Matrix Effects in Quantitative Assessment of Pharmaceutical Tablets Using Transmission Raman and Near-Infrared (NIR) Spectroscopy Anders Spare´n,a,* Madeleine Hartman,b Magnus Fransson,a Jonas Johansson,a Olof Svenssona a b

Pharmaceutical Development, AstraZeneca R&D Mo¨lndal, SE-43183 Mo¨lndal, Sweden Quality Assurance, AstraZeneca Sweden Operations, SE-151 85 So¨derta¨lje, Sweden

Raman spectroscopy can be an alternative to near-infrared spectroscopy (NIR) for nondestructive quantitative analysis of solid pharmaceutical formulations. Compared with NIR spectra, Raman spectra have much better selectivity, but subsampling was always an issue for quantitative assessment. Raman spectroscopy in transmission mode has reduced this issue, since a large volume of the sample is measured in transmission mode. The sample matrix, such as particle size of the drug substance in a tablet, may affect the Raman signal. In this work, matrix effects in transmission NIR and Raman spectroscopy were systematically investigated for a solid pharmaceutical formulation. Tablets were manufactured according to an experimental design, varying the factors particle size of the drug substance (DS), particle size of the filler, compression force, and content of drug substance. All factors were varied at two levels plus a center point, except the drug substance content, which was varied at five levels. Six tablets from each experimental point were measured with transmission NIR and Raman spectroscopy, and their concentration of DS was determined for a third of those tablets. Principal component analysis of NIR and Raman spectra showed that the drug substance content and particle size, the particle size of the filler, and the compression force affected both NIR and Raman spectra. For quantitative assessment, orthogonal partial least squares regression was applied. All factors varied in the experimental design influenced the prediction of the DS content to some extent, both for NIR and Raman spectroscopy, the particle size of the filler having the largest effect. When all matrix variations were included in the multivariate calibrations, however, good predictions of all types of tablets were obtained, both for NIR and Raman spectroscopy. The prediction error using transmission Raman spectroscopy was about 30% lower than that obtained with transmission NIR spectroscopy. Index Headings: Transmission Raman spectroscopy; Transmission near-infrared spectroscopy; Transmission NIR; Matrix effects; Particle size; Pharmaceutical tablets; Quantitative analysis.

INTRODUCTION The pharmaceutical industry is in constant need of new analytical tools to replace traditional techniques with ones better suited for the increasing demands of the industry. With the use of spectroscopic techniques, sample preparation is minimized or avoided and samples can be analyzed much faster, with the potential of physical and structural information being maintained. Received 18 July 2014; accepted 4 December 2014. * Author to whom correspondence should be sent. E-mail: anders. [email protected]. DOI: 10.1366/14-07645

580

Volume 69, Number 5, 2015

Near-infrared (NIR) spectroscopy has been used extensively in the pharmaceutical industry for many years.1–4 With its advantages of being fast, nondestructive, and sensitive to solid state properties, it has been applied in development as well as in manufacturing settings. In recent years, transmission sampling, rather than reflectance geometry, has been used for finished products due to increased sampling volume, leading to improved analytical accuracy. Near-infrared spectroscopy has some disadvantages related to the less selective spectral features of the overtone and combination bands used, and this limits its applicability in particular for lowdose products. Raman spectroscopy has been proposed as an alternative to NIR spectroscopy and has been used for the monitoring of synthesis,5,6 determination of drug substance (DS) polymorphic forms,7–9 blend monitoring,10 and quantitative determination of DS in whole tablets.11–17 Raman spectroscopy has the advantage of monitoring the fundamental vibrational modes and thus results in a superior chemical specificity compared with NIR. Until recently, the major disadvantage of quantitative Raman analysis of finished products was the limited sampling volume in the traditional Raman backscatter setup. This has been overcome using various sample presentation techniques, such as spinning the sample,7,18 using multiple fiber probes for extended laser spot sizes,19 or the application of alternative sampling geometries, such as transmission mode Raman spectroscopy,20–26 in which the diffuse Raman signal scattered from within the sample is collected. This has both the advantage of an increased sampling volume leading to a lowered analytical error as well as an increased optical path length, resulting in an increased Raman conversion probability. Transmission Raman analysis has been applied for both tablets and capsules and proved to be superior to backscatter Raman for quantitative applications.8,21–27 The influence of pharmaceutical material properties on optical spectroscopy techniques has been studied and shown to be of great importance.28–37 Dry pharmaceutical powders and powder blends are highly turbid media with significant elastic scattering and, in most wavelength bands, low absorption, resulting in potentially large effects on spectra and quantitative calibrations. Investigations of the influence of particle size in the literature have shown a varying degree of sensitivity to variation in particle size, although it has been suggested that transmission Raman is more sensitive than backscatter Raman owing to the significantly longer

