The need for new generation high power electronic devices capable of operating at extreme conditions has made silicon carbide (SiC) extensively investigated over the past two decades because its properties propose it as an extremely viable solution1. These properties include a higher breakdown field than that of silicon which permits much smaller drift regions, a higher thermal conductivity, which allows better heat dissipation and a wide band gap energy which enables higher operating temperatures, making SiC suitable for high-temperature, high-frequency and high-power electronic devices. In addition, its strong bonding and large optical phonon energies make SiC attractive also for applications in nonlinear optics, especially for ones that deal with short wavelength high optical power devices or high operating temperatures2.

Due to the one-dimensional polymorphism, SiC exists in a large variety of crystal structures known as polytypes which correspond to a unique stacking sequence of n successive Si-C bilayers. In such arrangements every bilayer can be viewed as a two-dimensional hexagonal lattice of vertical Si-C bonds. The large variety of polytypes implies a corresponding variety of properties such as the band gap energy. For example, SiC polytypes have a range of indirect band gaps3 from 2.39 eV in the cubic polytype (3C) to 3.33 eV in one of the hexagonal polytypes (2H). To date, the main focus of attention has been mainly drawn by 3C-SiC4,5 and two hexagonal forms, 4H and 6H-SiC6,7. In the case of 3C-SiC, this is because it has isotropic properties and it is the only SiC polytype that can be grown on a Si substrate.

Despite the progress made in SiC crystal growth techniques, the widespread application of SiC-based power devices is limited by the presence in both SiC bulk substrates and in the overgrown SiC epilayers of a high density of structural defects such as micropipes8, screw dislocations, basal plane dislocations and stacking faults9,10. In the case of 3C-SiC the most common defects are stacking faults (SFs), which are local regions of incorrect stacking of crystal planes, twin boundaries (TBs) which occur due to the twofold possibility to arrange the Si–C bilayer stacking along the c-axis on a (0001) hexagonal SiC substrate and double-positioning boundaries (DPBs) which are the dominant defects in 3C-SiC films with (111) orientation grown on a hexagonal substrate and form due to a 60° rotation of the initial 3C stacking on the (0001) hexagonal plane11,12. It has been shown that such extended defects are detrimental to the behavior of semiconductor devices. The understanding of their origin is thus crucial for a broad range of SiC-based applications and requires simple and readily available defect characterization methods for imaging defects present in the substrate or in the epilayer.

Various investigation techniques have been used to image and characterize different structural defects in SiC layers and devices, each technique having its own limitations. The optical beam induced current technique13, which is only suitable for device characterization was used to analyze n-type 4H-SiC nickel Schottky barrier diodes. The imaging of extended SFs in degraded SiC PiN diodes14 and the mapping of different defects in SiC15 was done by synchrotron white-beam X-ray topography. Although it is nondestructive in nature, the requirement of a synchrotron light source limits its use for routine characterization. X-ray diffraction (XRD) can yield information on polytypes and dislocations through indexing and peak-broadening analysis16, but achieving a spatial mapping capability is time-consuming. Transmission electron microscopy (TEM) can be used for the identification and evolution analysis of extended defects in 3C-SiC films17,18, but requires destructive sample preparation and severely limits the area of analysis to tens of microns or less. Scanning tunneling microscopy and atomic force microscopy (AFM) revealed atomic-scale images of 3C-SiC and structural defects of the films, respectively19, but are limited to surface studies. Among the investigation techniques used for defect characterization in SiC, nondestructive luminescence-based techniques helped to understand how defects in SiC are formed. The photoluminescence intensity mapping was used to identify three kinds of SFs in 4H-SiC epilayers20, while cathodoluminescence and lifetime mapping were used to identify the stressed diodes after extended forward current operation21. Electroluminescence was used for imaging luminescent defects in PiN diodes in situ, during forward biasing and ex situ, after degradation of the diodes22,23. Another optical technique, second harmonic generation (SHG) is known to be a rapid, noninvasive and sensitive investigation method used to identify different polytypes24 and structural defects in SiC epilayers25, to map microcrystalline inclusions of SiC polytypes in 6H-SiC epilayers26 and to probe the crystalline order of 3C-SiC films grown on (111) Si substrates by rotational anisotropy measurements27.

