Dent Mater 9:317-324, September, 1993

Determination of the degree of cure of dental resins using

Raman and FT-Raman spectroscopy
W. S. Shin 1, X.F. Li 1, B. Schwartz 1, S.L. Wunder 1, G.R. Baran 2

Department of Chemistry, Temple University, Philadelphia, PA, USA

2Department of Operative Dentistry, Temple University, Philadelphia, PA, USA

Abstract. FT-IR spectroscopy has traditionally been used to used to investigate the cure profiles of photoinitiated micro filled
determine the degree of conversion of dental resins. FT- composites (Roberts and Shaw, 1984). While IR spectroscopy has
Raman scattering provided an alternate method of obtaining provided much useful information on the degree of cure of dental
degrees of conversion for these systems and was particularly resins, there are several disadvantages in its use. In particular,
useful for measuring spectra of materials without any sample bulk samples must be prepared in thin sections unless MIR
preparation. Raman and FT-Raman spectroscopy gave iden- (Multiple Internal Reflection) or ATR (Attenuated Total Reflec-
tical results, but the latter technique was preferred for the highly tion) infrared spectroscopy techniques are employed; for brittle
fluorescent samples often encountered in commercial compos- samples, thin sections are not always easy to prepare. Further, for
ites. Linear calibration curves were obtained for the aromatic filled materials, the inorganic fraction typically has broad, in-
mixtures Bis-GMA/TEGDMA and BisphenoI-A/TEGDMA using tense absorbances which may interfere with the spectral bands of
C=C/O, and for the wholly aliphatic mixture EGDMA/EGDA interest of the resin.
using C=C/C=O, over a wide range of mole ratios. If both the Raman spectroscopy, like infrared spectroscopy, is a vibra-
mole and intensity ratios {C=C/O or C=C/C=O} were known for tional technique, and as such is sensitive to the vibrational modes
an uncured dental resin, then the degrees of conversion could of molecules (Szymanski, 1967). In dental resins, the vibrational
be obtained for the cured materials using Raman spectroscopy. bands of interest are typically the C=C double bond, the C=O
However, if the mole ratios for the uncured resin were unknown, vibration and the phenyl (O) ring modes of the monomers and
then the degree of conversion depended on the calibration crosslinked networks. For highly symmetrical molecules, the
curve, since the Raman scattering cross section of the vibra- quantum mechanical selection rules determine which modes of
tional modes depended on the molecules to which they were vibration will be IR or Raman active.
attached. For the unsymmetrical monomers and polymers used in den-
tal resins, most of the vibrations should have both infrared and
INTRODUCTION Raman activity. Nevertheless, there are still two important
The cure of dental resins is important from a practical as well as differences between the IR and Raman spectroscopy of these
a fundamental viewpoint. Investigations have shown a direct systems: (1) IR spectroscopy is an absorption technique, whereas
correlation between the degree of conversion of the resins and Raman spectroscopy is a scattering method and (2) intensities in
bulk properties such as hardness, wear, tensile and compressive IR measurements are determined by changes in the dipole too-
strength, etc. (Ruyter and Svendsen, 1978; Asmussen, 1982a; ments of the vibrations, whereas for Raman measurements, the
Ruyter, 1982; Ruyter and Oysaed, 1982; Wu, 1983; McKinney relevant quantity is the change in the polarizability tensor. These
and Wu, 1983; Ferracane, 1985; Ferracane and Greener, 1986; differences affect both the method of obtaining data from samples
Eliades et al., 1987). Of particular interest, for example, are the and the parameters which are necessary for calibration curves. In
kinetics of the curing reaction, the degree to which the thermal the IR method, it is assumed that the molar absorption coefficient
history of the cure affects the bulk properties, the influence of of a particular vibration is known or can be measured. Then,
filler on the cure, and information on the depth of cure. Thus, it using Beer's law, one can, for a given path length of sample, relate
would be useful to have a quick and reliable method ofdetermin- the intensity of the vibration to the concentration of the species
ing the degree of cure of dental resins. To date, infrared spectros- being measured. Ifa band can be found whose intensity does not
copy (Asmussen, 1982a; 1982b; Vankerckhoven et al., 1982; change, then it can be used as an internal standard, and the path
Ferracane and Greener, 1984; Ruyter and Oysaed, 1987; length of sample need not be determined. In Raman scattering,
Rueggeberg and Craig, 1988; Rueggeberg et al., 1990) and calo- the relevant quantity is the Raman scattering cross section of the
rimetry (Antonucci and Toth, 1983; Urabe et al., 1991) have been band of interest which depends on the intensity of the incoming
most frequently employed, but Raman scattering has also been light and on the polarizability tensor of the particular vibration