0003-7028/15/6905-0580/0 Q 2015 Society for Applied Spectroscopy

APPLIED SPECTROSCOPY

optical path length.34 Their suggestion, however, is not in agreement with the findings of Hu et al.,38 who concluded that if the sampling volume is much larger than the particle size, the Raman signal intensity should not be affected by particle size. A proposed method for handling sensitivity to particle size in quantitative Raman analysis is simply to make sure that calibrations are made on a size range relevant to the actual samples.34,38 In this paper, the influence of material properties on quantitative transmission Raman and NIR spectroscopy was further investigated. Ibuprofen tablets were manufactured according to an experimental design with variation in content of DS and particle size of both DS and mannitol, as well as tablet compression force. The aim was to study how the tablet matrix affects the method accuracy and to compare the performance of Raman and NIR spectroscopy for fast tablet analysis. This work was presented at SciX 2013.39

EXPERIMENTAL Chemicals. The DS ibuprofen was supplied by IOL Chemicals and Pharmaceuticals (Punjab, India) and sieved into three particle-size fractions with a d50 of approximately 71, 95, and 154 lm. Mannitol Parteck M100 (d50  91 lm) and Mannitol Parteck M200 (d50  211 lm) were purchased from Merck KGaA (Darmstadt, Germany), while Mannitol Pearlitol 400DC (d50  450 lm) was from Roquette (Lestrem, France). The particle sizes of the DS and the filler were determined with laser diffraction spectroscopy (Mastersizer 2000 Sirocco, Malvern Instruments Ltd, UK). Magnesium stearate and sodium hydroxide pellets were from AstraZeneca R&D (So¨derta¨lje, Sweden) and Scharlau (Barcelona, Spain), respectively. Tablet Manufacture. A two-level full factorial design in drug substance (d50  71, 154 lm) and filler particle size (d50  91, 450 lm) was set up for manufacturing the tablets, using Modde 9.0 (Umetrics MKS AB, Umea˚, Sweden). This design was repeated at five DS concentration levels (16, 18, 20, 22, and 24 w/w%). Twenty-three different powder blends of a batch size of 100 g were mixed according to the settings in Table I, where experiments 21–23 are center points. Tablets from all powder blends were manufactured at three compression forces (8, 12, and 16 kN), resulting in 69 different tablet types. About 50 tablets were manufactured for each tablet type, resulting in about 150 tablets for each experimental point in the design (Table I), in total around 3450 tablets. The manufacturing of the tablets was done in the following way. The DS and the filler were coarsely blended manually with a spoon for 15–20 s, and the mixture was placed in a Turbula 2TF blender (Glen Mills Inc., Switzerland) and mixed for 10 min at 46 rpm. The lubricant was added, and the mixture was coarsely blended manually for 15–20 s and then mixed in the blender for a further 2 min. Tablets were manufactured with a single punch press Korsch EK-0 (Korsch AG, Germany) equipped with flat, round 10 mm punches. Since the 23 different powder mixtures possessed different properties, e.g., flowability, that would affect the size and weight of the tablets, the fill

TABLE I.

Design of experiment for the manufacturing of tablets.

Experiment number

DS particle size d50 (lm)

Filler particle size d50 (lm)

DS concentration (w/w%)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

71 154 71 154 71 154 71 154 71 154 71 154 71 154 71 154 71 154 71 154 95 95 95

91 91 450 450 91 91 450 450 91 91 450 450 91 91 450 450 91 91 450 450 211 211 211

