Organic overlayer model of a dental composite analyzedby laser desorption postionization mass spectrometry andphotoemission

Manshui Zhou,1 Chunping Wu,1 Praneeth D. Edirisinghe,1 James L. Drummond,2 Luke Hanley1

1Department of Chemistry, University of Illinois at Chicago, Chicago, Illinois 60607-70612Department of Restorative Dentistry, University of Illinois at Chicago, Chicago, Illinois 60607-7061

Received 25 July 2005; accepted 24 August 2005Published online 12 December 2005 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/jbm.a.30591

Abstract: Some dental composites consist of a polymeriz-able resin matrix bound to glass filler particles by silanecoupling agents. The resin in these composites includesbisphenol A diglycidyl methacrylate (Bis-GMA) as well asother organic components. Silane coupling agents such as3-(trimethoxysilyl) propyl methacrylate (MPS) have beenused to improve the mechanical properties of the dentalcomposites by forming a covalent bond between the glassfiller particles and the resin. These resin–glass compositesundergo material property changes during exposure to theoral environment, but degradation studies of the commer-cial composites are severely limited by their chemical com-plexity. A simplified model of the dental composite has beendeveloped, which captures the essential chemical character-istics of the filler particle–silane–resin interface. This modelsystem consists of the resin matrix compound Bis-GMA

covalently bound via a methacryloyl overlayer to amor-phous silicon oxide (SiO2) surface via a siloxane bond. Scan-ning electron microscopy shows the porous characteristicand elemental composition of the SiO2 film, which approx-imately mimics that of the glass filler particles used in dentalcomposites. LDPI MS and XPS verify the chemistry andmorphology of the Bis-GMA-methacryloyl overlayer. Pre-liminary results demonstrate that LDPI MS will be able tofollow the chemical processes resulting from aging Bis-GMA-methacryloyl overlayers aged in water, artificial sa-liva, or other aging solutions. © 2005 Wiley Periodicals, Inc.J Biomed Mater Res 77A: 1–10, 2006

Key words: biodegradation; interface; dental composite;surface morphology; amorphous silicon oxide

INTRODUCTION

One popular dental composite consists of a poly-merizable resin matrix bound to glass filler particlesby silane coupling agents.1,2 The resin typically con-tains one or more monomers such as bisphenol Adiglycidyl methacrylate (Bis-GMA) and triethyleneglycol dimethacrylate. Silica or glass filler particles areused as reinforcements and to reduce overall polymer-ization shrinkage. Silane coupling agents such as3-(trimethoxysilyl) propyl methacrylate (MPS) havebeen used to improve the mechanical properties ofdental composites by forming a covalent bond be-tween the glass filler particles and the various meth-ylmethacrylate based compounds comprising the

resin. The hydrolytic stability of the composites is alsoimproved by silanes because of the hydrophobic na-ture of silane coupling agents.3

These resin–glass composites undergo materialproperty changes during exposure to the oral environ-ment. The detrimental effects of the degradation ofdental biomaterials include the release of syntheticcompounds into local tissue and their accumulationelsewhere in the body. The erosion process can alsoweaken the materials and reduce the longevity of therestoration.1,4–6 Mass spectrometry has been widelyused to examine the organic degradation productsleached from dental composites during aging stud-ies,7–9 and also showed that the Bis-GMA componentof dental composites can be hydrolyzed by salivaryenzymes.10

Silanes are used in these dental composites to im-prove mechanical properties, presumably by co-valently binding the resin to silica filler particles.However, prior work has generally not determinedwhether Bis-GMA was initially bound to the silanecoupling agent or to other resin components. Degra-

Correspondence to: L. Hanley; e-mail: LHanley@uic.eduContract grant sponsor: National Institute of Dental and

Craniofacial Research; contract grant number: DE-07979Contract grant sponsor: National Science Foundation;

contract grant number: CTS-0321202

© 2005 Wiley Periodicals, Inc.

dation of these dental composites in the oral environ-ment may proceed in part by degradation of thesecoupling agents. Hydrolysis of the SiOOOSi bond bywater is expected to weaken the polymer–filler inter-face during aging.11,12 Hydrolysis of the ester linkagethat serves as the silane–resin bond is also feasible andtherefore another potential, but largely unexplored,degradation mechanism of dental composites. Fur-thermore, some enzymes can form siloxane bondsfrom silanols, the reverse reaction to hydrolysis.13

These results indicate that the design of dental com-posites less susceptible to fracture requires a fullerunderstanding of the chemical nature of the aqueousand enzymatic hydrolytic degradation of the organicresin–silane–inorganic filler particle interface.

