Low resistivity of highly Si-doped n-type Al_0.62Ga_0.38N layer by suppressing self-compensation

Aluminum nitride (AlN) and aluminum gallium nitride (AlGaN) have ultrawide bandgaps and are key materials for the next generation of high-power electronic devices and deep ultraviolet (UV) light sources. Instead of traditional bulky low-pressure mercury lamps, AlGaN-based deep UV light emitting diodes (LEDs) are expected to be utilized in applications such as purification, sterilization, disinfection of medical instruments, and UV curing. Recent reports indicate that external quantum efficiencies (EQEs) over 10% were achieved for deep UV LEDs, with the emission wavelengths around 280 nm.^1–4) However, further improvements in EQEs are required for high-power applications because the large power losses lead to an increase in device size and the reduction of reliability due to heating during operation. Additionally, an improvement in wall-plug efficiency (WPE) is also desired to replace mercury lamps. The WPE is generally reduced by some loss components as follows: the non-radiative recombination in quantum wells,⁵⁾ the absorption loss through a p-type contact layer,^6,7) the hole injection loss from a p-type side,⁸⁾ and the electron injection loss due to the series resistance of n-type AlGaN (n-AlGaN). The reduction in the last-mentioned series resistance of n-AlGaN layers has been studied for application to UV LEDs with Al-rich n-AlGaN cladding layers.^9–14)

There are two possible sources for increasing the resistance of an n-AlGaN layer. First, in the high AlN mole fraction n-AlGaN, a silicon (Si) atom no longer plays as a shallow donor due to the emergence of a DX state with the increase of the AlN mole fraction, resulting in increasing ionization energies.¹⁵⁾ The wide range of ionization energies for DX states in AlN and AlGaN have been reported to be within 78 to 345 meV.^10,16–20) The second source is the self-compensation effect due to the complexes between cation vacancies–silicon (V_III–nSi) complexes.²¹⁾ Bryan et al.¹²⁾ indicated that V_III–nSi complexes were responsible for a high Si doping limit above 2 × 10¹⁹ cm⁻³ in Al_0.7Ga_0.3N layers and enhanced for samples with higher dislocation densities. Suppression of such compensation sources could significantly improve the conductivity of high AlN mole fraction n-AlGaN layers.^22,23) In addition, the correlation between electron mobility and the concentration of self-compensation sources has been identified and can play a key role in resistivity control.¹²⁾ To realize low-resistance n-AlGaN, it is also necessary to understand this suppression of self-compensation by controlling the growth conditions. In the present study, we successfully demonstrate the lowest resistance of n-Al_0.62Ga_0.38N on sapphire substrates at the high Si doping of 3 × 10¹⁹ cm⁻³ with less carrier compensation. The optical and electrical characterizations suggest that the suppression of V_III–nSi formation by controlling growth condition plays a key role for reducing resistance. Additionally, we demonstrate the mobility degradation due to V_III–nSi complexes, which can significantly contribute to the resistivity in the high Si doping regime.

Crystals of the AlN templates were grown using low-pressure metalorganic vapor phase epitaxy (MOVPE). Trimethylaluminum (TMAl), trimethylgallium (TMGa), and ammonia (NH₃) gas were utilized as Al, Ga, and N precursors, respectively. Monosilane (SiH₄) was utilized as a Si dopant. The surface of a 4-in. c-plane flat sapphire substrate (FSS) with a misorientation angle of 0.35° was thermally cleaned in a hydrogen ambient, followed by the low-temperature (LT) growth of a 0.3-μm-thick AlN and the subsequent high-temperature (HT) growth of a 3 μm thick AlN at approximately 1200 and 1400 °C, respectively. These LT-AlN layers were grown as a nucleation layer to suppress cracking and relax stress intentionally by providing roughness on the surface and air voids in the templates, while the HT-buffer layer reduced the number of threading dislocations.²⁴⁾ Subsequently, 1.3 μm thick n-Al_0.62Ga_0.38N layers were grown on AlN templates, where the AlN mole fraction was set to be around 0.62 by adjusting a TMAl/TMGa ratio for each growth condition.

