Resonance and off-resonance Raman spectroscopy and imaging are used to examine the spatial variation of the solid-state structure and electronic character of few-layer MoS

 flakes. Simultaneous acquisition of photoluminescence spectra with the Raman scattering provides complementary ways of rendering Raman and photoluminescence spectral images of thin-film MoS

	Graphene is probably the most well-known substance of the emerging class of materials known as two-dimensional (2D) crystals. These materials are constituted by monolayer and few-layered structures. In recent years, new inorganic 2D materials have emerged, including MoS

. These materials have attracted significant interest because of special electronic, optical, and optoelectronic properties in the monolayer and few-layer forms that are different from those manifested by the bulk form (1,2). One of the most significant differences of the 2D crystals is the transformation from an indirect band gap semiconductor in the bulk to a direct band gap semiconductor in the monolayer and few-layer crystals. Thus, the fabrication of optoelectronic devices in addition to familiar integrated electronic circuitry is envisioned for these materials. These optoelectronic characteristics have prompted substantial research to discover the means of fabrication and the physical characteristics of two-dimensional crystals to produce integrated electronic and optoelectronic devices (3).

	Many articles on nanomaterials emphasize the means to fabricate the materials, their properties, and their performance when constructed as devices (4). However, there is a great need for characterization of these materials because of their experimental nature and structural variability, often within one film. After all, none of these materials are of the nature of a compound that can be taken out of a reagent bottle. Consequently, there have been attempts to develop analytical methods for the reliable characterization of 2D crystals. You may have observed the spatially varying colors in reflected white light images of nanomaterials, and so there have been developments to use optical microscopy to rapidly identify the number of molecular layers that make up the 2D crystal (5). In this installment, we focus on Raman and photoluminescence spectroscopy and imaging for the characterization of these experimental materials.

	Two-dimensional crystals are not just small portions of the bulk materials carrying the same properties as those of the three-dimensional kind. Rather, their physical, electronic, and spectral characteristics can be significantly different. Specifically, MoS

 in its bulk form is an indirect bandgap semiconductor, whereas in its monolayer and few-layer forms it becomes a direct bandgap semiconductor (6). Splendani and coworkers (7) studied exfoliated MoS

 monolayer and few-layer flakes by photoluminescence and observed direct excitonic transitions. Exciting with 532-nm laser light they observed broad photoluminescence centered at 627 and 677 nm, whereas these emissions are completely absent when illuminating bulk MoS

. The room-temperature bandgap of bulk MoS

 is approximately 1.7 eV (corresponding to 730 nm) with complex excitonic features between 1.7 and 2.5 eV (8).

 crystal class, and factor group analysis predicts one 

 Raman active modes (8–14). The symmetry assignments and corresponding Raman band positions for bulk hexagonal MoS

). Furthermore, it is important to remember that some visible wavelengths of the laser light used to excite Raman scattering correspond to energies of MoS

 electronic transitions. The absorption spectrum of MoS

 reflects the band gap of 1.7 eV, but it also manifests fine structure with narrow absorption peaks at 1.9 eV (653 nm) and 2.1 eV (590 nm) related to d-to-d orbital transitions split by spin-orbit coupling and designated A1 and B1 excitons, respectively (8,15). Consequently, one can observe an excitation wavelength dependence of the first-order Raman band intensities as well as the appearance of Raman bands assigned to harmonic and combination modes when the laser excitation is of a wavelength that couples into these excitonic transitions (8,12,14,15).


	Figure 1: Raman spectra of few-layer MoS

 flake taken at the same location with different excitation wavelengths.

	
	Just as the monolayer and few-layer 2D crystal structure of MoS

 affects the electronic structure, particularly the nature of the bandgap, the vibrational modes are also affected. The first-order Raman bands listed above are for bulk MoS

. However, these bands shift to different energies progressively as the structure changes from approximately six layers down to a single monolayer. The 

 band has been observed to shift progressively from 32 cm

 in the single "trilayer" (16). The term 

 is not a planar compound; that is, the Mo and S atoms are not all in the same plane. So, a plane of S atoms can be envisioned on the surface with a plane of Mo atoms above that plane and another plane of S atoms above the Mo one; hence the term trilayer. Also, the band positions and respective separations of the 

 bands show progressive changes from the bulk to a single layer. The 

) band shifts to higher wavenumbers and the 

) band shifts to lower wavenumbers in progressing from the bulk to a single trilayer corresponding to a band separation of 25 cm


	Figure 2: Photoluminescence and Raman spectra of few-layer MoS

	The changes of Raman band position with the number of molecular layers described above are most readily observed when using a 532-nm or shorter wavelength excitation. The Raman spectra obtained when using 632.8-nm light to excite few-layer MoS

 appear quite different because of coupling with the aforementioned A1 excitonic transition and the subsequent resonance enhancement. The activity of the harmonic and combination modes yield a richer spectrum and the determination of the number of trilayers present is determined from both changes in band shape and relative intensities (18,19).