Previous work from this group analyzed the inor-ganic material changes on the surface of dental com-posites after aging in different media that simulate theoral environment.6 However, examination of the or-ganic degradation in commercial dental composites isseverely limited by their complex composition thatincludes a variety of monomers and proprietary addi-tives. This study therefore develops a simplifiedmodel of the dental composite that captures the essen-tial chemical characteristics of the filler particle–sila-

ne–resin interface. This model system consists of apolymerizable resin matrix compound Bis-GMA co-valently bound to amorphous silicon oxide (SiO2) sur-face via the silane coupling agent MPS, as shownschematically in Figure 1. The amorphous SiO2 surfaceserves as a model of the filler particle surface as it hassize features and morphology similar to many of theparticles found in real composites, and displays asimilar chemistry (hydroxyl terminated) and high sur-face area. One end of MPS should form SiOOOSibonds with hydroxyl groups on the surface of oxi-dized silicon, and the other end will strengthen thecomposite by forming covalent bonds with Bis-GMA.3

A home built laser desorption vacuum ultraviolet(VUV, 118 nm) postionization linear time-of-flightmass spectrometer (LDPI MS) is used for surface anal-ysis of the model system. The advantage of VUVpostionization in LDPI MS is that it enhances ioniza-tion yields with a minimum of fragmentation.14–16

LDPI MS is used to verify the preparation steps of themodel dental composite by chemical analysis of themethacryloyl overlayer, Bis-GMA-methacryloyl over-layer, and several control surfaces. X-ray photoelec-tron spectroscopy (XPS) is also used to confirm thebinding process, since it is sensitive to elemental and

Figure 1. Schematic of the steps used in preparation of the overlayer model of the dental composite system. Bis-GMA isbound to porous silicon oxide (SiO2) via the silane coupling agent MPS by photoinduced functionalization with initiators (seetext) to form a Bis-GMA-methacryloyl overlayer.

2 ZHOU ET AL.

functional group composition and can be readilyquantified.2 However, XPS suffers from a lack ofchemical resolution and typically cannot provide thedetailed chemical structure and molecular weight in-formation available to LDPI MS. Finally, scanningelectron microscopy is utilized to study the morphol-ogy of the oxidized silicon substrate.2

EXPERIMENTAL DETAILS

Materials and sample preparation

Bis-GMA, MPS, triethanolamine, 1-vinyl-2-pyrrolidinone(VP), 2�,4�,5�,7�-tetrabromo-fluorescein disodium salt (EosinY), coronene, and hydrofluoric acid (HF, 48% in water) areused as purchased (Aldrich Chemical Co., Milwaukee, WI).

N-type silicon wafers (orientation, 100; resistivity, 1–10 �cm; Wafer World, West Palm Beach, FL) are etched in a 1:1mixture of 50% aqueous HF and ethanol at 20 mA/cm2 for5 min under ambient (�0.5 mW/cm2) diffuse light, and thenthe etched wafers are cleaned with ethanol. The resultantamorphous silicon surfaces are oxidized in 31.6% H2O2 at50°C for an hour, and then stored in 1N HNO3 solution toproduce the amorphous SiO2 substrate.

The cleaned SiO2 surfaces are silanized in 2.0 wt % MPS/toluene solution under an argon atmosphere (no or tracewater) at 60°C for 96 h. The silanized substrates are thenwashed with toluene, methanol, and distilled water. Finally,the substrates are dried under an argon atmosphere thenbaked at 80°C for 12 h.

A 50 �L portion of stock solution of co-initiators preparedas 10 mM Eosin Y, 225 mM triethanolamine, and 37 mMVP17 in ethanol is directly added to 5.0 mL of 2.0 mg/mLBis-GMA/ethanol solution containing the silanized sub-strate. Then, the glass tube containing the substrate andreagents is purged with argon, sealed, gently stirred for 3–4h at room temperature, and then irradiated with visible light(Triad 2000 Visible Light Cure System, Dentsply Int., York,PA) for 20 min.18 The Bis-GMA-methacryloyl overlayersused for preliminary aging experiments are first stored indouble distilled water at 37°C for 2 h, then removed foranalysis.

Instrumentation

The home-built laser desorption postionization LDPI MSconsists of vacuum chamber with sample transfer load lock,sample holder, Xe cell, acceleration optics, and a 1.3 mfield-free flight tube with multichannel plate ion detector(detector assembly by R.M. Jordan Co., Grass Valley, CA)similar to the one previously described.19 Two-step acceler-ation is used with the sample at 3.0 keV, the first grid at 2.7keV, and the second grid at ground. The ion beam is focusedwith an Einzel lens at 1.2 keV and the extraction high voltagepulse delay is 500 ns. Data are recorded with a 500 MHzoscilloscope (54616B, Agilent Technologies, Palo Alto, CA).

The resolution m/�m—defined by the full width half max-imum at the m/z 495 peak—is 240 and the mass accuracy is700 ppm.

The VUV (118 nm, 10.49 eV) radiation is generated byfrequency tripling of the third harmonic of a Nd:YAG laser(Tempest 10 Hz 355 nm, New Wave Research, Fremont, CA)in pure Xe gas.14,19,20 The optimum Xe pressure for 118 nmproduction is found to be 1 Torr when 50 mJ of 355 nmradiation is focused into the tripling cell through a 27-cmfocal-length lens. The VUV radiation is focused onto theionization region in the time-of-flight MS by a MgF2 lens,which also serves as a vacuum window. The 118 nm laserbeam is brought into the chamber parallel to the samplesurface where it is tightly focused 2 mm above the desorp-tion region. For this reason, desorption scans are run hori-zontally to ensure overlap of the desorption and ionizationregion. The relative delay between the desorption and pos-tionization laser pulses is 3.0 �s. Coronene is used to opti-mize the ionization laser beam position and Xe gas pressure.The 355 nm desorption laser (Minilite, Continuum, SantaClara, CA) intensity is kept near threshold for signal pro-duction.