In the present work we propose SHG microscopy for the fast detection and identification of defects in SiC epilayers grown on hexagonal silicon carbide by the vapour-liquid-solid technique. By combining the SHG-based imaging with XRD and SHG rotational anisotropy the growth of 3C polytype on the 4H-SiC substrate was confirmed and the polytype of the imaged defects was identified. If in previous publications24,25,26,27 optical SHG and especially SHG rotational anisotropy were used to detect polytype inclusions, identify different SiC polytypes, or determine crystalline properties for SiC film surfaces, in the present work we emphasize on its potential in high resolution imaging and in structural defect identification. In the same time, because defects detection and identification in SiC require high resolution surface characterization, AFM combined with SHG microscopy were employed so as to provide the topography of the samples needed to locate the defects imaged by SHG.


XRD has been used to confirm the crystallinity and epitaxial quality of the SiC film (Fig. 1). The structure of the film was 3C-SiC as indicated by strong cubic (111) and (222) diffraction peaks at 35.6° and 75.3° respectively, confirming the growth of the 3C-SiC polytype. Since in the Bragg-Brentano configuration the diffraction data collected at each angle corresponds only to the structural planes parallel to the sample surface28, fig. 1 confirms that (111) 3C-SiC is parallel with the surface and hence with the (0001) 4H-SiC substrate on which the 3C-SiC was grown. It can thus be concluded that the 3C-SiC layer was grown epitaxially with preferred orientation and highly aligned with the substrate. This is actually consistent with the two-dimensional nucleation mechanism for the 3C polytype growth.

Figure 1
figure 1

X-ray diffractogram for the 3C-SiC epilayer.

The labeled peaks correspond to the cubic SiC crystal structure. No other foreign polytype-related reflection was observed in the XRD scans.

In order to evaluate the potential of SHG-based microscopy to image defects in SiC epilayers we simultaneously collected confocal reflection (Fig. 2a) and SHG images (Fig. 2b) from a 3C-SiC film. Fig. 2b displays a xy scan of a 250 × 250 μm2 surface area, which contains an unusual number of defects; this region was intentionally selected to show different defects in a single SHG image. Visible defects, which can be seen in the reflection image often indicate but do not guarantee the presence of structural defects in the epilayer, making the detection of all defects in the epilayer nearly impossible. SHG imaging reveals elements that cannot be distinguished in the reflection image, which demonstrates the complementarity of the two imaging methods and propose SHG microscopy as a suitable technique to image not only superficial, but also buried structural defects.

Figure 2
figure 2

Laser scanning microscopy images of 3C-SiC epilayer.

(a) confocal reflection image; (b) SHG image. Both images are simultaneously collected from the same surface region. The three insets for both confocal and SHG images confirm that a surface inspection using confocal microscopy alone is not enough because some defects revealed in the SHG image cannot be observed in the reflection image.

The potential of SHG microscopy stretches beyond simple imaging as the polarization of the SHG radiation can be exploited to obtain additional information about the crystalline structure of the sample26. In this regard, we have used SHG rotational anisotropy to identify the polytype for both defects and SiC epilayer. To determine the second harmonic (SH) rotational anisotropy of the samples illustrated in Fig. 2, xy scans were performed at different polarization angles for the incident laser (Fig. 3).

Figure 3
figure 3

SH rotational anisotropy.

By changing the polarization of the incident laser radiation, different SHG intensities are obtained. The corresponding SHG images at different polarization angles are displayed: (a) 50°; (b) 100°. The SHG intensity plot (c) is valid for both the epilayer and the defects. The plot for the epilayer intensity dependence with azimuthal angle is not shown due to lower intensity compared with that of the defects. The average SHG intensity from the defects is 10 times higher than the SHG intensity from the SiC epilayer. Having the same rotational symmetry, both epilayer and defects are identified as 3C-SiC.

