Raman Spectroscopic Discrimination of Estrogens

Abstract

Estrogens are a group of steroid compounds found in the human body that are eventually discharged and ultimately end up in sewer effluents. Since these compounds can potentially affect the endocrine system its detection and quantification in sewer water is important. In this study, estrogens such as estrone (E1), estradiol (E2), estriol (E3), and ethynylestradiol (EE2) were discriminated and quantitated using Raman spectroscopy. Simulated Raman spectra were correlated with experimental data to identify unique marker peaks, which proved to be useful in differentiating each estrogen molecules. Among these marker peaks are Raman modes arising from hydroxyl groups of the estrogen molecules in the spectral region 3200–3700 cm⁻¹. Other Raman modes unique to each of the estrogen samples were also identified, including peaks at 1722 cm⁻¹ for E1 and 2109 cm⁻¹ for EE2, which corresponds to their distinctive structures each containing a different set of functional groups. To quantify the components of estrogen mixtures, the intensities of each identifying Raman bands, at 581 cm⁻¹ for E1, 546 cm⁻¹ for E2, 762 cm⁻¹ for E3 and 597 cm⁻¹ for EE2, were compared and normalized against the intensity of a common peak at 783 cm⁻¹. Quantitative analysis yielded most results within an acceptable 20% error.

1. Introduction

Steroid hormones are compounds typically composed of 18–21 carbon atoms comprising a general structural scaffold formed by four fused rings: a benzene ring designated as ring A, two cyclohexane rings designated as rings B and C, and one cyclopentane ring designated as ring D (Figure 1).¹ Steroid hormones are classified according to differences in their bonding receptors and biological functions. Estrogens are steroid hormones found in the human body that are associated with the female reproductive system. The most common estrogens are estrone (E1), estradiol (E2), and estriol (E3),which are naturally occurring, and ethynylestradiol (EE2) which is synthetic. These estrogens are discharged from the body and eventualy finds it way into the sewer system. If not abated and treated properly, these estrogens, particularly E1, E2 and EE2, can contaminate drinking water sources and can potentially affect the endocrine systems of humans and other animals.²

Figure 1.

Chemical structures of the estrogens used in the study.

Different analytical techniques have been developed to detect the presence of estrogens. Chromatographic methods, for example, are often utilized to determine the trace levels of estrogens. Although gas chromatography (GC) was first used to quantitate estrogens,^3,4 liquid chromatography (LC) is now a more preferred technique due to its better limit of detection (LOD), simpler sample preparation and lower cost of derivatization needed for analysis.^5,6 However, these techniques still utilize extensive sample preparation and costly instrumentation. As a faster, more convenient, and more direct alternative to the chromatographic methods, spectroscopic techniques have been developed.

Vibrational spectroscopy, one of the established tools in analytical chemistry that is commonly used for structural determination and probing molecular interactions, can also be employed to obtain quantitiative and qualitative information of mixtures. There have been a few studies that has utilized vibrational spectroscopy to probe estrogen samples.⁷ Among these are studies using infrared (IR) and Raman spectroscopy to examine the structure of 17β-estradiol as well as several A-ring and 17α-ethynylestradiol derivatives.^8,9 Additionally, the role of the hydroxyl groups in the conformation of bare estradiol and the structures of its monohydrated clusters were monitored using IR-UV double resonance spectroscopy and density functional theory (DFT) calculations.¹⁰ The less frequent use of Raman spectroscopy in the study of estrogens is due mainly to its inherently weak signals and the presence of fluorescence interference brought about by the fluorescent contaminants in most samples. A specialized Raman techniques such as surface enhanced Raman spectroscopy and the like are often required.^11,12 Raman spectroscopy, however, also poses several advantages including minimal sample preparation and its non-destructive nature as well as its ease of use especially with the advent of portable Raman measuring devices.

In this study, Raman spectroscopy was used to qualitatively and quantatively analyze estrogen samples based on the intensity changes of several unique marker peaks relative to a normalizing peak. Results of the DFT-based Gaussian simulations were compared with the measured Raman spectra to assign vibrational modes associated with each peak.

2. Experimental section

2.1. Chemicals

The different estrogen samples used in the study have been purchased from the following suppliers: E1 from Tokyo Chemical Industry Co., Ltd (Portland, OR), E2 and EE2 from Sigma-Aldrich, and E3 from ICN Biomedicals (Irvine, CA).