The crystalline structure of AlN templates was evaluated using high resolution X-ray diffraction measurements. The typical full width at half maximum (FWHM) value of the AlN template grown on the FSS was symmetric and single peaked; 200 and 320 arcsec for (002) and (102) diffractions, respectively. Based on the empirical equation,²⁵⁾ the corresponding screw and edge dislocation densities of the AlN template were estimated to be 9 × 10⁷ and 1 × 10⁹ cm⁻², respectively. No pits were observed on the surfaces of AIN templates. The FWHM values of the X-ray diffraction peaks of n-AlGaN layers were typically 300 and 355 arcsec for (002) and (102) diffractions, respectively. The screw and edge dislocation densities of the n-AlGaN layer were estimated to be 1 × 10⁸ and 9 × 10⁸ cm⁻², respectively. If an electron trap exists at every c-lattice translation along edge dislocations (i.e. one trap site per 0.25 nm),^12,26,27) it can produce an electron compensation of 4 × 10¹⁶ cm⁻³ based on full trapping. This value of electron compensation may be overestimated if the occupancy of dislocation-related traps is less than unity, as is the case for a dislocation density of ∼10⁹ cm⁻², as indicated in Ref. 26. This indicates that carrier compensation due to threading dislocation is negligible in the high doping regime above 10¹⁸ cm⁻³. There were no obvious hillocks and no cracks on the surfaces of the n-AlGaN layers. The surface morphology of n-AlGaN layers featured three-dimensional islands with a root mean square roughness of 4–7 nm. The luminescence properties for n-AlGaN layers were examined using a cathodoluminescence (CL) system equipped with a scanning electron microscope (SEM) with the use of an acceleration voltage of 5 kV and a probe current of 1 nA at RT. These Si concentrations of the n-AlGaN layers were estimated by quadrupole type secondary ion mass spectrometry (SIMS), where the quantitative values were determined using a reference Al_0.62Ga_0.38N sample injected with Si-ions. Resistivities, carrier concentrations, and mobilities of the n-AlGaN layers were evaluated using the van-der-Pauw method and the Hall-effect measurement at RT. To perform these analyses, 7.5 mm square Hall devices were prepared by depositing 0.6 mm diameter Ti/Al/Ni electrodes at four corners. These electrodes were annealed at 700 °C in a nitrogen atmosphere. The I–V curves between the electrodes were linear. In addition, as the resistance between the electrodes of Hall devices is proportional to the sheet resistance, contact resistance is negligible in Hall-effect measurements.

We summarized the AlN mole fraction dependence of resistivity in n-AlGaN, including our results and those of other studies, as shown in Fig. 1.^{9–14,28–36)} The lowest resistance of n-AlGaN with an AlN mole fraction of 62% was achieved in this study. The lowest resistivity value of the Al_0.62Ga_0.38N layer reached 6.6 × 10⁻³ Ω cm at the Si concentration of 3.2 × 10¹⁹ cm⁻³. When this n-AlGaN layer was applied to a UV emitter, no visible resistance loss at the n-type spreading layer was observed.

