Analysis of Thin-Film Polymers Using Attenuated Total Internal Reflection–Raman Microspectroscopy Willie Tran,* Louis G. Tisinger, Luis E. Lavalle, Andre´ J. Sommer Miami University, Molecular Microspectroscopy Lab, Department of Chemistry and Biochemistry, Oxford, OH 45056 USA
Two methods commonly employed for molecular surface analysis and thin-film analysis of microscopic areas are attenuated total reflection infrared (ATR-IR) microspectroscopy and confocal Raman microspectroscopy. In the former method, the depth of the evanescent probe beam can be controlled by the wavelength of light, the angle of incidence, or the refractive index of the internal reflection element. Because the penetration depth is proportional to the wavelength of light, one could interrogate a smaller film thickness by moving from the mid-infrared region to the visible region employing Raman spectroscopy. The investigation of ATR Raman microspectroscopy, a largely unexplored technique available to Raman microspectroscopy, was carried out. A Renishaw inVia Raman microscope was externally modified and used in conjunction with a solid immersion lens (SIL) to perform ATR Raman experiments. Thin-film polymer samples were analyzed to explore the theoretical sampling depth for experiments conducted without the SIL, with the SIL, and with the SIL using evanescent excitation. The feasibility of micro-ATR Raman was examined by collecting ATR spectra from films whose thickness measured from 200 to 60 nm. Films of these thicknesses were present on a much thicker substrate, and features from the underlying substrate did not become visible until the thin film reached a thickness of 68 nm. Index Headings: Raman; Attenuated Total Reflection; ATR; Total internal reflection; TIR; Solid immersion lens; SIL; Microspectroscopy; Micrometer thin films.
INTRODUCTION Thin-film technology plays an important role in many products from precision, laser, and consumer optics to electronics, semiconductors, biofilms, tribology, and medical technology.1–5 Optical components ranging from spectrometer grade lenses to eyeglasses for corrective vision employ thin-film coatings, such as antireflection coatings.6 The medical industry employs thin-film membrane technology for filtration, dialysis, and electrolysis, among a host of many other applications.2 These films can vary in thickness anywhere from several micrometers to nanometers (single monolayers).6–8 The molecular characterization of these films is important to understanding their functional performance. However, thin transparent films supported on much thicker substrates present a challenge in that many of the current techniques used to characterize these films probe deeper into the surface than the thickness of the film. Because of this, spectra obtained using these methods are dominated with spectral features from the bulk of the substrate. In addition, the analysis of Received 29 January 2013; accepted 10 July 2014. * Author to whom correspondence should be sent. E-mail: tranw@ miamiOH.edu. DOI: 10.1366/13-07024
230
Volume 69, Number 2, 2015
microscopic defects in these films is important to elucidate failure and contamination mechanisms during their production and use. Two methods commonly employed for molecular surface analysis and thin-film analysis of microscopic areas are attenuated total reflection infrared (ATR-IR) microspectroscopy, and confocal Raman microspectroscopy.9–11 Both methods have gained popularity over the last 17 years; however, both have drawbacks when films whose thickness is less than 1 lm are studied. In a standard IR accessory using a germanium hemisphere, the limiting penetration depth at 700 wavenumbers is approximately 1 lm. For films that are thinner than 1 lm, spectral contamination from subsurface layers is expected. Further, spectral contamination becomes more problematic if the sub-layer has a stronger IR absorption than the surface film. The advent of confocal microscopy improved the depth resolution for Raman microspectroscopy by employing a pinhole aperture at the back image plane of the microscope, attenuating the light from the out-of-focus regions of the sample.12–15 The ability of researchers to sample along the z-axis of the microscope allowed studies on multilayered samples such as layered polymers.16 However, this approach has several limitations, brought to light by Everall’s investigations into the depth resolution of a commercial confocal Raman microscope.17,18 In these experiments, the depth of focus was found to be approximately three times greater than that predicted by theory. This difference was caused by refraction at the air–sample interface, which can be corrected using an immersion objective with a suitable index-matching fluid.18 However, the depth resolution using this approach is 1.25 lm, assuming a numerical aperture (NA) of 1.4 and an excitation wavelength of 632 nm. In addition, the confocal Raman microscope still collects light outside the confocal region due to its farfield optical configuration. A solution to these problems is to conduct Raman microspectroscopy using evanescent excitation. This approach has been referred to as ATR Raman or total internal reflection (TIR) Raman spectroscopy. Compared to IR wavelengths, Raman excitation wavelengths are an order of magnitude shorter, thereby yielding an order of magnitude smaller penetration depth. Second, Milster pointed out several years ago the benefits of using nearfield optics (evanescent energy) to increase the read/ write density in optical data storage systems.19 The same concept can be used for Raman spectroscopy where excitation is done evanescently and collection is done far field. Figure 1 demonstrates the concept of near-field excitation with far-field collection and the
0003-7028/15/6902-0230/0 Q 2015 Society for Applied Spectroscopy
APPLIED SPECTROSCOPY
FIG. 1. (left) Far-field illumination and collection by a confocal microscope. (right) Near-field illumination with far-field collection. Oversampling beyond the sample layer is avoided when illuminating in the near field. (Reproduced with permission from Ref. 20. Copyright McGraw-Hill 2010.)
corresponding sampling volume.20 Milster19 also provided a detailed review on the use of high-refractive-index (n) hemispheres (solid immersion lenses (SILs)) to focus and collect visible light in the optical recording industry, thereby increasing the read–write density of the devices. This method yielded an improvement in the xy resolution by n, while the axial resolution and collection efficiency were improved by n2. Poweleit et al. demonstrated the xy improvement for Raman imaging using a La SFN-9 glass SIL (nc = 1.868) with 488 nm excitation on patterned silicon.21 Sohn et. al. reported on using a zinc selenide (ZnSe) hemisphere to construct a microscope that could conduct both ATR IR microspectroscopy and Raman SIL microspectroscopy at the same time.22 Additional studies have shown near-field spectroscopy benefits Raman spectroscopy over far-field spectroscopy by providing surface enhancement, a reduced Rayleigh tail, and a zpolarization effect that can be used to obtain different information.23,24 The first report on ATR Raman was published by Ikeshoji et al. in 1973 and briefly explored the theoretical and experimental intensities of ATR Raman scattering.25 This study was followed by several reports demonstrating ATR Raman using high excitation powers at the sample with strong Raman-scattering samples such as carbon tetrachloride.26–28 Later, Tisinger and Sommer adapted Milster’s sampling method to acquire ATR Raman spectra of polydiacetylene (PDA) directly applied to the bottom of a ZnSe hemisphere.29,30 To our knowledge, this report was the first attempt at performing ATR Raman using a conventional Raman microscope rather than the macro-ATR Raman spectrometers used in previous studies. Tisinger and coworkers were able to
demonstrate the feasibility of conducting ATR Raman measurements using off-axis excitation and on-axis collection through a microscope. They also demonstrated that, just like ATR Fourier transform (FT)-IR measurements, ATR Raman measurements require intimate contact between the internal reflection element (IRE) and sample to obtain quality spectra.29–31 In 2004, Greene and Bain investigated a 2.5 lm thick polyethylene napthalate film on a 200 lm thick polyethylene terephthalate (PET) substrate using 532 nm excitation and 200 mW power output using a cubic zirconia IRE.32 In 2008, Ishizaki and Kim analyzed a 135 nm polystyrene film using ATR Raman.33 They employed a sapphire SIL as the IRE and 752 nm laser excitation with a 56 mW power output. In 2010, McKee and Smith developed a scanning angle TIR Raman spectrometer to investigate films cast onto a ZnSe IRE. The instrument could vary the angle of incidence to vary the sampling depth of the films.34 Recently, Michaels investigated poly(3,4-ethylenedioxythiophene)-poly(styrene sulfonate) (PEDOT:PSS) film on PET using a 785 nm excitation. Large intensity counts for the PEDOT:PSS film were observed due to PEDOT:PSS having a Raman scattering (the cross section at 440 cm1 is 40 times greater than the cross section of PET at 1726 cm1), making it a strong Raman scatterer.35 A comprehensive critical review of ATR Raman spectroscopy was published by Woods and Bain in 2012, reporting on work with solid–solid, solid– liquid, solid–air, liquid–liquid, and liquid–air interfaces, as well as depth profiling and scattering angle studies.36 A review of the literature shows that previous investigations employed large laser powers (50–200 mW)25–28,32,33 as well as strong Raman-scattering samples such as
APPLIED SPECTROSCOPY
231
FIG. 2. (a) The ATR sampling cell. (b) Side view of the ball bearing providing point contact.