	Figure 3: Photoluminescence and Raman spectra of few-layer MoS

 flake taken at different locations on the same flake.

Resonance Raman Scattering and Photoluminescence of Few-Layer MoS

 are typically heterogeneous, and so the Raman and photoluminescence spectra tend to spatially vary on the flake. Nevertheless, there are spectra that can be considered representative of few-layer MoS

 and they are shown in Figure 1. These spectra were acquired at the same location from the interior of a MoS

 flake, but with different excitation wavelengths. The spectrum acquired with 532-nm excitation is typical of those reported in the literature; it consists primarily of the 

 as well as the Si substrate Raman band at 520 cm

. In contrast, the spectrum generated with 632.8-nm excitation is striking insofar as it consists of the same first-order bands and those due to harmonics and combination modes. Remember that the absorption spectrum of MoS

 contains fine structure with narrow absorption peaks at 1.9 eV (653 nm) and 2.1 eV (590 nm) related to d-to-d orbital transitions split by spin-orbit coupling and designated A1 and B1 excitons, respectively (8,15). Consequently, the 632.8-nm excitation couples into the A1 transition producing a resonance enhancement of all of the additional modes observed in Figure 1. All of the bands in the spectrum excited with 632.8-nm light can be accounted for and have previously been assigned (8,12,15).


	Figure 4: Raman spectra of few-layer MoS

 flake taken at different locations on the same flake. These are the same locations from which the photoluminescence spectra in Figure 3 were obtained.

	
	The effect of excitation wavelength on the photoluminescence spectra of few-layer MoS

 flakes can be seen in the spectra shown in Figure 2. The spectra were acquired at the same location on the flake and are plotted on arbitrary intensity scales to better show the differences in relative strengths of both the photoluminescence and Raman signals. There are several differences to observe that relate to the electronic transitions and their coupling to the few-layer MoS

 flake phonons. The photoluminescence excited at 532 nm is substantially weaker relative to the Raman scattering when compared to the photoluminescence-Raman strength ratio of the 632.8-nm excited spectrum. Furthermore, the relative strengths of the 691- and 930-nm emissions are inverted for the 532- and 632.8-nm excitation wavelengths. An important aspect of this measurement is that one can simultaneously obtain a Raman spectrum and a photoluminescence spectrum from the same location. That fact holds important implications for the ability to simultaneously perform photoluminescence and Raman imaging of few-layer MoS


	Figure 5: Photoluminescence and Raman spectra of few-layer MoS

	As we noted earlier, the structure of few-layer MoS

 films vary spatially and the most prominent changes appear at the edges. Photoluminescence spectra excited with 532-nm light obtained near the center and edges of a few-layer MoS

 flake are shown in Figure 3. There are substantial differences in the emission spectra, specifically the emergence of an emission maximum at 687 nm and the absence of the 930-nm peak at the edge. Also, note that the Si Raman band from the flake edge is significantly stronger than its counterpart in the flake center spectrum. That difference can be attributed to either fewer layers at the edge, a change in the absorptivity of MoS

 at that location, or both. Note too that the second-order Raman bands that appear between 550 and 575 nm are strong at the flake edge, whereas they are absent in the spectrum from the flake center. These are the bands that appear prominently in the 632.8-nm excited Raman spectrum, so a change in the electronic properties of the flake edge is indicated. That portion of the Raman spectrum consisting of the 

 bands from these same locations are shown in Figure 4. We can see from the band positions and their separation of 24–25 cm

 that the thicknesses at these positions are greater than four trilayers (17). However, what is noteworthy is the slight broadening to lower wavenumber of the 

band at the flake edge such that the peak maximum is now at 406 cm

. The differences in the Raman and photoluminescence spectra may be an indication of a structural difference at this particular edge location. Furthermore, it appears that there may be a correlation between the Raman scattering and photoluminescence of few-layer MoS


	Figure 6: Raman spectra of few-layer MoS

 flake taken at different locations on the same flake. These are the same locations from which the photoluminescence spectra in Figure 5 were obtained.