2.2. Raman spectroscopy

Raman measurements were done at room temperature using a Jasco NRS-3100 confocal dispersive Raman spectrometer equipped with a macro-Raman measurement accessory (Easton, MD). Raman scattering was induced by a 12 mW 488 nm laser and collected on a thermoelectrically cooled charge coupled-device detector. The micro-Raman assembly permitted direct measurements of solid samples in quartz slides. Samples were measured with varying scan times and accumulations, and a constant slit size of 0.1 × 6 mm.

In addition, we also used a ‘home-built’ set-up to probe Raman spectra of the sample at the 3200–3700 cm⁻¹. Raman scattering was induced using a 488 nm line of Coherent Innova 400-K3-krypton ion laser (Coherent Radiation Inc., Palo Alto, CA). The detection system consists of Canon lens (Melville, NY) with a focal length of 55 mm to collect the Raman radiation, a SPEX 1877-0.6 m Triplemate spectrometer (Metuchen, NJ) and a liquid nitrogen-cooled charged-coupled device (CCD) detector (Princeton, NJ). Spectral resolution is 6 cm⁻¹ with ± 1 cm⁻¹ band position reproducibility. Samples were measured with varying scan times and accumulation, and a constant slit width of 0.1 mm.

The Raman spectra of eight mixtures, each containing a different combination of samples with two, three or four estrogens mixed in varying mass ratios to a total estrogen concentration of 40 mg/mL were also obtained. The summary of the different mixtures and respective mass ratios are in Table 1.

Table 1.

Summary of the quantitative analysis of the eight estrogen mixtures showing the theoretical and calculated relative amounts of the mixture components with their respective % error.

Mixture	Fraction of Component
	E1			E2			E3			EE2
	Theo	Calc	% error	Theo	Calc	% error	Theo	Calc	% error	Theo	Calc	% error
1	0.50	0.54	8.86							0.50	0.39	21.76
2	0.50	0.56	12.47	0.50	0.41	18.72
3	0.67	0.69	2.84							0.33	0.27	19.17
4				0.33	0.39	17.37	0.67	0.71	6.94
5	0.33	0.32	3.54	0.33	0.37	10.43				0.33	0.26	21.87
6	0.50	0.52	4.67				0.25	0.19	24.35	0.25	0.20	21.00
7	0.25	0.27	7.12	0.25	0.27	9.88	0.25	0.19	23.00	0.25	0.20	17.77
8	0.30	0.30	1.35	0.20	0.24	17.67	0.20	0.16	22.43	0.30	0.22	26.84

2.3. Computational details

Computational studies were carried out to aid in the assignments of the obtained Raman peaks. Calculations were done using Gaussian 09W density functional theory (DFT) approximation implementing the Becke’s three parameter exchange functional in combination with the Lee, Yang, and Parr correlation function (B3LYP).^1,13,14 The 6–31G(d) basis set was employed as a compromise between accuracy and applicability to large molecules.^15,16 Smaller basis sets are enough in DFT based calculations because the basic functions do not have to describe correlating orbitals. The geometry of each compound was optimized and the corresponding vibrational frequency was simultaneously obtained.

3. Results and Discussion

All four estrogen samples in this study have the fundamental scaffold structure of steroidal estrogens made up of a tetracyclic molecular framework of four fused rings; a phenol, two cyclohexane rings and a cyclopentane ring (Figure 1).¹⁸ These steroidal estrogens, also known as the C18 steroidal compounds, differ in the functional groups attached to the D-ring, specifically at positions C16 and/or C17. E1 has the standard C18 steroidal scaffold with a carbonyl group on C17 which compared to E2 has a hydroxyl group on the same position. E3 has the same structure as E2 with an additional hydroxyl group on C16 in a trans- orientation. EE2 also has a similar structure as with E2 but for the acetylene group (ethynyl) attached to C17. These structural differences should give rise to characteristic Raman bands that upon assignment with the aid of DFT simulations can be the basis of discrimination between the four estrogen samples.

The Raman spectra of all solid estrogen samples at different regions are shown in Figures 2, 3 and 4. Inspection of the experimental and calculated Raman spectra showed common peaks among the four samples arising from the C18 steroidal scaffold as well as distinct peaks arising from the differences in the structure of the four estrogen samples. Plausible peak assignment based on reported literature¹⁹ and DFT calculations are summarized in Table S1.