Dental Materials~September 1993 317

(Hendra et al., 1991). As in the case ofIR spectroscopy, it is useful The chemicals which have most frequently been used for
to have one band whose intensity can be used as an internal establishing calibration curves (for IR spectral analysis) are
standard. However, in the case of Raman scattering, the scatter- mixtures of: (1) Bis-GMA and TEGDMA (Ruyter and Svendsen,
ing power of a particular vibration is often not the same for the 1978; Ban et al., 1982; Vankerckhoven et al., 1982; Rueggeberget
same mode (such as the C=C, C=O or (I)) located on different a1.,1990); (2) Bisphenol A and TEGDMA (Ruyter and Svendsen,
molecules. 1978; Vankerckhovenetal., 1982; Rueggebergetal.,1990); and
The use of Raman spectroscopy has been limited due to the (3) proportions ofhydrogenatedand unhydrogenatedcommercial
poorer sensitivity of Raman spectroscopy compared to IR spec- bonding resin (V ankerckhoven et al. , 1982; Rueggeberg et al. ,1990 ).
troscopy and the existence ofbroadband fluoresence which inter- Because of limited access to the third of these calibration mix-
feres with the Raman signal ofinterest. Raman spectroscopy can tures and an interest in investigating calibration curves for
detect species present in concentrations of about 1%, whereas IR systems which did not have aromatic moieties, the first two of
spectroscopy can be used for concentrations of approximately these calibration systems were investigated along with mixtures
0.1%. Recently, a hybrid technique called FT-Raman spectros- ofEGDMA(ethylene glycoldimethacrylate) and EGDA(ethylene
copy has been developed which eliminates the disadvantages of glycol diacetate). (EGDMA, EGDA and Bisphenol-A were ob-
Raman scattering as a tool for characterization of impure poly- tained from Aldrich Chemical Co., Milwaukee, WI, USA). The
mer systems such as dental resins (Hirschfield and Chase, 1986; molar ratio ofC=C/(I) is 0/2 for Bisphenol-A. By mixing TEGDMA
Chase, 1986; 1987; Johnson et al., 1989). An important advance with either Bis-GMA or Bisphenol A, any molar ratio of C=C/O
was the use of an IR laser as the exciting source for the Raman could be achieved. Calibration solutions were prepared with
scattered light: the lower energy IR radiation does not excite the molar ratios for C=C/¢P ranging from 1/1 (pure Bis-GMA) to
electronic transitions in the sample which give rise to the un- 18.8/1 for the Bis-GMA/TEGDMA system and from 0 (pure
wanted fluorescence. The Raman scattered light then enters the Bisphenol A) to 7.8 for the Bisphenol A/TEGDMAsystem. In the
FTIR spectrometer as it would for an emission IR experiment former case, the molar ratio of C=C/O was equal to {1 + moles
(either as a parallel beam or condensed at an aperture) after the TEGDMA/molesBis-GMA},and in the latter case, the molar ratio
intense incident beam has been filtered out. The FTIR spectrom- ofC=C/(I)was equal to {molesTEGDMA/molesBisphenol-A}. For
eter is equipped with near IR optics to accomodate the wave- the EGDMA and EGDA system, the use of the C=C/C=O ratio to
length range of the incident signal. monitor conversion was investigated. The ratio of C=C/C=O was
The advantages of FT-Raman scattering are: (1) samples of 1/1 for EGDMAand 0/2 for EGDA. Mixtures ofthe two monomers
any thickness and geometry can be investigated because the were prepared with molar ratios of C=C/C=O ranging from 0
incident radiation can be focused on the sample and the scattered (pure EGDA) to l(pure EGDMA). In this case, the molar ratio of
radiation collected with suitable collection optics; (2) many inor- C=C/C=O was equal to the molar ratio of EGDMA/EGDA.
ganic fillers such as silica are poor Raman scatterers and thus FT-Raman spectra were obtained using a Mattson Cygnus
appear as weak, broad features in the Raman spectra; and (3) the 100 FTIR (Mattson Instruments, Madison, WI, USA) equipped
sensitivity of Raman scattering to the highly symmetrical C=C with near IR optics, i.e., a quartz beamsplitter and an InGaAs
vibration of the dimethacrylate resins used in dentistry. detector. The experimental details and a schematic of the
It is the purpose of this paper to investigate whether FT- experimental setup can be found in the paper of Johnson et al.
Raman scattering, like IR spectroscopy, can be used as a conve- (1989). The FT-Raman spectra of the uncured dental resins and
nient method of assessing the degree of cure in dental resins. In the calibration mixtures were obtained in a forward scattering
particular, calibration curves were obtained from the same mix- geometry, with the samples held in 3 mm-thick NMR tubes.
tures as were used for IR calibration curves, and the regions of Spectra were obtained with 8 cm-1resolution, and typically took
linearity for the two methods were compared. In addition, an 45 min of acquisition time for 2000 scans. Ordinarily, spectra
investigation was conducted to determine if a calibration system must be corrected for relative band intensities by ratioing against
could be established for a wholly aliphatic mixture. the instrument response profile. Spectra shown here were not
corrected for band intensities because the spectral lines of inter-
MATERIALS AND METHODS est (the C=C,¢Pand C=O) were sufficiently closein frequency (i.e.,
The resin formulations used in this study included: (1) Bis-GMA within 50 cm-1)that the instrument response was constant over
(2,2-Bis- [4-(2-hydroxy-3-methacryloxypropyloxy)phenyl] pro- this frequency interval.
pane) and TEGDMA (triethylene glycol dimethacrylate); (2) Raman data were also obtained on a Spex 1403 double
urethane dimethacrylate (UDMA) and 1,6-hexanediol monochrometer (Spex Industries, Metuchin, NJ, USA) equipped
dimethacrylate (HDDM); and (3) ethoxylated bisphenol-A with 1800-groove holographic gratings. Approximately 30 mW
dimethacrylate (EBPADMA). In addition, one composite resin from the 514.5 nm line of a 5-watt Coherent Argon ion laser
made from EBPADMAwith 3% silanated glass filler was inves- (Coherent, Palo Alto, CA, USA) was used to illuminate the
tigated. The ratio of the double bond to the phenyl ring vibra- samples, and spectra were recorded from 1500 cm-~to 1800 cm-~
tion, C=C/(I), is 1/1 for Bis-GMA and EBPADMA and 1/0 for at 2 cm-1resolution. It should be noted that the results obtained
TEGDMA, UDMA and HDDM. In this study, the settingkinetics on this scanning spectrometer (2 cm-~resolution) were identical
of 60/40 mixtures of Bis-GMA/TEGDMA, 80/20 mixtures of to those obtained from the FT-Raman instrument (8 cm-1resolu-
UDMA/HDDM and pure EBPADMA were monitored. Two of tion), as expected. There were no problems obtaining conven-
these systems (Bis-GMA/TEGDMAand EPBADMA) contained tional Raman spectra for the monomer systems, as no undesir-
both aromatic and aliphatic C=C groups, while the UDMA/ able fluorescence was observed. However, the polymerized den-
HDDM system was wholly aliphatic. (Allofthe pure resins, resin tal resins, particularly those heat-cured at elevated tempera-
mixtures and composite were obtained from ESSCHEM Co., tures, did fluoresce, making it difficult to obtain Raman spectra
Essington, PA, USA). on the grating instrument.