PDA29 and PEDOT:PSS.26 Previous methods also applied samples directly onto the IREs to improve the signal-tonoise ratio (SNR) and experimental success rate. However, this practice is not conducive to routine sample analysis;29,30,33,34 this sample alteration is undesirable because it inhibits a fundamental advantage of ATR and Raman spectroscopy—the ability to analyze a sample in its natural state. With the exception of Ishizaki and Kim, the previous work mentioned employed Raman microscopes. The purpose of the present study is to investigate the capabilities and limitations of ATR Raman microspectroscopy conducted using a commercial Raman microscope. The samples investigated were polystyrene films supported on a thick polycarbonate substrate using laser powers of less than 6 mW. This work was first presented at the 2007 meeting of the Federation of Analytical Chemistry and Spectroscopic Societies in Memphis, Tennessee, and later presented at the 2009 meeting of the Pittsburgh Conference in Chicago.37,38
EXPERIMENTAL Materials. Certified American Chemical Society (ACS)-grade toluene was purchased (Fisher Chemical, Fairlawn, NJ) and polystyrene beads were obtained (Sigma Aldrich Company, St. Louis, MO). Recordable compact disks (CD-Rs; Memorex) were used as the polycarbonate substrate on which the polystyrene was applied. The CD-R was cut into 2.54 3 2.54 cm substrates, and the aluminum reflective layer was removed. The film was cast on the side opposite where the reflective layer resided. Sample Preparation. The polystyrene (n = 1.58) was dissolved in toluene by a ratio of 4% weight of polymer to solvent. An EC101DT Digital Photo Resist Spinner (Headway Research, Inc.) was used to spin coat the polystyrene solution onto the polycarbonate substrate. The polycarbonate was mounted, and when the spin coater reached the desired revolutions per minute, 500 lL of the polystyrene-toluene solution was micropipetted onto the polycarbonate substrate. Various film thicknesses (60–200 nm) were obtained by varying the
232
Volume 69, Number 2, 2015
revolutions per minute from 2000 to 4400 on the spin coater. Film thicknesses were confirmed using both reflection-absorption IR spectroscopy and spectroscopic ellipsometry. Thickness confirmation samples were prepared simultaneously with the polystyrene-polycarbonate sample by spin coating polystyrene onto a cut 2.54 3 2.54 cm microscope slide (MirrIR Low-e, Chesterland, OH). Reflection–absorption IR spectra of the samples were collected using a Perkin-Elmer Spotlight 300 IR microscope interfaced to a Perkin-Elmer Spectrum One Fourier Transform spectrometer (Perkin-Elmer). The system employed a 100 3 100 lm liquid-nitrogen-cooled mercury cadmium telluride (HgCdTe) detector. Each spectrum collected represents the average of 32 individual scans possessing a spectral resolution of 4 cm1. The peak height of the polystyrene C–H stretch at 1493 cm1 was used to determine film thickness from a calibration curve.30 A PhE-101 Discrete k Ellipsometer (Micro Photonics Inc., Allentown, PA) was used to gather the spectroscopic ellipsometry data of the film thickness directly from the polystyrene-polycarbonate sample. The ellipsometer was calibrated with silicon dioxide–coated silicon wafer standards (J. A. Woolam Co., Inc., Lincoln, NE). Attenuated Total Internal Reflection Accessory. A ZnSe hemisphere measuring 1.5 mm in height and 3 mm in diameter (Spectral Systems, Hopewell Junction, NY) was employed as the IRE for this study. A custom-made cell was fabricated (Miami University Instrumentation Lab, Ohio) to hold the ZnSe IRE in place for sample contact. The cell, which can be seen in Fig. 2a, consisted of a 7.62 long (l) 3 0.54 wide (w) 3 0.3175 high (h) cm aluminum bottom plate, on which the sample was mounted. This plate consisted of two pegs on which the top plate would register as well as three threaded holes to secure the top plate. The top plate was a 3.81 (l) 3 0.54 (w) 3 0.3175 (h) cm aluminum plate fashioned with an 4 mm diameter hole in which the ZnSe hemisphere would sit. The surrounding area of the hole was beveled inward to 658, and a groove was provided to permit side laser illumination.
FIG. 3. Block diagram of the ATR Raman setup.
The top plate contained through holes for screws that were tightened to initiate contact between the sample and ZnSe hemisphere. A side perspective of the cell (Fig. 2b) demonstrates how point contact was achieved between the IRE and the sample. A ball bearing was placed in the hole in the bottom plate. As the top plate was tightened, the ball bearing provided point contact between the sample and hemisphere, ensuring intimate contact. Instrument Setup. Raman spectra were collected using an inVia Raman microscope (Renishaw). A 633 nm helium–neon (HeNe) laser was used as the excitation source; the beam path is shown in Fig. 3. The laser was conditioned using a Pellin–Broca prism, a polarizer, beam expander, and plasma line filter. Prior to entering the IRE at an angle of 518, the light was focused using a 35 mm focal length plano-convex lens (Edmund Optics, Barrington, NJ). The combination of the lens and the IRE produced an elliptical spot on the sample approximately 26 lm high and 36 lm long. The Ramanscattered radiation was collected using an infinitycorrected Olympus ULWD MSPlan 203 objective (0.40 N.A.; f = 180 mm; Olympus). A photograph of the ATR accessory on the microscope and excitation optic is shown in Fig. 4. The collected ATR Raman spectra represent 40 accumulations using a 30 s integration time (total scan time of 1 h) at 4 cm1 resolution. The incident power of the laser prior to entering the IRE was 5.2 mW. The Raman spectra were then analyzed using GRAMS/AI software (Thermo Scientific). Comparison of the spectra collected with and without the SIL was done using peak heights and peak areas for the strongest transitions in the spectra. The data for both were consistent with one another.
from 0.6 to 15 lm), high refractive index of 2.59 at 633 nm,39 and isotropic crystal structure. Zinc sulfide (ZnS) is an alternative material; it is harder than ZnSe but has a somewhat lower refractive index of 2.35 at 633 nm. Most important, ZnSe has no major absorptions above the 500 wavenumber shift. A comparison of the Raman spectra of ZnS and ZnSe is shown in Fig. 5. Comparison of Raman Spectra Collected with and without a Solid Immersion Lens. The hemisphere shape of the SIL minimizes Fresnel reflections because both the incident and exiting rays are normal to the hemisphere surface. In addition, the SIL yields an immersion effect that improves the depth of field and the focused beam diameter. The depth of field is given by: z¼
4k ðnSIL sinhÞ2
ð1Þ
RESULTS AND DISCUSSION Zinc selenide was chosen as the IRE due to its relatively high transmission in the visible range (. 65%
FIG. 4. Photograph of the ATR accessory on the microscope stage and excitation optic.
APPLIED SPECTROSCOPY
233
FIG. 5. Raman spectra of ZnS and ZnSe SILs at 633 nm excitation. The ZnSe peaks are limited to 600 cm1 and below, whereas ZnS has peaks interdispersed between 100 and 800 cm1.
where k is the excitation source wavelength, nSIL is the refractive index of the SIL, and h is the half-angle acceptance of the objective. Under the present conditions, the theoretical depth of field employing a ZnSe SIL at 633 nm and 0.40 N.A. objective is 2.36 lm, whereas without the hemisphere, it is 15.83 lm. The focused beam diameter, d, in the xy plane is given by: d¼
1:22k nSIL sinh
ð2Þ
This value is reduced from 1.93 to 0.75 lm when the SIL is employed. Immersion also increases the amount of Raman signal collected because of the improvement in the lightcollecting efficiency F, which is given by:40 2 F nSIL sin2 h
ð3Þ
Relative to a measurement conducted without the SIL, the amount of light collected under the present conditions is 6.71 times greater using the SIL. Proof of these concepts is shown in Fig. 6 and Table I. Figure 6 illustrates the Raman spectra of single-crystal silicon (Fig. 6a) and polycarbonate (Fig. 6b) that were collected without the SIL and by placing the SIL in contact with the surface of the samples. Qualitatively, the signal decreases for polycarbonate and increases for single-crystal silicon when the SIL is used. The Raman, SR, signal collected from a given sample is given by:41 SR Io NV rF
ð4Þ
where Io is the laser irradiance (in W/cm2), N is the number of particles per unit volume, V is the volume of uniformly illuminated sample, r is the Raman scatter-
234
Volume 69, Number 2, 2015
ing cross section, and F relates to the collection efficiency. For the same condensed-phase sample, N and r are constant. Table I lists values for Io, V, and F for each sample. In the case of polycarbonate, the improvement in laser irradiance (power density) and collected flux is exactly offset by the decrease in illumination volume when the SIL is employed. The remaining difference between the experimental and theoretical values can be accounted for by taking into account the Fresnel reflection losses as (1) the laser enters the hemisphere and traverses the IRE-sample interface and (2) the Raman-scattered light traverses the IRE-sample interface and exits the hemisphere. The agreement between the theoretical value and the experimentally obtained value indicates that the considerations involved in Eqs. 1–4 provide a good approximation for the events taking place. The only provisions are that the sample be isotropic, nonabsorbing, and optically transparent. Silicon, on the other hand, is different. Although the depth of field in air is 15.83 lm, the penetration depth allowed by silicon is only 2.82 lm.42,43 Hence, the disparity between the illumination volume in air and that using the SIL is not as great as it is for polycarbonate. This lessened disparity, along with the increased laser irradiance and collected flux, favors the Raman signal when the SIL is employed. The agreement between the theoretical and experimental values again validates the previous considerations. In the case where the sample characteristics are limiting the illumination volume, as in a very thin, optically transparent film or a sample with a shallow optical skin depth, the gain in irradiance and collected flux significantly improves the Raman signal when the SIL is used. Such is the case for single-crystal silicon,
TABLE I. Confocal parameters with and without the use of a SIL for polycarbonate and silicon. Polycarbonate Parameter
Air
SIL SIL/air
d (lm) 1.93 0.75 Z (lm) 15.83 2.36 46.33 1.03 Volume (lm3) 0.34 2.29 Power density (mW/lm2) Flux collected, F 0.16 1.07 Combined volume, power density, and flux Reflection losses (%) Theoretical considerations Experimental
FIG. 6. Raman spectra collected without the SIL and with the SIL. (a) Silicon. (b) Polycarbonate. The Raman spectrum of silicon collected using the SIL has the higher intensity and that of polycarbonate has the lower intensity.