XPS (AXIS 165, Kratos Analytical, Manchester, UK) isperformed using monochromatic 1486.6 eV Al K� radiation.The surface charge is neutralized using filament current of1.7 A, charge balance of 2.5 eV, and filament bias of 1.3 eV.Survey and core level scans are collected on each of at leastthree of every sample type. The XPS results for each of thethree samples are reported as average atomic percentages �standard deviation, as calculated by standard methods.

Data analysis methods

The energy scale is calibrated by reference to the COC/COH C1s peak component at 285.0 eV. Curve fitting of C1s

spectra is performed by using XPSpeak 4.1 software (http://www.phy.cuhk.edu.hk/�surface/XPSPEAK/, R.W.M. Kwok,Hong Kong, China). All spectral peaks are defined as a Gauss-ian/Lorentzian mixture and the background is corrected for bythe Shirley method. Quantitative analysis of surfaces by Elec-tron Spectroscopy software (QUASES, Tougaard APS,Odense, Denmark) is used to estimate the film morphologywithin the XPS sampling depth.21,22 QUASES relies on thephenomenon that the inelastically scattered energy lossstructure of XP spectra carries information on the depth oforigin of the detected electrons. The Si2s core level is used forQUASES analysis. The attenuation length (�) used inQUASES computations are approximated by inelastic meanfree path (IMFP) values determined using the TPP-2M equa-tion. The SiO2 and polymer cross-sections developed byTougaard21,22 are used for the analysis of the silicon oxidesurface, methacryloyl overlayer, and Bis-GMA-methacryloyloverlayer using IMFP values of 2.698, 2.992, and 2.992, re-spectively.23,24 Three samples are analyzed for each type ofanalysis.

Commercial software (ACD/MS Fragmenter, AdvancedChemistry Development, Toronto, ON, Canada) is used inelectron impact and protonation modes to analyze and iden-tify peak structures of the cations formed here by VUVpostionization. This software presents fragmentation ions

ORGANIC OVERLAYER MODEL OF A DENTAL COMPOSITE 3

from a given parent ion chemical structure as a genealogicaltree through the use of established rules of ion fragmenta-tion.

RESULTS AND DISCUSSION

The preparation and characterization of the modeldental composite are studied herein using scanningelectron microscopy, laser desorption postionizationmass spectrometry (LDPI MS), and XPS. The desiredchemical structures of each step in the preparation ofthe model system are depicted in Figure 1. First, a SiO2surface is prepared to simulate the surface of the silicafiller particles used in actual dental composites. Next,the methacryloyl siloxane MPS is covalently bound tothe SiO2 surface. Finally, the resin component Bis-GMA is bound to the SiO2 surface via MPS throughphotoinduced functionalization to simulate the fillerparticle–polymer matrix composite interface. The sev-eral steps, alternate strategies, and controls used inpreparing the model of the composite interface arecharacterized by scanning electron microscopy, LDPIMS, and XPS.

Formation of silicon oxide surface to simulatesilica filler particle

SiO2 surfaces are prepared from silicon wafersetched in concentrated HF/ethanol solution underambient light,25 followed by exposure to concentratedH2O2 to form oxide (SiOOOSi) and hydroxide(SiOOH) moieties. Cross-sectional scanning electronmicroscopy in Figure 2 (bottom) shows the amor-phous micrographs of hydrated SiO2 layer formed onthe surface with micron sized islands ranging from�0.2 to �5 �m diameter. These dimensions areroughly consistent with the glass filler particle sizesused in commercial dental composites.6 Figure 2 (top)shows electron micrographs of a porous silicon bandsandwiched between the bulk Si substrate and theamorphous SiO2 surface layer, in agreement withprior observations26 that the SiO2-porous Si exists as alayer on top of the porous Si layer, which itself is ontop of the bulk Si. However, differences in the etchingand oxidation processes may lead to a varying extentof oxidation and porosity compared with those previ-ously reported.26 Figure 3 shows the Si2p core levelspectra of the SiO2 surface. A predominant peak at103.0 eV is displayed due to SiO2, which is composedof silicon coordinated to four oxygen atoms.27 A small99.1 eV Si2p peak is observed on the blank SiO2 surfaceand is attributed to the underlying bulk Si. QUASES isused to estimate the film morphology by analyzing the

inelastically scattered energy loss structure of XP spec-tra that contains information on the depth of origin ofthe detected 154.0 eV Si2s core level electrons.21,22 Theenergy loss structure of the Si2s peak indicates that theamorphous SiO2 surface can be modeled as a contin-uous overlayer of �80 nm thickness (data not shown).Although a continuous overlayer seemingly contra-dicts with the porous morphology observed by elec-tron microscopy, the QUASES analysis is expected togive similar results for either morphology, since thelayer thickness is much greater than the IMFP of theSi2s emitted electrons.