318 Shin et aL/Cure of dental resins using Raman spectroscopy

 The spectra in the region of interest between 1500 cm -1 and
1800 cm 4 were analyzed using a curve fitting program
(Mattson Instruments) which allowed each vibration to be a
mixture of a Lorentzian and Gaussian peak shape. Since there
was no a priori reason to expect the vibrational modes for these
condensed phase systems to have a particular peak shape, the
program was run to optimize the Lorentzian and Gaussian
combination which gave the best fit between the calculated -p_
curve and the real data. The number of peaks in this spectral
region was fixed for a particular calibration system, as discussed
in more detail below. For systems containing aromatic vibra-
tions, two peaks were always observed at 1582 cm-1(weak) and at
1610 cm -1(strong). The C=C vibration was always observed as a
singlet at a frequency of 1640 cm 4. The C=O vibration was
observed as a poorly resolved doublet in the calibration mixtures
because the frequency of the C=O is affected by intra and inter-
1500 1650 1800
molecular hydrogen bonding. Thus in the data analysis, better
Raman Shift (cm -1)
fits of the calculated spectra to the real data were obtained by
considering the C=O as a doublet.
Unlike IR spectral data for these same systems, flat base-
lines were observed both above and below the spectral region
of interest, i.e., above the C=O frequency of approximately
1720 cm-1and below the frequency of the weak aromatic mode at © C=C
1582 cm-L In addition, unlike the case for the IR spectra, the
Raman spectra contained no underlyingvibrational modes which
interfered with the bands of interest. Data analysis of the
calibration mixtures involved: (1) drawing a straight baseline C=O
between 1500 cm -1 and 1800 cm 1 and curve fitting the bands
within this region to the appropriate number of vibrations for
the system under consideration; or (2) using two different
baseline methods without the curve fitting routine to obtain peak
height information, a straight baseline between 1500 cm 4 and
1800 cm 1, or two individual baselines underneath the peaks of
interest.
The curve fitting program had as variables the intensities,
frequencies, and halfwidths of the vibrations as well as the I
relative contribution of Lorentzian and Gaussian components to 1500 1650 1800
the peak shapes. The program was run to optimize these vari- Raman Shift (cm-1)
ables for a fixed number of peaks in order to minimize the
difference between the experimental and calculated spectra. Fig. I. Curve fitting analysis for the dental liquid EBPADMA with a straight baseline
drawn between 1500 cm 1 and 1800 cm 4. This spectrum was fitted with 4 peaks, the
Fig. 1 shows an example of the curve fitting analysis for the C=O vibration at 1730 cm 4, the C=C vibration at 1640 cm 4, and the two phenyl
dental monomer EBPADMA with a straight baseline drawn vibrations at 1609 cm 4 and 1582 cm 4. The vibrations are labeled in the figure. (top):
between 1550 cm ~and 1800 cm -1. The separation of this calcu- Comparison of experimental and calculated spectra. (bottom): Separation of
lated spectrum into four peaks is shown in Fig. 1. The intensities calculated spectrum into four peaks.
for the individual bands were then obtained from the peak areas
or from the peak heights. The peak heights obtained from the curing profile for one of the three monomer systems investigated,
data without using the curve fitting program were also measured namely 60/40 Bis-GMA/TEGDMA, is shown in Fig. 3 in the
using the two baseline methods mentioned above. Molar ratios spectral region between 1500 cm -1 and 1800 cm 1, with the
of C=C/O or C=C/C=O versus Intensityc=c/Intensity. (Ic=c/I,) or positions of the C=C (1640 cm-1), C=O (1720 cm-1) and phenyl
IntensitYc=c/Intensityc=o(Ic=c/Ic=o)were plotted for each baseline (1610 cm -1) vibrations marked. As expected, the C=C vibration
method using intensities from either integrated peak areas or decreased with respect to the phenyl group mode with increasing
peak heights. A linear regression analysis was used to determine time. At a temperature of 65 o (Fig. 3B), there was still high
the goodness of fit. residual double bond content, and heating at higher tempera-
tures (75°C, 1 h) was necessary to significantly increase conver-
RESULTS sion, as shown in spectrum D of Fig. 3.
Fig. 2 shows the Raman spectra of Bis-GMA, TEGDMA, Similar curing profiles of the EBPADMA and 80/20 UDMA/
HDDM, UDMA and EBPADMA in the spectral region between HDDM dental resins were obtained showing the same qualita-
1500 cm-1 and 1800 cm -1, with the positions of the C=C, O, and tive trends. In the case of the 80/20 UDMA/HDDM dental resin,
C=O vibrations marked. Table 1 contains the positions of the the intensity of the C=C vibration decreased with respect to that
relevant vibrations for all of the monomers investigated, along of the C=O band with increasing temperature of the cure. For the
with data on their nominal molecular weights. A representative EBPADMA dental resin, the FT-Raman and Raman spectra of

