Biomaterials 24 (2003) 5083–5089

ARTICLE IN PRESS

*Correspondin

7264.

E-mail addres

0142-9612/03/$ -

doi:10.1016/S014

Glial cell and fibroblast cytotoxicity study on plasma-depositeddiamond-like carbon coatings

Amarjit Singha, Gholamreza Ehteshamia, Stephen Massiaa,*, Jiping Hea,Robin G. Storerb, Gregory Rauppc

aHarrington Department of Bioengineering, Arizona State University, P.O. Box 879709, Tempe, AZ 85287-9709, USAbSchool of Chemistry, Physics and Earth Sciences, Flinders University, Adelaide, SA 5001, Australia

cDepartment of Chemical and Materials Engineering, Arizona State University, P.O. Box 876006, Tempe, AZ 85287-6006, USA

Received 27 November 2002; accepted 30 May 2003

Abstract

Diamond-like carbon films have been evaluated as coatings to improve biocompatibility of orthopedic and cardiovascular

implants. This study initiates a series of investigations that will evaluate diamond-like carbon (DLC) as a coating for improved

biocompatibility in chronic neuroprosthetic implants. Studies in this report assess the cytotoxicity and cell adhesion behavior of

DLC coatings exposed to glial and fibroblast cell lines in vitro. It can be concluded from these studies that DLC coatings do not

adversely affect 3T3 fibroblast and T98-G glial cell function in vitro. We also successfully rendered DLC coatings non-adhesive (no

significant fibroblast or glial cell adhesion) with surface immobilized dextran using methods developed for other biomaterials and

applications. Future work will further develop DLC coatings on prototype microelectrode devices for chronic neural implant

applications.

r 2003 Elsevier Ltd. All rights reserved.

Keywords: Diamond-like carbon films; Surface modification; Neural prosthetics; Immobilized dextran; Cell-resistant surface

1. Introduction

Diamond-like carbon (DLC) has attracted muchattention in recent years because of its hardness, wearresistance, chemical inertness, and low coefficient offriction. These material properties make DLC coatingsdesirable for reducing prosthetic wear debris andsubsequent failure in orthopedic implants. Therefore,recent research efforts have developed technologies forapplying DLC coatings to interfacial surfaces ofimplants [1]. DLC films can be deposited using plasmachemical vapor deposition, radio frequency sputtering,or ion beam-based methods [2,3].DLC films have been studied extensively for military

application as a single-layer antireflective coating forinfrared vision systems [4]. Thomson et al. [5] were thefirst to study biological effects on DLC films. Since thenthere has been a steady increase in the number of papers

g author. Tel.: +1-480-965-2448; fax: +1-480-727-

s: smassia@asu.edu (S. Massia).

see front matter r 2003 Elsevier Ltd. All rights reserved.

2-9612(03)00424-1

reporting on medical application and biological effectsof DLC coatings [3–7]. For the potential use of DLCand C–N films in orthopedic implants, Du et al. [7]studied the morphological behavior of osteoblasts cellson DLC coatings in vitro. They found that after aperiod of 2 weeks, the cells attached, spread, andproliferated on the DLC-coated surfaces withoutapparent impairment of cell physiology. The effect ofDLC coating on cellular metabolism was studied bymeasuring the production of three osteoblast-specificmarker proteins: alkaline phosphatase, osteocalcin andtype 1 collagen. The presence of DLC films had noadverse effect on these measured parameters. In vivostudies of DLC coated cobalt-chromium cylindersimplanted in intra-muscular locations in rats andtranscortical sites in sheep for a period of 90 days werewell tolerated confirming no signs of toxicity [8]. Linderet al. [9] recently reported on the biocompatibility ofDLC coating with respect to adhesion and activation ofprimary human monocytes and macrophages in vitroand concluded that DLC-coatings did not enhancemonocyte/macrophage adhesion or activation and

ARTICLE IN PRESSA. Singh et al. / Biomaterials 24 (2003) 5083–50895084

therefore would not be expected to enhance inflamma-tory responses in vivo [9]. For application in heart valveprostheses, DLC-coated titanium surfaces were recentlyreported by Jones et al. to reduce protein adsorptionand platelet attachment suggesting good in vivo hemo-compatibility properties [10]. Recently, these coatingshave been found to decrease thrombogenicity incoronary stent implants by reducing the release of metalions in the blood [11,12].In this report the cytotoxicity and cell adhesion

behavior of DLC coatings exposed to glial andfibroblast cell lines in vitro are described. Resultsfrom these studies provide information for determiningthe suitability of DLC coatings for neural implantapplications.