In fig. 3c a dominant fourfold rotational symmetry and an additional twofold symmetry are visible. In the case of bulk 3C-SiC, a twofold rotational symmetry appears in components of the nonlinear susceptibility tensor. According to Neumann's principle, this twofold symmetry shows up also in the SH electric field, resulting in a fourfold symmetry in the SHG intensity24. The 3C-SiC surface possesses a fourfold rotational crystal symmetry, which cannot be displayed in the SH rotational anisotropy due to the fact that the susceptibility tensor, being a third rank tensor, can only resolve symmetries up to threefold26. Thus, the contributions from the surface to the SHG intensity are purely isotropic when changing the laser polarization. The case of hexagonal SiC should also be taken into consideration because the 3C-SiC which is of interest in our case is grown on a 4H-SiC substrate. The surfaces of the hexagonal SiC polytypes that are cut parallel to the hexagonal SiC double layers, possess an overall sixfold symmetry which again gives rise only to an isotropic SH response. Only in the case of a misorientation of the surface away from the hexagonal planes an additional onefold contribution from the hexagonal SiC substrate will be introduced in the SH response.

In our case, the dominant fourfold symmetry confirms the presence of 3C-SiC on the sample, while the twofold symmetry is originating from an isotropic contribution either from the 3C-SiC surface or 4H-SiC substrate. The fourfold symmetry dominates the twofold symmetry, because bulk contributions tend to dominate the surface contributions due to the longer interaction length in the case of bulk when compared to surfaces consisting of only a few atomic layers26.

Both defects and the epilayer have the same fourfold symmetry, with the intensity of the epilayer ten times lower compared to the case of defects. This difference in intensity is an advantage that can be used when imaging a SiC sample to identify defects from the lower-intensity epilayer.

The differences between the experimental data and the theory may be explained by quantitatively analyzing the case when the laser polarization is fixed (Fig. 4). The SH radiation is expected to have the polarization parallel with that of the incident laser beam and therefore by rotating the analyzer before the microscope detector a cos2θ pattern should result. Successive SHG images of the SiC sample were recorded rotating the analyzer by 360° with 10° steps. SHG intensity as a function of the analyzer rotation angle (θ) is well consistent with the prediction based on a cos2θ pattern (Fig. 4). The deviation of the experimental data from the theoretical curve can be explained by the depolarization effects of galvanometric mirrors, the imaging objective and the transmission through the sample.

Figure 4
figure 4

The dependence of the SHG intensity on the rotation angle of the analyzer.

By rotating the analyzer before the detector with 10° steps, different intensities for the SHG are obtained. The experimental data fits well with the theoretical cos2θ function.

To further investigate the structure of the defects in the SiC films, SHG images and AFM topography in tapping mode were collected for the same areas on the sample surface (Fig. 5). Usually, morphological defects extend deep in the epilayer but form characteristic features on the surface. These surface features with distinct shapes can be easily identified when using optical microscopy imaging and can be linked with the corresponding defects that are buried below the sample surface. Triangular features are clearly discernible by confocal microscopy (Fig. 5a,d) and AFM (Fig. 5c,f), while SHG microscopy offers more details about the defects generating these features (Fig. 5b,e). SHG images confirm the extension of the surface features into the epilayer. Triangular features in the confocal image (Fig. 5a) prove to be hexagonal in the SHG image with a buried central defect originating from DPBs (Fig. 5b). The distinct equilateral triangles are suspected to arise from partial dislocations bounding triangular stacking faults that have been previously resolved by using AFM on as-grown 3C-SiC films29 and by using electron channeling contrast imaging30. Another source of the triangular features can be the existence of DPBs which preferentially expand along the (111) planes. Since the inclined (111) planes intersect the top (111) surface along <110> directions, the defects appear as equilateral triangles at the surface (Fig. 5) as also shown previously31.

Figure 5
figure 5

Combining AFM and laser scanning microscopy.

Confocal (a), (d), SHG (b), (e) and AFM (c), (f) images of different defects in the 3C-SiC epilayer. Because the optical and AFM images are not perfectly registered but the scan area has the same size, confocal images are used as a reference to locate different surface features. Triangular features on the surface can be identified in SHG images as either DPBs (b) or triangular SFs (e).