	To investigate this hypothesis, let's look at Figure 5, which shows spectra from center and edge positions taken on another few-layer MoS

 flake. There appears to be little difference between photoluminescence and Raman scattering, particularly the second order modes, as a function of flake position. Furthermore, the corresponding Raman 

 bands in Figure 6 are identical. Therefore, it seems that a correlation may exist between the Raman scattering and the photoluminescence of few-layer MoS

 flake obtained with 632.8-nm excitation. Red corresponds to the substrate Si and the green intensities correspond to the spatially varying strength of the Raman band at approximately 450 cm

	In the previous section, we compared photoluminescence and Raman spectra of MoS

 taken at the center and edge of few-layer flakes and saw evidence for spatial variation of the solid-state structure. Here, we apply Raman and photoluminescence imaging to compare the spectral and structural differences revealed through spectroscopy to the contrast observed when viewing the flakes with reflected white light microscopy. A collection of hyperspectral data from a few-layer MoS

 flake is shown in Figure 7. A reflected white light image of the flake appears in the lower right hand corner and a Raman image corresponding to the reflected light image appears to its left. The plot on the upper left consists of all of the Raman spectra acquired over the image area and the upper right hand plot is of the single spectrum associated with the cross-hair location in the Raman and reflected light images. The Raman data were acquired using 632.8-nm excitation and a 100× Olympus objective and by moving the stage in 200-nm increments over an area of approximately 13 μm × 13 μm. The Raman image is actually a rendering of signal strength for a particular Raman band as a function of position on the sample. In this particular case, the Raman image is rendered through a color coded plot of Raman signal strength over the corresponding color bracketed Raman shift positions in the two upper traces.

 flake obtained with 532-nm excitation. Red corresponds to the substrate Si and the green intensities correspond to the spatially varying strength of the Raman band at approximately 408 cm

	The red brackets surround the Raman band of Si at 520 cm

. Consequently, the area surrounding the few-layer MoS

 flake appears red. The green bracket covers the Raman scattering near 450 cm

, which is assigned to the second order longitudinal acoustic mode at the M point of the Brillouin zone (2LA(M)) of MoS

 (12). This and the other second order modes are prominent in the spectrum because of resonance enhancement. The Raman image in the lower left corner in Figure 7 is presented in a color overlay mode, and the substrate Si signal is attenuated by the MoS

 flake above it. Thus, there is little mixing of red with the green in the Raman image of the flake. The spatially varying shades of green are related to the overall signal strength of the 2LA(M) band under resonance enhancement. Note the rough correspondence of the Raman image of the few-layer MoS

 flake with the spatially varying contrast in the reflected white light image. Of course, the bright green areas correspond to much stronger Raman scattering at that location and the darker areas correspond to a weaker Raman signal. What cannot be seen in a still-frame Raman image is the spatial variation of the band shapes and relative intensities. Interactively changing a bracket selection for width and position in the hyperspectral data set, including selecting ratios of two brackets, allows one to render different resonance Raman images that can reveal spatially varying solid-state structure and electronic properties. The better approach to image the spatial variation of the band structure is through real-time imaging while sliding the color bracket selector over different Raman bands.


	Figure 9: Side-by-side comparison of Raman images of the same few-layer MoS

 flake based on different Raman bands and acquired using different excitation wavelengths. Left: 532-nm excitation; Right: 632.8-nm excitation.

	The spectra and image shown in Figure 7 were obtained using 632.8-nm (1.96-eV) excitation. That energy occurs right between two excitonic transitions in the absorption spectrum of bulk MoS

, the so-called A1 and B1 excitons at 1.88 eV and 2.06 eV, respectively (8,15,20,21). Thus, 632.8-nm excitation allows one to achieve resonance with the MoS

 A1 excitonic transition and thereby produce resonance Raman enhancement of first and second order vibrational modes of MoS

. Off-resonance Raman scattering can be obtained using 532-nm excitation. A collection of hyperspectral data from a few-layer MoS

 flake obtained with off-resonance 532-nm excitation is shown in Figure 8. The Raman data were acquired using 532-nm excitation and a 100× Olympus objective and by moving the stage in 200 nm increments over an area of approximately 13 μm × 13 μm. The spectra consist primarily of the 

 bands and the first-order optical phonon from Si at 520 cm

. Here, red corresponds to the Si substrate and green to the 

 band. Without resonance Raman enhancement, the spatially varying contrast in the Raman image is found to correspond to the signal strength of the 408 cm

 band without the changes in relative intensities and band shapes observed when excited with 632.8-nm light. All of the spectra in the image manifest band separations of 24–25 cm

, and so there is no indication of a single trilayer present in this particular MoS


	Figure 10: Raman image of few-layer MoS

 flake obtained with 632.8-nm excitation. Red corresponds to the substrate Si and the green intensities correspond to the spatially varying strength of the Raman band at approximately 380 cm

 flake using two excitation wavelengths and rendering the Raman images through the selection of different bands, a first-order mode and a resonantly enhanced second-order mode. The spatially varying intensity of the 

) band appears to correspond to varying few-layer MoS

 film thickness when excited at 532 nm. In contrast, the resonance Raman image of the same few-layer MoS

 flake reveals the spatially varying excitonic electronic structure. The Raman images are best viewed side-by-side in Figure 9. From this perspective, we are rendering images to enhance the contrast between off-resonance and resonance Raman scattering corresponding to the film thickness and excitonic electronic structure, respectively.


