Microstructure and thermoelectric properties of CrN and CrN/Cr₂N thin films

The main growth parameters for the reactive sputter-deposition of chromium nitride were the substrate deposition temperature (kept constant for 1 h before initiating the growth process) and the argon/nitrogen gas flow mixture. In an initial study of deposition temperature, we observed that low temperature deposition at 550 ${}^\circ {\rm C}$ or lower will lead to the formation of polycrystalline films with high electrical resistivity, while deposition temperatures higher than 700 ${}^\circ {\rm C}$ typically results in predominant growth of Cr₂N over cubic CrN consistent with earlier observations [49]. Thus, we limit the presentation of experiments below to the most important temperature ranges between 600 ${}^\circ {\rm C}$ and 700 ${}^\circ {\rm C}$.

The gas flow mixture of argon and nitrogen also affected the films. A 60% argon and 40% nitrogen gas flow mixture resulted in sharp CrN (1 1 1) XRD peaks. However, an 80% argon and 20% nitrogen gas flow mixture led to the formation of the Cr₂N phase. Thus, the 20% nitrogen gas mixture (11 sccm N₂, 43 sccm Ar) is used to study phase mixtures of CrN and Cr₂N while as the 40% nitrogen gas mixture (21 sccm N₂, 33 sccm Ar) results in single phase CrN. The aim of this research is to study the thermoelectric properties of chromium nitride showing this phase variation: CrN, Cr₂N and CrN/Cr₂N.

Figure 1(a) shows $\theta -2\theta$ diffractograms for a set of CrN thin films deposited at 600 ${}^\circ {\rm C}$, 650 ${}^\circ {\rm C}$, and 700 ${}^\circ {\rm C}$ under the conditions of a 40% nitrogen gas mixture and a grounded substrate. In addition, XPS measurements show that all the studied CrN films are sub-stoichiometric, with N  =  45% and Cr  =  55% (considering a 3% error bar). XPS spectra for all samples are available in the supplementary material. These values match the elastic recoil detection analysis (ERDA) values obtained by Kerdsongpanya et al [8] for similar samples deposited in the same PVD system. As for the single phase Cr₂N films, the approximate stoichiometry is N  =  69% and Cr  =  30%. Oxygen contamination of all films was measured to be less than 2%.

Zoom In Zoom Out Reset image size

The diffraction pattern shows the CrN (1 11) peak located at $2\theta =37.49{}^\circ$ corresponding to a 0.23 ${\rm nm}$ lattice spacing. The peak located at $2\theta =41.69{}^\circ$ is the sapphire (0 0 0 6) substrate peak. The inset shows a zoomed image of the CrN (1 1 1) peak. The fringes are so called Laue oscillations (layer thickness fringes) which appear only for very smooth films, and the separation of these correspond to the film thickness.

Figure 1(b) shows $\theta -2\theta$ diffractograms for a similar set of samples when deposited with 20% nitrogen gas flow mixture. At 600 ${}^\circ {\rm C}$, a Cr₂N (0 0 0 2) phase impurity peak (with a smaller intensity compared to the CrN peak) is seen at approximately $2\theta \approx 40{}^\circ$. Increasing the deposition temperature to 650 ${}^\circ {\rm C}$ decreases the Cr₂N (0 0 0 2) peak intensity, while an increase to 700 ${}^\circ {\rm C}$ will result in single-phase Cr₂N.

Figures 2(a) and (b) shows rocking curve measurements of single phase CrN (1 1 1) deposited at 700 °C comparing the full width at half maximum (FWHM) of the (1 1 1) and (2 2 2) peaks. The FWHM for CrN (1 1 1) is 0.015° compared to the 0.041° for CrN (2 2 2), indicating the existence of mosaicity. Figure 2(c) is a pole figure of the CrN {1 1 1} peaks (single phase CrN, deposited at 700 °C) and includes six equally distanced and discrete poles surrounding the central pole at $\psi =70.5{}^\circ$ which is the angle between {1 1 1} crystal planes in a cubic crystal structure. A single crystal fcc structures deposited in the [1 1 1] direction should only show three surrounding poles, indicating a twin domain epitaxial growth (where the domains are rotated $60{}^\circ$ with respect to each other) [50], a well-known feature for cubic material deposited on sapphire.