whose optical skin depth is approximately 2.82 lm at 633 nm. Figure 7 illustrates Raman spectra of an 80 nm thick polystyrene film on polycarbonate collected both in air and using the hemisphere. The spectrum collected in air does not exhibit the symmetric stretch of the styrene ring at the 1001 wavenumber shift, but it does show a feature at the 1006 wavenumber shift for polycarbonate. The spectrum collected using the SIL clearly shows the ring stretch at the 1001 wavenumber shift for styrene. The result is interesting in that the thickness of the styrene film constitutes only 3.5% of the depth of field given in Table I. The increase in the styrene transition intensity is not warranted based on the depth of field and styrene film thickness. However, those values do not take into account the molecular orientation within the film or the effects of refraction. It is well known that polystyrene films of this thickness have the styrene rings oriented parallel to the surface normal.44 Without the SIL, marginal rays enter the sample at an angle of 158. As such, the electric field vector is oriented nearly perpendicular to the styrene rings. With the SIL, marginal rays enter the sample at an angle of 418, with the result that the electric field vector interacts more with the styrene rings. Finally,
0.02 6.71 6.71
Silicon Air 1.93 2.82 8.26 0.34 0.16
Silicon SIL/air 0.75 2.36 1.03 2.29 1.07
1.00 43 0.57 0.56
0.12 6.71 6.71 5.62 43 3.20 3.28
Koyama et al. showed that the collection efficiency exceeds that predicted by theory when a SIL is used to collect fluorescence from dye-doped polystyrene spheres of 200 nm diameter.45 Use of a Solid Immersion Lens with Evanescent Excitation. Up to this point, most of the discussion has focused on the use of a SIL and far-field collection. As such, strong transitions from the subsurface polycarbonate are still observed. By separating excitation to near field (evanescent excitation), one can investigate much thinner films, thereby eliminating the parasitic signal from the subsurface layers. Figure 1 illustrates the differences between the two situations. The figure on the left-hand side shows far-field illumination and collection, with the resulting sample illumination volume being a cylinder with height z and diameter d. For films whose thickness is less than z, subsurface contamination is expected. The right-hand side of Fig. 1 illustrates near-field (evanescent) excitation with farfield collection. In this case, if the film thickness is less than z, no subsurface contamination is expected until the film thickness approaches the penetration depth (the depth to which the evanescent excitation penetrates), dp, which is given by the well-known Harrick equation:46 dp ¼
k
2pnSIL sin2 h n 2s=SIL
1=2
ð5Þ
where n2s/SIL is the refractive index of the IRE and n2s/SIL is the ratio of the refractive index of the sample to the SIL. The theoretical sampling depth of ATR Raman microspectroscopy using the same parameters above is 80 nm and represents a 303 improvement in sampling depth compared to far-field illumination and collection using the SIL. Two objectives of this research were to verify the penetration depth and to determine how long it would take to collect ATR spectra from a moderate Ramanscattering sample using a conventional Raman microprobe. The chosen sample system was thin films (thickness in nanometers) of polystyrene cast onto thick substrates of polycarbonate. Reference spectra of the bulk materials obtained under identical conditions demonstrated that the Raman transition for polycarbon-
APPLIED SPECTROSCOPY
235
FIG. 7. Raman spectra of an 80 nm thick polystyrene film on a polycarbonate substrate (PS) collected without the SIL (bottom spectrum) with the use of a zinc selenide SIL (top spectrum).
FIG. 8. Comparison of Raman spectra collected using the SIL and evanescent excitation on 85, 68, and 60 nm thick polystyrene films on a much thicker polycarbonate substrate. * marks the 1001 cm1 polystyrene reference peak; à marks the polycarbonate peaks found at 1110 and 890 cm1. Also shown are reference spectra for polystyrene (top spectrum) and polycarbonate (bottom spectrum).
236
Volume 69, Number 2, 2015
FIG. 9. The non-normalized polystyrene spectra at 85, 68, and 60 nm thicknesses.
ate located at the 890 cm1 shift is five times stronger than the symmetric stretch of the aromatic ring located at the 1001 cm1 shift in the Raman spectrum of polystyrene. Initial attempts to collect ATR spectra proved that intimate contact was an issue. This problem was overcome by inserting a small ball bearing into the sample holder, as shown in Fig. 2b. The ball bearing provided point contact exactly in the position where it was needed. Various films of polystyrene on polycarbonate, ranging in thickness from 60 to 170 nm were collected in 35 min. The spectra of the thicker films showed good SNR, and those that are thinner exhibited a SNR of approximately 10. Figure 8 shows the spectra of the thinner films along with those of bulk polystyrene and polycarbonate. The features associated with polycarbonate are not observed until the polystyrene film thickness is less than 74 nm. At thicknesses less than 74 nm (not shown), the transitions associated with polycarbonate are observed at the 1600 and 880 cm1 shifts. Figure 9 illustrates the non-normalized polystyrene spectra featured in Fig. 8. Two critical factors we have observed in the success of the measurements are intimate contact with the IRE and the quality of the IRE surface. TABLE II. Center and peak area of polystyrene peak of interest. Sample
Center
Areaa
Polystyrene1 Polystyrene2 Polystyrene3 Polystyrene4 Polystyrene5
1002.0 1002.3 1002.0 1002.0 1002.1
36 38 37 37 37
673 218 938 877 833
Heighta 4231 4257 4461 4219 4831
a Peak height and peak area values collected from the polystyrene peak of interest at 1002 cm1 using GRAMS/AI.
To put the measurements in perspective, the amount of sample giving rise to a spectrum is on the order of 0.82 fL, which translates to 0.82 fg on the basis of the density of polystyrene. These values were derived from an optical analysis of the microprobe and a film thickness equal to the 80 nm penetration depth. In addition, the spectra were generated using an incident power of 7 lW/lm2.
REPRODUCIBILITY Table II contains the values of five Raman spectra of the polystyrene–polycarbonate sample obtained in succession, without adjustments to the setup, to test the reproducibility of ATR Raman microspectroscopy. The peak height and peak area values were collected from the polystyrene peak of interest at 1002 cm1 using GRAMS/AI. As shown, there are no significant deviations between peak intensities and peak areas among the five analyses. Relative standard deviations are 1.58% for the peak area and 6% for the peak height.