	Figure 11: Raman image of few-layer MoS

 flake obtained with 532-nm excitation. Red corresponds to the substrate Si and the green intensities correspond to the spatially varying strength of the Raman band at approximately 380 cm

	Of course, another way to render off- and in-resonance Raman images of the few-layer MoS

 flakes is to view them through the bracketing of the same Raman band. A collection of data from a few-layer MoS

 flake is shown in Figure 10. The Raman data were acquired using 632.8-nm excitation and a 100× Olympus objective and by moving the stage in 500-nm increments. The surrounding Si in red is not rectangular because the mapping was performed to follow an irregular shape; this approach allows one to maximize mapping efficiency by analyzing the object of interest without needlessly analyzing the surrounding area. Here the mapping was performed in resonance with 632.8-nm excitation and the 

) band was bracketed to render the Raman image in the lower left corner of Figure 10. The single Raman image reveals the spatially varying signal strength of the 

 band; however, it does not reveal the spatially varying complexity of band structure and relative strengths of the first- and second-order modes. A portion of the spectrum corresponding to the cross hairs can be seen in the upper right hand corner. This same MoS

 flake was mapped with 532-nm excitation, and the entire hyperspectral data set is shown in Figure 11. Here, we render a Raman image by bracketing the same 

 and Si bands as in Figure 10. So now we have a direct comparison of the resonance and off-resonance Raman images of a few-layer MoS

 flake based on the same first-order Raman band. The cross hair has been placed in the exact same position of Figures 10 and 11 for a comparison of the resonance and off-resonance Raman spectra from the same location in the flake. The spectral differences are quite striking. Even when bracketing the same Raman bands, the resonance and off-resonance Raman images that are rendered are sufficiently different to reveal the spatial heterogeneity of the few-layer MoS


	Figure 12: Raman image of few-layer MoS

. Note the strain components in the upper-right-hand corner spectrum.

	Thus far, the 532-nm excited spectra that we examined have consisted of bands that are fairly sharp and symmetric. However, with sufficiently high spectral resolution one can observe band broadening and asymmetry, thereby indicating strain, at certain locations within a few-layer MoS

 flake and even throughout an entire flake. A Raman image obtained with 532-nm excitation is shown in Figure 12. The Raman data were acquired using a 100× Olympus objective and by moving the stage in 300-nm increments over an area of approximately 13 μm × 13 μm. The red brackets surround the Raman band of Si at 520 cm

 flake appears red. The green bracket covers the Raman scattering of the 

 rendered Raman image comports well with the contrast observed in the reflected white light image. There is a direct correlation of the shapes and locations of the contrasting regions in the reflected white light image with the Raman image. Nevertheless, the reflected light bright and dark regions do not necessarily correlate directly to strong Raman scattering at the same location on the flake. Sometimes, a dark area in the reflected white light image corresponds to strong Raman scattering as in the diagonal streak in the northwest corner of the flake. In addition, note the spectrum in the upper-right-hand corner associated with the cross hair location in the images. You see a strain component to the low energy side of both the 

) bands. Strain that can be clearly seen in the spectra from a Raman image does not reveal itself in the reflected white light image.


	Figure 13: Raman image of few-layer MoS

 flakes, strain can be observed in the Raman spectra along edges and even throughout the entire flake. A good example of this is shown in Figure 13. The Raman data were acquired using 532-nm excitation and a 100× Olympus objective and by moving the stage in 300-nm increments. The Raman spectrum at the cross hairs, near the center of the flake, consists of sharp symmetric bands. However, there is a substantial number of bands in the hyperspectral data set (upper-left-hand frame) that are broadened and shifted. Those "strain" spectra were found almost entirely at the upper-left edge of the large flake and throughout the small yellow flake in the Raman image. Raman spectra obtained from the center and edge of the large flake and the center of the small flake in the lower-left corner are shown in Figure 14. Note that both the 

 bands shift and broaden in the same direction to a lower wavenumber. Therefore, we can properly conclude that these spectral changes indicate strain rather than a reduction in the number of MoS


	Figure 14: Raman spectra of few-layer MoS

 from different locations on the flake shown in Figure 13. Note the substantial shifts and broadening corresponding to strain.

 few-layer flake in Figure 13 exhibited strain along a portion of the perimeter. An interesting fact is that very often the structure of the perimeter of few-layer MoS

 flakes is quite different from that of the interior, manifesting differences in both Raman scattering and photoluminescence. A collection of data from a few-layer MoS

 flake is shown in Figure 15. The Raman data were acquired using 532-nm excitation and a 100× Olympus objective and by moving the stage in 300-nm increments. A reflected white light image of the flake appears in the lower-right-hand corner and a Raman image corresponding to the reflected light image appears to its left. Here we have used the green brackets to highlight the strain free 

 band and the blue brackets to cover the lower wavenumber broadened side of the 

 band to reveal locations of strain in the flake. Note that the entire perimeter of the flake appears blue in the Raman image because of the presence of strain. It is noteworthy that strain that can be clearly seen in the spectra from a Raman image does not reveal itself in the reflected white light image.