Dental Materials~September 1993 319

 TABLE 1: MOLECULARWEIGHTS AND FREQUENCIESOF CH31 CH~ CH 3
C H ~ = C - CII- O - ( C H 2 C H 2 =- O - ~ - C - ~ - IO - ( C H zCH z b- O - C -IBC = C H
THE C=C, C=O AND PHENYL VIBRATIONS FOR = - ~ O
2

DENTAL RESIN MONOMERS

Monomer Molecular Weight phenyl (cm-~) C=C(cm-') C=O(cm -1)
CH3 CH~ CH~ CH~
I I L
CH2=C-C-O- CH22-O-C-NHCHz-CH-CH - C - CH 2 - N H - C - O - ( C H z ) 2 - O - C - C = C H
Bis-GMA 512 1610 1640 1718 II JI I II II 2
o o ~ oD
_>,
TEGDMA 286 1640 1714 CH~ CH 3
' ,
~,H2=C-ICI-O-(CH2).-O-C - C= CH2
UDMA 470 1640 E
1716 O O C
CH~
CH3-c-o-(CH2CHzO)3-C- C = CH2/f
~
1,6-HDDM 254 1640 1716
EBPADMA* 422 1610 1640 1718 CH3 OH CH 3 OH CH 3 --
CH~= CI _ C _ O _ C H 2 _ CIH _ C H 2 _ O _ ~ C _ ~ _~O _ C H 2 _ C IH _ C H 2 _ O _ C _ CI = CH 2
II I II

L_ f J 2,
O CH3 O
BisphenoI-A 288 1618
EGDA 146 1738
1500 1650 1800

EGDMA 198 1640 1718 Raman Shift (cm-1)

* The average number of ethylene oxide units is 5.