16 16 16 16 18 18 18 18 20 20 20 20 22 22 22 22 24 24 24 24 20 20 20

depth was adjusted for every new powder blend in order to manufacture tablets in the same weight range. This was done manually by altering the fill depth of the punches and weighing the tablets repeatedly until the desired tablet weight was achieved. Subsequent collection of tablets within the experimental point was made. All tablets had a nominal weight of 300 mg and a thickness of 2.9–3.5 mm. The compression force was adjusted manually, and data were collected with the data acquisition system Wintab3 (AstraZeneca R&D, Mo¨lndal, Sweden). Powder mixtures were made in random order, immediately before tablet manufacturing. Raman Measurements. Transmission Raman measurements were made with an AccuRA transmission Raman system (Horiba Scientific, Lille, France) equipped with a 300 mW, 785 nm diode laser, and an x,y-stage for automatic sample presentation of up to 32 tablets. The incident laser beam diameter was 4.3 mm. All tablets were measured for 30 s, using 10 acquisitions, i.e., a total acquisition time of 5 min. The measurements were carried out in random order, both between and within experiments in the design. Near-Infrared Measurements. Transmission NIR spectra were measured with an ABB Bomem MB160 Fourier transform (FT) NIR spectrometer (ABB-Bomem Inc., Canada). The incident light beam was led through a hole with a diameter of approximately 3 mm. Thirty-two scans per sample were recorded, using a resolution of 16 cm1, giving a measurement time of approximately 30 s per tablet. Reference Measurements. Individual tablets were weighed and dissolved in 0.1 M sodium hydroxide in 100 mL volumetric flasks and placed on a mechanical shaker, where they were vigorously shaken for 45 min and then filled to volume. All sample solutions were filtered through an Acrodisc 37 mm syringe filter with

APPLIED SPECTROSCOPY

581

1 lm glass fiber membrane (Pall Corp., USA) to dispose of the magnesium stearate, which is highly insoluble at high pH values. It was confirmed that no considerable loss of DS occurred during the filtering step. Immediately after sample preparation, ultraviolet (UV) spectra were measured in the spectral region 200– 800 nm, using a Cary 50 Bio (Varian, Australia), equipped with a flow-through cuvette with a path length of 5 mm. The average background signal in the region 500–800 nm was subtracted from the absorbance at 264 nm, and the background corrected absorption at 264 nm was used to determine the ibuprofen concentration. The samples were measured in random order. Standard solutions were treated in the same way as the samples, except for the filtering. This step was left out since there was no magnesium stearate present in the standards. Unlike the sample solutions, the ibuprofen that was used for the standard solutions was derived from a non-sieved fraction. In order to check possible drift of the system, one standard solution was repeatedly measured every sixth sample, but no significant drift was detected. Data Evaluation. Multivariate evaluations, such as principal component analysis (PCA) and orthogonal partial least squares regression (OPLS), were made in Simca Pþ 13.0 (Umetrics MKS AB, Umea˚, Sweden). Trellis score plots were made in Spotfire 5.0.1 (TIBCO, Boston, MA).

RESULTS AND DISCUSSIONS Qualitative Evaluation of Matrix Effects in Raman and Near-Infrared Spectra. Six tablets from each tablet type in the design, in total 414 tablets, were measured with transmission NIR and Raman spectroscopy. The NIR and Raman raw spectra in Fig. 1 clearly show that on top of the variation caused by changes in DS and mannitol concentrations, there was a large offset effect between different tablets in the design, both in transmission NIR (Fig. 1a) and Raman (Fig. 1b) spectra. This was a first indication of an effect caused by the variation in the tablet matrix within the design, which will be further evaluated in this paper. Principal component analysis was made on the raw transmission NIR and Raman spectra (no spectral pretreatment; spectra were only mean centered) of two tablets from each experimental point, in total 138 tablets. The spectral regions of NIR and Raman spectra used were 7105–13 100 cm1 and 134.6–2238.5 cm1, respectively. PCA trellis score plots for NIR and Raman spectra, respectively, are shown in Figs. 2 and 3. For both NIR and Raman spectroscopy, the tablets clustered with respect to the filler particle size and tablet compression force. For Raman, clustering according to the particle size of the DS was also observed. These observations indicate that both the filler particle size and the compression force affected NIR and Raman spectra, while the effect of the DS particle size was more pronounced for Raman spectra than for NIR. It is not surprising that matrix properties of the tablets affected transmission NIR spectra—on the contrary, this is a well-known phenomenon within the NIR community. We here report large matrix effects also on transmission

