AlisonConstantin

Abstract

In this project, we apply the theoretical learnings from class in laboratory experiments. Our goal was to identify the material of two chips that looked identical to our eyes. Using a 532-nm laser, a grating spectrometer, and a CCD, we were able to perform Raman spectroscopy to identify the materials. Lastly, we thought of a way how we could do the same experiment at significantly smaller financial cost.

Introduction

Figure 1: Photo of the two chips we are trying to identify. One is diamond and the other is 4H-SiC.

Materials characterization is an essential part of the development of any device, from the nanoscopic to macroscopic and from prototyping to mass-production. Various techniques facilitate investigations of material composition and to troubleshoot problems caused by effects such as strain. In our case, we recently found two unmarked transparent chips (Figure 1). Our lab works with two different materials that can both be clear: diamond and 4H-SiC. Thus, we set out to identify which material comprised each chip.

Many different methods can be used to distinguish diamond from 4H-SiC, from hardness testing to X-ray diffraction. We opted to perform Raman spectroscopy as it is a non-invasive, low-energy characterization technique. As we discuss in further detail below, diamond and 4H-SiC both have large enough Raman shifts that we expect to be able to discern the Raman scattering spectral lines from the excitation laser with the appropriate spectral filters and a commercial spectrometer.

Background

Figure 2: Previously reported Raman shifts for diamond and 4H-SiC. (a) Spectrum showing the main Raman line for diamond at 1332 cm^-1. Plot from Ref. [2]. (b) Two main Raman lines for 4H-SiC are located at 777 cm^-1 and 967 cm^-1. Plot from Ref. [3].

Raman scattering is a specific kind of scattering in which a photon incident upon a material gains or loses energy when scattered. These two processes are respectively known as anti-Stokes and Stokes Raman scattering. For Raman scattering with crystals, the energy is gained from or lost to vibrational modes in the material. This effect leads to peaks in the spectrum of scattered light corresponding to different vibrations present in the crystal [1]. Each crystal has its own unique Raman spectrum.

Here we focus on the Raman lines resulting from Stokes scattering, where energy is lost. Raman shifts are typically reported in terms of wavenumbers in units of cm^-1. When multiplied by Planck's constant $h$ and the speed of light $c$, the Raman shift in cm^-1 can be converted into the energy change experienced by the photon. Recalling that the energy of a photon of wavelength $\lambda$ is ${\frac {hc}{\lambda }}$, we can find that the wavelength at which we would expect to see Raman emission ($\lambda _{Raman}$) is

$\lambda _{Raman}=\left({\frac {1}{\lambda _{excite}}}-\Delta _{Raman}\right)^{-1}$

where $\lambda _{excite}$ is the wavelength of the excitation light incident on the material and $\Delta _{Raman}$ is the Raman shift wavenumber.

Figure 2 shows the previously reported Raman spectra of 4H-SiC and diamond. 4H-SiC has two major lines around 777 cm^-1 and 967 cm^-1. The main Raman shift associate with diamond is 1332 cm^-1. From this information and the equation above, we can compute where we expect Raman spectral lines to appear on our spectrometer, when we excite our samples with $\lambda _{excite}=532$ nm. The expected Raman lines are written in Table 1.

Table 1: Expected wavelengths of Raman scattering lines for diamond and 4H-SiC, with 532-nm excitation. Raman shifts from Refs. [2] and [3].

Material	$\Delta _{Raman}$ (cm-1)	Raman line (nm)
Diamond	1332	573
4H-SiC	777, 967	555, 561

Methods

Figure 3: Schematic diagram of the setup used to guide the excitation light (in green) to the sample and collect the emitted and scattered light (in red). 532-nm excitation light passes through a 532/3 bandpass filter, is transmitted through a beamsplitter (blue square), and is directed to an objective. The objective lens focuses the excitation light onto the sample. The emitted and scattered light is then collected by the objective lens, reflected by the beamsplitter, passed through a 532-notch filter and 550 long-pass filter, and collected into an optical fiber. The light guided by this collection fiber can be outcoupled to a detector of choice (e.g., a spectrometer).

Figure 4: Schematic illustration of a spectrometer (top) with diffraction grating (lower right) and a CCD (lower left). Figure adapted from Refs. [4] and [5].

Excitation and detection

To measure the Raman spectra, we excited our sample with 532-nm laser light. We spectrally filter the excitation light with a bandpass filter centered around 532 nm with a width of 3 nm (532/3). The green excitation light is directed to and focused on the sample of interest using a confocal setup represented schematically in Figure 3. The scattered and emitted light from the sample is collected by the objective and directed toward a collection optical fiber. Before focusing into the optical fiber, the light passes through two chromatic filters: a 532-notch and a 550 long-pass filter. The 532-notch filter transmits all light except that in a narrow band around 532 nm. The 550 long-pass filter transmits light with a wavelength 550 nm. This allows us to pre-select what wavelengths we study with our detector of choice. The output of the detection fiber will be sent to a spectrometer, which we shall detail in the next section.