Fig. 2. Raman spectra of monomers used in dental resin formulations in the spectral
light-cured (130 mW of a filtered tungsten source at room tem- region between 1500 cm4 and 1800 cml: (A) Bis-GMA, (B) TEGDMA, (C) HDDM,
perature) samples were compared with samples heat-cured for (D) UDMA,2,4,4 Isomer and (E) EBPADMA.
1 h. The decrease in the C=C mode with respect to the phenyl
vibration was greater for the heat-cured than the light-cured
sample, as expected. For light-cured systems, it was advanta-
geous to use the FT-Raman system rather than a conventional
Raman setup because the infrared laser beam used to monitor the
cure for the former would not further cure the sample in the beam.
The 514.5 nm radiation from the argon ion laser used in the C=C C=O
scanning Raman system continued to cure the sample in the
beam. The FT-Raman spectra of a 100°C heat-cured sample of
EBPADMA and a 100°C heat-cured sample of EBPADMA with
3% silanated glass filler were identical within experimental
error.
Raman spectra between 1500 cm4 and 1800 cm-~of represen-
tative mixtures of Bis-GMA]TEGDMA, Bisphenol A/TEGDMA,
and EGDMA/EGDAare shown in Figs. 4, 5 and 6, respectively.
The frequency of the C=C vibration for all of the monomers
investigated appeared at 1640 cm-1and the FWHM (full width at t
half maximum) was 10 cm-1, which was narrow for a liquid 1500 1650 1800
sample. Similarly, the frequency ofthe major phenyl vibration at Raman Shift (cm-1)
1610 cm4 was relatively constant and had a FWHM of approxi-
mately 15 cm-L Fig. 3. FT-Raman spectra in the spectral region 1500 cm1 to 1800 cm-1 of 60•40 Bis-
GMA/TEGDMA: (A) uncured dental resin, (B) heat-cured for 30 rain at 65°C, (C) heat-
The C=O mode was affected by both intra and intermolecular cured for 1 h at 65°C, (D) heat-cured for 1 h at 75°C.
hydrogenbonding. For TEGDMA, HDDM, EBPADMA,EGDMA
and EGDA, which did not hydrogen bond, the FWHM of their
C=O vibrations, seen in spectra B, C, and E of Fig. 2 and spectra standard if, for example, the presence of moisture were to signif-
A and D of Fig. 6, was approximately 15 cm-L By contrast, in the icantly affect the hydrogen bonding and thus, the intensity of
systems for which hydrogen bonding occurred, the C=O appeared the vibration. In this study, the presence of moisture was not
as a broadened doublet (- 30 cm4). The presence ofboth C=O and controlled.
OH groups in Bis-GMA, and the presence of both NH and C=O The frequency of the C=O vibration was also affected by
groups in UDMA, accounted for the hydrogen bonding which intramolecular resonance effects and therefore occurred at differ-
occurred for these monomers. The broadened C=O doublet ent frequencies in molecules where this effect was present or
arising from this interaction was observed in the spectra labelled absent (Szymanski, 1967). All of the monomers investigated,
A and D of Fig. 2 for Bis-GMA and UDMA, respectively. When except EGDA, contained a C=O vibration adjacent to a C=C
TEGDMA was mixed with Bisphenol-A, the hydrogen bonding vibration, and the frequency of the C=O for these systems oc-
between the C=O group of TEGDMA and the OH group of curred at 1716 + 2 cm-1. In EGDA(spectrum D, Fig. 6), where this
Bisphenol-A resulted in the broadened doublets observed in resonance structure did not exist, the frequency of the C=O
spectra B and C of Fig. 5. The effects of hydrogen bonding on vibration occurred at 1738cm-1. When EGDMA was mixed with
the C=O vibration may affect its usefulness as an internal EGDA (spectra B and C, Fig. 6), a broadened doublet occurred due

328 Shin et aL/Cure of dental resins using Raman spectroscopy

 CH3
CH2= C_C_O_CH2_
I
OH CH~ OH
CI _CH2_O ~ _ C I ~ _ O _ C H 2 _ C_
CH3
I CH2_O _C_ CI =CH
©
O CH~ O
o,.3 A
A/_ \
-c~ = C=C C=O
03
E g
E

CH2--C-C-O-(CH~CH20)3-C-C = CH~ \
CH3 CH3
I I
,~H2=C-C-O - CH2CH20 6-C C=CH /~
o
.... o
)L_J I
A 1650 1800
1500
I
Raman Shift (cm -1)
1500 1650 1800
Raman Shift (cm-1)
Fig. 4. Raman spectra between 1500 cm -~ and 1800 cm -~ of Bis-GMA/TEGDMA Fig. 5. Raman spectra between 1500cm 4 and 1800 cm 1 of BisphenoI-A/TEGDMA
calibration solutions, with molar ratios [C=C/~] : (A) -, pure TEGDMA; (B) 5.1; calibration solutions, with molar ratios [C=C/~] : (A) = -, pure TEGDMA; (B) 1.8;
(C) 1.8; and (D) 1, pure Bis-GMA. (C) 0.2; (D) 0, pure BisphenoI-A.