2. Experimental methods

2.1. DLC film deposition

The deposition system is a inductive coupled plasma(ICP) source, Hartley-type oscillator, operating at afrequency of 0.56MHz and is able to produce, in largevolume, a low-temperature plasma with high electrondensities (5� 1012 cm�3) and uniformity. A schematicdiagram of the Flinders University ICP facility is shownin Fig. 1. The deposition chamber is a double-walledwater cooled stainless steel cylinder with the innerdiameter 32 cm and a height of 20 cm capable ofhandling 4 in silicon wafers. The chamber has fourdiagnostic ports arranged at right angles to accommo-date various probes, such as Langmuir and magneticprobes. The top plate of the plasma chamber is made ofquartz, 32 cm in diameter and 1.25 cm thick and acts as adielectric window to couple RF energy to the plasma.

Fig. 1. Schematic diagram of the Flinders University Inductive

Coupled Plasma (ICP) apparatus utilized for deposition of DLC films.

The RF field is induced by a spiral planar coil of 17turns of 0.635 cm diameter water-cooled copper tube.The overall diameter of the coil is 28 cm. The coil is heldat a distance of 3mm above the quartz plate.Borosilicate glass slides (22� 22mm2) were used assubstrates for film deposition. Prior to deposition, thesubstrates were sputtered-cleaned in argon plasma andthen switched to high-purity methane gas for DLC filmdeposition. The unique feature of this system is that it ispossible to have E- and H-mode driven plasma. At lowinput power, the source operates in a capacitive regime,with the predominant electrostatic power to plasmacoupling. This regime is called the E-mode [13,14].When the input power reaches a certain threshold, thedischarge transits to electromagnetic H-mode, withdominant coupling inductive [13,14]. In this regime theplasma is very dense having higher electron density.

2.2. Cleaning and sterilization of samples

For the biocompatibility test, DLC-coated glass slides(1 cm2) were cleaned using a published protocol [15]:Samples were: (1) immersed in acetone in ultrasonicbath for 2min; (2) rinsed with 95% ethanol andimmersed in 95% ethanol in an ultrasonic bath for20min; (3) rinsed with deionized water (DI), immersedin a solution of 2% Liquinox brand detergent (Alconox,Inc.) in an ultrasonic bath for 20min; and (4) extensivelyrinsed with DI water, with one final immersion underultrasound for 15min. The samples then were place on asheet of aluminum foil and dried under the cell culture(sterilized) hood overnight. Then the coverslips werewrapped in the same aluminum foil and autoclaved forabout half an hour at 100–150�C.

2.3. Dextran coating methods for DLC films

All samples were cleaned and sterilized as describedabove and then placed in a sterile hood. All solutionsutilized for dextran coating were filter sterilized.Dextran was immobilized to DLC coatings to modulatecell adhesion using previously described methods [16].Aminated DLC surfaces were prepared by immersion in0.01% aqueous poly l-Lysine (PlL) solution then wasincubated overnight. Periodate-oxidized dextran wasdissolved in 0.2m sodium phosphate buffer (pH 9,0.02 g/ml). Immediately following surface aminationprocedures in the cleanroom, oxidized dextran solution(2ml) was added to sterile six-well multiwell dishescontaining surface-aminated substrates. The substrateswere allowed to incubate at room temperature for 16 hon a rocker platform and protected from light. Follow-ing incubation, the reaction mixture was decanted fromthe culture wells, and replaced by fresh 0.1m solution ofsodium borohydride, (NaBH4) to reduce Schiff basesformed and to quench any free unreacted aldehyde

ARTICLE IN PRESSA. Singh et al. / Biomaterials 24 (2003) 5083–5089 5085

groups present on the oxidized dextran chain. Thesubstrates were allowed to incubate for 2 h on the rockerplatform. The NaBH4 solution was then decanted andthe substrates were rinsed gently several times withdeionized water to remove unbound dextran.