582

Volume 69, Number 5, 2015

Raman spectra, in accordance with similar effects reported.30,32–36 Quantitative Evaluation of Matrix Effects in Raman and Near-Infrared Spectra. Since both transmission Raman and NIR spectra were heavily affected by the tablet matrix, it is of great importance to study how that may affect quantitative assessment of the DS concentration in tablets. To test whether the factors varied in this experimental design (DS particle size, filler particle size, and compression force) also had an effect on the ability of the Raman and NIR models to predict the DS content, several calibration models based on subsets of the design containing tablets with different matrix properties were built. Both transmission Raman and NIR spectra were pretreated with standard normal variate transformation (SNV)40 to correct for baseline shifts and slope, and mean centered before OPLS41 was applied. The spectral regions of Raman and NIR spectra used were 134.6–2238.5 cm1 and 7105–13 100 cm1, respectively. The performance of the calibration models was evaluated by calculating the root mean square error of prediction (RMSEP) and the mean bias of an independent test set, which consisted of tablets that were not included in any of the calibration models. A summary of calibration statistics from the different models can be found in Table II. An efficient way of testing the robustness of a calibration model is to stress it by extracting fractions of the full factorial design used in this work. Hence the robustness of calibration models based on tablets exclusively made of small filler particles, small DS particles, or tablets compacted with low compression force was challenged with test sets consisting of samples with other matrix characteristics, i.e., large DS particles. Figure 4 shows the effect of including only tablets with small particle size of the filler in the calibration and predicting tablets with different filler particle sizes. This resulted in a systematic underestimation of the DS content in the predicted tablets, indicated with both a large RMSEP and a large negative mean bias (Table II). The pattern and the magnitude of the effect were similar for both transmission Raman and NIR spectroscopy. The opposite effect was seen when tablets with small DS particles were included in the calibration, while tablets with medium-size and large DS particles were included in the test set (Fig. 5). In this case, the discrepancy led to an overestimation of the DS content, detected using the large RMSEP and a positive mean bias (Table II). The effect was larger for transmission Raman than for NIR spectroscopy, but the absolute size of the effect was somewhat smaller than the effect of particle-size variation for the filler. This may be due to both a larger range of particle-size variations of the filler compared with that of the DS used in this study and an effect of the larger weight ratio of the filler in the tablets. When the calibrations were based solely on tablets manufactured with a low compression force and the prediction set consisted of tablets manufactured with medium and high compression forces, there was a slight positive mean bias in DS content for transmission Raman spectroscopy and a small negative mean bias for transmission NIR spectroscopy (Fig. 6). The former

FIG. 1.

Transmission (a) NIR and (b) Raman raw spectra of whole tablets.

APPLIED SPECTROSCOPY

583

FIG. 2. PCA analysis of raw transmission NIR spectra of whole tablets. Trellis score plot of components 1 and 2, colored with respect to mannitol particle size (d50, lm), while marker shape is related to compression force (kN). The trellis is divided with respect to ibuprofen particle size (d50, lm).

was somewhat more sensitive to changes in compression force than the latter. It is well known that, for example, particle-size variations and other matrix effects in solid samples affect the offset and the slope of NIR spectra. Here we show that the variations in the tablet matrix have similar effects on transmission Raman spectra. This results in an effect of the tablet matrix on the predicted concentration of the DS. A closer inspection of the Raman spectra reveals that it is not only the offset and the slope of the spectra that are affected by the matrix, but also the heights of characteristic spectral peaks of the DS and the filler, and the latter effect still remained after spectral pretreatment. An example of this is shown in Fig. 7, in which eight SNV pretreated transmission Raman spectra of tablets with the same concentration and particle size of the DS, but different particle sizes of the filler, are displayed. This explains why the predicted content of DS is affected by matrix effects, such as particle-size variations. When all the variations in the tablet matrix (DS and filler particle size and compression force) were included in the calibration, the predictions on an independent test set with similar matrix variations gave a low RMSEP and a very low mean bias (Fig. 8). No groupings with respect to matrix composition could be seen (Fig. 8). Transmis-

584

Volume 69, Number 5, 2015

sion Raman spectroscopy (Fig. 8a) gave a slightly better performance than did transmission NIR spectroscopy (Fig. 8b), with RMSEPs for the independent test set of 0.54 and 0.76 w/w% DS, respectively. There is no clear evidence that this difference should be related to the concentration of the drug substance. As a complement to the above analysis based on Raman spectra only, models containing both Raman spectra and tablet matrix information, such as weight, height, and density of the tablets, DS particle size, filler particle size, and tablet compression force, were evaluated. These models did not, however, show any improved prediction properties. Previous studies on the effect of variations in the sample matrix on Raman spectra have given somewhat ambiguous results. There seems to be good agreement that an increase in the particle size of powders leads to a decrease in Raman intensity,28,29,38 which, according to Wang et al., is not in line with predictions from the Kubelka–Munk model for the Raman signal.29 Hu et al. found that a large sampling volume made backscatter Raman spectroscopy less sensitive to particle-size variations than did a small sampling volume.38 Their results contradict the later findings of Townshend et al., that quantitative assessment with transmission Raman spectroscopy was more sensitive to a mismatch in