Table I shows the XPS elemental analysis of the SiO2surface. Only oxygen and silicon peaks arising fromthe SiO2 surface appear, except for a 4% trace of car-bon (see Fig. 4 for C1s core level). The elemental com-position ratio of oxygen to silicon is close to 2, whichconfirms the existence of the SiO2 layer on the surface.Energy dispersive spectra recorded with electron mi-

Figure 2. Cross-sectional scanning electron micrographs ofthe porous silicon oxide surface. Top image is cross-sectionalview of SiO2 on top of bulk Si (image scale �60 �m across).Bottom image is top view of porous SiO2 (image scale �15�m across).

4 ZHOU ET AL.

croscopy also confirm that the elemental ratio of oxy-gen to silicon is close to 2 (data not shown).

Formation of covalently bound methacryloyloverlayer on SiO2

The amorphous SiO2 surfaces are silanized in dryMPS/toluene solution, and then the cleaned sub-strates are baked under inert atmosphere for 12 h.Table I shows that the resultant methacryloyl over-layer displays an increase in the intensity of the C1speak from XPS and a corresponding decrease in O1sand Si2p signal, consistent with the formation of asilane overlayer on the SiO2 substrate. This drop isalso seen in Figure 3, which compares the Si2p corelevel spectra of the SiO2 surface and the methacryloyloverlayer. The results of peak fitting the componentsof the C1s core level spectra of the methacryloyl over-layer are shown in Table I. The 285.0 eV bindingenergy component due to aliphatic/aromatic carbons

mainly present in the methacryloyl overlayer is themost intense component, while the lower intensitycomponents at 286.6 and 288.8 eV binding energies areattributed to the COO and OOCAO functionalgroups, respectively.28

The methacryloyl overlayer morphology is modeledthrough analysis of the Si2s peak energy loss structureas an overlayer with an exponentially decaying Sicontent from 33% at the SiO2 interface to 1% at the airinterface with a decay length at 1.6 nm. The 1.6 nmdecay length slightly exceeds the �1.2 nm extendedmolecular chain of MPS.29 Overall, this modeled mor-phology is consistent with the presence of a one to twolayer thick methacryloyl overlayer on the SiO2 surfacewith the Si end bound to the oxide.

Figure 5(a) shows the laser desorption postioniza-tion mass spectra (LDPI MS) of the methacryloyl over-layer analyzed above by XPS (see Fig. 4 and Table I).This methacryloyl overlayer forms a SiOOOSi bondwith the silicon oxide surface under these conditions.Thus, LDPI MS displays no parent ion peak for themethacryloyl overlayer, and only the various frag-

Figure 3. Si2p core levels by XPS of the (a) SiO2 surface, (b)methacryloyl overlayer, and (c) Bis-GMA-methacryloyloverlayer.

TABLE IBinding Energies, Elemental Compositions, and C1 Component Peak Assignments from X-Ray Photoelectron Spectra

(XPS) of SiO2 Surface, Methacryloyl Overlayer and BisGMA-Methacryloyl Overlayer

Sample

C1s (%) at 285 eV

O1s (%)at 532 eV

Si2p (%)at 103 eV

N1s (%) at399 eV

Br3d (%)at 69 eV

Components of C1s Peak Fits (%)

Total C1s(%)

COC at 285eV

COO at 286.6eV

CAO at 288.8eV

SiO2 layer – – – 4.2 � 1.6 65.6 � 2.9 30.2 � 1.3 – –Methacryloyl overlayer 63.6 � 6.2 23.0 � 6.7 13.4 � 0.5 46.9 � 8.9 39.6 � 5.6 13.5 � 3.2 – –Bis-GMA-methacryloyl

overlayer 47.8 � 3.1 46.6 � 2.3 5.6 � 0.8 71.6 � 2.2 25.3 � 1.9 3.1 � 0.3 1.3 � 0.2 0.4 � 0.0

Figure 4. C1s core levels by XPS of the (a) SiO2 surface, (b)methacryloyl overlayer, and (c) Bis-GMA-methacryloyloverlayer.

ORGANIC OVERLAYER MODEL OF A DENTAL COMPOSITE 5

ment ions depicted schematically in Figure 5(a). Thehighest mass peak from the methacryloyl overlayeroccurs at m/z 127 and is formed by cleavage of theCOSi surface bond during the laser desorption step.This observation is in agreement with previous stud-ies that alkylsiloxane monolayers on Si undergo SiOCbond cleavage during LDPI MS.30 Typical methacry-late ion peaks also appear at m/z 41, 69, 85, and 99 andare assigned to H2CAC(CH3), H2CAC(CH3)CO,H2CAC(CH3)CO2

, and H2CAC(CH3)CO2CH3, re-

spectively.31

It is well known that water plays an important rolein the formation of siloxane self-assembled monolay-ers and multilayers. Excess water results in facile po-

lymerization in solution and polysiloxane depositionof the surface, whereas drier conditions can lead topreferential formation of monolayers.32,33 Thus, theeffect of varying water content in the MPS/toluenesolution is examined here using LDPI MS. Figure 5(c)shows the LDPI MS of MPS physisorbed on the SiO2surface from a MPS/toluene solution containing 5%water, with the schematic of ion formation under theseconditions depicted below this (lowest) spectrum. TheMPS parent ion peak at m/z 248 is observed along witha siloxane fragment ion peak at m/z 121 and a proton-ated methacrylic peak at m/z 164, while the ion peak atm/z 216 is from loss of a neutral methanol from MPS,together with the ion peak at m/z 161 resulting fromfurther loss of its isopropene group. No hydrolyzedMPS parent ion peak (m/z 206) appears, but loss of aneutral water from MPS leads to the appearance of m/z188, consistent with the expected hydrolysis reactionupon nucleophilic substitution reaction between waterand MPS. The results indicate that MPS hydrolyzes inaqueous solution and immediately undergoes homo-condensation to possibly form a new compound (seebelow).