Figure 2.

Presented are the Raman spectra in the 2800–3000 cm⁻¹ region of the estrogen samples highlighting the prominent hydroxyl Raman modes. The spectra above 3200 cm⁻¹ was zoomed-in to highlight the weak -OH Raman modes.

Figure 3.

Presented are the Raman spectra in the 1700–1800 cm⁻¹ region of the estrogen samples showing distinctive carbonyl Raman mode in E1. The inset shows the ethynyl Raman mode of EE2 at ca. 2100 cm⁻¹. Consistent with their chemical structures, none of the other sample exhibit peaks at ca. 2100 cm⁻¹.

Figure 4.

Raman spectra of the four estrogen samples in the 700 cm⁻¹ region with labeled distinct peaks for each estrogen molecule and common peak used in the quantitative analysis of the estrogen mixtures.

3.1.1. 2800–3700 cm⁻¹ Region

The 3200 – 3700 cm⁻¹ region presents the best spectroscopic window to discriminate the estrogen samples due to the O-H stretching band that is prominent in this region (Figure 2, Table S1). It is, however, important to note that although the bands due to the O-H stretching are of medium and strong intensity in the infrared spectrum, they are inherently weak in the Raman spectrum. Despite which we were able to discriminate each estrogen samples.

The different environments around the hydroxyl groups, which influence the Raman bands in this region, can provide variation in the Raman spectra. E1 has a peak centered at ca. 3370 cm⁻¹, E2 displayed a very weak band centered at ca. 3504 cm⁻¹ while E3 revealed one at around 3553 cm⁻¹. For EE2, a sharp peak at 3315 cm⁻¹, a broader peak at 3547 cm⁻¹ and a weaker peak at ca. 3507 cm⁻¹ can be observed. The sharp 3315 cm⁻¹ peak, however, typically observed around 3280–3349 cm⁻¹ can be assigned to the terminal alkyne (–C≡CH).²⁰ The differences revealed here are unusual as it is expected for these samples to exhibit Raman band characteristic of the –OH vibration that is consistent with their chemical structures.

An inspection of estrogen structures would reveal that the number of peaks observed, except for E1, is not consistent to predictions based on its structure. For instance, the spectrum of E3 yields only one distinct band, but the structure clearly predicts at least three modes representing the three hydroxyl groups in the structure. In addition, E2, which is expected to have two modes, showed only one very weak peak. DFT simulation (which yield peaks in the simplified gas phase model that is shifted from those observed in solution experiments) placed the phenolic-OH stretch mode at ca. 3750 cm⁻¹ for E1, E2, E3 and EE2 (Figure S1). For E2, the D-ring-OH mode stretch was predicted to be at 3747 cm⁻¹, this same Raman mode was predicted to be at 3740 cm⁻¹ for EE2 (Figure S1). The experimental Raman spectrum of E2 was not able to resolve these modes and thus appearing as one weak band. For E3, the two D-ring-OH stretch were predicted to be at 3741 and 3734 cm⁻¹ in the gas phase model. In the experimental spectrum of E3 these peaks were not resolved and appeared as a single broad band at ca. 3547 cm⁻¹ which is blue-shifted compared to that observed for E1. This blue-shift is easily explained by the arrangement of the estrogen molecules within their respective crystal structures.^21,22 A close inspection of the crystal structure for E1 showed hydrogen bonding between the phenolic-OH group and the carbonyl O of a neighboring E1 molecule.²³ The crystal structure of E2, E3 and EE2 showed more extensive head-tail hydrogen bonding between the A-ring-OH and D-ring-OH groups which contribute to the stability of the crystal.^24–26

The aromatic-CH stretch mode that lays within the 3000 – 3200 cm⁻¹ region is also useful in at least discriminating E1 from E2, E3 and EE2 samples (Figure 2, Table S1). The E1 spectra reveal three aromatic-CH stretch modes at 3062, 3075 and 3093 cm⁻¹ (3166, 3191 and 3208 cm⁻¹ in the gas phase simulation) while E2, E3 and EE2 samples exhibited two such modes at ca. 3062 and 3075 cm⁻¹ (3202 and 3218 cm⁻¹ simulated peaks) (Figure S1). The slight difference in the peak position of the aromatic-CH stretch mode for the different estrogen samples can be attributed to the differences in the packing of the molecules as revealed by their respective crystal structures. Clearly the structure of E1, which has a more planar D-ring, stands out from the rest of the estrogen samples.