	Figure 15: Raman image of few-layer MoS

 flake obtained with 532-nm excitation. Red corresponds to the substrate Si and the green intensities correspond to the spatially varying strength of the Raman band at centered at 408 cm

. The blue bracket to the low energy side of the 408 cm

 band reveals the strain component clearly seen at the perimeter of the flake in the Raman image.

	We find that the use of 532-nm excitation is most useful for identifying strain through the broadening and shifting of the 

 bands. However, the use of 632.8-nm excitation in resonance with the MoS

 A1 excitonic transition also reveals structural differences at the perimeter. A collection of data from a few-layer MoS

 flake is shown in Figure 16. The Raman data were acquired using 632.8-nm excitation and a 100× Olympus objective and by moving the stage in 200-nm increments over an area of approximately 8 μm × 8 μm. A reflected white light image of the flake appears in the lower-right-hand corner and a Raman image corresponding to the reflected light image appears to its left. The plot on the upper left consists of all of the Raman spectra, including a photoluminescence component, acquired over the image area. The upper-right-hand plot is of the single spectrum associated with the crosshairs location in the region showing photoluminescence. The Raman image is rendered using the green color brackets to capture the 

 band and the blue brackets surround the broad photoluminescence. Even in the resonance Raman image, the 

 band manifests a greater intensity at the perimeter than it does in the flake interior. Also, we see a light blue region in the upper-right corner of the flake corresponding to the emission centered at 678 nm (1050 cm

 Raman shift). This region appears light blue in the spectral image because there are contributions from 

 Raman scattering (green) and the 678-nm photoluminescence (blue) from the same locations. So, with the selection of the appropriate excitation wavelength, one can simultaneously perform resonance Raman and photoluminescence spectral imaging of few-layer MoS


	Figure 16: Raman image of few-layer MoS

 flake obtained with 632.8-nm excitation. Red corresponds to the substrate Si and the green intensities correspond to the spatially varying strength of the Raman band at centered at 408 cm

. The blue bracket captures the photoluminescence, which can be seen primarily in the northeastern quadrant of the flake in the Raman and photoluminescence image.

	We compared photoluminescence and Raman spectra of MoS

 taken at the center and edge of few-layer flakes and saw evidence for spatial variation of the solid-state and electronic structure. We applied Raman, resonance Raman, and photoluminescence hyperspectral imaging to observe spatially varying structural differences that are not revealed through reflected white light microscopy. The selection of the appropriate excitation wavelength allows one to simultaneously perform resonance Raman and photoluminescence spectral imaging of few-layer MoS

. That ability should allow materials scientists to better design and fabricate electronic and optoelectronic devices based on few-layer MoS

	(1) D. Jariwala, V.K. Sangwan, L.J. Lauhon, T.J. Marks, and M.C. Hersam,

	(4) H. Li, J. Wu, Z. Yin, and H. Zhang, 

	(5) H. Li, J. Wu, X. Huang, G. Lu, J. Yang, X. Lu, Q. Xiong, and H. Zhang, 

	(6) K.F. Mak, C. Lee, J. Hone, J. Shan, and T.F. Heinz, 

	(7) A. Splendiani, L. Sun, Y. Zhang, T. Li, J. Kim, C.-Y. Chim, G. Galli, and F. Wang, 

	(10) S. Jimenez Sandoval, D. Yang, R.F. Frindt, and J.C. Irwin, 

	(14) D.O. Dumcenco, K.Y. Chen, Y.P. Wang, Y.S. Huang, and K.K. Tiong, 

	(15) B.C. Windom, W.G. Sawyer, and D.W. Hahn, 

	(16) Y. Zhao, X. Luo, H. Li, J. Zhang, P.T. Araujo, C.K. Gan, J. Wu, H. Zhang, S.Y. Quek, M.S. Dresselhaus, and Q. Xiong, 

	(17) C. Lee, H. Yan L.E. Brus, T.F. Heinz, J. Hone and S. Ryu, ACS Nano 4 2695-2700 (2010).