to the frequency difference of 20 cm-1between the frequency of CH3-C-O-

II
C H 2 C H 2 - O - C - CH 3
II
the C=O vibration of EGDA (1738 cm-1) and that of EGDMA o o
(1718cm 1). Thus it was not meaningful to measure a peak height
for this system.
In order to obtain intensities for the vibrations of interest,
peak areas or peak heights were measured using the previously
discussed methods of analysis. For systems in which there were (D C=O
no aromatic goups, the spectral vibrations of interest, namely the
C=O and C=C modes were well separated and had no interference
from overlapping modes. In the case of the UDMA/HDDM
mixture, both the C=O and C=C vibrations appeared at the same I
frequency for UDMA and HDDM, the width of the vibrations o o /\ "1
in the mixture was narrow, and a curve fitting procedure using I I
1500 1650 1800
two peaks could be used. The frequencies of the C=O and
Raman Shift (cm -1)
C=C vibrations for the mixture EGDMA/EGDA were also well
separated, and peak areas were easily measured for the C=C and
C=O vibrations since they did not overlap, using a flat baseline Fig. 6. Raman spectra between 1500 cm 1 and 1800 cm 1 of EGDMAJEGDA calibration
solutions, with molar ratios [C=C/C=O]: (A) 1, pure EGDMA, (B) 0.4; (C) 0.1;
drawn from above to below the frequencies of interest. However, (D) 0, pure EGDA.
since the C=O appeared as a doublet, a curve fitting procedure
was used which separated the vibrations into three peaks; the
1640 cm 1C=C vibration and the {1718 cm-1 and 1738 cm-G C=O DISCUSSION
doublet. For the systems which contained all three vibrations, The use of Raman scattering and other spectroscopic methods for
C=C, C=O, and O, there was some band overlap between the C=C monitoring the dental resins requires that the amount of double
and • modes. A curve fitting procedure was used which separated bonds which are present after cure be quantified to determine the
the vibrations into four or five peaks: the 1640 cm-~C=C vibra- percent conversion. For infrared spectroscopy, the approxima-
tion, the {1610 cm 1 and 1582 cm -1}phenyl doublet and the C=O tion is made that the molar absorptivities for the C=C and phenyl
singlet or doublet. Although the 1582 cm-~vibration was weak, vibrations are the same for any molecule which contains these
including it in the fitting program improved the goodness of fit groups. The ratio of the aliphatic C=C absorption to that of the
between the measured and calculated spectra. A flat baseline aromatic absorption is calculated and the aromatic absorption is
was drawn from above to below the frequencies of interest. used as an internal standard. The percent conversion is then
Calibration curves using peak areas or peak heights from the obtained using the formula (Ferracane et al., 1984; Rueggeberg et
curve fitting program, as well as from the two baseline methods, al., 1990):
were obtained for the three mixtures. The results of the linear
DC% = 100- C/U (1)
regression analysis used to determine the goodness of fit are
presented in Table 2. The peak area ratio for C=C/C=O versus the where C = equivalent molar ratio of the cured specimen and U =
molar ratio for the EDGMA/EGDA system is plotted in Fig. 7. The equivalent molar ratio of the uncured specimen. This is the most
peak height ratios obtained from the curve fitting routine, plotted general formula, where both the molar ratios of the uncured and
as a function of the mole ratio of C=C/¢p for both the Bis-GMA/ cured material are unknown and must be determined from the
TEGDMA and Bisphenol-A/TEGDMA mixtures are shown in calibration curve. In the case where the molecular composition
Fig. 8. of the uncured dental resin is known, i.e., both the molar ratio and

Dental Materials~September 1993 321

 TABLE 2: LINEAR REGRESSION ANALYSIS OF Bis-GMA/TEGDMA, due to the underlying absorbances on top
BisphenoI-A/TEGDMA AND EGDMA/EGDA CALIBRATION SOLUTIONS of which the vibrational modes of interest
were sitting (Rueggeberg et a/.,1990). In
Mixture Method of Baseline Intensity Ratios r2 Raman spectroscopy, this was not as much
Analysis Method
of a concern because there were no under-
lying vibrational modes, so that intensity
EGDMNEGDA curve fit linear (C----C)AREA/(C----O)ARE
A 0.998
ratios were easily obtained. As seen in
Table 2, the regression analyses used for
the different baseline methods and using
Bis-GMA/TEGDMA curve fit linear (C=C)AREA/(I~)ARE A 0.981 either areas or peak heights for intensities
curve fit yielded similar values for the coefficient of
linear (C=C)pEAK H'r/((~)PEAKHT 0.998
linear regression parameter (r2) when the
straight (C=C)PEAK H'r/(([{))PEAKHT 0.997 data were fit to a straight line. However,
with decreasing C=C content, the C=C vi-
individual (C=C)PEAK HT/((~)PEAKHT 0.985 bration will be sitting on the base of the
mode. Therefore, data analysis methods
which better separate the peaks, i.e., which
BisphenoI-A/TEGDMA curve fit linear (C=C),REA/((t))AREA 0.958 use the curve deconvolution procedures,
may give more accurate results.
curve fit linear (C=C)PEAKHT/(F)PEAKHT 0.986 A second concern when using IR spectra
straight (C=C)PEAKHT/(F)PEAK,T 0.988 to obtain degrees of conversion is the com-
position range where linearity between
individual (C----C)PEAK
HT/(F)PEAKHT 0.968 absorbance ratios and mole ratios is main-
r2= coefficient of linear regression analysis tained. The results of the present study
indicated that the same calibration mix-
tures were linear over roughly the same
range of compositions in both the IR and
TABLE 3: INTENSITY RATIOS OF C=C/C=O AND C=C/¢ OBTAINED USING THE CURVE Raman measurements, that is for C=C/O
FITTING PROGRAM AND USING EITHER PEAK AREAS OR PEAK HEIGHTS mole ratios of up to 7/1.
Monomer (Ic=c/Ic=o)*REA (Ic=c/Ic=o)PKHT (Ic=c/I~)*RE* (Ic=c/IF)PKHT
A comparison of the calibration curves
of Bis-GMA/TEGDMA and Bisphenol-A/
TEGDMA 1.07 1.72 TEGDMA when the method of data analy-
sis was the same (i.e., peak areas or the
Bis-GMA 0.99 2.15 0.73 0.96
same baseline method used in both cases)
UDMA 0.81 1.94 indicated that although both calibration
1,6 HDDM 0.97 1.63 mixtures yielded linear plots, they did not
have the same slopes, so that different
EBPADMA 1.28 1.59 0.78 1.01 percent conversions were obtained depend-
EGDMA 1.21 ing on the calibration mixture employed.
The linearity observed indicated that, as
methyl acrylate 1.29 2.51 for IR spectroscopy, when the molar ratios
ethyl dimethacrylate 4.17 6.41
of the bands of interest and their intensity
ratios (i.e., Ic=c/I¢) were known for the
benzyl cinnamate 12.68 9.3 2.18 2.36 uncured dental resins, as in this study, Eq.
cinnamyl acetate 41.86 25.18 1.86 0.99 2 could be used to determine the percent
conversion. Any consistent method of data
ethyl sorbate 7.62 13.95
analysis could be employed to obtain per-
cent conversion, and details about whether
intensity ratio of the uncured dental resin are known, this the C=O vibration was a singlet or a doublet were not important
formula can be simplified, so that (Ferracane and Greener, 1984): since the shape of the bands did not change with conversion.
However, for an unknown dental resin, the percent conversion
DC% = 100 - Intensity (C=C/(I))curedspecimen results will depend on the calibration mixture used to convert
(2) intensity ratios to mole ratios. As for the case of IR spectroscopy,
Intensity (C=C/¢P)...... dspecimen
Eq. 1 can be employed in Raman scattering only if a calibration
Here, the straight line drawn between two points, one deter- curve provides a relationship between the equivalent molar
mined by the intensity and C=C/¢p mole ratio of the uncured ratios and the relevant Raman intensities. In the case of Raman
material, and the other by the zero point for 100% conversion, is scattering, the calibration curves were not the same because of
used as the calibration curve. the difference in scattering power of the same vibrational mode
When percent conversion is monitored using IR spectroscopy, residing on different molecules, rather than the particular method
much attention has been paid to various baseline procedures of baseline analysis (as is the case for IR spectroscopy). The ratio
used to evaluate the intensity ratio of the C=C to ¢P vibrations, of Ic=c/I~ for a series of molecules is presented in Table 3, where