Other defects visible by using SHG microscopy were also hexagonal-shaped. The size of such defects exceeds the usual AFM field-of-view dimensions and were only investigated using SHG (Fig. 6). The hexagonal defects visible in the SHG image are DPBs which appear due to the growth of the 3C polytype on a hexagonal substrate. Similar hexagonal features were observed under an optical microscope32 after potassium hydroxide (KOH) etching of the SiC sample.

Figure 6
figure 6

Hexagonal defects in SHG microscopy.

SHG image of hexagonal defects associated to DPBs which appeared due to the growth of cubic polytype on a hexagonal substrate.


The difference between the lattice parameters of the film and those of the substrate produces a lattice mismatch strain in the epilayer, with high influence near the film surface when imaging films of reduced thickness. This relative displacement of the elements causes local stress in the crystal lattice, that along with dislocations and stacking faults produces micro-strain across the film thickness, resulting in strain-induced SHG.

The presented experiment shows that SHG can be regarded as a viable characterization tool of 3C-SiC layers. The ten times difference in SHG intensity between defects and background allows their rapid identification and a fast data acquisition, rotation of the sample being no longer necessary for finding the best sample position for optimal SHG intensity. SHG enables determining the SiC polytype by rotational anisotropy studies that can be performed by rotating the polarization angle of the incident laser. Full SiC wafers can also be imaged even with a relatively reduced field-of-view, by using automated stitching algorithms33 to create an image of the entire wafer after acquiring several SHG images to cover the whole wafer.

The nondestructive defect mapping method based on the principle of optical second harmonic generation that we introduced enables the rapid and in-depth identification of structural defects such as stacking faults, dislocations and double positioning boundaries in SiC epilyers. A major advantage of the SHG-based defect assessment method is that it can be extended to characterize any structural defects in SiC epilayers of different polytypes. Unlike TEM, X-ray topography or KOH etching, SHG-based techniques are nondestructive and require minimal sample preparation. Since the proposed SHG imaging technique is noninvasive and rapid, it can be used in situ in a production line to provide rapid feedback to processing engineers by highlighting areas within the SiC epilayer with structural defects. Such tasks could be performed also in automated scenarios, speeding the inspection process even more, by combining SHG microscopy with computer vision methods18 that can replicate analysis and inspection chores performed by human operators.


(111) 3C-SiC layers were grown by vapour-liquid-solid mechanism12,34 in a homemade vertical cold-wall quartz reactor using silicon-gallium (Si-Ga) melts on commercial on-axis (0001) 4H-SiC wafers. The surface pretreatment included a slight hydrogen etching at 1450°C followed by a deposition of a Si layer on the seed at 1000°C. A 13 SiGa composition was used for the melt, adjusting the growth temperature just above the melting point of the Si-Ga alloy at 1200°C, in order to limit Ga loss by evaporation. The growth time was 30 minutes, resulting a film with a thickness of 900 nm. More details about the fabrication procedure are given in previous publications35,36.

The XRD investigations were performed with a APD2000 (GNR Analytical Instruments Group, Italy) diffractometer in a Bragg-Brentano configuration.

SHG imaging was carried out by using a 50 mW laser beam, with 80 fs pulses at 790 nm provided by a Spectra-Physics (now Newport Corporation, USA) Ti:Sapphire laser with a pulse repetition rate of 80 MHz. A Leica TCS SP confocal laser scanning microscope with a 40×, 0.7 NA objective, was used to simultaneously acquire both confocal reflection and SHG images in a transmission configuration. A bandpass filter with the central wavelength at 390 nm (FB390-10, Thorlabs) placed before the detector ensured that only the SHG signal was detected. The SHG rotational anisotropy was measured by obtaining images through the rotation of an achromatic half-wave plate (AHWP05M-980, Thorlabs) placed before the microscope while maintaining a fixed position for the analyzer in front of the detector. Because both 3C- and 4H-SiC are wide band gap materials, with 2.39 eV and 3.26 eV band gap energies, the wavelength used in our experiment (790 nm) and the corresponding wavelength of the SHG (395 nm) allowed for investigation in the volume of the sample and in a transmission configuration because no absorbtion of either the incident beam, nor the SHG ocurred in the sample.

A Q-Scope 350 AFM (Quesant, USA) was used for AFM imaging in tapping mode.