	(18) K. Golasa, M. Grzeszczyk, R. Bozek, P. Leszczynski, A. Wysmolek, M. Potemski, and A. Babinski, S

	(19) B. Chakrabory, H.S.S. Ramakrishna Matte, A.K. Sood, and C.N.R. Rao, 

	(20) R. Coehoorn, C. Haas, and R.A. de Groot, 

	(21) G.L. Frey, R. Tenne, M.J. Matthews, M.S. Dresselhaus, and G. Dresselhaus, 

 is a Raman applications manager at Horiba Scientific, in Edison, New Jersey, where he works with Fran Adar. David is sharing authorship of this column with Fran. He can be reached at: 

Articles

Molecular Spectroscopy Workbench

 Resonance and off-resonance Raman spectroscopy and imaging are used to examine the spatial variation of the solid-state structure and electronic character of few-layer MoS2 flakes. Simultaneous acquisition of photoluminescence spectra with the Raman scattering provides complementary ways of rendering Raman and photoluminescence spectral images of thin-film MoS2.

Resonance Raman and Photoluminescence Spectroscopy and Imaging of Few-Layer MoS2

	(Image Credit: Steve Allen/Getty Images)

 salary survey shows that the average salary has remained essentially static overall, which is welcome news after last year's drop. But not all sectors and groups are doing better.
	

	It can be hard to gauge whether or not your employment conditions are fair, and whether you are earning as much as you deserve - or as much as your colleagues do, be they down the hall or across the country.

 salary survey helps answer those questions, and also provides an overall sense of the job market for spectroscopists. This year, the top-level result is fairly positive, but the benefits are not enjoyed by everyone. And a few details hint at some additional underlying concerns and trends.

Salaries Stable Overall, but Not for Everyone

	This year, our survey results show overall stability in the spectroscopy job market, with a very slight increase (0.9%) in the average salary overall, from $81,990 in 2014 to $82,733 in 2015. Following last year's 7.3% drop (Table I), this year's small uptick is welcome news. The improvement, however, was not shared evenly across all segments and groups (Table II).

	For example, even though the highest average salary is found in industry (58% of our sample), this year's average of $85,219 is down 6.4% from last year. In academia (26% of our sample), average salaries are also down - 6.9% lower than 2014. On the other hand, salaries for spectroscopists working at government or nationally funded laboratories (15% of our survey) went up 5.8%, but the average for this group ($79,741), is still a chunk lower than in industry.

	We also see uneven results when we look at different age groups and education levels. As expected, respondents holding a PhD degree earn the highest average salary, but this group showed a disturbing drop in average salary between 2014 and 2015- from $97,936 to $88,724 (down 9.4%). This decrease may be partly explained by an increase in our sample of younger respondents 

31% of our respondents this year are under 40, compared to only 20% last year

which likely includes many newly minted PhDs. But the idea that spectroscopists with more experience earn more did not hold true across all age groups. Respondents in the age range of 50–60 earned the highest salary, and showed a whopping 11% increase in average salary, from $86,003 to $95,161. But those over 60 saw a significant drop, from $100,117 to $90,209 (down 9.9%), putting their average below that of the 50–60 age group.

	There is also a great deal of variation in the average salaries according to job role, as shown in Table IV. The best-paid job role is technical director (2.7% of respondents) with an average salary of $122,134. Laboratory directors and managers, who represent the largest proportion of respondents (17.5%) also do quite well, with an average salary of $87,006. The lowest-paid job role is post-doctoral researcher (2% of respondents) with an average salary of $32,005.

More Part-Time Workers and More Post-Docs

	In addition to the fact that not all sectors and groups benefited from this year's average stabilization of salaries, a few other disturbing trends appear in our results (Table III). First, full-time employment has decreased over the past three years, from 92.8% in 2012 to 86.5% in 2015 (or from 94.7% to 88.9%, if we include full-time contractors). Second, in academia, the proportion of post-doctoral researchers and graduate students has increased from 0.3% in 2012 to 4.5% in 2015. Third, unemployment, although still low among our respondents, has tripled from 0.6% in 2012 to 1.9% in 2015. In addition, the gender gap among our respondents has widened and is discussed in further detail later.

	The other negative trend seen in our survey - though by no means new, nor unique to spectroscopy - is increasing workloads. The majority of respondents (64%) reported that workloads have increased over the past year (Figure 1), mainly because of increased business, new equipment, or staffing cuts, or a combination of those factors. A handful said the main cause was "increased responsibilities," though it was unclear whether those new responsibilities included a promotion or a raise. Quite a few academics said they have to teach larger or more classes than in the past - in some cases because of administrative goals to make faculty "more productive" - and some mentioned that they are doing more research with the same funding, meaning they cannot afford to hire as many post-docs to share the workload.


	Figure 1: The most common reasons for increased workloads.

Satisfied, Nevertheless - Though a Bit Restless

	Surprisingly, although 64% of respondents reported a heavier workload, only 53% said their stress levels had increased. And despite the increased workloads and the other concerns revealed by the survey, respondents remain happy overall. More than three-quarters of respondents report that they are either satisfied (58%) or very satisfied (18%) with their jobs.