The rest of this region, 2800 – 3000 cm⁻¹, which can be attributed to the aliphatic-CH stretch modes, is less useful in discriminating the estrogen samples. Experimental results (Figure 2, Table S1) showed peaks that can potentially differentiate each estrogen samples such as the one seen at 2953 and 3005 cm⁻¹ for E1, 2815 cm⁻¹ for E2 and 2827 cm⁻¹ for E3. There is also the peak around ~2845 cm⁻¹ observed in both E1 and E3. But these peaks are rendered useless due to spectral crowding. Variations in the characteristic peaks are due to a number of intermolecular and intramolecular interactions which exert influence on the structure, like for example steric effects or the formation of H-bonds. For instance, E1 has a more rigid structure due to the presence of the carbonyl group and fewer sites for H-bonding. The presence of ethynyl in EE2 can provide steric effect and the presence of an extra hydroxyl can provide another site for H-bonding. These factors may influence the strength and angles of existing chemical bonds, resulting in different vibrational frequencies. This has been observed in the IR spectra of two structurally similar polyunsaturated C₂₀ fatty acids (PUFAs) eicosapentaneoic acid and arachidonic acid.²⁷

This particular region is more useful in that we can identify common peaks observed experimentally in all estrogens, such as those around ~2970, ~ 2940, ~2920, ~2890 and ~2860 cm⁻¹. These bands were due to the common steroid backbone (hydrocarbon network) found in all estrogens.

3.1.2. 1000–2200 cm⁻¹Region

The 1700–2200 cm⁻¹ region can be used to exclusively identify two of the estrogen samples (Figure 3, Table S1). A prominent peak at 2109 cm⁻¹ was observed on the EE2 arising from the C≡C stretching mode found of its ethylene group that usually appear at the 2100–2250 cm⁻¹ region, gas phase DFT-based simulation placed this band at 2217 cm⁻¹ (Figure S2). This functional group is unique to EE2 and hence can be used as a discriminant in identifying EE2. On the other hand, another prominent peak was observed for E1 at 1722 cm⁻¹ which was observed in the simulated spectra at 1837 cm⁻¹ (Figure S2). This pertains to the carbonyl group (C=O), which is observed usually experimentally observed within the 1720–1740 cm⁻¹ for ketones. This functional group is unique to E1 and can therefore be used as a marker for E1 samples.

In the 1000–1700 cm⁻¹ region, aromatic ring deformation modes, in-plane –OH bending modes, and aliphatic/aromatic –CH bending modes are usually observed and can be used for rapid discrimination among estrogens. The peaks experimentally observed at ~1600 cm⁻¹ for all the estrogen samples were assigned to the ring (C=C) stretching mode. While all the estrogens have similar ring structures subtle differences in structural detail yield notable disparities in the observed peak positions. One can imagine that while the A ring structure of each estrogen sample is exactly the same they may be influenced by groups could form interactions. The x-ray crystal structures of these molecules depict head to tail arrangement so one can picture the functional groups on the D-ring interacting with those on ring A.^23–26 For instance, the –C=C– ring stretch mode for EE2, which appears at 1628 cm⁻¹, may be influenced by the terminal alkyne (–C≡CH) on its D-ring. Similarly, the carbonyl group on the D-ring of E1 shifts the –C=C– ring stretch mode to 1622 cm⁻¹. The hydroxyl group(s) on the D-ring of E2 (1618 cm⁻¹) and E3 (1614 cm⁻¹) further red shifts –C=C– ring stretch mode.

Two other interesting –C=C– ring stretch modes at ca. 1503 and 1250 cm⁻¹, have C-O and C-OH character, respectively. Again, considering the heat-to-tail arrangements of the molecules, it’s interesting to see how the ca. 1503 and 1250 cm⁻¹ modes are influenced by the surrounding group. The E1, E2 and EE2 samples revealed a peak at 1503 cm⁻¹ while a similar mode for the E3 sample is blue shifted to 1511 cm⁻¹. As for the 1250 cm⁻¹ ring-OH mode, they all have varying positions, 1259 cm⁻¹ for E1, 1254 cm⁻¹ for E2, 1256 cm⁻¹ for E3 and 1251 cm⁻¹ for EE2. The differences observed here can again be due to the nature of compounds and the interactions they make.

