E.
Despagnet-Ayoub
*a,
W. W.
Kramer
a,
W.
Sattler
b,
A.
Sattler
a,
P. J.
LaBeaume
c,
J. W.
Thackeray
c,
J. F.
Cameron
c,
T.
Cardolaccia
c,
A. A.
Rachford
c,
J. R.
Winkler
*a and
H. B.
Gray
*a
aBeckman Institute, California Institute of Technology, Pasadena, California 91125, USA. E-mail: despagne@caltech.edu
bThe Dow Chemical Company, Formulation Science, Core R&D, 400 Arcola Road, Collegeville, Pennsylvania 19426, USA
cDow Electronic Materials, 455 Forest Street, Marlborough, Massachusetts 01752, USA
First published on 3rd November 2017
The products from the 193 nm irradiation of triphenylsulfonium nonaflate (TPS) embedded in a poly(methyl methacrylate) (PMMA) film have been characterized. The analysis of the photoproduct formation was performed using chromatographic techniques including HPLC, GPC and GC-MS as well as UV-vis and NMR spectroscopic methods. Two previously unreported TPS photoproducts, triphenylene and dibenzothiophene, were detected; additionally, GPC and DOSY-NMR spectroscopic analyses after irradiation suggested that TPS fragments had been incorporated into the polymer film. The irradiation of acetonitrile solutions containing 10% w/v PMMA and 1% w/v TPS in a 1 cm-path-length cuvette showed only a trace amount of triphenylene or dibenzothiophene, indicating that topochemical factors were important for the formation of these molecules. The accumulated evidence indicates that both products were formed by in-cage, secondary photochemical reactions: 2-(phenylthio)biphenyl to triphenylene, and diphenylsulfide to dibenzothiophene.
We studied the photochemistry of the triphenylsulfonium (TPS) cation, a photoacid generator employed in photolithography,2 in both solution and rigid media employing 193 nm irradiation as well as other excitation sources. In earlier work on TPS reactions in solution, it was established that both in-cage and cage-escape products3–5 are formed, most likely by the heterolysis or homolysis of sulfur–carbon bonds. The heterolysis pathway produces a phenyl cation and diphenylsulfide, whereas a diphenylsulfinyl radical cation6 and a phenyl radical are the expected products from homolysis.3,7 Recombination reactions of the different fragments yielded 2-, 3- and 4-(phenylthio)biphenyl with one equivalent of acid. After cage-escape, the fragments reacted with the solvent or with fragments from another TPS molecule or other contaminants, generating diphenylsulfide, biphenyl, and acid (Scheme 1). An investigation of crystalline or finely ground solid onium salt photochemistry revealed decomposition pathways similar to those found in solution, but with increased amounts of in-cage products, likely owing to the greatly reduced fragment diffusion.8,9 There have been few prior reports of the photochemistry of TPS in polymeric matrices,10 only one of which with results from experiments involving 193 nm irradiation.11
As TPS is normally embedded in polymers in the photolithographic process, it is of importance to elucidate the photodecomposition pathways of this photoacid generator in confined environments, with the goal of establishing whether or not these pathways are qualitatively different from those observed in solution.
Notably, in our work on the 193 nm photochemistry of triphenylsulfonium nonaflate in multiple polymeric matrices, we found two products that were not observed in liquid phase experiments: the new topochemistry12 product is triphenylene, a desulfurized TPS derivative. Such discovery will also have a great impact on organic electronics; the introduction of TPS in polymer light emitting devices allowed the improvement of conductivity by charge injection. Therefore, the formation of polycyclic aromatic compounds during the irradiation process is an attractive characteristic as it will certainly favor even more the conductivity of these polymer films.13
Irradiation was performed using one of three different light sources. The 193 nm irradiation employed a home-built nitrogen-purged exposure chamber equipped with a D2-lamp and a 193 nm interference filter (20 ± 5 nm FWHM, Tmax = 15%) and a 1 cm water filter (emission spectrum: see Fig. S1 in the ESI,† dose <10 μW as measured by using a high-sensitivity thermal power sensor (ThorLabs S401C)). 254 nm radiation was produced using a krypton fluoride excimer mirror to direct the output from a Xe arc lamp through two 1 cm-path-length water infrared filters and a 254 nm interference filter. The irradiation power measured with a photodiode (ThorLabs Standard Photodiode Power Sensor (S120C), Si, 200–1100 nm) was 360 μW (360 μJ s−1) (emission spectrum: see Fig. S2 in the ESI†). Coated discs were placed in front of the light source for 3 h (Xe-lamp) or 3 days (3 d) (D2-lamp). Pulsed laser excitation (266 nm, 8 ns) was from the fourth harmonic of a Q-switched Nd:YAG laser (Spectra-Physics Quanta-Ray PRO-Series) operating at 10 Hz. A mechanical shutter was used to select individual pulses at a reduced repetition rate (1 Hz).