	Nevertheless, our data show that spectroscopists are changing jobs more than in the past, as is generally true across the economy. One-third of our respondents are planning to look for a new job over the next year. The drivers for this job seeking are a mix of dissatisfaction and ambition. Of those planning to look for a new job, 59% are seeking a higher salary, and 53% are looking for a new challenge. Several other reasons were given by those seeking new jobs:

		Desire for a change in employer type, 21%

	On the flip side, some respondents are staying put despite job dissatisfaction. Common reasons included positive factors such as good colleagues, hours, and salary. A few mentioned negative reasons, such as fears of facing age discrimination if they try to find another job. A handful confessed that they are hanging in there until they retire.

	The location of a job is another key factor when deciding whether it is time to move on. Many of our respondents are staying in jobs they are not satisfied with because the location is convenient. And although these results suggest that many people would not readily make a big geographic move to improve their employment situation, there are some important regional variations in average salaries and job opportunities across and outside the United States, as seen below and in Table V.

	In the United States, the Northeast (22% of respondents) has held its top salary position for the second year in a row, albeit with a dip in average salary from $106,318 in 2014 to $103,504 in 2015. The Southwest (16% of respondents) remains in second place for salaries, but the average in this region has decreased for several years, falling 8.5% in the last year, from $90,271 to $82,546.

	The Northwest (5% of respondents) has risen to the third position, with an average salary reported of $79,027 - an 8% increase compared to 2014 ($73,510). The Southeast (18% of respondents) has fallen to fourth place, with a reported average salary of $77,381 (down from $82,175 in 2014). The Midwest (20% of respondents) is in last place, with an average salary of $77,193 in 2015 (up from $73,510 in 2014).

	A fifth of respondents (20%) were from locations outside the United States, primarily in Europe (Denmark, France, Germany, Spain, and the UK), Australia and New Zealand, China, India, Japan, and Mexico. The average salary of these respondents - $61,006 - is markedly lower than in the United States. Multiple factors are responsible for this disparity, including cost of living and the variations in how health care and other core services are provided.

	There is still a persistent gap in the average salaries of men and women. This year, the average salary for male respondents is $83,199 (down 0.2% from 2014), compared to $74,584 for women (down 6%). In 2014, the difference in average salary for men and women was $3995; this has more than doubled, to $8615.


	Figure 2: Respondents in positions of management. Inside circle: Female respondents. Outside circle: Male respondents.

	Men also continue to dominate management, with more than half of male respondents (52%) in a position of management compared to only 35% of female respondents (Figure 2). And male managers make more than female managers - $86,616 for men versus $76,926 for women (Table VI).

	Figure 3 shows that 40% of male respondents hold a PhD compared to 34% of female respondents; however, male respondents with a PhD have an average salary of $94,108 compared to $75,652 for female PhD holders (Table VI). A higher level of education doesn't seem to have the same impact for women as it does for men.


	Figure 3: Highest education level attained according to gender.

	The biggest pay gap between the sexes is seen in academia, where the average salary is $67,918 for men and $50,428 for women. In government and nationally funded laboratories, average salaries are $84,694 and $76,926, respectively. Industry is the one sector where the average salary for women ($89,172) exceeds that of men ($85,877).

	Despite the only very modest increase in average salary overall, the salary drops for many groups, and the uncertainty of the employment market, half of respondents would recommend their career to a recent graduate, and only 10% would not. Interestingly, however, 40% of our participants responded "maybe" to this question. Given the high levels of job satisfaction expressed by our respondents, this ambivalence about recommending this path to young people may reflect concerns about the future of spectroscopy careers.

	We also asked respondents for the best career advice that they had received. Many common and expected themes emerged, such as "work hard," "be honest," and "love what you do." A selection of other responses appears in a box on below.

	This year's survey shows a modest overall improvement in average salaries, but in many sectors and groups, the averages actually went down. In addition, the percentage of full-time workers decreased, workloads went up, and the pay gap between men and women widened.

	Yet in spite of it all, job satisfaction levels among spectroscopists remain high, perhaps reflecting a repeating refrain in the career advice shared by our respondents: "Love what you do."

 is an Associate Editor, based in Chester, UK. Direct correspondence to 

The latest Spectroscopy salary survey shows that the average salary has remained essentially static overall, which is welcome news after last year's drop. But not all sectors and groups are doing better.

2015 Salary Survey: Modest Gains, but Not for Everyone

This installment evaluates the application requirements for the determination of 15 elemental impurities in pharmaceutical materials as described in the new 

) Chapters <232> and <233> and offers suggestions about which atomic spectroscopic technique might be the most suitable to use.

