Transmission Raman Measurements Using a Spatial Heterodyne Raman Spectrometer (SHRS)

Abstract

A spatial heterodyne Raman spectrometer (SHRS) was used to measure transmission Raman spectra of highly scattering compounds. Transmission Raman spectral intensities of ibuprofen were only 2.4 times lower in intensity than backscatter Raman spectra. The throughput was about eight times higher than an f/1.8 dispersive spectrometer, and the width of the area viewed was found to be seven to nine times higher, using 50.8 mm and 250 mm focal length collection lenses. However, the signal-to-noise (S/N) ratio was two times lower for the SHRS than the f/1.8 dispersive spectrometer, apparently due to high levels of stray light.

Keywords

Transmission raman spectroscopy TRS raman spectroscopy fourier transform raman FT Raman spatial heterodyne raman spectrometer SHRS SHS spatial heterodyne raman

Introduction

Transmission Raman spectroscopy (TRS) provides nondestructive, non-invasive, and unique molecular identification of the bulk contents of a heterogeneous sample.^1,2 Transmission Raman spectroscopy has been shown to be a useful technique in various areas of research such as cancer diagnostics,³ pharmaceutical qualification and quantification,⁴ and process control applications.⁵ Stone and Matousek demonstrated the ability to distinguish between benign and malignant markers of breast cancer tumors at clinically relevant depths using TRS.³ Eliasson et al. conducted feasibility studies using TRS for quantitative analysis of pharmaceutical samples. The work demonstrated TRS, in combination with chemometrics, could be used to measure the active pharmaceutical ingredient concentration with ∼1–2% relative error despite strong Raman signals from the capsule shell.⁴ Macleod and Matousek reviewed TRS studies pertaining to using TRS for process control of pharmaceutical products.⁵ The review describes how TRS can be used for fast, quantitative, nondestructive process control measurements with high specificity of the bulk components for heterogeneous samples.

The TRS technique takes advantage of the diffuse scattering of Raman photons through a sample to increase the Raman signal from the bulk relative to the surface (i.e., reduced sub-sampling). Sub-sampling is the bias to surface layer constituents in backscatter Raman of highly scattering samples, which limits the ability to use backscatter Raman to obtain spectra that are representative of the bulk for highly scattering samples. In TRS, diffuse scattering of light increases interactions within the sample leading to spectra that are more representative of the bulk. However, this comes at the price of reducing the Raman signal compared to backscatter Raman.

In TRS, Raman scattered photons exit the sample in a very large area, in a spot about six times the sample thickness.^6–8 Typical dispersive spectrometers have small entrance slits and are not efficient at collecting light from a large area sample, which leads to low collection efficiency and reduced sensitivity in TRS in comparison to backscatter Raman. The backscatter/transmission (B/T) ratio is a measure of signal loss and is typically a large number for TRS, 10–100.^1–2,4,9 To increase the sensitivity in TRS measurements, signal enhancement techniques can be employed or the collection area can be increased. Signal enhancement has been achieved through the use of unidirectional coupling mirrors to recycle the Raman and laser photons lost on the non-collection sides of the sample.^10,11 Most TRS studies have used fiber optic bundles to increase the collection area, but the small diameter of the fibers still limits the total area that can be viewed, and typical B/T ratios using optical fibers for light collection are 10 to 18.^1–4,12 Fourier transform (FT) Raman has also been demonstrated for TRS measurements as the throughput should be larger than a typical dispersive spectrometer. However, Pelletier et al., using FT Raman for TRS, reported a 20–100 times decrease in the transmission Raman intensity compared to the backscatter intensity, citing self-absorption at near-infrared (NIR) wavelengths and poor Raman collection efficiency.⁹

Recently, we demonstrated a new type of FT Raman spectrometer, the spatial heterodyne Raman spectrometer (SHRS), that has very high light throughput and a wide field of view (FOV).¹³ The SHRS is well suited for measuring large sample areas and could be useful for TRS measurements. The purpose of this paper is to demonstrate TRS using the SHRS and to make comparisons to a more conventional, high throughput dispersive spectrometer.