CONCLUSION Previous research on ATR Raman employed strong Raman scatterers, large power outputs, and long collection times. The purpose of the present study was to investigate the capabilities and limitations of ATR Raman microspectroscopy conducted using a commercial Raman microscope. The results of the study verify the theoretical penetration depth of 80 nm and show that spectra with reasonable SNR can be collected in approximately 0.5 h. More important, no modifications to the Raman microscope system need to be made because the incident illumination is external to the instrument. Two critical factors in the success of the
APPLIED SPECTROSCOPY
237
measurements are intimate contact with the IRE and the quality of the IRE surface. ACKNOWLEDGMENTS The authors gratefully acknowledge financial support from Kodak, Procter and Gamble, Perkin Elmer, and Renishaw. We also acknowledge the Miami University Instrumentation Laboratory for the construction of the ATR accessory. 1. D. Ristau, H. Ehlers. ‘‘Thin Film Optical Coatings’’. In: F. Tra¨ger, editor. Springer Handbook of Lasers and Optics. New York: Springer ScienceþBusiness Media, LLC, 2007. Pp. 373-396. 2. G. Ozaydin-Ince, A.M. Coclite, K.K. Gleason. ‘‘CVD of Polymeric Thin Films: Applications in Sensors, Biotechnology, Microelectronics/Organic Electronics, Microfluidics, MEMS, Composites and Membranes’’. Rep. Prog. Phys. 2012. 75: 1-40. 3. N. Setter, D. Damjanovic, L. Eng, G. Fox, S. Gevorgian, S. Hong, A. Kingon, H. Kohlstedt, N.Y. Park, G.B. Stephenson, I. Stolitchnov, A.K. Taganstev, D.V. Taylor, T. Yamada, S. Streiffer. ‘‘Ferroelectric Thin Films: Review of Materials, Properties, and Applications’’. J. Appl. Phys. 2006. 100(5): 051606. doi: http://dx.doi.org/10.1063/1. 2336999. 4. Y. Fu, H. Du, W. Huang, S. Zhang, M. Hu. ‘‘TiNi-Based Thin Films in MEMS Applications: A Review’’. Sensor. Actuator. A. 2004. 112: 395408. 5. S.K. Arya, S. Saha, J.E. Ramirez-Vick, V. Gupta, B. Shekhar. S.P. Singh ‘‘Recent Advances in ZnO Nanostructures and Thin Films for Biosensor Applications: Review’’. Anal. Chim. Acta. 2012. 737: 1-21. 6. U. Schulz, U.B. Schallenberg, N. Kaiser. ‘‘Antireflection Coating Design for Plastic Optics’’. Appl. Optics. 2002. 41(16): 3107-3110. 7. Specialty Coating Systems. ‘‘SCS Equipment’’. 2014. http:// scscoatings.com/equipment/index.aspx [accessed Aug 1 2012]. 8. MEMSnet. ‘‘MEMS Thin Film Deposition Processes’’. http:// memsnet.org/mems/processes/deposition.html [accessed Aug 12012]. 9. J.E. Katon. ‘‘Infrared Microspectroscopy’’. In: F.M. Mirabella, editor. Modern Techniques in Applied Molecular Spectroscopy. New York: Wiley-Interscience, 1998. Pp. 267-290. 10. A.J. Sommer. ‘‘Raman Microspectroscopy’’. In: F.M. Mirabella, editor. Modern Techniques in Applied Molecular Spectroscopy. New York: Wiley-Interscience, 1998. Pp. 291-322. 11. P.M. Fredericks. ‘‘Depth Profiling of Polymers by Vibrational Spectroscopy’’. In: N.J. Everall, J.M. Chalmers, P.R. Griffths, editors. Vibrational Spectroscopy of Polymers: Principles and Practice. Hoboken, NJ: John Wiley and Sons, 2007. Pp. 179-200. 12. N.J. Everall. ‘‘Modeling and Measuring the Effect of Refraction on the Depth Resolution of Confocal Raman Microscopy’’. Appl. Spectrosc. 2000. 54(6): 773-782. 13. D.J. Gardiner, M. Bowden, P.R. Graves. ‘‘Novel Applications of Raman Microscopy [and Discussion]’’. Philos. Tr. R. Soc. A. 1986. 320: 295-306. 14. K.P.J. Williams, G.D. Pitt, D.N. Batchelder, B.J. Kip. ‘‘Confocal Raman Microspectroscopy Using a Stigmatic Spectrograph and CCD Detector’’. Appl. Spectrosc. 1994. 48(2): 232-235. 15. R. Tabaksblat, R.J. Meier, B.J. Kip. ‘‘Confocal Raman Microspectroscopy: Theory and Application to Thin Polymer Samples’’. Appl. Spectrosc. 1992. 46(1): 60-68. 16. G.J. Puppels, W. Colier, J.H.F. Olminkhof, C. Otto, F.F.M. de Mul, J. Greve. ‘‘Description and Performance of a Highly Sensitive Confocal Raman Microspectrometer’’. J. Raman Spectrosc. 1991. 22: 217-225. 17. N.J. Everall ‘‘Confocal Raman Microscopy: Why the Depth Resolution and Spatial Accuracy Can Be Much Worse Than You Think’’. Appl. Spectrosc. 2000. 54(10): 1515-1520. 18. N.J. Everall. ‘‘Confocal Raman Microscopy: Performance, Pitfalls, and Best Practice’’. Appl. Spectrosc. 2009. 63(9): 245A-262A. 19. T.D. Milster. ‘‘Near-Field Optics: A New Tool for Data Storage’’. Proc. IEEE. 2000. 88(9): 1480-1490. 20. H.J. Gulley-Stahl, A.J. Sommer, A.P. Evan. ‘‘Evanescent Wave Imaging’’. In: G. Srinivasan, editor. Vibrational Spectroscopic Imaging for Biomedical Applications. New York: McGraw-Hill, 2010. P. 117, Fig. 4.10. 21. C.D. Poweleit, A. Gunther, S. Goodnick, J. Menendez. ‘‘Raman Imaging of Patterned Silicon Using a Solid Immersion Lens’’. Appl. Phys. Lett. 1998. 73(16): 2275-2277.
238
Volume 69, Number 2, 2015
22. A.F. Sohn, L.G. Tisinger, A.J. Sommer. ‘‘Combined ATR Infrared Microspectroscopy and SIL Raman Microspectroscopy’’. Microsc. Microanal. 2004. 10(Suppl 2): 1316-1317. doi:10.1017/ S143927604884757. 23. C.L. Jahncke, H.D. Hallen, M.A. Paesler. ‘‘Nano-Raman Spectroscopy and Imaging with the Near-field Scanning Optical Microscope’’. J. Raman Spectrosc. 1996. 27: 579-586. 24. H.D. Hallen, M.A. Paesler, C.L. Jahncke. ‘‘Raman Spectroscopy: Probing the Border Between Near-Field and Far-Field Spectroscopy’’. In: S. Jutamulia, T. Asakura, editors. Far- and Near-Field Optics: Physics and Information Processing. Proc. SPIE. 1998. 3467. doi:10.1117/12.326819. 25. T. Ikeshoji, Y. Ono, T. Mizuno. ‘‘Total Reflection Raman Spectra; Raman Scattering Due to the Evanescent Wave in Total Reflection’’. Appl. Optics. 1973. 12(10): 2236-2237. 26. T. Takenaka, T. Nakanaga. ‘‘Resonance Raman Spectra of Monolayers Adsorbed at the Interface Between Carbon Tetrachloride and an Aqueous Solution of a Surfactant and a Dye’’. J. Phys. Chem. 1976. 80(5): 475-480. 27. M. Futamata, P. Borthen, J. Thomassen, D. Schumacher, A. Otto. ‘‘Application of an ATR Method in Raman Spectroscopy’’. Appl. Spectrosc. 1994. 48(2): 252-260. 28. M. Futamata. ‘‘Highly-Sensitive ATR Raman Spectroscopy Using Surface-Plasmon-Polariton’’. Internet J. Vib. Spectrosc. 2000. 4(2): 9-23. 29. L.G. Tisinger, A.J. Sommer. ‘‘Attenuated Total Internal Reflection (ATR) Raman Microspectroscopy’’. Microsc. Microanal. 2004. 10(Suppl 2): 1318-1319. doi:10.1017/S1431927604884794. 30. L.G. Tisinger. Investigations in Quantitative Infrared Using Attenuated Total Reflectance. [Ph.D. Dissertation]. Oxford, OH: Miami University, 2002. 31. L. Lewis, A.J. Sommer. ‘‘Attenuated Total Internal Reflection Microspectroscopy of Isolated Particles: An Alternative Approach to Current Methods’’. Appl. Spectrosc. 1999. 53(4): 375-380. 32. P.R. Greene, C.D. Bain. ‘‘Total Internal Reflection Raman Spectroscopy’’. Spectrosc. Europe. 2004. 16: 8-15. 33. F. Ishizaki, M. Kim. ‘‘Near-Infrared Attenuated Total Reflection Raman Spectroscopy for Polymer Surface Observation’’. Jpn. J. Appl. Phys. 2008. 47(3): 1621-1627. doi:10.1143/JJAP.47.1621. 34. K.J. McKee, E.A. Smith. ‘‘Development of a Scanning Angle Total Internal Reflection Raman Spectrometer’’. Rev. Sci. Instrum. 2010. 81(4): 043106. doi:10.1063/1.3378682. 35. C.A. Michaels. ‘‘Surface-Sensitive Raman Microscopy with Total Internal Reflection Illumination’’. J. Raman Spectrosc. 2010. 41: 1670-1677. doi:10.1002/jrs.2610. 36. D.A. Woods, C.D. Bain. ‘‘Total Internal Reflection Raman Spectroscopy’’. Analyst. 2012. 137: 35-48. doi:10.1039/c1an15722a. 37. A.J. Sommer. ‘‘Attenuated Total Internal Reflection Raman Microspectroscopy’’. Paper presented at: FACSS 2007. Memphis, TN; Oct 14-18 2007. Paper 613. 38. W. Tran, A.J. Sommer. ‘‘Analysis of Thin Film Polymers Using ATRRaman Microspectroscopy’’. Paper presented at: Pittsburgh Conference and Exposition on Analytical Chemistry and Applied Spectroscopy. Chicago, IL; March 2009. Abstract 2610-7. 39. Crystran Ltd. ‘‘Zinc Selenide (ZnSe)’’. 2011. http://www.crystran.co. uk/uploads/files/181.pdf [accessed Dec 7 2011]. 40. M. Pluta. Advanced Light Microscopy. Vol. 1. New York: Elsevier Science, 1988. 41. G.J. Rosasco. ‘‘Raman Microprobe Spectroscopy’’. In: R.J.H. Clark, R.E. Hester editors. Advances in Infrared and Raman Spectroscopy. London: Heyden and Son, 1980. Vol. 7, Chap. 4, p. 226. 42. I.D. Wolf. ‘‘Micro-Raman Spectroscopy to Study Local Mechanical Stress in Silicon Integrated Circuits’’. Semicond. Sci. Tech. 1996. 11: 139-154. 43. I.D. Wolf, J. Jimenez, J.P. Landesman, C. Frigeri, P. Braun, E.D. Silva, E. Calvet. Raman and Luminescence Spectroscopy for Microelectronics: Catalogue of Optical and Physical Parameters. ‘‘Nostradamus’’ Project SMT4-CT-95-2024. Luxembourg: European Commission, 1998. 44. D.A. Fischer, G.E. Mitchell, A.T. Yeh, G.L. Gland. ‘‘Functional Group Orientation in Surface and Bulk Polystyrene Studied by Ultra Soft Xray Absorption Spectroscopy’’. Appl. Surf. Sci. 1998. 133: 58-64. 45. K. Koyama, M. Yoshita, M. Baba, T. Suemoto. ‘‘High Collection Efficiency in Fluorescence Microscopy with a Solid Immersion Lens’’. Appl. Phys. Lett. 1999. 75: 1667-1669. 46. N.J. Harrick. Internal Reflection Spectroscopy. New York: Wiley Interscience, 1967.