	Since the introduction of the first commercially available atomic absorption (AA) spectrophotometer in the early 1960s, there has been an increasing demand for better, faster, easier-to-use, and more flexible trace-element instrumentation. A conservative estimate shows that in 2015, the market for atomic spectroscopy instruments will be close to $1 billion in annual revenue - and that doesn't include aftermarket sales and service costs, which could increase this number by an additional $400–500 million. As a result of this growth, we have seen a rapid emergence of more-sophisticated equipment and easier-to-use software. Moreover, with an increase in the number of manufacturers of atomic spectroscopy instrumentation and its sampling accessories, together with the availability of newer techniques like microwave-induced plasma optical emission spectroscopy (MIP-OES) and laser-induced breakdown spectroscopy (LIBS), the choice of which technique to use is often unclear. It also becomes even more complicated when budgetary restrictions allow only one analytical technique to be purchased to solve a particular application problem.

	To select the best technique for a particular analytical problem, it is important to understand exactly what the problem is and how it is going to be solved. For example, if the requirement is to monitor copper at percentage levels in a copper plating bath and it is only going to be done once per shift, flame atomic absorption would adequately fill this role. Alternatively, when selecting an instrument to determine 15 elemental impurities in pharmaceutical products according to 

) Chapters <232> and <233>, this application is probably better suited for a multielement technique such as inductively coupled plasma–optical emission spectrometry (ICP-OES) or perhaps inductively coupled plasma–mass spectrometry (ICP-MS), depending on the materials being analyzed.

	So, when choosing a technique it is important to understand not only the application problem, but also the strengths and weaknesses of the technology being applied to solve the problem. However, there are many overlapping areas between the major atomic spectroscopy techniques, so it is highly likely that for some applications, more than one technique would be suitable. For that reason, it is important to go through a carefully thought out evaluation process when selecting a piece of equipment.

	The main intent of this installment is to look at the application requirements for the determination of 15 elemental impurities in pharmaceutical materials as described in the new 

 Chapters <232> and <233> (1,2). The goal is to offer some insight as to which might be the most suitable atomic spectroscopic approach, particularly for new users who do not have a strong background or training in the real-world application of these techniques. We will not discuss either the MIP-OES or LIBS techniques in this evaluation. They are both very useful techniques for many sample types and have been described at length in the open literature (3). However, for the purpose of space and clarity, we will be focusing on the four most commonly used atomic spectroscopy techniques: flame atomic absorption (FAA), electrothermal atomization (ETA), ICP-OES, and ICP-MS.

	To set the stage, let's first take a brief look at their fundamental principles. Simplified schematics of each of these techniques are shown in Figure 1.


	Figure 1: Simplified schematic of AA, ICP-OES, and ICP-MS.

	This is predominantly a single-element technique for the analysis of liquid samples that uses a flame to generate ground-state atoms. The sample is aspirated into the flame via a nebulizer and a spray chamber. The ground-state atoms of the sample absorb light of a particular wavelength, either from an element-specific, hollow cathode lamp or a continuum source lamp. The amount of light absorbed is measured by a monochromator (optical system) and detected by a photomultiplier tube or solid-state detector, which converts the photons into an electrical signal. As in all atomic spectroscopy techniques, this signal is used to determine the concentration of that element in the sample, by comparing it to calibration or reference standards.

	FAA typically uses a liquid sample flow of about 2–5 mL/min and is capable of handling in excess of 10% total dissolved solids, although for optimum performance it is best to keep the solids down below 2%. For the majority of elements, its detection capability is 1–100 ppb with an analytical range up to 10–1000 ppm, depending on the absorption wavelength used. However, it is not really suitable for the determination of the halogens, nonmetals like carbon, sulfur, and phosphorus, and has very poor detection limits for the refractory, rare earth, and transuranic elements. Sample throughput for 15 elements per sample is typically 10 samples per hour.

	This is also mainly a single-element technique, although multielement instrumentation is now available. It works on the same principle as FAA, except that the flame is replaced by a small heated tungsten filament or graphite tube. The other major difference is that in ETA, a very small sample (typically, 50 μL) is injected onto the filament or into the tube, and not aspirated via a nebulizer and a spray chamber. Because the ground-state atoms are concentrated in a smaller area, more absorption takes place. The result is that ETA offers detection capability at the 0.01–1 ppb level, with an analytical range up to 10–100 ppb.

This installment evaluates the application requirements for the determination of 15 elemental impurities in pharmaceutical materials as described in the new United States Pharmacopeia (USP) Chapters  and  and offers suggestions about which atomic spectroscopic technique might be the most suitable to use.

Spectroscopy-03-01-2015

Resonance Raman and Photoluminescence Spectroscopy and Imaging of Few-Layer MoS2

2015 Salary Survey: Modest Gains, but Not for Everyone

Determining Elemental Impurities in Pharmaceutical Materials: How to Choose the Right Technique

Trends in Spectroscopy: A Big-Picture Look at Recent Innovations and Future Directions

App of the Month: 1HCSC

Vol 30 No 3 Spectroscopy March 2015 Regular Issue PDF