Experimental

Spatial Heterodyne Raman Spectrometer

Figure 1 (top) shows a schematic of the SHRS. The basic spectrometer design has been previously described.^13–16 For TRS, the SHRS uses a 25 mm cube beam splitter (CM1-BS013, ThorLabs) and two 25 mm, 150 grooves/mm gratings, blazed at 500 nm. A grating mount (DGM-1; Newport Corp.) was used to manually control the grating rotation and tilt, to precisely set the Littrow wavelength. A 105 mm focal length, f/2.8 lens (AF Micro-Nikkor; Nikon) was used to image the gratings at unity magnification onto a liquid nitrogen cooled charge-coupled device (CCD) detector with 1340 × 1300, 20 -μ pixels (VersArray; Princeton Instruments). The CCD fringe image was recorded using Winspec (32-bit, version 2.5.22.0) and data processing was performed using Matlab (MathWorks, version R2013a) and IGOR Pro (WaveMetrics, version 6.03A2) software. In some cases, additional images were used for background corrections of the fringe images. This has been described previously.¹⁴

A 532 nm diode-pumped continuous wave (CW) laser (Millennia Pro 2 s; Spectra-Physics) was used as the excitation source for all SHRS spectra presented and was also used as a reference wavelength to set the Littrow angle of the gratings to 532 nm. Raman light was collected from the sample and directed into the SHRS using either an f/5 achromatic, 50.8 mm diameter, 250 mm focal length, MgF₂ coated lens (PAC088; Newport Corp.) or an f/2 achromatic, 25 mm diameter, 50.8 mm focal length, MgF₂ coated lens (49766; Edmund Optics), L₁ in Figure 1. The collimated light was filtered using various combinations of three filters: a 600 nm shortpass filter (10SWF-600-B; Newport Corp.) to block out-of-band light and 532 nm longpass (LP03-532RE-25; Semrock) and 532 nm holographic filters (Supernotch; Kaiser Optical Systems, Inc.) for laser line rejection. These filters limited the SHRS aperture to 22.8 mm.

Kaiser Holospec

The Kaiser Holospec f/1.8 Holographic Imaging spectrometer (Kaiser Optical Systems, Inc.) is a very fast, commercially available visible Raman spectrometer and is used as a “worst case” comparison for the SHRS. The spectrometer was equipped with a low frequency 532 nm Stokes grating (HSG-532-LF; Kaiser Optical Systems, Inc.) giving a spectral range of 50–2400 cm^–1 with linear dispersion of 3.1 nm/mm. The slit width was either 25 or 100 µm giving a nominal spectral resolution of 3–11 cm^–1. For the Holospec FOV measurements, a ∼150 mW, 532 nm diode laser (Solid state DP55L) was used, whereas for throughput and signal-to-noise (S/N) studies the Spectra Physics laser was used. As shown in Figure 1 (bottom), a 25 mm diameter, f/2 achromatic focusing lens was used to f/# match the Kaiser spectrometer. Collection lenses were identical to those used in the SHRS set-up. Note: for both the dispersive and SHRS instruments, when the larger diameter, longer focal length lens was used for light collection, about the same amount of light was lost to the spectrometer, by coupling mismatch to the 25 mm diameter focusing lens, in the case of the dispersive system, or coupling to the 22.8 mm aperture of the SHRS. Since the mismatch is the same for both spectrometers, it should not affect the results. With the dispersive instrument, a thermo-electrically cooled CCD detector with 1340 × 400, 20 -μ pixels (Pixus 400; Princeton Instruments) was used. Spectra were collected and analyzed with Winspec, Matlab, and IGOR Pro software.

Samples

Potassium perchlorate, sodium nitrate, sodium nitrite, and ammonium nitrate were purchased from Sigma-Aldrich at 99% + purity and used “as is.” Acetaminophen (Good Sense®, Extra Strength 500 mg caplets) was purchased as an over-the-counter pharmaceutical and the thin coating was scraped off before analysis in the hopes of reducing fluorescence from the coating. Removing the coating was later found to have little effect on fluorescence in the Raman spectrum. Ibuprofen, 99%, was purchased from Acros Organics. Ibuprofen and perchlorate pellets were made by grinding the samples using a mortar and pestle, then pressed with a hydraulic pellet press (Carver Laboratory Equipment, model 3912) with a 13 mm stainless steel pellet dye.