2.4. Cell lines

The cell cultures investigated in this study were 3T3fibroblasts (ATCC #CRL-6476) and a glial-like (glio-blastoma) T98-G cell line (ATCC #CRL-1690). 3T3cells were maintained in DMEM with 10% FBS andT98-G cells were maintained in MEM with 10% FBS.All cell culture media and reagents were obtained fromLife Technologies, Inc.

2.5. Cell seeding

Six-well culture plates were initially coated with a0.5% pHEMA (in 95% ethanol) solution to reduce cellattachment to well surfaces. Following thorough airdrying of pHEMA-coated culture plates under thesterile hood, cleaned and sterile material samples wereplaced in each well. Approximately 2ml of cell suspen-sion in media (15,000 cells/ml) was added to each well ofthe culture dish. The cultures plates were then incubatedat 37�C, 5% CO2 for 24 h.

2.6. Cytotoxicity assay

Glial cell and fibroblast cytotoxicity was evaluatedusing a Live/Dead Viability/Cytotoxicity Kit (L-3224,Molecular Probes) and previously described methods[17]. Cells were seeded into sample wells as described inthe ‘‘Cell seeding’’ section. Stained samples wereexamined at 100�magnification via epifluorescencemicroscopy (Leica DM IRB inverted microscopeequipped with an Optronics HBO 100 mercury lightsource) to visualize both viable (fluorescein filter set) andnon-viable (rhodamine filter set) cells. Digital imageswere acquired using image analysis software (Image-Pros Plus 4.1). The number of number live (green-staining) and dead (orange-staining) cells were assessedin three images per sample well. A percentage of viablecells was calculated for each image [(number of live cells/total number of cells)� 100]. An average percentage wascalculated for the three sample wells (nine images). Amean percentage was determined for each sample groupfrom values obtained from three independent experi-ments. Comparisons between sample groups were madeusing one-factor ANOVA (a ¼ 0:05).

2.7. Cell adhesion assay

3T3 and T98-G cells were seeded into sample wells asdescribed in the ‘‘Cell seeding’’ section and incubated

for 24 h. Following incubation, samples were fixed(3.8% formaldehyde in PBS, 5min) and stained (0.1%aqueous toluidine blue, 5min). Stained cells were thenexamined using phase contrast or stereomicroscopy(Leica) at 100�magnification. Three random 100�fields were selected for each substrate for analysis. Theextent of cell adhesion was determined for each captureddigital image by calculating a percentage of cell areacoverage using digital image analysis software. Finaldata were presented as a percentage of control adhesion.The percentage of control cell area was calculated bymultiplying the ratio of percentage area coverage on allDLC substrates to percentage cell area coverage ontissue culture plastic (a reference material) by 100 (%control cell area coverage on tissue culture plas-tic=%100 by definition). The mean percentage ofcontrol adhesion was determined from triplicate inde-pendent experiments. Comparisons between samplesgroups were made using one-factor ANOVA.

3. Results and discussion

3.1. Characterization and validation of DLC films

In our experiments, the films were deposited on glasssubstrates in the E-mode regime of the methane gasplasma in which the source operates in the capacitiveregime. The deposited films were smooth, partiallyopaque, and mechanically hard as they were scratchresistant and not soft like graphite carbon. Films werecharacterized for optical properties by measuringrefractive index in the wavelength 400 to 800 nm usingSOPRA ellipsometer. Fig. 2 shows the variation ofrefractive index from 2.549 at 400 nm to 2.07 at 800 nm.Films are partially opaque in the visible spectrum, sohigher refractive index is expected than in the infraredwhere these films are transparent. It appears from thegraph that the value of n would be 2.00 at 10.06 mmwavelength, which is characteristic of DLC film used asa single-layer AR coating on Ge substrate [4,18]. Therefractive index data for the plasma deposited filmsobtained in the E-mode is consistent with reportedplasma-deposited DLC films [2,4]. However, in theelectromagnetic H-regime, which is characteristic ofinductive coupling, there was no film deposition in awide range of pressure and gas flow. Instead ofdeposition, etching of glass and silicon was noticeablyobserved. Typical deposition conditions were 30 mTorrpressure and 10–15 sccm flow of methane. The plasmadriven in H-mode is very bright and intense character-ized by 2–3 orders of magnitude higher electron densitythan the E-mode plasma. The H-mode plasma is verysimilar to helicon and ECR operated plasma. Similareffect of etching was observed in an ECR microwavesource when substrate is positioned near the ECR zone