Figure 5(b) further examines physisorbed MPS onthe SiO2 surface, displaying the LDPI MS after pump-ing in vacuum for 12 h. This spectrum displays nopeaks related to the siloxane, indicating either a com-plete sublimation of MPS or a homocondensation. Ifhomocondensation is faster than siloxane (SiOOOSi)formation, the hydrolyzed MPS may no longer beavailable for reaction with the hydroxyl surface. Thehydrolyzed MPS could then homocondense to formwhat has previously been postulated to be the eight-membered siloxane ring shown schematically in Fig-ure 5(b),34 a physisorbed compound that should beeasily removed by rinsing in ethanol.

Formation of Bis-GMA-methacryloyl overlayerbound to SiO2

Bis-GMA is bound to the methacryloyl overlayer onSiO2 by photoinduced functionalization. Figure 4 dis-plays the C1s core level XPS for the Bis-GMA-methac-ryloyl overlayer, which can be compared with themethacryloyl overlayer C1s also shown there. The C1sbinding energy of the OOCAO shifts upwards by0.4 eV from 288.8 eV for the methacryloyl overlayerto 289.2 eV for the Bis-GMA-methacryloyl overlayerdue to electron localization that occurs upon the lossof the terminal CAC bonds at the ends of Bis-GMAand MPS when they are coupled.35 An increase in thenumber of carbon atoms adjacent to multiple oxygenatoms may also contribute to the upward shift inbinding energy for this OOCAO C1s component thatis observed for the Bis-GMA-methacryloyl overlayer.

Figure 5. Laser desorption postionization mass spectra(LDPI MS) the SiO2 surface silanized in 2.0 wt % MPS/toluene solutions under various conditions (a) no or tracewater (forming the methacryloyl overlayer), (b) 5% waterand pumped in vacuum for 12 h, and (c) 5% water andanalyzed immediately after preparation (adsorbed MPS).

6 ZHOU ET AL.

Table I shows that the Bis-GMA-methacryloyl over-layer displays an increase in the intensity of the C1speak from XPS and a corresponding decrease in bothO1s and Si2p intensities. These changes are consistentwith the formation of a Bis-GMA overlayer on top ofthe methacryloyl overlayer. Furthermore, the appear-ance of N1s (1%) at 399 eV and Br3d (0.4%) at 69 eV aredue to traces of the initiator triethanolamine (TEA),the photosensitize Eosin Y, and/or the accelerator1-vinyl 2-pyrrolindinone, which are partially trappedin the overlayer during photoinduced functionaliza-tion. A trace peak appears at 496 eV in the XPS surveyscan of the Bis-GMA-methacryloyl overlayer that istentatively attributed to �1% of either Sn (oxidized) orZn (Auger line), presumably present as an impurity inone of the initiators. The mechanism of photoinducedfunctionalization has been proposed to proceed byphotoexcitation of the Eosin Y triplet state, which un-dergoes a redox reaction with triethanolamine to forma triethanolamine radical that initiates the covalentbinding of Bis-GMA and the methacryloyl overlayer.17

Figure 3 compares the Si2p core level spectra of theSiO2 surface and the methacryloyl overlayer with thatof the Bis-GMA-methacryloyl overlayer. The bulk Sipeak becomes smaller and eventually disappearsupon treatment with MPS and Bis-GMA, as the over-layer thickness increases beyond the XPS samplingdepth. The binding energies of SiOOx shift downwardby 0.4 eV, from 103.0 eV for the methacryloyl over-layer to 102.6 eV for the SiO2 surface, corresponding toa slight reduction in the average number of oxygenatoms bound to each silicon atom, consistent with ashift from SiO2 to siloxane.27 However, the Si2p oxidecomponent binding energy in the Bis-GMA-methacry-loyl overlayer remains nearly the same, with the onlychange likely due to loss of some SiO2 signal as theoverlayer thickness grows (and the XPS samplingdepth shifts outwards from SiO2). The energy lossstructure of the Si2s peak is fit by a Bis-GMA-methac-ryloyl film morphology, which is �80% of a �4 nmthickness and �20% of a �0.6 nm thickness, with nosilicon found in the film. This model is consistent withformation of nearly complete Bis-GMA-methacryloylmonolayer with the only silicon in the overlayer lo-cated at the SiO2 interface.