FIG. 3. PCA analysis of raw transmission Raman spectra of whole tablets. Trellis score plot of components 1 and 3, colored with respect to ibuprofen particle size (d50, lm), while marker shape is related to compression force (kN). The trellis is divided with respect to mannitol particle size (d50, lm).

TABLE II. A summary of the calibration models and their performance.

Model ID

Calibration set

Test seta

OPLSCb

R2Yc

Q2Yd

RMSEP test set (w/w%)

TRe

M1

0.97

0.97

0.54

0.17

M2

2

0.98

0.98

2.54

2.41

TR

M3

All variation 69 samples Large DS particles size 78 samples Large filler particle size 78 samples High compression force 92 samples All variation 69 samples Large DS particles size 78 samples Large filler particle size 78 samples High compression force 92 samples

4

TR

All variation 69 samples Small DS particle size 60 samples Small filler particle size 60 samples Low compression force 46 samples All variation 69 samples Small DS particle size 60 samples Small filler particle size 60 samples Low compression force 46 samples

1

0.99

0.99

3.04

2.58

3

0.96

0.96

1.30

1.07

5

0.98

0.95

0.76

0.10

2

0.97

0.97

1.40

0.91

2

0.96

0.95

2.45

1.93

3

0.93

0.92

0.95

0.32

Analytical technique

TR

M4 f

TNIR

M5

TNIR

M6

TNIR

M7

TNIR

M8

Bias test set (w/w%)

a

Center points included. OPLSC = number of OPLS components, determined using cross-validation. R2Y = explained variance in Y data. d Q2Y = explained cross-validated variance in Y data. e Abbreviation for transmission Raman spectroscopy. f Abbreviation for transmission NIR spectroscopy. b c

APPLIED SPECTROSCOPY

585

FIG. 4. Predictions on an independent test set made with OPLS calibrations for transmission (a) Raman and (b) NIR spectra. Calibration set, tablets with small mannitol particles (green); test set, tablets with large mannitol particles (red) and medium-size particles (blue). Models used for transmission Raman and NIR spectroscopy, respectively, M3 and M7 (Table II).

FIG. 5. Predictions on an independent test set made with OPLS calibrations for transmission (a) Raman and (b) NIR spectra. Calibration set, tablets with small DS particles (green); test set, tablets with large (red) and medium-size DS particles (blue). Models used for transmission Raman and NIR spectroscopy, respectively, M2 and M6 (Table II).

particle size of the filler between the calibration and test sets than was Raman spectroscopy in backscatter geometry.34 The present study confirms that transmission Raman spectroscopy is indeed sensitive to variations in the particle size of the drug substance and of the filler, as well as of other matrix effects, such as tablet hardness (here expressed as compression force). We believe that in backscatter mode, the effect of sampling volume is a statistical phenomenon related to the number of particles of the drug substance. In transmission mode, we believe that the statistical effect is less pronounced while the variable particle size is rather affecting the elastic scattering coefficient and thus the optical path length. This is in agreement with the findings of Townshend et al.34 In addition to using UV spectroscopy as a reference method, a direct comparison of all measurements made with transmission Raman and NIR spectroscopies was made (Fig. 9). In general, the predicted values from the two techniques agreed well. This suggests that, in the absence of wet chemical reference values for all samples, NIR spectroscopy could be used as an alternative, rough reference method to calibrate Raman

spectroscopy, or the other way around. In this case no clustering with respect to matrix composition could be seen (Fig. 9).

586

Volume 69, Number 5, 2015

CONCLUSIONS 









Both transmission Raman and NIR spectroscopy were sensitive to tablet matrix effects, such as variations in particle size of the DS or the filler and compression force. Good predictions were obtained when all matrix variations in Raman and NIR data were included in the calibrations. The Raman model including all matrix variations had slightly better prediction performance on an independent test set than did the corresponding NIR model. Transmission NIR and Raman spectroscopy had about the same sensitivity to particle-size variations of the filler. Raman transmission spectroscopy was somewhat more sensitive to variations of particle size of the DS and compression force than was NIR transmission spectroscopy.