Two methods commonly employed for molecular surface analysis and thin-film analysis of microscopic areas are attenuated total reflection infrared (ATR-IR) microspectroscopy and confocal Raman microspectroscopy. In the former method, the depth of the evanescent probe beam can be controlled by the wavelength of light, the angle of incidence, or the refractive index of the internal reflection element. Because the penetration depth is proportional to the wavelength of light, one could interrogate a smaller film thickness by moving from the mid-infrared region to the visible region employing Raman spectroscopy. The investigation of ATR Raman microspectroscopy, a largely unexplored technique available to Raman microspectroscopy, was carried out. A Renishaw inVia Raman microscope was externally modified and used in conjunction with a solid immersion lens (SIL) to perform ATR Raman experiments. Thin-film polymer samples were analyzed to explore the theoretical sampling depth for experiments conducted without the SIL, with the SIL, and with the SIL using evanescent excitation. The feasibility of micro-ATR Raman was examined by collecting ATR spectra from films whose thickness measured from 200 to 60 nm. Films of these thicknesses were present on a much thicker substrate, and features from the underlying substrate did not become visible until the thin film reached a thickness of 68 nm. Index Headings: Raman; Attenuated Total Reflection; ATR; Total internal reflection; TIR; Solid immersion lens; SIL; Microspectroscopy; Micrometer thin films.
INTRODUCTION Thin-film technology plays an important role in many products from precision, laser, and consumer optics to electronics, semiconductors, biofilms, tribology, and medical technology.1–5 Optical components ranging from spectrometer grade lenses to eyeglasses for corrective vision employ thin-film coatings, such as antireflection coatings.6 The medical industry employs thin-film membrane technology for filtration, dialysis, and electrolysis, among a host of many other applications.2 These films can vary in thickness anywhere from several micrometers to nanometers (single monolayers).6–8 The molecular characterization of these films is important to understanding their functional performance. However, thin transparent films supported on much thicker substrates present a challenge in that many of the current techniques used to characterize these films probe deeper into the surface than the thickness of the film. Because of this, spectra obtained using these methods are dominated with spectral features from the bulk of the substrate. In addition, the analysis of Received 29 January 2013; accepted 10 July 2014. * Author to whom correspondence should be sent. E-mail: tranw@ miamiOH.edu. DOI: 10.1366/13-07024
230
Volume 69, Number 2, 2015
microscopic defects in these films is important to elucidate failure and contamination mechanisms during their production and use. Two methods commonly employed for molecular surface analysis and thin-film analysis of microscopic areas are attenuated total reflection infrared (ATR-IR) microspectroscopy, and confocal Raman microspectroscopy.9–11 Both methods have gained popularity over the last 17 years; however, both have drawbacks when films whose thickness is less than 1 lm are studied. In a standard IR accessory using a germanium hemisphere, the limiting penetration depth at 700 wavenumbers is approximately 1 lm. For films that are thinner than 1 lm, spectral contamination from subsurface layers is expected. Further, spectral contamination becomes more problematic if the sub-layer has a stronger IR absorption than the surface film. The advent of confocal microscopy improved the depth resolution for Raman microspectroscopy by employing a pinhole aperture at the back image plane of the microscope, attenuating the light from the out-of-focus regions of the sample.12–15 The ability of researchers to sample along the z-axis of the microscope allowed studies on multilayered samples such as layered polymers.16 However, this approach has several limitations, brought to light by Everall’s investigations into the depth resolution of a commercial confocal Raman microscope.17,18 In these experiments, the depth of focus was found to be approximately three times greater than that predicted by theory. This difference was caused by refraction at the air–sample interface, which can be corrected using an immersion objective with a suitable index-matching fluid.18 However, the depth resolution using this approach is 1.25 lm, assuming a numerical aperture (NA) of 1.4 and an excitation wavelength of 632 nm. In addition, the confocal Raman microscope still collects light outside the confocal region due to its farfield optical configuration. A solution to these problems is to conduct Raman microspectroscopy using evanescent excitation. This approach has been referred to as ATR Raman or total internal reflection (TIR) Raman spectroscopy. Compared to IR wavelengths, Raman excitation wavelengths are an order of magnitude shorter, thereby yielding an order of magnitude smaller penetration depth. Second, Milster pointed out several years ago the benefits of using nearfield optics (evanescent energy) to increase the read/ write density in optical data storage systems.19 The same concept can be used for Raman spectroscopy where excitation is done evanescently and collection is done far field. Figure 1 demonstrates the concept of near-field excitation with far-field collection and the
0003-7028/15/6902-0230/0 Q 2015 Society for Applied Spectroscopy
APPLIED SPECTROSCOPY
FIG. 1. (left) Far-field illumination and collection by a confocal microscope. (right) Near-field illumination with far-field collection. Oversampling beyond the sample layer is avoided when illuminating in the near field. (Reproduced with permission from Ref. 20. Copyright McGraw-Hill 2010.)
corresponding sampling volume.20 Milster19 also provided a detailed review on the use of high-refractive-index (n) hemispheres (solid immersion lenses (SILs)) to focus and collect visible light in the optical recording industry, thereby increasing the read–write density of the devices. This method yielded an improvement in the xy resolution by n, while the axial resolution and collection efficiency were improved by n2. Poweleit et al. demonstrated the xy improvement for Raman imaging using a La SFN-9 glass SIL (nc = 1.868) with 488 nm excitation on patterned silicon.21 Sohn et. al. reported on using a zinc selenide (ZnSe) hemisphere to construct a microscope that could conduct both ATR IR microspectroscopy and Raman SIL microspectroscopy at the same time.22 Additional studies have shown near-field spectroscopy benefits Raman spectroscopy over far-field spectroscopy by providing surface enhancement, a reduced Rayleigh tail, and a zpolarization effect that can be used to obtain different information.23,24 The first report on ATR Raman was published by Ikeshoji et al. in 1973 and briefly explored the theoretical and experimental intensities of ATR Raman scattering.25 This study was followed by several reports demonstrating ATR Raman using high excitation powers at the sample with strong Raman-scattering samples such as carbon tetrachloride.26–28 Later, Tisinger and Sommer adapted Milster’s sampling method to acquire ATR Raman spectra of polydiacetylene (PDA) directly applied to the bottom of a ZnSe hemisphere.29,30 To our knowledge, this report was the first attempt at performing ATR Raman using a conventional Raman microscope rather than the macro-ATR Raman spectrometers used in previous studies. Tisinger and coworkers were able to
demonstrate the feasibility of conducting ATR Raman measurements using off-axis excitation and on-axis collection through a microscope. They also demonstrated that, just like ATR Fourier transform (FT)-IR measurements, ATR Raman measurements require intimate contact between the internal reflection element (IRE) and sample to obtain quality spectra.29–31 In 2004, Greene and Bain investigated a 2.5 lm thick polyethylene napthalate film on a 200 lm thick polyethylene terephthalate (PET) substrate using 532 nm excitation and 200 mW power output using a cubic zirconia IRE.32 In 2008, Ishizaki and Kim analyzed a 135 nm polystyrene film using ATR Raman.33 They employed a sapphire SIL as the IRE and 752 nm laser excitation with a 56 mW power output. In 2010, McKee and Smith developed a scanning angle TIR Raman spectrometer to investigate films cast onto a ZnSe IRE. The instrument could vary the angle of incidence to vary the sampling depth of the films.34 Recently, Michaels investigated poly(3,4-ethylenedioxythiophene)-poly(styrene sulfonate) (PEDOT:PSS) film on PET using a 785 nm excitation. Large intensity counts for the PEDOT:PSS film were observed due to PEDOT:PSS having a Raman scattering (the cross section at 440 cm1 is 40 times greater than the cross section of PET at 1726 cm1), making it a strong Raman scatterer.35 A comprehensive critical review of ATR Raman spectroscopy was published by Woods and Bain in 2012, reporting on work with solid–solid, solid– liquid, solid–air, liquid–liquid, and liquid–air interfaces, as well as depth profiling and scattering angle studies.36 A review of the literature shows that previous investigations employed large laser powers (50–200 mW)25–28,32,33 as well as strong Raman-scattering samples such as
APPLIED SPECTROSCOPY
231
FIG. 2. (a) The ATR sampling cell. (b) Side view of the ball bearing providing point contact.