Results and Discussion

Spatial Heterodyne Raman Spectrometer

The SHRS is similar to a Michelson interferometer with stationary gratings used in the place of moving mirrors (Figure 1). The Raman signal from the sample is collimated and directed into the interferometer, where it is split into two beams and directed to the gratings in each arm of the interferometer. The gratings are tilted at a specific angle, the Littrow angle θ_L, so that at one wavelength (i.e., the Littrow wavelength) the light is exactly retro-reflected. All other wavelengths are heterodyned to the Littrow wavelength and form a wavelength dependent interference pattern of vertical fringes which is imaged onto the CCD.^17,18 The columns of the fringe image are typically integrated to make the fringe cross-section, and the FT of the fringe cross-section produces the Raman spectrum. In the interferogram, the fringe spacing is defined by:

f = 4 (σ - σ L) \tan θ L

(1)

where f is in fringes per centimeter, σ is the wavenumber of the Raman band, and σ_L is the Littrow wavenumber. The intensity of the fringe pattern is a function of position x on the detector, given by:

I (x) = \int 0 \infty B (σ) {1 + \cos [8 π (σ - σ L) x \tan θ L]} d σ

(2)

where B(σ) is the input spectral intensity at wavenumber σ, and the FT of I(x) yields the Raman spectrum.^17–20

Unlike a dispersive spectrometer, the resolution of the SHRS is not a strong function of the entrance aperture so large apertures can be used to provide very high throughput, without sacrificing spectral resolution. The resolving power (R) of the SHRS is equal to the total number of grating grooves illuminated, about 6840 for the system used here with a 22.8 mm clear aperture (limited by the filters). The resolving power gives a nominal 2.6 cm^–1 spectral resolution,¹⁷ about the same as the f/1.8 spectrometer with 25 µm slit. The spectral resolution of the SHRS was measured with a Hg lamp to be ∼8 cm^–1. The resolution of the SHRS can be affected by non-optimal camera focusing, poor collimation of input beam, optical defects in gratings, and poor quality imaging optics.

The Etendue of a spectrometer is a function of the area viewed and the collection solid angle. The collection solid angle of the SHRS, Ω_S, is defined in Eq. 3.¹⁷

Ω s = \frac{2 π}{R}

(3)

For the system used here, the collection solid angle is about 9.2 × 10^–4sr, giving an acceptance angle (or FOV) of ∼3 × 10^–2 radians (i.e., 1.7°). For small angles, the width of the area viewed by the SHRS is the product of the instrument acceptance angle and the distance to the sample. Thus, for the 250 mm and 50.8 mm focal length lenses, the width of the area viewed on the sample would be 7.6 mm and 1.5 mm, respectively. The Etendue of the SHRS with a 22.8 mm aperture is then 0.38 mm²·sr. For the f/1.8 dispersive spectrometer with a 25 µm slit, the Etendue is 0.048 mm²·sr, about eight times lower than the SHRS.

Representative Transmission Raman Spectroscopy Spectra Measured with the Spatial Heterodyne Raman Spectrometer

Figure 2 shows transmission Raman spectra measured using the SHRS of some highly scattering samples, an over-the-counter acetaminophen tablet, ammonium nitrate, sodium nitrate/sodium nitrite mixture, and potassium perchlorate, as well as the corresponding fringe image cross-sections for each sample (upper left inset). The TRS spectra of the unground salts were measured with the samples in 1 cm quartz cuvettes. The over-the-counter acetaminophen tablet was measured as-is, after scraping off the thin coating. The measured spectral resolution for the sample Raman bands were in the range of 8.4–10 cm^–1; a little larger than the measured resolution of the instrument.

The fringe visibility (FV; see Eq. 4) of the interferograms shown in the inset are ∼0.10, 0.17, 0.18, and 0.36 for the acetaminophen, ammonium nitrate, sodium nitrate/sodium nitrite, and potassium perchlorate, respectively. Fringe visibility is a measure of the modulation efficiency of the interferometer and low values can indicate poor interferometer alignment or high levels of background light outside the range of the interferometer. Low FV can also be the result of non-uniform illumination on the CCD and this can be corrected by flat field corrections. However, no flat field or any other corrections were done to the images for the data shown. The indicated values are less than typical FV values reported for visible FT Raman systems, suggesting high background light and non-optimal alignment of the SHRS interferometer.²¹

FV = \frac{(I max - I min)}{(I max + I min)}

(4)

High levels of background light in the SHRS seem to be the main source of the relatively low FV values measured. Background sources can include unblocked ambient room light, sample fluorescence, and unblocked laser light. Making the measurements in a dark room eliminated ambient light. Fluorescence was not observed in the Raman spectra of the high purity salt samples. However, the acetaminophen spectrum shows some fluorescence, which is not uncommon for organic samples measured using a 532 nm laser. The high background in the fringe cross-sections is likely from unblocked laser light scattering inside the instrument and reaching the detector. Poor FV increases noise in the SHRS spectra and is a current limitation of the SHRS used for these studies.