Zoom In Zoom Out Reset image size

Figure 3 shows the SEM images of the samples corresponding to the diffraction patterns in figure 1. These images show a temperature-activated change in the surface morphology. The insets in figures 3(a)–(c) present the AFM measurements performed on single phase CrN (samples Cr40-600, Cr40-650 and Cr40-700). The SEM and AFM results show that the film deposited at 600 ${}^\circ {\rm C}$ is composed of many small triangular grains growing in the (1 1 1) direction as determined by the XRD pattern and with an approximate size of 50 ${\rm nm}$. As the temperature increases towards 700 ${}^\circ {\rm C}$, SEM images show large micrometer-sized grains and grain boundaries which originates from the twin domain nature of the film. Figures 3(d)–(f) show the SEM images of the samples deposited with a 20% nitrogen gas mixture (samples Cr20-600, Cr20-650 and Cr20-700). These images show that Cr20-600, which has a larger amount of Cr₂N mixed inside the CrN matrix (compared to Cr20-650) leads to a dented surface and multiple surface features. On the other hand, Cr20-700 (which is single phase Cr₂N) has less of such features.

Zoom In Zoom Out Reset image size

Table 1 lists sample features according to their processing parameters, compound phase, roughness analysis, electrical resistivity, and charge carrier density values. Although the samples Cr40-600, Cr40-650 and Cr40-700 (deposited with a 40% N₂ gas mixture) are single phase and nearly stoichiometric (N/Cr ~ 95%), the four-point-probe measurements show a large variation in the electrical resistivity depending on the deposition temperature. Films deposited at 600 ${}^\circ {\rm C}$ (sample Cr40-600) show a resistivity of 300 ${\rm m}\,\Omega \,{\rm cm}$, while as films deposited at 700 ${}^\circ {\rm C}$ (sample Cr40-700) show a resistivity of 0.75 ${\rm m}\,\Omega \,{\rm cm}$ (400 times lower). Samples Cr20-600, Cr20-650 and Cr20-700 (deposited with a 20% N₂ gas mixture) show a competing relationship between film deposition temperature and electrical resistivity, as the low nitrogen mixture during the deposition process leads to the formation of the metallic secondary phase Cr₂N. Although single phase Cr₂N has a relatively low electrical resistivity of 0.05 ${\rm m}\Omega {\rm cm}$, the mixture of CrN and Cr₂N does not decrease the electrical resistivity below that of sample Cr40-700 with a resistivity of 0.75 ${\rm m}\,\Omega \,{\rm cm}$.

In order to study the secondary phase Cr₂N inside the film, TEM images of single phase CrN (sample Cr40-700) are compared with the CrN/Cr₂N mixture films (sample Cr20-650 and Cr20-600). Figure 4(a) shows a bright field TEM image for CrN films deposited at 700 ${}^\circ {\rm C}$ with a 40% nitrogen gas flow mixture (Cr40-700). Figure 4(b) shows the HRTEM image and the selected area electron diffraction (SAED) pattern of the same film, exhibiting epitaxial growth. The observed lattice spacing of 0.23 ${\rm nm}$ corresponds to the CrN d₍₁₁₁₎ lattice spacing in the XRD diffraction pattern. Figure 4(c) shows a TEM image of sample Cr20-650 deposited at 650 ${}^\circ {\rm C}$ with a 20% nitrogen gas flow mixture. The HRTEM image in figure 4(d) shows that these deposition parameters lead to the formation of 10  ×  15 ${\rm nm}$ nanoinclusions which can be seen residing in the CrN matrix.

Zoom In Zoom Out Reset image size

Figure 4(e) shows a TEM image of a CrN/Cr₂N film deposited at a lower temperature (sample Cr20-600). The corresponding HRTEM image (figure 4(f)) shows that a decrease in the deposition temperature has caused an increase in the Cr₂N grain size in the film and results in 20  ×  50 ${\rm nm}$ Cr₂N nanoinclusions randomly dispersed in the CrN matrix. These changes also include crystal defects and grain boundaries.