PDA29 and PEDOT:PSS.26 Previous methods also applied samples directly onto the IREs to improve the signal-tonoise ratio (SNR) and experimental success rate. However, this practice is not conducive to routine sample analysis;29,30,33,34 this sample alteration is undesirable because it inhibits a fundamental advantage of ATR and Raman spectroscopy—the ability to analyze a sample in its natural state. With the exception of Ishizaki and Kim, the previous work mentioned employed Raman microscopes. The purpose of the present study is to investigate the capabilities and limitations of ATR Raman microspectroscopy conducted using a commercial Raman microscope. The samples investigated were polystyrene films supported on a thick polycarbonate substrate using laser powers of less than 6 mW. This work was first presented at the 2007 meeting of the Federation of Analytical Chemistry and Spectroscopic Societies in Memphis, Tennessee, and later presented at the 2009 meeting of the Pittsburgh Conference in Chicago.37,38
EXPERIMENTAL Materials. Certified American Chemical Society (ACS)-grade toluene was purchased (Fisher Chemical, Fairlawn, NJ) and polystyrene beads were obtained (Sigma Aldrich Company, St. Louis, MO). Recordable compact disks (CD-Rs; Memorex) were used as the polycarbonate substrate on which the polystyrene was applied. The CD-R was cut into 2.54 3 2.54 cm substrates, and the aluminum reflective layer was removed. The film was cast on the side opposite where the reflective layer resided. Sample Preparation. The polystyrene (n = 1.58) was dissolved in toluene by a ratio of 4% weight of polymer to solvent. An EC101DT Digital Photo Resist Spinner (Headway Research, Inc.) was used to spin coat the polystyrene solution onto the polycarbonate substrate. The polycarbonate was mounted, and when the spin coater reached the desired revolutions per minute, 500 lL of the polystyrene-toluene solution was micropipetted onto the polycarbonate substrate. Various film thicknesses (60–200 nm) were obtained by varying the
232
Volume 69, Number 2, 2015
revolutions per minute from 2000 to 4400 on the spin coater. Film thicknesses were confirmed using both reflection-absorption IR spectroscopy and spectroscopic ellipsometry. Thickness confirmation samples were prepared simultaneously with the polystyrene-polycarbonate sample by spin coating polystyrene onto a cut 2.54 3 2.54 cm microscope slide (MirrIR Low-e, Chesterland, OH). Reflection–absorption IR spectra of the samples were collected using a Perkin-Elmer Spotlight 300 IR microscope interfaced to a Perkin-Elmer Spectrum One Fourier Transform spectrometer (Perkin-Elmer). The system employed a 100 3 100 lm liquid-nitrogen-cooled mercury cadmium telluride (HgCdTe) detector. Each spectrum collected represents the average of 32 individual scans possessing a spectral resolution of 4 cm1. The peak height of the polystyrene C–H stretch at 1493 cm1 was used to determine film thickness from a calibration curve.30 A PhE-101 Discrete k Ellipsometer (Micro Photonics Inc., Allentown, PA) was used to gather the spectroscopic ellipsometry data of the film thickness directly from the polystyrene-polycarbonate sample. The ellipsometer was calibrated with silicon dioxide–coated silicon wafer standards (J. A. Woolam Co., Inc., Lincoln, NE). Attenuated Total Internal Reflection Accessory. A ZnSe hemisphere measuring 1.5 mm in height and 3 mm in diameter (Spectral Systems, Hopewell Junction, NY) was employed as the IRE for this study. A custom-made cell was fabricated (Miami University Instrumentation Lab, Ohio) to hold the ZnSe IRE in place for sample contact. The cell, which can be seen in Fig. 2a, consisted of a 7.62 long (l) 3 0.54 wide (w) 3 0.3175 high (h) cm aluminum bottom plate, on which the sample was mounted. This plate consisted of two pegs on which the top plate would register as well as three threaded holes to secure the top plate. The top plate was a 3.81 (l) 3 0.54 (w) 3 0.3175 (h) cm aluminum plate fashioned with an 4 mm diameter hole in which the ZnSe hemisphere would sit. The surrounding area of the hole was beveled inward to 658, and a groove was provided to permit side laser illumination.
FIG. 3. Block diagram of the ATR Raman setup.
The top plate contained through holes for screws that were tightened to initiate contact between the sample and ZnSe hemisphere. A side perspective of the cell (Fig. 2b) demonstrates how point contact was achieved between the IRE and the sample. A ball bearing was placed in the hole in the bottom plate. As the top plate was tightened, the ball bearing provided point contact between the sample and hemisphere, ensuring intimate contact. Instrument Setup. Raman spectra were collected using an inVia Raman microscope (Renishaw). A 633 nm helium–neon (HeNe) laser was used as the excitation source; the beam path is shown in Fig. 3. The laser was conditioned using a Pellin–Broca prism, a polarizer, beam expander, and plasma line filter. Prior to entering the IRE at an angle of 518, the light was focused using a 35 mm focal length plano-convex lens (Edmund Optics, Barrington, NJ). The combination of the lens and the IRE produced an elliptical spot on the sample approximately 26 lm high and 36 lm long. The Ramanscattered radiation was collected using an infinitycorrected Olympus ULWD MSPlan 203 objective (0.40 N.A.; f = 180 mm; Olympus). A photograph of the ATR accessory on the microscope and excitation optic is shown in Fig. 4. The collected ATR Raman spectra represent 40 accumulations using a 30 s integration time (total scan time of 1 h) at 4 cm1 resolution. The incident power of the laser prior to entering the IRE was 5.2 mW. The Raman spectra were then analyzed using GRAMS/AI software (Thermo Scientific). Comparison of the spectra collected with and without the SIL was done using peak heights and peak areas for the strongest transitions in the spectra. The data for both were consistent with one another.
from 0.6 to 15 lm), high refractive index of 2.59 at 633 nm,39 and isotropic crystal structure. Zinc sulfide (ZnS) is an alternative material; it is harder than ZnSe but has a somewhat lower refractive index of 2.35 at 633 nm. Most important, ZnSe has no major absorptions above the 500 wavenumber shift. A comparison of the Raman spectra of ZnS and ZnSe is shown in Fig. 5. Comparison of Raman Spectra Collected with and without a Solid Immersion Lens. The hemisphere shape of the SIL minimizes Fresnel reflections because both the incident and exiting rays are normal to the hemisphere surface. In addition, the SIL yields an immersion effect that improves the depth of field and the focused beam diameter. The depth of field is given by: z¼
4k ðnSIL sinhÞ2
ð1Þ
RESULTS AND DISCUSSION Zinc selenide was chosen as the IRE due to its relatively high transmission in the visible range (. 65%
FIG. 4. Photograph of the ATR accessory on the microscope stage and excitation optic.
APPLIED SPECTROSCOPY
233
FIG. 5. Raman spectra of ZnS and ZnSe SILs at 633 nm excitation. The ZnSe peaks are limited to 600 cm1 and below, whereas ZnS has peaks interdispersed between 100 and 800 cm1.
where k is the excitation source wavelength, nSIL is the refractive index of the SIL, and h is the half-angle acceptance of the objective. Under the present conditions, the theoretical depth of field employing a ZnSe SIL at 633 nm and 0.40 N.A. objective is 2.36 lm, whereas without the hemisphere, it is 15.83 lm. The focused beam diameter, d, in the xy plane is given by: d¼
1:22k nSIL sinh
ð2Þ
This value is reduced from 1.93 to 0.75 lm when the SIL is employed. Immersion also increases the amount of Raman signal collected because of the improvement in the lightcollecting efficiency F, which is given by:40 2 F nSIL sin2 h
ð3Þ
Relative to a measurement conducted without the SIL, the amount of light collected under the present conditions is 6.71 times greater using the SIL. Proof of these concepts is shown in Fig. 6 and Table I. Figure 6 illustrates the Raman spectra of single-crystal silicon (Fig. 6a) and polycarbonate (Fig. 6b) that were collected without the SIL and by placing the SIL in contact with the surface of the samples. Qualitatively, the signal decreases for polycarbonate and increases for single-crystal silicon when the SIL is used. The Raman, SR, signal collected from a given sample is given by:41 SR Io NV rF
ð4Þ
where Io is the laser irradiance (in W/cm2), N is the number of particles per unit volume, V is the volume of uniformly illuminated sample, r is the Raman scatter-
234
Volume 69, Number 2, 2015
ing cross section, and F relates to the collection efficiency. For the same condensed-phase sample, N and r are constant. Table I lists values for Io, V, and F for each sample. In the case of polycarbonate, the improvement in laser irradiance (power density) and collected flux is exactly offset by the decrease in illumination volume when the SIL is employed. The remaining difference between the experimental and theoretical values can be accounted for by taking into account the Fresnel reflection losses as (1) the laser enters the hemisphere and traverses the IRE-sample interface and (2) the Raman-scattered light traverses the IRE-sample interface and exits the hemisphere. The agreement between the theoretical value and the experimentally obtained value indicates that the considerations involved in Eqs. 1–4 provide a good approximation for the events taking place. The only provisions are that the sample be isotropic, nonabsorbing, and optically transparent. Silicon, on the other hand, is different. Although the depth of field in air is 15.83 lm, the penetration depth allowed by silicon is only 2.82 lm.42,43 Hence, the disparity between the illumination volume in air and that using the SIL is not as great as it is for polycarbonate. This lessened disparity, along with the increased laser irradiance and collected flux, favors the Raman signal when the SIL is employed. The agreement between the theoretical and experimental values again validates the previous considerations. In the case where the sample characteristics are limiting the illumination volume, as in a very thin, optically transparent film or a sample with a shallow optical skin depth, the gain in irradiance and collected flux significantly improves the Raman signal when the SIL is used. Such is the case for single-crystal silicon,
TABLE I. Confocal parameters with and without the use of a SIL for polycarbonate and silicon. Polycarbonate Parameter
Air
SIL SIL/air
d (lm) 1.93 0.75 Z (lm) 15.83 2.36 46.33 1.03 Volume (lm3) 0.34 2.29 Power density (mW/lm2) Flux collected, F 0.16 1.07 Combined volume, power density, and flux Reflection losses (%) Theoretical considerations Experimental
FIG. 6. Raman spectra collected without the SIL and with the SIL. (a) Silicon. (b) Polycarbonate. The Raman spectrum of silicon collected using the SIL has the higher intensity and that of polycarbonate has the lower intensity.