Backscatter/Transmission Comparison

Figure 1 shows the sample arrangements for transmission (Figure 1a) and backscattered (Figure 1b) Raman measurements. In TRS, the laser is brought in from the backside of the sample, and laser and Raman photons are scattered diffusely through the sample and collected on the opposite side. The TRS sampling geometry averages the Raman signal from all depths in the sample and reduces surface layer bias. For highly diffuse scattering samples, TRS can lead to reduced scattered laser light in the Raman spectrum.^1,22

Figure 3 shows backscatter and transmission Raman spectra of a 13 mm diameter, 4.4 mm thick ibuprofen pellet measured with the SHRS. The measured vibrational frequencies in the Raman spectra shown match the theoretical and experimental data of Jubert et al.²³ The transmission Raman intensity was found to be only 2.4 times lower than the backscatter Raman intensity. This B/T ratio is much smaller than typical values reported for TRS which are in the range of 10–100.^1,2,4,9 The smaller ratio reported here results from the larger FOV of the SHRS.

Width of the Area Viewed

Figures 4 and 5 show the experimentally measured width of the area viewed by the SHRS and the f/1.8 dispersive system using 250 mm and 50.8 mm focal length collection lenses, respectively. The width of the area viewed was determined by moving a razor edge across the sample and measuring the Raman intensity of the sample at each position. The diameter of the area viewed was taken as the distance between the 5% and 95% values in the fitted curve. A 7 mm thick × 13 mm diameter KClO₄ pellet was used for these measurements, except for the SHRS 250 mm focal length lens experiment, where the powder sample was held in a 10 mm cuvette. The diameter of the illuminated spot on the samples, facing the spectrometer, was larger than the area viewed by either spectrometer. A 100 µm slit was used in the dispersive spectrometer for these studies to achieve the largest possible area viewed, for a “worst case” comparison to the SHRS. With a 100 µm slit, the spectral resolution of the dispersive system is ∼10 cm^–1 and the calculated width of the area viewed, with 250 mm and 50.8 mm focal length lenses, should be 500 µm and 100 µm, respectively.

The measured width of the area viewed by the SHRS was 7.2 mm and 1.3 mm with the 250 mm and 50.8 mm focal length lenses, respectively, very close to the predicted values of 7.6 and 1.5 mm. The width of the area viewed by the dispersive system was 840 µm and 180 µm with the 250 mm and 50.8 mm focal length lenses, respectively, about twice the predicted values. The difference in the width of the area viewed for the dispersive system is attributed to the thickness and misalignment of the razor edge at the sample surface. These errors are more pronounced in the dispersive spectrometer, because the width of the area viewed is very small. Therefore, the measured width of the area viewed by the SHRS is about nine and seven times larger than the width of the area viewed by the f/1.8 dispersive spectrometer, using 250 mm and 50.8 mm focal length lenses, respectively.

Throughput

As shown above, the Etendue of the SHRS is about eight times larger than the dispersive spectrometer, assuming the entire 8 mm high slit of the dispersive spectrometer is illuminated. Figure 6 shows Raman spectra of ibuprofen, five measured with the SHRS and five measured with the f/1.8 dispersive system, using identical collection optics. The reproducibility and S/N ratio are so high that each of the five overlap exactly on this scale. The dispersive spectrometer slit width was set to 0.025 mm to match the resolution of the SHRS. For these measurements, all experimental parameters were the same except for the CCD gain, which was 1 and 4 photoelectrons/count for the dispersive system and the SHRS, respectively. In the inset, the spectra are shown over the full wavelength range and the SHRS spectral intensity is shown multiplied by four to compensate for lower CCD gain. The average baseline subtracted peak intensity (i.e., counts) for the 830 cm^–1 band of ibuprofen was two times higher for the SHRS than the dispersive system (the area of the bands gave similar result). Correcting for the lower gain of the SHRS detector, the measured light throughput of the SHRS was about eight times larger than the dispersive system, which matches the calculated values.