322 Shin et al./Cure of dental resins using Haman spectroscopy

 by inter or intramolecular interactions (such as hydrogen bond-
ing) which gave rise to band broadening of the C=O vibration.
0 This effect could be eliminated by the use of integrated areas to
II
calculate the intensity of this vibration. A more serious problem
©
II was that the C=O mode was not a good internal standard if the
©
calibration mixture and the dental resin were not the same.
o
Values of Ic=c/Ic=ofor several monomers are shown in Table 2. It
can be seen that there are very large differences in the ratio
among the molecules. As was the case for Ic=c/I., the values are
most similar when the difunctional methacrylate molecules used
in dental resins were considered. However, since the mole ratio
a f of C=C/C=Ois equal to 1 for all ofthese molecules, they cannot be
used to obtain a calibration curve. Thus, for wholly aliphatic
©l I
mixtures, Eq. 2, rather than Eq. 1, should be used.
O.O0 0.50 1 .O0
Raman and FT-Raman spectroscopy provided alternative
Molar Ratio C:C/C=O methods of determining percent conversion for dental resin
Fig. 7. Calibration curve for mixtures of EGDMA/EGDA. The intensity ratio of the systems. The principle advantage of the Raman technique was
C=C/C=O vibrations was obtained using peak areas. The molar ratio of C=C/C=O was that the material could be investigated without any sample
equal to the mole fraction of EGDMA. preparation. In addition, inorganic glass fillers did not spectrally
interfere with the bands used to determine percent conversion for
10 -- the dental resins. The advantages of the near IR laser used in
3, conventional Raman scattering lay in the ability to avoid un-
(- wanted background fluorescence and also the ability to investi-
©
gate visible light-cured samples without further cure. The disad-
II
©
vantage ofthe Raman scattering method was the limited number
0
of Raman systems in the dental community. However, since the
FT-Raman technique uses an FTIR spectrometer, the FT-Raman
8E
system may become more commonly available as an accessory in
E
o~ the future.
3:
ACKNOWLEDGMENTS
¢
n This investigation was supported in part by Research Grant DE
09530 from the National Institute ofDental Research, Bethesda,
o 5 lO MD, USA. The dental resin monomers and calibration monomers
were kindly supplied by George Cowperthwaite of ESSCHEM
Molar Rat,o C=C/Phenyl Corporation.