Figure 6(b) displays the LDPI MS of Bis-GMA-methacryloyl overlayer prepared under the same con-ditions used for the overlayers analyzed above byXPS. The spectrum shows that the Bis-GMA-methac-ryloyl overlayer also fragments at the COSi bondduring laser desorption to form the larger ion at m/z640 depicted schematically at the bottom of Figure 6.The other major observed ions are readily attributed tological fragmentation peaks of the Bis-GMA-methac-ryloyl overlayer. For example, the m/z 531 ion formsdue to the cleavage of the first ether bond adjacent tothe SiO2 surface, the m/z 471 ion results from the

cleavage of the second carboxyl bond, and the m/z 413forms due to the cleavage of the COC bond adjacentto the first alcohol group. The m/z 312 ion results fromaddition of hydrogen to the m/z 311 ion, which itselfforms by cleavage of the second ether bond and thesecond carboxyl bond (shown schematically at thebottom of Fig. 6). The m/z 41, 69, and 85 ions aretypical for a methacrylate group, whereas the m/z 143and 157 ions are due to the cleavage of the secondether bond.

Overall, the LDPI MS of Figure 6(b) is consistentwith formation of a Bis-GMA-methacryloyl overlayercovalently bound to the SiO2 surface whose MS differsentirely from that of adsorbed Bis-GMA. For compar-ison, Figure 7(a,b) shows the LDPI MS of Bis-GMAadsorbed either (a) directly on bare SiO2 or (b) themethacryloyl overlayer without photolysis or the ini-tiator compounds. These two Bis-GMA controls givesimilar mass spectra, which both display ions at m/z143, 495, and 486, with the only difference betweenthem being the appearance of the parent ion peak atm/z 512 [Fig. 7(b)]. Loss of neutral methanol and waterfrom Bis-GMA leads to the formation of the m/z 486

Figure 6. Analysis of Bis-GMA-methacryloyl overlayer by(a) displays the LDPI MS of this overlayer after aging indistilled water for 2 h, (b) LDPI MS, and (c) direct laserdesorption (LD) MS.

ORGANIC OVERLAYER MODEL OF A DENTAL COMPOSITE 7

and 495 ions, respectively, as shown schematically atthe bottom of Figure 7. The m/z 143 ion is attributed tothe same cleavage discussed above for the covalentlybound Bis-GMA (see Fig. 7).

Photolysis of the terminal CAC bond of Bis-GMAcan lead to reaction not only with the terminal CACbond of the methacryloyl overlayer but also with theterminal CAC of a second Bis-GMA, leading in theformer case to Bis-GMA polymerization. Such Bis-GMA polymers can be detected with LDPI MS andthereby distinguished from the Bis-GMA-methacry-loyl overlayer. Figure 8 displays the LDPI MS of an-other control produced by photopolymerization ofBis-GMA physisorbed directly onto the amorphousSiO2 surface (without any MPS treatment), prior towashing with ethanol. Photopolymerization leads to aseries of high molecular weight ions above m/z 1000 ina repeating mass distribution typical of polymersshown schematically at the bottom of Figure 8. Forexample, the m/z 1259 ion forms by cleavage of the

bonds labeled 1 and 4 in Figure 8, while further cleav-age at bonds 2 and 3 leads to formation of ions at m/z1183, and 1109, respectively. Similarly, the ion peak atm/z 1495 is due to the cleavage of bonds 5 and 9, andfurther cleavage at bonds 6, 7, and 8 leads to theformation of ion peaks at m/z 1418, 1344, and 1328,respectively. It might be expected that the mass spec-trum peak spacings would be larger in Figure 8 toreflect the mass of the intact Bis-GMA monomer, butthe long chain of Bis-GMA readily fragments at bonds2 and 3 (and analogous bonds throughout the poly-mer) during LDPI, causing the observed peak spac-ings at 71 and 74 mass units corresponding toCH3CCOO and CH2CHOHCH2O group losses, re-spectively. The interpretation of the polymer structurematches the LDPI MS shown in Figure 8. The dataindicates that competition occurs between self-poly-merization of Bis-GMA and the reaction of Bis-GMAbinding to the methacryloyl overlayer. However, theself-polymerized products do not affect the LDPI MSdetection after washing with ethanol, which com-pletely removes any polymerized products shown inFigure 8.

Figure 8. LDPI MS of Bis-GMA physisorbed on siliconoxide surface with photopolymerization (without any MPStreatment), prior to remove of polymerized material by eth-anol.

Figure 7. LDPI MS of adsorbed Bis-GMA on (a) SiO2 sur-face without photopolymerization and (b) methacryloyloverlayer without photopolymerization.

8 ZHOU ET AL.

Feasibility of monitoring degradation of the modelsystem by LDPI MS

LDPI MS can provide a direct and detailed chemicalanalysis of the surface bound species. The chemicalstructures of the parent and fragment ions can be usedto deduce the original chemical composition of theorganic overlayer before and after aging. Figure 6(a)shows the LDPI MS spectrum of a Bis-GMA-methac-ryloyl overlayer aged in water at 37°C for 2 h anddisplays many of the same features as does the unagedsample [Fig. 6(b)]. It thus appears that this aging pe-riod is too short to completely remove the overlayer,which remains at least in part on the surface after briefexposure to water. However, these short aging resultsare preliminary and future work will examine theaging process in detail using LDPI MS.