Signal-to-Noise

Using the 830 cm^–1 Raman band of ibuprofen in the spectra shown in Figure 6, the S/N ratio was compared for both systems. It is difficult to compare absolute values of the S/N ratio between the two systems, because the noise is distributed differently. To compare relative noise levels, the ratio of replicate spectra for each system were compared (solid and dashed lines, labeled “Ratios” in Figure 6). The magnitude of the noise in the ratio spectra was four to five times higher for the spectra measured using the SHRS than the dispersive system. However, the signal was approximately two times higher using the SHRS system, thus the S/N ratio was about 2–2.5 times higher using the dispersive system. The throughput of the SHRS was eight to nine times higher than the dispersive system and this might be expected to lead to a higher S/N ratio of ∼3. The actual S/N ratio of the SHRS is lower for a couple of reasons. For one thing, the FV of the system was lower than expected because of high background light. As shown in the Figure 2 inset, only a small part of the total collected light contributed to the interferogram, ∼4% in the case of the potassium perchlorate spectrum. The remaining ∼96% still contributes to noise in the spectrum. If the background could be reduced to zero, the S/N ratio could potentially be improved about fivefold. Another reason for the lower than expected S/N ratio in the SHRS has to do with the large number of Raman bands in the spectra that were used for comparison. The area of the 830 cm^–1 band is about 10% of the integrated area of all the bands in the spectral region shown. This is almost a “worst case” in terms of S/N ratio. The noise is distributed equally in an interferometer, so the noise of the 830 cm^–1 band is expected to be approximately three times higher than a dispersive system for equal signal levels. This factor of three almost cancels out the higher SHRS throughput.

Conclusions

A SHRS has been used to measure transmission Raman spectra of several highly scattering samples. The results demonstrate a larger throughput is achieved with the SHRS in comparison to a fast, f/1.8 dispersive spectrometer. The measured sample area viewed is also seven to nine times larger using the SHRS, and the larger area viewed leads to a backscatter/transmission ratio of about 2.4, which is very low for a TRS measurement. The S/N ratio of the SHRS is slightly less than the f/1.8 dispersive spectrometer for identical measurement conditions. The S/N ratio of the SHRS used for these comparisons was limited by low FV, in large part caused by high background signals. More careful attention to the design of the spectrometer to reduce stray light should substantially improve the S/N ratio and sensitivity of the spectrometer.

Footnotes

Acknowledgments

We would like to give special thanks to Dr. Pavel Matousek for insightful discussions.

Conflict of Interest

The authors report there are no conflicts of interest.

Funding

We would like to thank the National Science Foundation (grant no. CHE-1308211); NASA (grant no. NNX14AI34G); the Research Corporation of the University of Hawaii (RCUH)/NAVSEA/DOD (contract no. ZA01339); and the Office of Naval Research (grant no. N00014-12-1-0087) for supporting this work.

References

Matousek

Parker

A.W.

“Bulk Raman Analysis of Pharmaceutical Tablets”. Appl. Spectrosc 2006; 60(12): 1353–1357.

Johansson

Sparen

Svensson

Folestad

Claybourn

“Quantitative Transmission Raman Spectroscopy of Pharmaceutical Tablets and Capsules”. Appl. Spectrosc 2007; 61(11): 1211–1218.

Stone

Matousek

“Advanced Transmission Raman Spectroscopy: A Promising Tool for Breast Disease Diagnosis”. Cancer Res 2008; 68(11): 4424–4430.

Eliasson

Macleod

N.A.

Jayes

L.C.

Clarke

F.C.

Hammond

S.V.

Smith

M.R.

Matousek

“Non-Invasive Quantitative Assessment of the Content of Pharmaceutical Capsules Using Transmission Raman Spectroscopy”. J. Pharm. Biomed. Anal 2008; 47(2): 221–229.

Macleod

N.A.

Matousek

“Emerging Non-Invasive Raman Methods in Process Control and Forensic Applications”. Pharmaceut. Res 2008; 25(1): 2205–2215.

Everall

Matousek

MacLeod

Ronayne

K.L.

Clark

I.P.

“Temporal and Spatial Resolution in Transmission Raman Spectroscopy”. Appl. Spectrosc 2010; 64(1): 52–60.

Everall

Priestnall

Dallin

Andrews

Lewis

Davis

Owen

George

M.W.

“Measurement of Spatial Resolution and Sensitivity in Transmission and Backscattering Raman Spectroscopy of Opaque Samples: Impact on Pharmaceutical Quality Control and Raman Tomography”. Appl. Spectrosc 2010; 64(5): 476–484.