While there are a number of peaks observed in this region due to O-H in-plane bending, C-H (aromatic and aliphatic) in-plane bending, and C-C stretching they do not stand out, as the peaks have weak intensities and are also not well resolved. There are distinct peaks at 1464, 1227 and 1193 cm⁻¹ for E1; 1327, 1287, 1181, 1133 and 1073 cm⁻¹ for E2; 1160 cm⁻¹ for E3 and 1115 for EE2. Peaks that are shared by two or three estrogens can also be found in this region, such as the peak at ~1255, ~1170 and ~1000 cm⁻¹ (E1, E2 and EE2), ~1240 cm⁻¹ (E2 and E3), ~1185 and ~1135 cm⁻¹ (E1 and E2), ~1105 cm⁻¹ (E1 and E3) and ~1060 cm⁻¹ (E1, E3 and EE2). Most of the peaks observed in this region may not clearly discriminate each estrogen samples but taken together with the other more prominent peaks strengthens the validity of the qualitative identification.

3.1.3. Below 1000 cm⁻¹Region

In the spectral region below 1000 cm⁻¹ (Figure 4, Table S1), the usual modes observed are out-of-plane bending of -OH and -CH functional groups and ring deformations that arise from the C18 steroidal scaffold structure common to all estrogen samples used. The sample E1 exhibited a distinct peak at 584 cm⁻¹ corresponding to the ring deformation on the B and D ring of its structure. For E2, distinct peaks can be observed at 947 and 542 cm⁻¹, which corresponds to ring deformation vibrational modes. For E3, distinct modes were recorded at 993, 767 and 588 cm⁻¹. EE2 exhibited the most number of distinct vibrational modes due to its ethynyl functional group appearing within this spectral window.

This spectral window presents a great advantage for use in analytical applications. Relatively, peak intensities in this region are more intense compared to those of the hydroxyl Raman modes at the 3500 cm⁻¹ region. Other intense peaks that can be of useful in analytical applications can be found in the 1700 and 2100 cm⁻¹ regions. However, these peaks are isolated and would require multiple spectral measurements in order to cover all shifts containing distinct peaks that can be used to qualitatively identify the presence of these estrogen molecules. The 400–1000 cm⁻¹ region therefore presents the best candidate for use in qualitative and quantitative estrogen analyses.

3.2.1. Qualitative Analysis of Estrogen Mixtures

The ability of Raman spectroscopy to detect estrogen molecules in a sample mixture was carried out by using a solution mixture of estrogens of different proportions. The preferred solvent was DMF due to the estrogen samples’ greater solubility and stability in less polar solvents. Other solvents that were considered were ethanol and dimethylsulfoxide (DMSO). Poor solubility and decomposition of the estrogens in ethanol upon exposure to laser radiation were the major difficulties encountered in obtaining high-quality sample spectra. DMSO, on the other hand, allows better solubility, but intense solvent peaks covered distinct estrogen bands, which then hinders qualitative identification of components of estrogen mixtures. The 3200 cm⁻¹ region as well as the 2200 cm⁻¹ and 1700 cm⁻¹ spectral regions provided several distinct and isolated peaks that would be ideal for analysis. However, upon taking the spectra of the estrogens in solution with DMF, overwhelming DMF signals covered much of the sample estrogen Raman bands in this region, making analysis impossible. It is also impractical to select modes at different spectral windows since this will require another layer of peak normalization during analysis. It was therefore more ideal to use the lower 400–1000 cm⁻¹ region where all estrogen samples in solution showed isolated distinct peaks and a common peak within the same spectral window. In addition to these, the position of the DMF peaks relative to the estrogen peaks of interest allows the elimination of a solvent subtraction step, which makes analysis faster and simpler.

Eight estrogen mixtures were prepared with different components and ratios, as shown in Table 1. Distinct peaks ideal for analysis were 581 cm⁻¹ for E1, 542 cm⁻¹ for E2, 767 cm⁻¹ for E3 and 598 cm⁻¹ for EE2. Sample spectra of the different mixtures are shown in Figure 6. Determination of the estrogen mixture composition is straightforward and is done by simply locating each of the marker bands for each specific estrogen.

Figure 6.

Raman spectra of the eight estrogen mixtures used in the qualitative and quantitative analysis of its estrogen components.