FIG. 6. Predictions on an independent test set made with OPLS calibrations for transmission (a) Raman and (b) NIR spectra. Calibration set, tablets made with low compression force (green); test set, tablets made with high (red) and intermediate compression forces (blue). Models used for transmission Raman and NIR spectroscopy, respectively, M4 and M8 (Table II).

FIG. 8. Predictions on an independent test set made with OPLS calibrations for transmission (a) Raman and (b) NIR spectra. Calibration set: all tablet matrix variations included (tablets T1), independent test set: all tablet matrix variations included (tablets T2). The straight lines indicate slope 1. Models used for transmission Raman and NIR, respectively, M1 and M5 (Table II). 



FIG. 7. SNV pretreated transmission Raman spectra of the same concentration (16 w/w%) and the same DS particle size (d50 = 154 lm), but different mannitol particle sizes (blue = large particle size; green = small particle size). Mannitol peak at 880 cm1, ibuprofen peak at 837 cm1.

The Raman and NIR models predicted similar content in the tablets, when all matrix variations in Raman and NIR data were included in the calibrations. This work shows that transmission Raman spectroscopy has a similar sensitivity to sample matrix effects as transmission NIR spectroscopy and highlights the importance of including matrix variations in quantitative calibrations. The latter could preferably be done

APPLIED SPECTROSCOPY

587

11.

12.

13. 14.

15.

16.

17.

18.

19. 20. FIG. 9. Comparison of predicted concentrations of DS (%w/w), using the OPLS calibrations for transmission Raman and NIR spectroscopy presented in Fig. 8 (M1 and M5 in Table II). The straight line indicates slope 1.

by using statistical experimental design whenever possible.

21.

22.

23.

ACKNOWLEDGMENT Magnus Swenson is acknowledged for supervising the tablet manufacturing.

24. 25.

1. M. Blanco, J. Coello, H. Iturriaga, S. Maspoch, C. de la Pezuela. ‘‘Near-Infrared Spectroscopy in the Pharmaceutical Industry’’. Analyst. 1998. 123(8): 135R-150R. 2. B.F. MacDonald, K.A. Prebble. ‘‘Some Applications of Near-Infrared Reflectance Analysis in the Pharmaceutical Industry’’. J. Pharm. Biomed. Anal. 1993. 11(11-12): 1077-1085. 3. M. Blanco, M. Alcala´. ‘‘Use of Near-Infrared Spectroscopy for OffLine Measurements in the Pharmaceutical Industry’’. In: K.A. Bakeev, editor. Process Analytical Technology. Chichester, UK: John Wiley and Sons, 2005. Pp. 362-391. 4. Y. Roggo, P. Chalus, L. Maurer, C. Lema-Martinez, A. Edmond, N. Jent. ‘‘A Review of Near Infrared Spectroscopy and Chemometrics in Pharmaceutical Technologies’’. J. Pharm. Biomed. Anal. 2007. 44(3): 683-700. 5. O. Svensson, M. Josefson, F.W. Langkilde. ‘‘Reaction Monitoring Using Raman Spectroscopy and Chemometrics’’. Chemom. Intell. Lab. Syst. 1999. 49(1): 49-66. 6. O. Svensson, M. Josefson, F.W. Langkilde. ‘‘The Synthesis of Metoprolol Monitored Using Raman Spectroscopy and Chemometrics’’. Eur. J. Pharm. Sci. 2000. 11(2): 141-155. 7. F.W. Langkilde, J. Sjo¨blom, L. Tekenbergs-Hjelte, J. Mrak. ‘‘Quantitative FT-Raman Analysis of Two Crystal Forms of a Pharmaceutical Compound’’. J. Pharm. Biomed. Anal. 1997. 15(6): 687-696. 8. A. Aina, M.D. Hargreaves, P. Matousek, J.C. Burley. ‘‘Transmission Raman Spectroscopy as a Tool for Quantifying Polymorphic Content of Pharmaceutical Formulations’’. Analyst. 2010. 135(9): 2328-2333. 9. C.M. McGoverin, M.D. Hargreaves, P. Matousek, K.C. Gordon. ‘‘Pharmaceutical Polymorphs Quantified with Transmission Raman Spectroscopy’’. J. Raman Spectrosc. 2012. 43(2): 280-285. 10. G.J. Vergote, T.R.M. De Beer, C. Vervaet, J.P. Remon, W.R.G. Baeyens, N. Diericx, F. Verpoort. ‘‘In-Line Monitoring of a

588

Volume 69, Number 5, 2015

26.