Oelkrug

Ostertag

Kessler

R.W.

“Quantitative Raman Spectroscopy in Turbid Matter: Reflection or Transmission Mode?” Anal. Bioanal. Chem 2013; 405(10): 3367–3379.

Pelletier

M.J.

Larkin

Santangelo

“Transmission Fourier Transform Raman Spectroscopy of Pharmaceutical Tablet Core”. Appl. Spectrosc 2012; 66(4): 451–457.

10.

Pelletier

M.J.

“Sensitivity-Enhanced Transmission Raman Spectroscopy”. Appl. Spectrosc 2013; 67(8): 829–840.

11.

Matousek

“Raman Signal Enhancement in Deep Spectroscopy of Turbid Media”. Appl. Spectrosc 2007; 61(8): 845–854.

12.

Shin

Chung

“Wide Area Coverage Raman Spectroscopy for Reliable Quantitative Analysis and its Applications”. Analyst 2013; 138(12): 3335–3346.

13.

Gomer

N.R.

Gordon

C.M.

Lucey

Sharma

S.K.

Carter

J.C.

Angel

S.M.

“Raman Spectroscopy Using a Spatial Heterodyne Spectrometer: Proof of Concept”. Appl. Spectrosc 2011; 65(8): 849–857.

14.

Lamsal

Angel

S.M.

“Deep-Ultraviolet Raman Measurements Using a Spatial Heterodyne Raman Spectrometer (SHRS)”. Appl. Spectrosc 2015; 69(5): 525–534.

15.

N. Lamsal, S.M. Angel, S.K. Sharma, T.E. Acosta. “Visible and UV Standoff Raman Measurements in Ambient Light Conditions Using a Gated Spatial Heterodyne Raman Spectrometer”. Abstract 1459, 46th LPSC 2015. Woodland TX; March 16–20, 2015.

16.

Lamsal

Sharma

S.K.

Acosta

T.E.

Angel

S.M.

“Ultraviolet Stand-off Raman Measurements Using a Gated Spatial Heterodyne Raman Spectrometer”. Appl. Spectrosc 2016; 70(4): 666–675.

17.

J.M. Harlander. Spatial Heterodyne Spectroscopy: Interferometric Performance at any Wavelength Without Scanning. [Ph.D. Thesis]. Madison, WI: University of Wisconsin, 1991.

18.

Harlander

J.M.

Roesler

F.L.

Chakrabarti

“Spatial Heterodyne Spectroscopy: A Novel Interferometric Technique for the FUV”. In: Siegmund

O.H.W.

Hudson

H.S.

(eds). EUV, X-Ray, and Gamma-Ray Instrumentation for Astronomy. Proceedings of SPIE Volume 1344, Bellingham, WA: SPIE, 1990, pp. 120–131.

19.

Harlander

J.M.

Roesler

F.L.

Cardon

J.G.

Englert

C.R.

Conway

R.R.

“Shimmer: A Spatial Heterodyne Spectrometer for Remote Sensing of Earth’s Middle Atomsphere”. Appl. Opt 2002; 41(7): 1343–1352.

20.

Harlander

Reynolds

Roesler

“Spatial Heterodyne Spectroscopy for the Exploration of Diffuse Interstellar Emission-Lines at Far-Ultraviolet Wavelengths”. Astrophys. J 1992; 396(2): 730–740.

21.

Savoie

Beauchesne

Levesque

“Fourier Transform Raman Spectroscopy in the Visible Region”. Microchim. Acta 1988; 95(1): 223–225.

22.

Everall

Hahn

Matousek

Parker

A.W.

Towrie

“Picosecond Time-Resolved Raman Spectroscopy of Solids: Capabilities and Limitations for Fluorescence Rejection and the Influence of Diffuse Reflectance”. Appl. Spectrosc 2001; 55(12): 1701–1708.

23.

Jubert

Legarto

M.L.

Massa

N.E.

Tevez

L.L.

Okulik

N.B.

“Vibrational and Theoretical Studies of Non-Steroidal Anti-Inflammatory Drugs Ibuprofen [2-(4-isobutylphenyl) propionic acid]; Napoxen [6-methoxy-α-methyl-2-naphthalene acetic acid] and Tolmetin Acids [1-methyl-5-(4methylbenzoyl)-1H-pyrrole-2-acetic acid]”. J. Mol. Struct 2006; 783(1–3): 34–35.