3.2.2. Quantitative Analysis of the Estrogen Mixtures

Within the same 400–1000 cm⁻¹ spectral window, an isolated common peak at 783 cm⁻¹ can serve as a normalizing peak in order to bring peak intensities on the same scale. Normalization was carried out using the following formula where I is the Raman intensity at the specified wavenumber:

$$I_{581,\mathit{\text{norm}}}=\frac{I_{581}}{I_{783}}{I}_{546,\mathit{\text{norm}}}=\frac{I_{546}}{I_{783}}$$

$$I_{762,\mathit{\text{norm}}}=\frac{I_{762}}{I_{783}}{I}_{597,\mathit{\text{norm}}}=\frac{I_{597}}{I_{783}}$$

Concentration of the different components were calculated as follows where r_estrogen is the ratio of the estrogen component and I_norm,mix is the normalized intensity of the distinct peak of the component from the mixture Raman spectra and I_norm,std is the normalized intensity of the distinct peak of the component from the single component 40 mg/mL solution in DMF:

$$r_{\mathit{\text{estrogen}}}=\left(\frac{I_{\mathit{\text{norm}},\mathit{\text{mix}}}}{I_{\mathit{\text{norm}},\mathit{\text{std}}}}\right)$$

Table 1 summarizes the calculated component ratios of the estrogen mixtures. Upon setting the acceptable relative % error to 25 %, all but one measurement (mixture 8, EE2) fell within acceptable error range. The measured ratio of E1 had the least % error while EE2 had the greatest. Several factors can affect the accuracy of measurements including fluorescence interference from possible contaminants.¹² Even a small amount of fluorescence interference can alter the baseline that can result in a lower estimate of peak intensities. Also, the position of the peaks relative to the solvent peaks (DMF) also plays a big factor in increasing % error. In order to reduce peak overlap, the peaks for analysis are ideally isolated. The EE2 peak at 597 cm⁻¹ is only about 80 cm⁻¹ away from the very intense DMF peak at ca. 680 cm⁻¹, which can result in a lower estimate of the peak intensity. This ultimately leads to a lower ratio estimate, which is consistent with the data presented for all solutions mixtures containing EE2. E3 comes in second in worse % error, mainly due to the weak E3 peak at 762 cm⁻¹ that resulted in a poor estimate of the peak intensity after being overwhelmed by the very intense DMF peaks. E2 has a slightly better % error with a maximum of 18.72 %. The E2 peak at 546 cm⁻¹ slightly overlaps with the E3 peak at 534 cm⁻¹ and EE2 peak at 542 cm⁻¹. This led to a higher than expected estimate of the E2 ratio for mixtures 4, 5, 7 and 8 where either E3 or EE2 are also present in solution. Mixture 2 did not follow this trend since only E1 and E2 are present. Analysis of the E1 sample registered the best average % error of 5.84 % because the distinct E1 peak is relatively intense and isolated from other sample and solvent peaks.

Determination of the absolute concentrations of components is possible with the use of an internal standard that will serve as a common peak for normalization of distinct estrogen peaks. However, this comes with caveats, such as a standard peak intensity that is comparable to the sample peaks and that must be isolated from sample peaks and solvent peaks. This is a major difficulty considering the distribution of sample Raman bands in the spectra.

4. Conclusions

The experimental and calculated Raman spectra of the four most common estrogens were obtained and compared with one another. Common bands are found in all samples due to the common backbone structures. Bands that are unique to each estrogen were identified and used as discriminant or marker. For instance, the presence of hydroxyl group(s) gave band(s) at different wavenumbers at the 3200–3700 cm⁻¹ region which is unique for a given estrogen. In addition, bands in other regions that are unique to each estrogen were observed, such as 581 cm⁻¹ in E1, 542 cm⁻¹ in E2, 767 cm⁻¹ in E3 and 598 cm⁻¹ in EE2. These bands were successfully used in identifying the presence of each estrogen sample in a given mixture as well as in the assessing relative concentration of the estrogens with certain limitations to accuracy. With further fine-tuning, it is possible to be able to develop a more accurate and quick analysis of estrogens and other similar molecules for various analytical applications.

Figure 5.

Raman spectra of the estrogen samples dissolved in DMF at 40 mg/mL marking the distinct and common peaks for each estrogen sample. Peak marked with (*) is a DMF Raman mode.