whose optical skin depth is approximately 2.82 lm at 633 nm. Figure 7 illustrates Raman spectra of an 80 nm thick polystyrene film on polycarbonate collected both in air and using the hemisphere. The spectrum collected in air does not exhibit the symmetric stretch of the styrene ring at the 1001 wavenumber shift, but it does show a feature at the 1006 wavenumber shift for polycarbonate. The spectrum collected using the SIL clearly shows the ring stretch at the 1001 wavenumber shift for styrene. The result is interesting in that the thickness of the styrene film constitutes only 3.5% of the depth of field given in Table I. The increase in the styrene transition intensity is not warranted based on the depth of field and styrene film thickness. However, those values do not take into account the molecular orientation within the film or the effects of refraction. It is well known that polystyrene films of this thickness have the styrene rings oriented parallel to the surface normal.44 Without the SIL, marginal rays enter the sample at an angle of 158. As such, the electric field vector is oriented nearly perpendicular to the styrene rings. With the SIL, marginal rays enter the sample at an angle of 418, with the result that the electric field vector interacts more with the styrene rings. Finally,
0.02 6.71 6.71
Silicon Air 1.93 2.82 8.26 0.34 0.16
Silicon SIL/air 0.75 2.36 1.03 2.29 1.07
1.00 43 0.57 0.56
0.12 6.71 6.71 5.62 43 3.20 3.28
Koyama et al. showed that the collection efficiency exceeds that predicted by theory when a SIL is used to collect fluorescence from dye-doped polystyrene spheres of 200 nm diameter.45 Use of a Solid Immersion Lens with Evanescent Excitation. Up to this point, most of the discussion has focused on the use of a SIL and far-field collection. As such, strong transitions from the subsurface polycarbonate are still observed. By separating excitation to near field (evanescent excitation), one can investigate much thinner films, thereby eliminating the parasitic signal from the subsurface layers. Figure 1 illustrates the differences between the two situations. The figure on the left-hand side shows far-field illumination and collection, with the resulting sample illumination volume being a cylinder with height z and diameter d. For films whose thickness is less than z, subsurface contamination is expected. The right-hand side of Fig. 1 illustrates near-field (evanescent) excitation with farfield collection. In this case, if the film thickness is less than z, no subsurface contamination is expected until the film thickness approaches the penetration depth (the depth to which the evanescent excitation penetrates), dp, which is given by the well-known Harrick equation:46 dp ¼
k
2pnSIL sin2 h n 2s=SIL
1=2
ð5Þ
where n2s/SIL is the refractive index of the IRE and n2s/SIL is the ratio of the refractive index of the sample to the SIL. The theoretical sampling depth of ATR Raman microspectroscopy using the same parameters above is 80 nm and represents a 303 improvement in sampling depth compared to far-field illumination and collection using the SIL. Two objectives of this research were to verify the penetration depth and to determine how long it would take to collect ATR spectra from a moderate Ramanscattering sample using a conventional Raman microprobe. The chosen sample system was thin films (thickness in nanometers) of polystyrene cast onto thick substrates of polycarbonate. Reference spectra of the bulk materials obtained under identical conditions demonstrated that the Raman transition for polycarbon-
APPLIED SPECTROSCOPY
235
FIG. 7. Raman spectra of an 80 nm thick polystyrene film on a polycarbonate substrate (PS) collected without the SIL (bottom spectrum) with the use of a zinc selenide SIL (top spectrum).
FIG. 8. Comparison of Raman spectra collected using the SIL and evanescent excitation on 85, 68, and 60 nm thick polystyrene films on a much thicker polycarbonate substrate. * marks the 1001 cm1 polystyrene reference peak; à marks the polycarbonate peaks found at 1110 and 890 cm1. Also shown are reference spectra for polystyrene (top spectrum) and polycarbonate (bottom spectrum).
236
Volume 69, Number 2, 2015
FIG. 9. The non-normalized polystyrene spectra at 85, 68, and 60 nm thicknesses.
ate located at the 890 cm1 shift is five times stronger than the symmetric stretch of the aromatic ring located at the 1001 cm1 shift in the Raman spectrum of polystyrene. Initial attempts to collect ATR spectra proved that intimate contact was an issue. This problem was overcome by inserting a small ball bearing into the sample holder, as shown in Fig. 2b. The ball bearing provided point contact exactly in the position where it was needed. Various films of polystyrene on polycarbonate, ranging in thickness from 60 to 170 nm were collected in 35 min. The spectra of the thicker films showed good SNR, and those that are thinner exhibited a SNR of approximately 10. Figure 8 shows the spectra of the thinner films along with those of bulk polystyrene and polycarbonate. The features associated with polycarbonate are not observed until the polystyrene film thickness is less than 74 nm. At thicknesses less than 74 nm (not shown), the transitions associated with polycarbonate are observed at the 1600 and 880 cm1 shifts. Figure 9 illustrates the non-normalized polystyrene spectra featured in Fig. 8. Two critical factors we have observed in the success of the measurements are intimate contact with the IRE and the quality of the IRE surface. TABLE II. Center and peak area of polystyrene peak of interest. Sample
Center
Areaa
Polystyrene1 Polystyrene2 Polystyrene3 Polystyrene4 Polystyrene5
1002.0 1002.3 1002.0 1002.0 1002.1
36 38 37 37 37
673 218 938 877 833
Heighta 4231 4257 4461 4219 4831
a Peak height and peak area values collected from the polystyrene peak of interest at 1002 cm1 using GRAMS/AI.
To put the measurements in perspective, the amount of sample giving rise to a spectrum is on the order of 0.82 fL, which translates to 0.82 fg on the basis of the density of polystyrene. These values were derived from an optical analysis of the microprobe and a film thickness equal to the 80 nm penetration depth. In addition, the spectra were generated using an incident power of 7 lW/lm2.
REPRODUCIBILITY Table II contains the values of five Raman spectra of the polystyrene–polycarbonate sample obtained in succession, without adjustments to the setup, to test the reproducibility of ATR Raman microspectroscopy. The peak height and peak area values were collected from the polystyrene peak of interest at 1002 cm1 using GRAMS/AI. As shown, there are no significant deviations between peak intensities and peak areas among the five analyses. Relative standard deviations are 1.58% for the peak area and 6% for the peak height.
CONCLUSION Previous research on ATR Raman employed strong Raman scatterers, large power outputs, and long collection times. The purpose of the present study was to investigate the capabilities and limitations of ATR Raman microspectroscopy conducted using a commercial Raman microscope. The results of the study verify the theoretical penetration depth of 80 nm and show that spectra with reasonable SNR can be collected in approximately 0.5 h. More important, no modifications to the Raman microscope system need to be made because the incident illumination is external to the instrument. Two critical factors in the success of the
APPLIED SPECTROSCOPY
237