27.

28.

29.

30. 31.

32.

33.

34.

Pharmaceutical Blending Process Using FT-Raman Spectroscopy’’. Eur. J. Pharm. Sci. 2004. 21(4): 479-485. I. Jedvert, M. Josefson, F. Langkilde. ‘‘Quantification of an Active Substance in a Tablet by NIR and Raman Spectroscopy’’. J. Near Infrared Spectrosc. 1998. 6(1): 279-289. R. Szostak, S. Mazurek. ‘‘Quantitative Determination of Acetylsalicylic Acid and Acetaminophen in Tablets by FT-Raman Spectroscopy’’. Analyst. 2002. 127(1): 144-148. R. Szostak, S. Mazurek. ‘‘FT-Raman Quantitative Determination of Ambroxol in Tablets’’. J. Mol. Struct. 2004. 704(1-3): 229-233. S. Mazurek, R. Szostak. ‘‘Quantitative Determination of Diclofenac Sodium and Aminophylline in Injection Solutions by FT-Raman Spectroscopy’’. J. Pharm. Biomed. Anal. 2006. 40(5): 1235-1242. S. Mazurek, R. Szostak. ‘‘Quantitative Determination of Captopril and Prednisolone in Tablets by FT-Raman Spectroscopy’’. J. Pharm. Biomed. Anal. 2006. 40(5): 1225-1230. S. Mazurek, R. Szostak. ‘‘Quantitative Determination of Diclofenac Sodium in Solid Dosage Forms by FT-Raman Spectroscopy’’. J. Pharm. Biomed. Anal. 2008. 48(3): 814-821. S. Mazurek, R. Szostak. ‘‘Quantification of Atorvastatin Calcium in Tablets by FT-Raman Spectroscopy’’. J. Pharm. Biomed. Anal. 2009. 49(1): 168-172. A.T.G. De Paepe, J.M. Dyke, P.J. Hendra, F.W. Langkilde. ‘‘Rotating samples in FT-Raman Spectrometers’’. Spectrochim. Acta, Part A. 1997. 53(13): 2261-2266. H. Owen, D.J. Strachan, J.B. Slater, J.M. Tedesco. U.S. Patent US2005140973. Filed 2004. Issued 2005. P. Matousek, A.W. Parker. ‘‘Bulk Raman Analysis of Pharmaceutical Tablets’’. Appl. Spectrosc. 2006. 60(12): 1353-1357. J. Johansson, A. Spare´n, O. Svensson, S. Folestad, M. Claybourn. ‘‘Quantitative Transmission Raman Spectroscopy of Pharmaceutical Tablets and Capsules’’. Appl. Spectrosc. 2007. 61(11): 1211-1218. N.A. Macleod, C. Eliasson, P. Matousek. ‘‘Hidden Depths? New Techniques for Sub-Surface Spectroscopy’’. Spectrosc. Eur. 2007. 19(5): 7-10. C. Eliasson, N.A. Macleod, L.C. Jayes, F.C. Clarke, S.V. Hammond, M.R. Smith, P. Matousek. ‘‘Non-Invasive Quantitative Assessment of the Content of Pharmaceutical Capsules Using Transmission Raman Spectroscopy’’. J. Pharm. Biomed. Anal. 2008. 47(2): 221229. N.A. Macleod, P. Matousek. ‘‘Deep Noninvasive Raman Spectroscopy of Turbid Media’’. Appl. Spectrosc. 2008. 62(11): 291A-304A. A. Spare´n, J. Johansson, O. Svensson, S. Folestad, M. Claybourn. ‘‘Transmission Raman Spectroscopy for Quantitative Analysis of Pharmaceutical Solids’’. Am. Pharm. Rev. 2009. 12(1): 62, 66-71. 73. M. Fransson, J. Johansson, A. Spare´n, O. Svensson. ‘‘Comparison of Multivariate Methods for Quantitative Determination with Transmission Raman Spectroscopy in Pharmaceutical Formulations’’. J. Chemom. 2010. 24(11-12): 674-680. M.D. Hargreaves, N.A. MacLeod, M.R. Smith, D. Andrews, S.V. Hammond, P. Matousek. ‘‘Characterization of Transmission Raman Spectroscopy for Rapid Quantitative Analysis of Intact MultiComponent Pharmaceutical Capsules’’. J. Pharm. Biomed. Anal. 2011. 54(3): 463-468. M.V. Pellow-Jarman, P.J. Hendra, R.J. Lehnert. ‘‘The Dependence of Raman Signal Intensity on Particle Size for Crystal Powders’’. Vibr. Spectrosc. 1996. 12(2): 257-261. H. Wang, C.K. Mann, T.J. Vickers. ‘‘Effect of Powder Properties on the Intensity of Raman Scattering by Crystalline Solids’’. Appl. Spectrosc. 2002. 56(12): 1538-1544. P. Matousek. ‘‘Raman Signal Enhancement in Deep Spectroscopy of Turbid Media’’. Appl. Spectrosc. 2007. 61(8): 845-854. T.R.M. De Beer, W.R.G. Baeyens, Y.V. Heyden, J.P. Remon, C. Vervaet, F. Verpoort. ‘‘Influence of Particle Size on the Quantitative Determination of Salicylic Acid in a Pharmaceutical Ointment Using FT-Raman Spectroscopy’’. Eur. J. Pharm. Sci. 2007. 30(3-4): 229-235. P. Matousek, N. Everall, D. Littlejohn, A. Nordon, M. Bloomfield. ‘‘Dependence of Signal on Depth in Transmission Raman Spectroscopy’’. Appl. Spectrosc. 2011. 65(7): 724-733. Z.-P. Chen, L.-M. Li, J.-W. Jin, A. Nordon, D. Littlejohn, J. Yang, J. Zhang, R.-Q. Yu. ‘‘Quantitative Analysis of Powder Mixtures by Raman Spectrometry: The Influence of Particle Size and Its Correction’’. Anal. Chem. 2012. 84(9): 4088-4094. N. Townshend, A. Nordon, D. Littlejohn, M. Myrick, J. Andrews, P. Dallin. ‘‘Comparison of the Determination of a Low-Concentration Active Ingredient in Pharmaceutical Tablets by Backscatter and

