175.

18. Uchida M, Kim H, Kokubo T, Miyaji F, Nakamura T. Bone-like apatite formation induced on zirconia gel in a simulatedbody fluid and its modified solutions. J Am Ceram Soc2001;84:2041–2044.

19. Kim HM, Himeno T, Kawashita M, Kokubo T, Nakamura T.The mechanism of biomineralization of bone-like apatite onsynthetic hydroxyapatite: An in vitro assessment. J R SocInterface 2004;1:17–22.

20. Canham LT, Reeves CL. Apatite nucleation on low porositysilicon in acellular simulated body fluids. Mater Res SocSymp Proc 1996;414:189–194.

21. LeGeros RZ. Calcium Phosphates in Oral Biology and Medi-cine. New York: Karger; 1991.



22. Weiner S, Sagi I, Addadi L. Choosing the crystallization pathless traveled. Science 2005;309:1027–1028.

23. Wopenka B, Pasteris JD. A mineralogical perspective on theapatite in bone. Mater Sci Eng C 2005;25:131–143.

24. de Aza PN, Guitian F, Santos C, de Aza S, Cusco R, Artus L.Vibrational properties of calcium phosphate compounds, Part2: Comparison between hydroxyapatite and b-tricalciumphosphate. Chem Mater 1997;9:916–922.

25. Kohn MJ, Rakovan JF, Hughes JM. Phosphates: Geochemical,Geobiological, and Materials Importance. Washington, DC:Mineralogical Society of America; 2002.

26. Kokubo T. Design of bioactive bone substitutes based on bio-mineralization process. Mater Sci Eng C 2005;25:97–104.

27. Brown PW, Fulmer M. The effects of electrolytes on the ratesof hydroxyapatite formation at 25 and 388C. J Biomed MaterRes 1996;31:395–400.

28. Yamagishi K, Onuma K, Suzuki T, Okada F, Tagami J, OtsukiM, Senawangse P. A synthetic enamel for rapid tooth repair.Nature 2005;433:819.

29. Sawada K. The mechanisms of crystallization and transfor-mation of calcium carbonates. Pure Appl Chem 1997;69:921–928.



Documents

Mechanism of calcium disilicide-induced calcification of crystalline silicon surfaces in simulated body fluid under zero bias