The absorbance k-value was determined using a 320 nm thick film, soft-baked at 95 °C for 60 s on 200 mm silicon wafers, and a J. A. Woollam WVASE32 vacuum ultraviolet variable angle spectroscopic ellipsometer using polarized light. Changes to the polarized light phase and amplitude after it passed through the coated film were observed to cover 180–900 nm using a range of 1.4 eV to 6.875 eV in increments of 0.0375 eV and three incident angles of 65, 70, and 75 degrees. These raw data were fit to a model using J. A. Woollam software to determine the real and imaginary optical constants of the film at each wavelength.
The detection of the remaining TPS in the reaction mixture employed a Zorbax SB-C18 column (5 μm, 4.6 × 150 mm) with UV-detection at 214 nm. The eluent used was 50:50 (v/v) CH3CN/(H2O + 0.1% TFA) for 15 min, and then held at 100% CH3CN for 10 min to wash the column.
Gas chromatography-mass spectrometry (GC-MS) was conducted on a MSD5972/GC5890 with an HP6890 autosampler and an electron ionization (EI) method. The column was a DB-5 30 m × 0.25 mm × 0.25 μm film. The injector was set at 250 °C and the detector at 280 °C. The procedure: 50 °C for 2 min; 15 °C for 15 min; and then 270 °C for 3 min.
Gel permeation chromatography (GPC) employed a Waters Alliance system with an e2695 Separations Module using differential refractive index and photodiode array detection and THF (1.2 mL min−1) as the elution solvent (30 °C).
DOSY-NMR performed in DMSO-d6 was a DgcsteSL (DOSY gradient compensated stimulated echo Spin Lock) experiment15 with a diffusion delay of 50 ms, a difference in gradient pulse of 5 ms and an array of 15 gradient values.
HPLC was employed to separate and identify the different photoproducts present in the irradiated TPS-PMMA films. The HPLC trace (Fig. 2) shows the presence of several compounds in the reaction mixture. Many of these peaks were assigned by comparing their retention times in HPLC and GC-MS with those of previously reported photoproducts 2-, 3- and 4-(phenylthio)biphenyl (2-BiPhSPh, 3-BiPhSPh, and 4-BiPhSPh), biphenyl (PhPh), and diphenylsulfide (PhSPh).3,5,18 However, two peaks with retention times of ∼43 and ∼51 min did not match any known photoproducts. The product with the retention time of 43 min was identified as dibenzothiophene by GC-MS and by spiking an HPLC sample with a small amount of the pure substance. The product with retention time of 51 min was isolated by prep-HPLC and characterized by NMR spectroscopy. The 1H NMR spectrum of the product showed only two downfield signals (8.87 and 7.70 ppm), consistent with a highly symmetric aromatic ring system. GC-MS of the product gave a molecular weight of 228 m/z. Based on the 1H NMR and GC-MS analyses, we concluded that the photoproduct was triphenylene. We further confirmed this assignment by comparing the UV-vis spectrum, HPLC retention time, and 1H NMR spectrum with those of pure triphenylene.
As neither dibenzothiophene nor triphenylene has been previously identified as a product of TPS photolysis, additional control experiments were conducted in order to validate the assignment. The influence of the coating solvent as well as the identity of the polymer and irradiation wavelength were investigated by monitoring the appearance of triphenylene by UV-vis spectroscopy of the irradiated films. Triphenylene formation was not affected by the use of PGMEA or cyclohexanone as the coating solvent in place of THF, and the use of polymers other than PMMA had no effect on triphenylene formation. Indeed, the UV-vis spectra of the irradiated films of 1% w/v TPS in PEGME and poly(tBMA-co-HEMA-co-GBLMA) with different molecular weights (Mw = 4600; 14000; 5300 and 8600) (Fig. S4 and S5 in the ESI†) clearly demonstrated the presence of triphenylene, as evidenced by the sharp vibronic feature at 260 nm. Additionally, we investigated the effect of the irradiation wavelength by testing the TPS-PMMA film after exposure to 254 nm light from a Xe-lamp. The in situ UV-vis spectra of these films under 254 nm irradiation (Fig. S6 in the ESI†) show triphenylene generation (260 nm signal), reaching a maximum at 3 h exposure followed by a small decrease in band intensity. Triphenylene was observed in all cases.