ARTICLE IN PRESS

3.5

3.25

3

2.75

2.5

2.25

2

00.4 0.5 0.6 0.7 0.8

Wavelength (µm)

Nµm

N

Fig. 2. Ellipsometric measurement of refractive index over wavelength range from 400 to 800nm. Y-axis: refractive index (N); X-axis: wave-

length (mm).

A. Singh et al. / Biomaterials 24 (2003) 5083–50895086

or in the vicinity of the magnetic field. Because of thehigh electron density, atomic hydrogen is producedwhich etches away the depositing carbon film and aswell as glass substrate.The unique properties of diamond-like carbon films is

the result of the tetrahedral coordination between the C-atom represented by sp3 bonding, while bondingbetween C atom in graphite is trigonal or sp2. It ispossible to determine sp3 to sp2 ratio by calculating Zeff ;the effective number of valency electrons per carbonatom taking part in the inter-band transition, in theenergy range 02EZ by using the sum rule operation onthe e2 data. Zeff can be calculated using

Zeff ¼ 0:766A=d

Z EZ

0

Ee2ðEÞ dE; ð1Þ

where A is the mean atomic weight, E is the energy (eV),d is the density (kg/m3) and e2 is the dielectric constantrelated to the refractive index via a relation e2 ¼ 2nk;where k is the imaginary part of the complex refractiveindex often known as extinction coefficient. The carbonatom has four valency electrons that are available forinterband transitions. Taking the sum rule up to 8 eVwhich separates the p and s transitions, Zeff gives thenumber of electrons per carbon atom. All transitionsbelow 8 eV are p-p�; the experimental data (n, k andoptical band gap) is used to calculate Zeff which is thenused to determine the number of p electrons from theratio of Zeff of the films to Zeff for graphite. Thecalculation method is given in Ref. [19] which alsoprovides the Zeff data on graphite used in this calcula-tion.Comparison with Zeff for graphite allows us to

determine the relative concentration of sp3 to sp2 ratio.The sp3 concentration in plasma deposited filmsobtained in the E-mode plasma was found to be 80%and the remaining 20% bonds were sp2. Hydrogen freediamond-like carbon films obtained by sputtering areknown to have 75% or higher sp3 bonds [18], whileplasma-deposited films from methane or other hydro-carbon sources are believed to have slightly higher

concentrations of sp3 bonds (B85%) [3,19]. Thus ourfilms have sufficient number of sp3 bonds whichcontribute to mechanical hardness and a very smoothsurface which are characteristics of diamond-likecarbon.

3.2. Cytotoxicity analyses

The percent viability values for glial cells andfibroblasts cultured on DLC-coated substrates werecalculated from experimental data that was collectedusing the cytotoxicity assay described in the materialsand methods section. The results indicated that cellviability was not significantly different (po0:05) frompositive control values (cells cultured on tissue cultureplastic, 9875% viability). Thus DLC coatings wereconsidered not toxic for cultured glial cells andfibroblasts.

3.3. Cell adhesion studies

Cell adhesion and spreading was determined on allsubstrates and expressed as a percentage of control cellarea coverage on tissue culture plastic reference sub-strate. Morphology of Adherent 3T3 fibroblasts onDLC and poly-lysine coated DLC (aminated DLC)substrates was similar to cells routinely cultured ontissue culture plastic (Figs. 3A and 3B). 3T3 adhesionand spreading was robust on DLC and aminated DLCsubstrates (Fig. 4; 133.1724.3% control cell areacoverage on DLC, 112.471.5% control on aminatedDLC). These results further indicate that DLC coatingsdo not adversely affect 3T3 fibroblast adhesion, spread-ing, and function in comparison to normal cultureconditions on tissue culture plastic. Surface immobiliza-tion of dextran on aminated DLC significantly(po0:001) reduced 3T3 cell adhesion and spreading(Figs. 3C and 4; 0.4170.32% control). This resultsuggests that adequate surface immobilization of dex-tran was achieved to eliminate 3T3 cell adhesion similar

ARTICLE IN PRESS

(A) (B)

(C)

Fig. 3. Microscopic images of adherent and spread 3T3 fibroblasts on (A) DLC films, (B) DLC films with adsorbed poly-lysine (aminated DLC), and

(C) dextran-coated DLC films. Scale bar=50mm (A, C); scale bar=100mm (B).