Spectrometer and CCD

From the fiber we send a collimated beam to the spectrometer (see Figure 4) and focus it onto the input slit of the spectrometer using a lens. From the first mirror, the light gets reflected onto a large parabolic mirror, then onto a grating, where diffraction spatially separates the light by wavelength. From the second parabolic mirror the light gets focused down onto the CCD, which is comprised of 2000 x 256 pixels. When the light reaches the chip of the CCD, the light is distributed by wavelength across a line. By integrating the number of detection events per pixel in a column on the CCD, we can infer the intensity of light as a function of wavelength. For wavelengths 500 to 950 nm, our CCD has a quantum efficiency of 50% to >90%, making it well-suited to our wavelength range of interest. At low temperatures the read noise is as low as 4 electrons and the dark current as low as 0.0006 electrons per pixel per second [4]. This means that even the weakest light sources, such as single-photon sources, can be detected with this CCD.

Results

Figure 5: Spectra collected for (a) chip 1 and (b) chip 2 with setup described in the Methods Section.

Figure 6: Broadband spectra acquired for (a) chip 1 and (b) chip 2. Chip 1 exhibits few spectral features in the studied wavelength band other than the two Raman lines. The spectrum for chip 2 reveals not only the characteristic diamond Raman line at 573 nm, but also strong emission from NV^- centers, a common color center in diamond.

The spectra collected with the setup described above in the Methods Section for chips 1 and 2 are shown in Figure 5. Chip 1 has two prominent lines around 555 nm and 561 nm, while chip 2 has just one prominent line around 573 nm. Referring back to Table 1, these results indicate that chip 1 is 4H-SiC and chip 2 is diamond.

We can further confirm the identity of chip 2 by examining the broadband spectra shown in Figure 6. For chip 1, there are no significant features aside from the previously observed Raman lines. However, for chip 2, there is a large spectral feature that starts with a peak around 630 nm and rises into a large bump around 700 nm. This emission turns out to be from nitrogen-vacancy (NV^-) centers in diamond [6], which are common defects in diamond. Therefore, we can conclude that chip 2 is undeniably diamond because the Raman line is characteristic of diamond and photoluminescence spectrum observed is that of a defect found in diamond.

Conclusions

Figure 7: Spectral response of cones and Raman lines of 4H-SiC (yellow) and diamond (cyan). LMS cones are represented in red, green, and blue respectively.

In summary, we were able to distinguish the two materials using Raman spectroscopy. Our measurements agreed well with results reported in literature. Furthermore, we have observed NV^- centers in diamond, a common diamond color center. Such color centers exist in many different materials and are very popular in the diamond jewelry business.

Figure 8: Human eye LMS cone responses for diamond (diamonds) and 4H-SiC (circles) Raman peaks. LMS cones are represented in red, green, and blue respectively.

We believe that we could simplify this experiment and reduce cost significantly. We could use a lower-cost green light source, such as an LED or a laser pointer, which should suffice as an excitation source with a laser-line filter. For detection, instead of using a spectrometer, we could use bandpass filters around the Raman lines of the different materials we are investigating and a CMOS detector.

Furthermore, it is possible to use a phone camera or the human eye to distinguish the two materials. Exciting the chips with a green laser as before and blocking the green light with a long-pass filter will allow us to distinguish the two chips from each other. We use ISETCam [7] and MATLAB to plot the spectral response of the human eye's cones together with the Raman lines of 4H-SiC (yellow) and diamond (cyan). In Figure 8 we then plot the resulting human eye LMS cone responses for the Raman peaks of diamond (diamonds) and 4H-SiC (circles). For the two Raman lines from 4H-SiC, the relative responses of the L and M cones are similar (within ~5%). For the Raman line from diamond, the L cone has a >20% larger response than the M cone. Therefore, with proper filtering of the excitation light, the human eye would be able to perceive a difference between the Raman-scattered light of 4H-SiC and diamond.

References

[1] W. Guozhen, Raman Spectroscopy: An Intensity Approach (World Scientific, 2017).

[2] S. A. Solin and A. K. Ramdas, Phys. Rev. B 1, 1687 (1970).

[3] M.-Y. Li et al., CrystEngComm 18, 3347 (2016).

[4] “Andor - Oxford Instruments, iDus 416 Series,” https://andor.oxinst.com/products/idus-spectroscopy-cameras/idus-416(2019), accessed: 2019-11-21.

[5] “Shamrock 750,” https://andor.oxinst.com/products/kymera-and-shamrock-spectrographs/shamrock-750(2019), accessed: 2019-11-21.

[6] R. Schirhagl, K. Chang, M. Loretz, and C. L. Degen, Annu. Rev. Phys. Chem. 65, 83 (2013).

[7] Joyce Farrell and Brian Wandell 2019, ISETCam, "https://github.com/ISET/isetcam".

Contributions

Alison and Constantin did the project together, experiment, data analysis, and write-up.