Thermoelectric measurements (i.e. the Seebeck coefficient and the electrical resistivity) as a function of temperature were measured to study the effect of the secondary phase Cr₂N when mixed with CrN films which includes Cr40-700, Cr20-600, and Cr20-650. In addition, the Seebeck coefficient and electrical resistivity of single phase Cr₂N was also measured as a reference (sample Cr20-700). Figure 5(a) shows the temperature dependent electrical resistivity of all measured samples. For single phase CrN and CrN with small-sized nanoinclusions (Cr40-700 and Cr20-650) the resistivity increases with increasing temperature (metallic-like behavior) and peaks near 700 K before showing a decrease. This metal-like behavior is unexpected for semiconductors. For the CrN/Cr₂N films with 20 ${\rm nm}\times$ 50 ${\rm nm}$ nanoinclusions (Cr20-600), the resistivity peaks at temperatures near 550 K before showing the decrease.

Zoom In Zoom Out Reset image size

Figure 5(b) presents the Seebeck coefficients as a function of temperature for the investigated samples. The results show that small-sized (10 ${\rm nm}$  ×  15 ${\rm nm}$ particles) Cr₂N phase impurities does not have a large effect on the Seebeck coefficient and the values are approximately the same as single phase CrN with a  −230 $\mu \,{\rm V}{{{\rm K}}^{-1}}$ Seebeck coefficient at 600 K. The Seebeck coefficient of samples with large-sized (20 nm  ×  50 ${\rm nm}$ particles) Cr₂N impurities is measured to be approximately  −160 $\mu \,{\rm V}{{{\rm K}}^{-1}}$ and is almost constant with temperature. All measurements show negative Seebeck coefficients characteristic of n-type semiconducting behavior.

Figure 5(c) shows the power factor for all of the mentioned samples. The results show power factors as high as 5.0  ×  10⁻³ ${\rm W}{{{\rm m}}^{-1}}\,{{{\rm K}}^{-2}}$ at room temperature for single phase and low impurity CrN films and 1.5  ×  10⁻³ ${\rm W}{{{\rm m}}^{-1}}\,{{{\rm K}}^{-2}}$ for the film with large amounts of Cr₂N. The power factor decreases with temperature (because of the electrical resistivity increase) and drops to 2.5  ×  10⁻³ ${\rm W}{{{\rm m}}^{-1}}\,{{{\rm K}}^{-2}}$ near 800 K. Additional measurements on single phase CrN do not show any major changes in either the thermoelectric properties or the XRD pattern.

Table 1 also includes the temperature-dependent Hall carrier concentration results for the mentioned samples at room temperature and 800 K. As expected, there is a steady increase in the samples charge carrier concentration depending on the phase, starting from 0.39  ×  10¹⁹ ${\rm c}{{{\rm m}}^{-3}}$ for CrN and increasing up to 8.17  ×  10²² ${\rm c}{{{\rm m}}^{-3}}$ for Cr₂N (room temperature measurements). These results stem from the respective semiconducting and metallic properties.

Finally, modulated thermoreflectance microscopy was used to measure the thermal conductivity of single phase CrN and single phase Cr₂N at room temperature. The results show that epitaxial and semiconducting CrN (with a reduced number of defects and grain boundaries) has a thermal conductivity of approximately 4.0 ${\rm W}{{{\rm m}}^{-1}}\,{{{\rm K}}^{-1}}$ while as metallic Cr₂N showed a value of approximately 12.0 ${\rm W}{{{\rm m}}^{-1}}\,{{{\rm K}}^{-1}}$.

It is well known that higher deposition temperatures lead to enhanced adatom mobility and grain boundary diffusion resulting in grain coalescence and coarsening [51]. As a result, higher deposition temperatures could lead to a decrease in the defect density and grain boundaries (which affect electron mobility by scattering) leading to better electrical conductivity. This could explain the large power factor of sample Cr40-700 compared to previously reported power factors of polycrystalline bulk samples [52]. It should be noted that the CrN (1 1 1) peak belonging to Cr40-600, Cr40-650 and Cr40-700 all show Laue oscillations (layer thickness fringes). The Laue oscillation on a specific diffraction peak appears when the film presents a smooth surface and this can be used to calculate the film thickness [53]. By using the equation $t=\frac{\lambda \left( {{n}_{2}}-{{n}_{1}} \right)}{2\left( \sin {{\theta }_{2}}-\sin {{\theta }_{1}} \right)}$ for two consecutive fringes, film thickness can be calculated where $t$ is film thickness, $\lambda$ is the x-ray wavelength, $\theta$ is the diffraction angle of the fringe and $n$ is the fringe number. In our case, Cr40-600 is estimated to be 282 ${\rm nm}$, Cr40-650 is estimated to be 284 ${\rm nm}$, and Cr40-700 is estimated to be 289 ${\rm nm}$, consistent with film thicknesses determined from cross-sectional SEM images.