The photoproducts from the 193 and 254 nm irradiation of TPS-PMMA films were quantified by HPLC. Irradiation for 3 h with 254 nm radiation led to quantitative conversion of TPS, whereas irradiation for 3 d with 193 nm radiation resulted in 90% conversion (Table 1). In each case the main product was the in-cage molecule, 2-(phenylthio)biphenyl (2-BiPhSPh), in addition to smaller amounts of triphenylene; substantial amounts of cage-escape products (diphenylsulfide (PhSPh), dibenzothiophene) are observed. At both irradiation wavelengths, the photoproduct distribution favored in-cage products over cage-escape ones (77% in both cases), as expected due to limited diffusion through the polymer film.9
The material recovered only accounts for 47% (or 27%) of the initial TPS after the 193 (or 254) nm exposure despite ≥90% consumption of the starting material. This finding suggests that many TPS fragments reacted with PMMA. Indeed, GPC analysis of the PMMA-TPS film after irradiation showed increased near-UV absorption (260 nm) for irradiated samples compared to PMMA itself (Fig. 3), indicating that aromatic TPS fragments were incorporated in the polymer matrix. Furthermore, a DOSY-NMR experiment was performed on an irradiated film showing aromatic signals matching the diffusion rate of the PMMA polymer (Fig. S7 in the ESI†).
Fig. 3 UV-vis spectra of high MW GPC fractions of films before and after irradiation with and without TPS. |
Solution experiments showed lower TPS conversion compared to in-film experiments (3 d under 193 nm irradiation: 7% vs. 90%; 3 h under 254 nm irradiation: 25% vs. 100%). To confirm that detectable quantities of triphenylene or dibenzothiophene are not produced in solution at higher TPS conversion, longer irradiation was conducted at 254 nm. The TPS consumption was monitored by HPLC and the reaction was stopped after 48 h as the TPS consumption had reached 88%. Shown in Fig. 4 are the time traces of the consumption of TPS and the yields of the various photoproducts over the course of the reaction. Again, acetanilide, PhSPh, 2-BiPhSPh, and 4-BiPhSPh, but biphenyl also appeared at longer irradiation times. The ratio of in-cage to cage-escape products remained constant throughout the reaction. After 24 hours of irradiation (72% TPS consumption), benzene, dibenzothiophene, and triphenylene could be detected, but only in trace amounts (<1% yields). The total material recovered after 48 hours of irradiation accounts for 73% of the starting TPS, suggesting that reactions between TPS fragments and PMMA occur in solution as well.
Fig. 5 shows the GPC traces obtained for each of the films. Due to the small differences in the areas of cleaved wafers, the GPC traces were normalized at a retention time of 12 minutes corresponding to the PMMA fraction. As expected, TPS (ca. retention time = 14.8 minutes) decreases with increasing exposure, whereas the photoproducts increase (retention times between 15 and 16 minutes) with increasing exposure dose. The use of a photodiode array detector on the GPC allowed for the definite determination of photoproducts. Fig. 6 shows the UV-vis spectra obtained at a retention time of 15.8 minutes, corresponding to the elution of triphenylene (Fig. S11: UV-vis spectra obtained at a retention time of 15.1 minutes, corresponding to the elution of 2-BiPhSPh; see the ESI†). As was observed in the studies described above, triphenylene is produced with increasing exposure dose. Notably, triphenylene is produced at doses as low as 20 mJ cm−2. Significant production of triphenylene occurs with increasing exposure as clearly demonstrated by the UV-vis spectra. In order to quantify the photoproducts, we applied the same HPLC method as described above. The relative amounts of photoproducts are plotted in Fig. 7, demonstrating the rapid production of 2-BiPhSPh with later production of secondary photoproducts such as triphenylene. As these data were obtained in a DOW semiconductor fabrication facility, with quantified exposure doses, they demonstrate the relevance of this topophotochemistry to the lithographic world. At doses in the range of common exposure regimes, triphenylene is produced with a loss of sulfur.