Exposure of the model dental composite system todifferent aging solutions should lead to hydrolysisthat cleaves ester (OACOO), ether (COOOC),and/or SiOOOC bonds. It would be difficult, at best,for XPS to detect subtle changes in either the ratio ofCOO to OOCAO groups or their C1s binding ener-gies. However, the fragments of the Bis-GMA-methac-ryloyl overlayer that would be cleaved by hydrolysisshould be readily detectable by LDPI MS, includingthe siloxane SiOOOC bond that leads to the ion at m/z640, the ether COOOC bond that leads to the m/z 531ion, and the ester OACOO bond that leads to the m/z471 ion (see Fig. 6).

Comparison of LDPI MS with direct laserdesorption MS

The direct laser desorption MS (LD, without VUVpostionization) of a Bis-GMA-methacryloyl overlayeris displayed in Figure 6(c) and the only useful peak itdisplays is m/z 143, attributed above to a methacrylatefragment. The signal intensity of LDPI MS is �15�higher than that of direct laser desorption due to thepostionization laser desorbed neutrals from theseoverlayers, which leads to higher production of ionsfor mass spectrometric analysis, as discussed previ-ously.15 The porous SiO2 substrates utilized here arealso used to facilitate desorption/ionization of ad-sorbed organic and biomolecular analytes in directlaser desorption.36 Silylated porous substrates haveadditionally been used for that purpose, but the silox-ane layer itself was not detected in direct laser desorp-tion experiments.37 Previous work has shown severalexamples where organic monolayers give more usefulsignal by LDPI MS than direct laser desorption andother desorption/ionization methods.15,38

CONCLUSIONS

A simplified model of the dental composite hasbeen developed, which captures the essential chemicalcharacteristics of the filler particle–silane–resin inter-face. This model system consists of the resin matrixcompound Bis-GMA covalently bound via themethacryloyl overlayer to the SiO2 surface via a silox-ane bond. Scanning electron microscopy shows theporous characteristic and elemental composition ofthe SiO2 film, which generally mimics that of the glassfiller particles used in dental composites. LDPI MSand XPS verify the chemistry and morphology of theBis-GMA-methacryloyl overlayer. Preliminary resultsdemonstrate that LDPI MS will be able to follow thechemical processes resulting from aging Bis-GMA-methacryloyl overlayers aged in water, artificial sali-va,6 or other aging solutions. Alternate analysis strat-egies are also being explored to improve the signal tonoise of the Bis-GMA-methacryloyl overlayer spectra,including chemical derivatization followed by VUVpostionization.16 Detailed aging studies on the Bis-GMA-methacryloyl overlayer by LDPI MS are cur-rently underway and will be the subject of futurepublications.

The XPS measurements were made on equipment locatedin the Research Resources Center at UIC.

References

1. Ferracane JL. Current trends in dental composites. Crit RevOral Biol Med 1995;6:302–318.

2. Jones FH. Teeth and bones: Applications of surface science todental materials and related biomaterials. Surf Sci Rep 2001;42:75–205.

3. Liu Q, Ding J, Chambers DE, Debnath S, Wunder SL, BaranGR. Filler-coupling agent-matrix interactions in silica/polym-ethylmethacrylate composites. J Biomed Mater Res 2001;57:384–393.

4. Drummond JL. Cyclic fatigue of composite restorative materi-als. J Oral Rehabil 1989;5:509–520.

5. Drummond JL, Khalaf MA, Randolph G. In vitro aging ofcomposite restorative materials. Clin Mater 1988;3:209–221

6. Zhou M, Drummond JL, Hanley L. Barium and strontiumleaching from aged glass particle/resin matrix dental compos-ites. Dent Mater 2005;21:145–155.

7. Geurtsen W. Substances released from dental resin compositesand glass ionomer cements. Eur J Oral Sci 1998;106:687–695.

8. Spahl W, Budzikiewicz H, Geurtsen W. Determination ofleachable components from four commercial dental compos-ites by gas and liquid chromatography/mass spectrometry. JDent 1998;26:137–145.

9. Michelsen VB, Lygre H, Skalevik R, Tveit AB, Solheim E.Identification of organic elutes from four polymer-based dentalfilling materials. Eur J Oral Sci 2003;111:263–271.

10. Jaffer F, Finer Y, Santerre JP. Interaction between resin mono-mers and commercial composite resins with human salivaderived esterases. Biomaterials 2002;23:1707–1719.

ORGANIC OVERLAYER MODEL OF A DENTAL COMPOSITE 9

11. Xiao S-J, Textor M, Spencer ND, Sigrist H. Covalent attachmentof cell-adhesive, (Arg-Gly-Asp)-containing peptides to tita-nium surfaces. Langmuir 1998;14:5507–5516.

12. Lateef SS, Boateng S, Hartman TJ, Crot CA, Russell B, HanleyL. GRGDSP peptide bound silicone membranes withstand me-chanical flexing in vitro and display enhanced fibroblast adhe-sion. Biomaterials 2002;23:3159–3168.

13. Bassindale AR, Brandstadt KF, Lane TH, Taylor PG. Enzyme-catalysed siloxane bond formation. J Inorg Biochem 2003;96:401–406.