Transmission Raman Spectrometry’’. Anal. Chem. 2012. 84(11): 4671-4676. 35. N. Townshend, A. Nordon, D. Littlejohn, J. Andrews, P. Dallin. ‘‘Effect of Particle Properties of Powders on the Generation and Transmission of Raman Scattering’’. Anal. Chem. 2012. 84(11): 4665-4670. 36. D. Oelkrug, E. Ostertag, R.W. Kessler. ‘‘Quantitative Raman Spectroscopy in Turbid Matter: Reflection or Transmission Mode?’’. Anal. Bioanal. Chem. 2013. 405(10): 3367-3379. 37. N. Kellichan, A. Nordon, P. Matousek, D. Littlejohn, G. McGeorge. ‘‘Variation in the Transmission Near-Infrared Signal with Depth in Turbid Media’’. Appl. Spectrosc. 2014. 68(3): 383-387.

38. Y. Hu, H. Wikstro¨m, S.R. Byrn, L.S. Taylor. ‘‘Analysis of the Effect of Particle Size on Polymorphic Quantitation by Raman Spectroscopy’’. Appl. Spectrosc. 2006. 60(9): 977-984. 39. A. Spare´n, O. Svensson, M. Hartman, M. Fransson, J. Johansson. ‘‘Pharmaceutical Tablet Matrix Effects in Quantitative Transmission Raman Spectroscopy (abstract 70)’’. Paper presented at: SciX 2013. Hyatt Regency Milwaukee and Wisconsin Center, Milwaukee, WI; September 29–October 4, 2013. 40. R.J. Barnes, M.S. Dhanoa, S.J. Lister. ‘‘Standard Normal Variate Transformation and De-trending of Near Infrared Diffuse Reflectance Spectra’’. Appl. Spectrosc. 1989. 43(5): 772-777. 41. J. Trygg, S. Wold. ‘‘Orthogonal Projections to Latent Structures (OPLS)’’. J. Chemom. 2002. 16(3): 119-128.

APPLIED SPECTROSCOPY

589