0

20

40

60

80

100

120

140

160

DLC DLC + amine DLC dextran

% C

on

tro

l Cel

l Are

a co

vera

ge

Fig. 4. 3T3 fibroblast adhesion and spreading on DLC, Aminated

DLC (poly-lysine adsorbed on DLC), and Dextran-coated DLC. All

data is expressed as a percentage of control cell area coverage on tissue

culture plastic. Error bars represent standard deviations of mean

values determined from three independent experiments.

A. Singh et al. / Biomaterials 24 (2003) 5083–5089 5087

to what has been observed on other dextran-coatedmaterials [16,20,21].Morphology of Adherent T98-G glial cells on DLC

and poly-lysine coated substrates was similar to cellsroutinely cultured on tissue culture plastic (Figs. 5A andB). T98-G cell adhesion and spreading was robust onDLC and aminated DLC substrates (Fig. 6;147.1715.2% control cell area coverage on DLC,178.1737.2% control on aminated DLC). These results

further indicate that DLC coatings do not adverselyaffect T98-G glial cell adhesion, spreading, and functionin comparison to normal culture conditions on tissueculture plastic. Surface immobilization of dextran onaminated DLC significantly (po0:001) reduced T98-Gcell adhesion and spreading (Figs. 5C and 6;1.4171.23% control). This result suggests that adequatesurface immobilization of dextran was achieved toeliminate T98-G cell adhesion similar to what has beenobserved on other dextran-coated materials [16,20,21].

4. Conclusions

We have studied the cytotoxicity of plasma-depositedDLC films on glass substrates as a first step towarddeveloping DLC coatings for neural implant applica-tions. We utilized fibroblast and glial cell lines sincethese cells are representative of cells that will beencountered in the neural implant environment. Fromthese cell viability and adhesion studies, it can beconcluded that DLC coatings do not adversely affect3T3 fibroblast and T98-G glial cell function in vitro. Wealso successfully rendered DLC coatings non-adhesive(no significant fibroblast or glial cell adhesion) withsurface immobilized dextran using methods developedfor other biomaterials and applications [16,20,21]. Theseresults demonstrate that dextran-based bioactive, cell-selective coatings could be developed for DLC filmsusing previously described methods [16,20,21]. Future

ARTICLE IN PRESS

(A) (B)

(C)

Fig. 5. Microscopic images of adherent and spread T98-G glial cells on (A) DLC films, (B) DLC films with adsorbed poly-lysine (aminated DLC),

and (C) Dextran-coated DLC films. Scale bar=50mm (A, C); scale bar=100mm (B).

0

40

80

120

160

200

DLC DLC + amine DLC dextran

% C

on

tro

l Cel

l Are

a C

ove

rag

e

Fig. 6. T98-G glial cell adhesion and spreading on DLC, Aminated

DLC (poly-lysine adsorbed on DLC), and Dextran-coated DLC. All

data is expressed as a percentage of control cell area coverage on tissue

culture plastic. Error bars represent standard deviations of mean

values determined from three independent experiments.

A. Singh et al. / Biomaterials 24 (2003) 5083–50895088

work will further develop DLC coatings for chronicneural implant applications.

Acknowledgements

The work was supported in part by DARPA Bio-Info-Micro Program Grant # MDA972-00-1-0027. Part

of this research was funded by an internal grant fromFlinders University, Adelaide, Australia.

References

[1] Dearnaley PAA. A Review of metallic, ceramic and surface-

treated metals used for bearing surfaces in human joint

replacements. Proc Inst Mech Eng 1999;213:107–35.