When the films are deposited with a 20% nitrogen gas mixture, deposition temperature also affects film phase purity. The film deposited at 600 ${}^\circ {\rm C}$ has large-sized Cr₂N grains compared to the film deposited at 650 ${}^\circ {\rm C}$. On the other hand, the film deposited at 700 ${}^\circ {\rm C}$ is single phase Cr₂N. This discrepancy may originate from the slightly lower deposition temperature of 600 ${}^\circ {\rm C}$ (${{k}_{B}}{{T}_{873\,{\rm K}}}=0.0752\,{\rm eV}$) compared to that of 650 ${}^\circ {\rm C}$ (${{k}_{B}}{{T}_{923\,{\rm K}}}=0.0795\,{\rm eV}$) which could prevent the nitrogen species incorporation into the octahedral spacing due to the 0.0043 ${\rm eV}$ difference. Thus, the nitrogen deficiency in the film will lead to the preferred formation of Cr₂N. A preliminary formation energy estimate (based on 0 K first principles calculations) from the Materials Project website [54] for CrN is  −0.527 ${\rm eV}$, while for Cr₂N is  −0.523 ${\rm eV}$ (a 0.004 ${\rm eV}$ difference). Increasing the temperature from 600 ${}^\circ {\rm C}$ to 650 ${}^\circ {\rm C}$ will cover this energy difference for the nitrogen species to reside in the interstitial spacing, but the shortage of sufficient nitrogen gas flow will still cause a small amount of Cr₂N to remain in the film. However, films deposited at 700 ${}^\circ {\rm C}$ in a nitrogen deficient ambience will form single phase Cr₂N as the preferred phase.

Based on previous studies regarding the effects of nanoinclusions in thermoelectric material [55, 56], it can be hypothesized that the metallic Cr₂N inclusions can act as a charge carrier injector. This should lead to a reduction in the electrical resistivity. Despite this, the film deposited at 600 ${}^\circ {\rm C}$ (which has larger nanoinclusions and a 7 times larger charge carrier density) has a higher resistivity compared to the film deposited at 650 ${}^\circ {\rm C}$. The relative TEM image (figure 4(e)) shows that the film with large-sized nanoinclusions also includes multiple crystal grains and grain boundaries which can scatter the charge carriers and increase the electrical resistivity.

However, the most intriguing result is the anomalous electrical resistivity of CrN and CrN/Cr₂N (figure 5(a)). Unlike normal semiconducting behavior where resistivity decreases with temperature, the electrical resistivity of under-stoichiometric CrN and CrN/Cr₂N films increases with temperature. This behavior contrasts with the Seebeck coefficient values which matches that of n-type semiconducting. A likely explanation is that the nitrogen vacancies are acting as donors by greatly increase the charge carrier density, creating non-zero quasi-metallic density of states at the Fermi level for the sample [57, 58]. In this case, a rise in the temperature will probably enhance electron–electron scattering resulting in a small decrease in the conductivity values.

Thermal conductivity measurements show that sample Cr40-700 has a room temperature thermal conductivity of 4.0 ${\rm W}{{{\rm m}}^{-1}}\,{{{\rm K}}^{-1}}$. This is the upper value in CrN thermal conductivity where electron and phonon scattering are minimized. Thus, an estimated $zT$ value at room temperature for single phase CrN thin films would be less than 0.4.

As for the Cr₂N film (sample Cr20-700), its metallic properties result in a near-zero Seebeck coefficient which is not suitable for thermoelectrics. However, Cr₂N also shows a relatively low thermal conductivity value with good electrical conductivity and can be further studied for other applications such as diffusion barriers [59] or hard coatings [60].