14. Trevor JL, Lykke KR, Mencer DE, Pellin MJ, Hanley L. Surfaceanalysis of biotinylated self-assembled monolayers. AnalChem 1997;69:4331–4338.

15. Hanley L, Kornienko O, Ada ET, Fuoco E, Trevor JL. Surfacemass spectrometry of molecular species. J Mass Spectrom 1999;34:705–723.

16. Edirisinghe PD, Lateef SS, Crot CA, Hanley L, Pellin MJ, Cala-way WF, Moore JF. Derivatization of surface bound peptidesfor mass spectrometric detection via threshold single photonionization. Anal Chem 2004;76:4267–4270.

17. Cruise GM, Hegre OD, Scharp DS, Hubbell JA. A sensitivitystudy of the key parameters in the interfacial photopolymer-ization of poly(ethylene glycol) diacrylate upon porcine islets.Biotechnol Bioeng 1998;57:655–665.

18. Kim K, Shin K, Kim H, Kim C, Byun Y. In situ photopolymer-ization of a polymerizable poly(ethylene glycol)-covered phos-pholipid monolayer on a methacryloyl-terminated substrate.Langmuir 2004;20:5396–5402.

19. Savina MR, Lykke KR. Microscopic chemical imaging withlaser desorption mass spectrometry. Anal Chem 1997;69:3741–3746.

20. Van Bramer SE, Johnston MV. Tunable, coherent vacuum ul-traviolet radiation for photoionization mass spectrometry.Appl Spectrosc 1992;46:255–261.

21. Tougaard S. Surface nanostructure determination by X-rayphotoemission spectroscopy peak shape analysis. J Vac SciTechnol A 1996;14:1415–1423.

22. Tougaard S. Accuracy of the non-destructive surface nano-structure quantification technique based on analysis of the XPSand AES peak shape. Surf Interface Anal 1998;26:249–269.

23. Anderson TH, Tougaard S, Larsen NB, Almdal K, Johannsen I.Surface morphology of PS-PDMS diblock copolymer films. JElectron Spectrosc Relat Phenom 2001;121:93–110.

24. Grosvenor AP, Kobe BA, McIntyre NS. Studies of the oxidationof iron by water vapour using X-ray photoelectron spectros-copy and QUASES. Surf Sci 2004;572:217–227.

25. Lewis WG, Shen Z, Finn MG, Siuzdak G. Desorption/ioniza-tion on silicon (DIOS) mass spectrometry: Background andapplications. Int J Mass Spectrom Ion Processes 2003;226:107–116.

26. Winton MJ, Russell SD, Gronsky R. Observation of competingetches in chemically etched porous silicon. J Appl Phys 1997;82:436–441.

27. Brookes PN, Fraser S, Short RD, Hanley L, Fuoco E, Roberts A,Hutton S. The effect of ion energy on the chemistry of air-agedpolymer films grown from the hyperthermal polyatomic ionSi2OMe5

. J Electron Spectrosc Relat Phenom 2001;121:281–297.

28. Vickerman JC. Surface Analysis—The Principal Techniques.Chichester: Wiley; 1997.

29. Kurth DG, Bein T. Monomolecular layers and thin films ofsilane coupling agents by vapor-phase adsorption on oxidizedaluminum. J Phys Chem 1992;96:6707–6712.

30. Kornienko O, Ada ET, Tinka J, Wijesundara MBJ, Hanley L.Organic surface analysis by two laser ion trap mass spectrom-etry. Part II: Improved desorption/photoionization configura-tion. Anal Chem 1998;70:1208–1213.

31. Vasilopoulou M, Boyatzis S, Raptis I, Dimotikallic D, Argitis P.Evaluation of poly(hydroxyethyl methacrylate) imaging chem-istries for micropatterning applications. J Mater Chem 2004;14:3312–3320.

32. Brandriss S, Margel S. Synthesis and characterization of self-assembled hydrophobic monolayer coatings on silica colloids.Langmuir 1993;9:1232–1240.

33. Ulman A. Formation and structure of self-assembled monolay-ers. Chem Rev 1996;96:1533–1554.

34. Posthumus W, Magusin PCMM, Brokken-Zijp JCM, Tinne-mans AHA, van der Linde R. Surface modification of oxidicnanoparticles using 3-methacryloxypropyltrimethoxysilane. JColloid Interface Sci 2004;269:109–116.

35. Tepavcevic S, Choi Y, Hanley L. Growth of novel polythio-phene and polyphenyl films via surface polymerization byion-assisted deposition. Langmuir 2004;20:8754–8761.

36. Wei J, Buriak JM, Siuzdak G. Desorption/ionization mass spec-trometry on porous silicon. Nature 1999;399:243–246.

37. Trauger SA, Go EP, Shen Z, Apon JV, Compton BJ, BouvierESP, Finn MG, Siuzdak G. High sensitivity and analyte capturewith desorption/ionization mass spectrometry on silylated po-rous silicon. Anal Chem 2004;76:4484–4489.

38. King BV, Savina MR, Tripa CE, Calaway WF, Veryovkin IV,Moore JF, Pellin MJ. Single photon ionisation of self assem-bled monolayers. Nucl Instrum Methods Phys Res B 2002;190:203–206.

10 ZHOU ET AL.