Fig. 8. Calibration curves for mixtures of Bis-GMA/I"EGDMA (-) and BisphenoI-N Received May 18, 1992/Accepted March 27, 1993
TEGDMA (+). The intensity ratios of the C=C/F vibrations were obtained using peak
heights and a flat baseline. For the Bis-GMA/TEGDMA system, the molar ratio of
C=C/O is equal to {1 + moles TEGDMA/moles Bis-GMA}, and for the BisphenoI-A/ Address correspondence and reprint requests to:
TEGDMA system, the molar ratio of C=C/O is equal to {moles TEGDMA/moles S. L. Wunder
BisphenoI-A}. Department of Chemistry 016-00
Temple University
Philadelphia, PA 19122 USA
peak heights or areas were used to obtain intensities. From this
data, it is clear that better conversion data were obtained when
the calibration mixture contained molecules which were similar REFERENCES
to those used in dental resins. The intensity ratio of Ic=JI, was Antonucci JM, Toth EE (1983). Extent of polymerization of
much closer for the two dental resin monomers than for the two dental resins by differential scanning calorimetry. JDent Res
monomers not in this group. Thus, for the purposes of obtaining 23:704-707.
values of percent conversion for dental resins, the calibration Asmussen E (1982a). Restorative resins: hardness and strength
curve obtained from the Bis-GMA/TEGDMA system will give vs. quantitity of remaining double bonds. Scand J Dent Res
more accurate results than those obtained from the Bisphenol-A/ 90:484-489.
TEGDMA system. This was also found to be the case for the IR Asmussen E (1982b). Factors affecting the quantity ofremaining
calibration systems by Rueggeberg et al. (1990). double bonds in restorative resin polymers. Scand JDent Res
A wholly aliphatic calibration mixture was also considered. 90:490-496.
The results indicated that the same conclusions could be drawn Ban S, Murakami M, Mizumoto A, Kito M and Hasegawa J
for these systems. For the case of a dental resin where the initial (1982). Infrared spectroscopy of composite resins. Aichi-
mole ratio of C=C/C=O was not known, the values of percent Gakuin J Dent Sci 20:307-313.
conversion obtained from the EGDMA/EGDA calibration curve Chase DB (1986). Fourier transform Raman spectroscopy. J A m
may be inaccurate for two reasons. The first was that introduced Chem Soc 108: 7485-7491.

Dental Materials~September 1993 323

Chase DB (1987). Fourier transform Raman spectroscopy. Anal Res 63:293, Abstr. No. 1098.
Chem 59: 881-889. Rueggeberg FA, Craig RG (1988). Correlation of parameters
Eliades GC, Vougiouklakis G, Caputo AA (1987). Degree of used to estimate monomer conversion in a light-cured compos-
double bond conversion in light-cured composites. Dent Mater ite. J Dent Res 67:932-937.
3:19-25. Rueggeberg FA, Hashinger DT, Fairhurst CW (1990). Calibra-
Ferracane JL (1985). Correlation between hardness and degree tion of FTIR conversion analysis of contemporary dental resin
of conversion during the setting reaction of unfilled dental composites. Dent Mater 6:241-249.
restorative resins. Dent Mater 1:11-14. Ruyter IE (1982). Methacrylate based on composite restorative
Ferracane JL, Greener EH (1984). Fourier transform infrared materials. Acta Odont Scand 40:359-376.
analysis of degree of polymerization in unfilled resins-meth- Ruyter IE, Oysaed H (1982). Compressive creep of light-cure
ods comparison. JDent Res 63:1093-1095. resin based restorative materials. Acta Odont Scand 40:319-
Ferracane JL, Greener EH (1986). The effect ofresin formulation 324.
on the degree of conversion and mechanical properties of Ruyter IE, Oysaed H (1987). Composites for use in posterior
dental restorative resins. J Biomed Mater Res 20:121-31. teeth: Composition and conversion. JBiomed MaterRes 21:
Hendra P, Jones C, Warnes G (1991). Fourier transform Raman 11-23.
spectroscopy: instrumentation and chemical applications. Ruyter IE, Svendsen SA (1978). Remaining methacrylate
New York: Ellis Harwood. groups in composite restorative materials. Acta Odont Scand
Hirschfield T, Chase DB (1986). FT-Raman spectroscopy: devel- 40:359-376.
opment and justification. AppISpectr 40:133-137. Szymanski HA, Editor (1967). Raman spectroscopy: theory and
Johnson C, Mao J, Wunder SL (1989). FT-Raman spectroscopy practice. New York: Plenum Press.
of the curing reactions ofpolyimides. In: Feger C, Khojasteh Urabe H, Wakasa K, Yamaki M (1991). Cure performance of
MM and McGrath JE, editors. Polyimides: Materials, Chem- multifunctional monomers to photoinitiators: a thermo-
istry and Characterization. Amsterdam Elsevier Science analytical study on Bis-GMA-based resins. J Mater Sci 26:
Publishers B.V., 347-361. 3185-90.
Mc Kinney JE, Wu W (1983). Effect of degree of cure on hardness Wu W (1983). Degree of cure and wear resistance of dental
and wear on three commercial dental composites. JDentRes composites. JDent Res 62: 671, Abstr. No. 1048.
62:285, Abstr. No. 1047. Vankerckhoven H, Lambrechts P, Van Beylen M, Davidson CL,
Roberts TA, Shaw DJ (1984). Cure profiles of photoinitiated VanHerle G (1982). Unreacted methacrylate groups on the
microfilled composites by laser Raman spectroscopy. J Dent surfaces of composite resins. JDentRes 61:791-795.

324 Shin et at~Cure of dental resins using Raman spectroscopy