[2] Angus JC, Koidl P, Domitz S. Carbon thin films. In: Mort J,

Jansen F, editors. Plasma deposited thin films. Boca Raton, FL:

CRC Press; 1986. p. 89–123.

[3] Singh A, Lavigne P. Deposition diamondlike carbon films by low

energy ion beam and dc magnetron sputtering. Surf Coating

Technol 1991;47:188–98.

[4] Lettington AH, Smith C. Optical properties and applications of

diamond-like carbon coatings. Diamond Rel Mater 1992;1:805–9.

[5] Thomson LA, Law FC, Rushton N. Biocompatibility of

diamond-like carbon coating. Biomaterials 1991;12:37–40.

[6] Dion I, Roques X, Baquey C, Baudet E, Basse Cathalinat B,

Moore N. Hemocompatibility of diamond-like carbon coating.

Biomed Mater Eng 1993;3:51–5.

[7] Du C, Su XW, Cui FZ, Zhu XD. Morphological behavior of

osteoblasts on diamond-like carbon coating and amorphous C–N

film in organ culture. Biomaterials 1998;19:651–8.

[8] Allen MB, Myer B, Rushton N. In vitro and in vivo investigations

into biocompatibility of diamond-like carbon (DLC) coatings for

orthopedic applications. J Biomed Mater Res 2001;58:319–28.

[9] Linder S, Pinkowski W, Aepfelbacher M. Adhesion, cytoskeletal

architecture and activation status of primary human macrophages

on a diamond-like carbon coated surface. Biomaterials

2002;23:767–73.

[10] Jones MI, McColl IR, Grant DM, Parker KG, Parker TL.

Protein adsorption and platelet attachment and activation, on

TiN, TiC, and DLC coatings on titanium for cardiovascular

applications. J Biomed Mater Res 2000;52:413–521.

ARTICLE IN PRESSA. Singh et al. / Biomaterials 24 (2003) 5083–5089 5089

[11] Gutensohn K, Beythien C, Bau J, Fenner T, Grewe P, Koester R,

Padmanaban RK, Kuehnl P. In vitro analyses of diamond-like

carbon coated stents. Reduction of metal ion release, platelet

activation, and thrombogenicity. Thromb Res 2000;99:577–85.

[12] De Scheerder I, Szilard M, Yanmin H, Ping XB, Verbeken E,

Neerinck D, Demeyere E, Coppens W, Van De Werf F.

Evaluation of the biocompatibility of two new diamond-like

stent coatings (Dylyn) in a porcine coronary stent model.

J Invasive Cardio 2000;12:389–94.

[13] Turner MM, Leiberman MA. Hysteresis and the E-to-H

transition in radiofrequency inductive discharges. Plasma Sources

Sci Technol 1999;8:313–24.

[14] El-Fayami IM, Jones IR. Theoretical and experimental investiga-

tions of the electromagnetic field within a planar coil, inductively

coupled RF plasma source. Plasma Sources Sci Technol

1998;7:162–78.

[15] Rowland SA, Shalaby SW, Latour RA, von Recum AF.

Effectiveness of cleaning surgical implants—Quantitative analysis

of contaminant removal. J Appl Biomat 1995;6:1–7.

[16] Massia SP, Holecko MM, Ehteshami MM. In Vitro assessment of

bioactive coatings for neural implant applications. J Biomed

Mater Res, in press.

[17] Trudel J, Massia SP. Synthesis and cytotoxicity of photocross-

linked dextran and hyaluronan-based hydrogels. Biomaterials

2002;23:3299–308.

[18] Saviddes N, Window B. Diamond-like amorphous carbon film

prepared by magnetron sputtering. J Vac Sci Technol

1985;A3:2386.

[19] Saviddes N. In: Koidl P, Oelhafen P, editors. Amorphous

hydrogenated carbon films E-MRS, vol. XVII. Strasbourg,

France: EMRS European Material Research Society; 1987.

p. 275.

[20] Massia SP, Letbetter DS, Stark J. Surface-immobilized

dextran limits cell adhesion and spreading. Biomaterials

2000;21:2253–61.

[21] Massia SP, Stark J. Immobilized RGD peptides on surface-

grafted dextran promote biospecific cell attachment. J Biomed

Mater Res 2001;56:390–9.