measurements are intimate contact with the IRE and the quality of the IRE surface. ACKNOWLEDGMENTS The authors gratefully acknowledge financial support from Kodak, Procter and Gamble, Perkin Elmer, and Renishaw. We also acknowledge the Miami University Instrumentation Laboratory for the construction of the ATR accessory. 1. D. Ristau, H. Ehlers. ‘‘Thin Film Optical Coatings’’. In: F. Tra¨ger, editor. Springer Handbook of Lasers and Optics. New York: Springer ScienceþBusiness Media, LLC, 2007. Pp. 373-396. 2. G. Ozaydin-Ince, A.M. Coclite, K.K. Gleason. ‘‘CVD of Polymeric Thin Films: Applications in Sensors, Biotechnology, Microelectronics/Organic Electronics, Microfluidics, MEMS, Composites and Membranes’’. Rep. Prog. Phys. 2012. 75: 1-40. 3. N. Setter, D. Damjanovic, L. Eng, G. Fox, S. Gevorgian, S. Hong, A. Kingon, H. Kohlstedt, N.Y. Park, G.B. Stephenson, I. Stolitchnov, A.K. Taganstev, D.V. Taylor, T. Yamada, S. Streiffer. ‘‘Ferroelectric Thin Films: Review of Materials, Properties, and Applications’’. J. Appl. Phys. 2006. 100(5): 051606. doi: http://dx.doi.org/10.1063/1. 2336999. 4. Y. Fu, H. Du, W. Huang, S. Zhang, M. Hu. ‘‘TiNi-Based Thin Films in MEMS Applications: A Review’’. Sensor. Actuator. A. 2004. 112: 395408. 5. S.K. Arya, S. Saha, J.E. Ramirez-Vick, V. Gupta, B. Shekhar. S.P. Singh ‘‘Recent Advances in ZnO Nanostructures and Thin Films for Biosensor Applications: Review’’. Anal. Chim. Acta. 2012. 737: 1-21. 6. U. Schulz, U.B. Schallenberg, N. Kaiser. ‘‘Antireflection Coating Design for Plastic Optics’’. Appl. Optics. 2002. 41(16): 3107-3110. 7. Specialty Coating Systems. ‘‘SCS Equipment’’. 2014. http:// scscoatings.com/equipment/index.aspx [accessed Aug 1 2012]. 8. MEMSnet. ‘‘MEMS Thin Film Deposition Processes’’. http:// memsnet.org/mems/processes/deposition.html [accessed Aug 12012]. 9. J.E. Katon. ‘‘Infrared Microspectroscopy’’. In: F.M. Mirabella, editor. Modern Techniques in Applied Molecular Spectroscopy. New York: Wiley-Interscience, 1998. Pp. 267-290. 10. A.J. Sommer. ‘‘Raman Microspectroscopy’’. In: F.M. Mirabella, editor. Modern Techniques in Applied Molecular Spectroscopy. New York: Wiley-Interscience, 1998. Pp. 291-322. 11. P.M. Fredericks. ‘‘Depth Profiling of Polymers by Vibrational Spectroscopy’’. In: N.J. Everall, J.M. Chalmers, P.R. Griffths, editors. Vibrational Spectroscopy of Polymers: Principles and Practice. Hoboken, NJ: John Wiley and Sons, 2007. Pp. 179-200. 12. N.J. Everall. ‘‘Modeling and Measuring the Effect of Refraction on the Depth Resolution of Confocal Raman Microscopy’’. Appl. Spectrosc. 2000. 54(6): 773-782. 13. D.J. Gardiner, M. Bowden, P.R. Graves. ‘‘Novel Applications of Raman Microscopy [and Discussion]’’. Philos. Tr. R. Soc. A. 1986. 320: 295-306. 14. K.P.J. Williams, G.D. Pitt, D.N. Batchelder, B.J. Kip. ‘‘Confocal Raman Microspectroscopy Using a Stigmatic Spectrograph and CCD Detector’’. Appl. Spectrosc. 1994. 48(2): 232-235. 15. R. Tabaksblat, R.J. Meier, B.J. Kip. ‘‘Confocal Raman Microspectroscopy: Theory and Application to Thin Polymer Samples’’. Appl. Spectrosc. 1992. 46(1): 60-68. 16. G.J. Puppels, W. Colier, J.H.F. Olminkhof, C. Otto, F.F.M. de Mul, J. Greve. ‘‘Description and Performance of a Highly Sensitive Confocal Raman Microspectrometer’’. J. Raman Spectrosc. 1991. 22: 217-225. 17. N.J. Everall ‘‘Confocal Raman Microscopy: Why the Depth Resolution and Spatial Accuracy Can Be Much Worse Than You Think’’. Appl. Spectrosc. 2000. 54(10): 1515-1520. 18. N.J. Everall. ‘‘Confocal Raman Microscopy: Performance, Pitfalls, and Best Practice’’. Appl. Spectrosc. 2009. 63(9): 245A-262A. 19. T.D. Milster. ‘‘Near-Field Optics: A New Tool for Data Storage’’. Proc. IEEE. 2000. 88(9): 1480-1490. 20. H.J. Gulley-Stahl, A.J. Sommer, A.P. Evan. ‘‘Evanescent Wave Imaging’’. In: G. Srinivasan, editor. Vibrational Spectroscopic Imaging for Biomedical Applications. New York: McGraw-Hill, 2010. P. 117, Fig. 4.10. 21. C.D. Poweleit, A. Gunther, S. Goodnick, J. Menendez. ‘‘Raman Imaging of Patterned Silicon Using a Solid Immersion Lens’’. Appl. Phys. Lett. 1998. 73(16): 2275-2277.
238
Volume 69, Number 2, 2015
22. A.F. Sohn, L.G. Tisinger, A.J. Sommer. ‘‘Combined ATR Infrared Microspectroscopy and SIL Raman Microspectroscopy’’. Microsc. Microanal. 2004. 10(Suppl 2): 1316-1317. doi:10.1017/ S143927604884757. 23. C.L. Jahncke, H.D. Hallen, M.A. Paesler. ‘‘Nano-Raman Spectroscopy and Imaging with the Near-field Scanning Optical Microscope’’. J. Raman Spectrosc. 1996. 27: 579-586. 24. H.D. Hallen, M.A. Paesler, C.L. Jahncke. ‘‘Raman Spectroscopy: Probing the Border Between Near-Field and Far-Field Spectroscopy’’. In: S. Jutamulia, T. Asakura, editors. Far- and Near-Field Optics: Physics and Information Processing. Proc. SPIE. 1998. 3467. doi:10.1117/12.326819. 25. T. Ikeshoji, Y. Ono, T. Mizuno. ‘‘Total Reflection Raman Spectra; Raman Scattering Due to the Evanescent Wave in Total Reflection’’. Appl. Optics. 1973. 12(10): 2236-2237. 26. T. Takenaka, T. Nakanaga. ‘‘Resonance Raman Spectra of Monolayers Adsorbed at the Interface Between Carbon Tetrachloride and an Aqueous Solution of a Surfactant and a Dye’’. J. Phys. Chem. 1976. 80(5): 475-480. 27. M. Futamata, P. Borthen, J. Thomassen, D. Schumacher, A. Otto. ‘‘Application of an ATR Method in Raman Spectroscopy’’. Appl. Spectrosc. 1994. 48(2): 252-260. 28. M. Futamata. ‘‘Highly-Sensitive ATR Raman Spectroscopy Using Surface-Plasmon-Polariton’’. Internet J. Vib. Spectrosc. 2000. 4(2): 9-23. 29. L.G. Tisinger, A.J. Sommer. ‘‘Attenuated Total Internal Reflection (ATR) Raman Microspectroscopy’’. Microsc. Microanal. 2004. 10(Suppl 2): 1318-1319. doi:10.1017/S1431927604884794. 30. L.G. Tisinger. Investigations in Quantitative Infrared Using Attenuated Total Reflectance. [Ph.D. Dissertation]. Oxford, OH: Miami University, 2002. 31. L. Lewis, A.J. Sommer. ‘‘Attenuated Total Internal Reflection Microspectroscopy of Isolated Particles: An Alternative Approach to Current Methods’’. Appl. Spectrosc. 1999. 53(4): 375-380. 32. P.R. Greene, C.D. Bain. ‘‘Total Internal Reflection Raman Spectroscopy’’. Spectrosc. Europe. 2004. 16: 8-15. 33. F. Ishizaki, M. Kim. ‘‘Near-Infrared Attenuated Total Reflection Raman Spectroscopy for Polymer Surface Observation’’. Jpn. J. Appl. Phys. 2008. 47(3): 1621-1627. doi:10.1143/JJAP.47.1621. 34. K.J. McKee, E.A. Smith. ‘‘Development of a Scanning Angle Total Internal Reflection Raman Spectrometer’’. Rev. Sci. Instrum. 2010. 81(4): 043106. doi:10.1063/1.3378682. 35. C.A. Michaels. ‘‘Surface-Sensitive Raman Microscopy with Total Internal Reflection Illumination’’. J. Raman Spectrosc. 2010. 41: 1670-1677. doi:10.1002/jrs.2610. 36. D.A. Woods, C.D. Bain. ‘‘Total Internal Reflection Raman Spectroscopy’’. Analyst. 2012. 137: 35-48. doi:10.1039/c1an15722a. 37. A.J. Sommer. ‘‘Attenuated Total Internal Reflection Raman Microspectroscopy’’. Paper presented at: FACSS 2007. Memphis, TN; Oct 14-18 2007. Paper 613. 38. W. Tran, A.J. Sommer. ‘‘Analysis of Thin Film Polymers Using ATRRaman Microspectroscopy’’. Paper presented at: Pittsburgh Conference and Exposition on Analytical Chemistry and Applied Spectroscopy. Chicago, IL; March 2009. Abstract 2610-7. 39. Crystran Ltd. ‘‘Zinc Selenide (ZnSe)’’. 2011. http://www.crystran.co. uk/uploads/files/181.pdf [accessed Dec 7 2011]. 40. M. Pluta. Advanced Light Microscopy. Vol. 1. New York: Elsevier Science, 1988. 41. G.J. Rosasco. ‘‘Raman Microprobe Spectroscopy’’. In: R.J.H. Clark, R.E. Hester editors. Advances in Infrared and Raman Spectroscopy. London: Heyden and Son, 1980. Vol. 7, Chap. 4, p. 226. 42. I.D. Wolf. ‘‘Micro-Raman Spectroscopy to Study Local Mechanical Stress in Silicon Integrated Circuits’’. Semicond. Sci. Tech. 1996. 11: 139-154. 43. I.D. Wolf, J. Jimenez, J.P. Landesman, C. Frigeri, P. Braun, E.D. Silva, E. Calvet. Raman and Luminescence Spectroscopy for Microelectronics: Catalogue of Optical and Physical Parameters. ‘‘Nostradamus’’ Project SMT4-CT-95-2024. Luxembourg: European Commission, 1998. 44. D.A. Fischer, G.E. Mitchell, A.T. Yeh, G.L. Gland. ‘‘Functional Group Orientation in Surface and Bulk Polystyrene Studied by Ultra Soft Xray Absorption Spectroscopy’’. Appl. Surf. Sci. 1998. 133: 58-64. 45. K. Koyama, M. Yoshita, M. Baba, T. Suemoto. ‘‘High Collection Efficiency in Fluorescence Microscopy with a Solid Immersion Lens’’. Appl. Phys. Lett. 1999. 75: 1667-1669. 46. N.J. Harrick. Internal Reflection Spectroscopy. New York: Wiley Interscience, 1967.