Fig. 5 GPC traces (absorbance at 210 nm, normalized at 12 minute retention time) of extracted films from 0, 10, 20, 30, 40 and 50 mJ cm−2 exposure ladder study. |
Fig. 6 UV-vis spectra (corrected via the normalization factor taken in Fig. 5) obtained at 15.8 minute retention time from GPC traces, corresponding to the triphenylene fraction. |
Fig. 7 Photoproduct quantification by HPLC analysis. Concentrations have been normalized to the initial TPS consumption, which is set to 1 at an exposure dose of 0 mJ cm−2. |
The Dill C-parameter is a commonly used term to quantify the rate of a photochemical reaction of a photoacid generator (PAG) in a photoresist, and is defined as the first-order photodecomposition decay constant (i.e., M = M0 e−CE, where M = PAG concentration, M0 = initial PAG concentration, C = Dill C-parameter and E = exposure energy).19 Assuming that the destruction of TPS is first order and that each photochemical event leads to one photoacid produced, then the C-parameter can be determined by fitting the equation [H+] = 1 − e−CE, where [H+] is the normalized acid concentration (which is assumed to be equal to 1 − [TPS]/[TPS]0), C is the C-parameter and E is the exposure energy. From the data shown in Fig. 7, from exposure energies of 0 to 20 mJ cm−2, C is determined to be approximately 0.06 cm2 mJ−1, which is consistent with previous literature values.20
Fig. 8 Reaction profiles: PMMA/TPS film exposed to 266 nm pulsed laser irradiation. The concentration has been normalized versus TPS, which is set to 1 at an exposure dose of 0 mJ cm−2. |
The consumption of TPS plateaued after exposure to 658 mJ cm−2 (300 shots) at 83% conversion.21 The major photoproduct at all doses was the ortho-isomer, 2-(phenylthio)biphenyl (19.6%), as expected given the diffusion constraints in the polymer matrix. A small amount of the para-isomer 4-(phenylthio)biphenyl (1.8%) was also formed, but very little of the meta-isomer (0.6%) was observed. These are the primary products of TPS photochemistry, arising from the cleavage of the S-aryl bond followed by in-cage fragment recombination that strongly favors the formation of the ortho-isomer over the para-isomer, both being electronically preferred compared to the meta-isomer (Scheme 3).
As expected, the formation of the cage-escape photoproduct, diphenylsulfide (maximum of 7.1%), was less favored compared to in-cage products. Diphenylsulfide was produced by loss of a phenyl radical (homolysis pathway) or a phenyl cation (heterolysis pathway, Scheme 1). Biphenyl can be formed by the recombination of two phenyl radicals or more likely by secondary photolysis of one or more (phenylthio)biphenyl isomers.3 A small induction period was observed in the formation of biphenyl consistent with secondary photolysis. The reaction with the polymer matrix is possible as only 42% of the TPS consumed led to molecular photoproducts. Longer exposure of the film allowed the formation of dibenzothiophene (1%) and triphenylene (3%). The long induction periods before the formation of these products (especially for dibenzothiophene) suggest that secondary photolysis was in play (Scheme 3).
The formation of the two main photoproducts, 2-(phenylthio)biphenyl and diphenyl sulfide, reached plateaus after 658 mJ cm−2 exposure, in accord with the TPS consumption curve (Fig. 8); after that, both products were slowly consumed. Indeed, upon the exposure of PMMA/diphenylsulfide and PMMA/2-BiPhSPh films to 266 nm pulsed laser irradiation, dibenzothiophene and triphenylene were obtained as photolysis products, respectively. No intermediates were observed during the irradiation of a PMMA-2-BiPhSPh film, only primarily triphenylene, in accord with an in-cage mechanism. Furthermore, the irradiation of PMMA films containing 3- or 4-(phenylthio)biphenyl did not produce any triphenylene. We conclude that the presence of an ortho phenyl substituent is essential for the formation of triphenylene.
Most importantly, our findings highlight the role of the polymer in the microlithography process. Before our work in films, neither triphenylene nor dibenzothiophene had been observed in TPS photodecomposition, the former photoproduct resulting from the complete loss of sulfur.3,5,18 Moreover, although these photoproducts are generated via successive photolysis processes, our work demonstrates that this sequence can occur with low exposure under lithographically relevant conditions.
Footnote |
† Electronic supplementary information (ESI) available: (i) Emission spectra of the different irradiation sources, (ii) UV-Vis spectra over time under 193 and 254 nm irradiation, (iii) UV-Vis spectra of TPS films with different polymer matrixes, (iv) UV-Vis spectra and HPLC trace of the solution and wafer experiments, (v) NMR spectra of the cross-experiment, and (vi) DOSY-NMR experiment. See DOI: 10.1039/c7pp00324b |
This journal is © The Royal Society of Chemistry and Owner Societies 2018 |