Zoom In Zoom Out Reset image size

Figure 2(a) shows the Si concentration dependence of resistivity in n-AlGaN with an AlN mole fraction of 62%. The resistivity of the n-AlGaN layer decreased with increasing Si concentration, up to 3 × 10¹⁹ cm⁻³. The lowest resistivity of the n-AlGaN layer was doped at a Si concentration of 3 × 10¹⁹ cm⁻³. As Si atoms were further doped (>3 × 10¹⁹ cm⁻³), the resistivity increased extremely due to self-compensation. Figures 2(b) and 2(c) show the Si concentration dependencies of carrier concentration and mobility at RT, respectively. The correlation between carrier concentration and mobility is shown in the supplementary information (Fig. S1), available online at stacks.iop.org/APEX/13/025504/mmedia. The highest Si ionization ratio was 97% at 2 × 10¹⁹ cm⁻³ of Si concentration, indicating negligible carrier compensation. In the lower doping region (<2 × 10¹⁹ cm⁻³), the carrier concentrations nearly lay on the straight line of the 1: 1 relation for the majority of samples. The sample with carbon atoms of 4.4 × 10¹⁸ cm⁻³ close to the Si concentrations, as indicated by black arrows in Figs. 2(b) and 2(c), had a considerably lower carrier concentration than the Si one, which is possibly caused by electron compensation due to carbon on nitrogen site (C_N) states. The corresponding sample also had an extremely lower mobility, owing to the impurity scattering by C_N states. This feature agreed with the "mobility collapse", as reported by Kaess³⁷⁾ due to the ionized impurity scattering by carbon atoms. Moreover, both the carrier concentration and mobility dropped in the higher doping region above 3 × 10¹⁹ cm⁻³ due to the self-compensation effect.

Zoom In Zoom Out Reset image size

To clarify the compensation sources, CL analyses were performed for n-AlGaN layers fabricated by various growth conditions, as shown in Fig. 3. The CL spectra consisted of three luminescence bands: the near band edge (NBE) emission at the energy of 4.8–4.9 eV, which was close to the optical bandgap of Al_0.62Ga_0.38N,²²⁾ and two broad luminescence bands around 3.2 and 2.4 eV. Bryan et al.¹²⁾ clarified that the 3.2 and 2.4 eV bands are attributable to a C_N and V_III–nSi complexes, respectively. Based on this, the NBE and C_N bands were dominant for all n-AlGaN layers with Si concentrations less than 3 × 10¹⁹ cm⁻³, as shown in Fig. 3(a). Above this doping range, the 2.4 eV band due to V_III–nSi complexes rapidly increased as the Si concentration increased, whereas the NBE and the 3.2 eV band were simultaneously reduced. Note that the rapid increase of the 2.4 eV band correlated to the increase in resistivity of high Si doping, as shown in Fig. 2, indicating carrier compensation due to V_III–nSi complexes. This trend is consistent with the previous report.¹²⁾ For samples doped with a Si of 2 × 10¹⁹ cm⁻³, we also investigated the impact of the V/III ratio and growth temperature on the formation of V_III–nSi complexes as well as the 2.4 eV band. Figures 3(b) and 3(c) indicated that the formation of V_III–nSi was enhanced by the higher V/III ratio and the higher growth temperature. A temperature that is too high (1170 °C) leads to not only the increase of the 2.4 eV band, but also to the decrease in carrier concentration and mobility, as indicated by the white arrows in Figs. 2(b) and 2(c). Additionally, the V/III ratio of 1600 suppressed both the 3.2 and 2.4 eV bands, while enhancing NBE emission. In reality, the sample with the lowest resistivity was achieved using a V/III ratio of 1600, indicating that the optimal V/III ratio exists. These results provide a guideline towards the growth condition to suppress the formation of C_N and V_III–nSi defects.

Zoom In Zoom Out Reset image size

To investigate the correlation between V_III–nSi complexes and their electrical properties, carrier concentrations and mobilities were plotted as a function of the intensity ratio of the 2.4 eV band to the NBE for samples doped with Si of 2 × 10¹⁹ cm⁻³, as shown in Fig. 4. These data involve the samples used in Figs. 3(b) and 3(c). There were no significant changes in the carrier concentrations in the range of 2.4 eV–band intensity ratios below 20, indicating that the concentrations of the V_III–nSi defects were significantly lower than those of the shallow Si donors. However, gradual mobility degradation was observed with an increase in the 2.4 eV–band intensity ratio, suggesting ionized impurity scattering due to the V_III–nSi complexes. Moreover, the ionized-impurity-limited electron mobility is inversely proportional to the square of the number of charges (Z) in an ionized state.²⁶⁾ The Si donor is a single charge, whereas V_III–nSi complexes can take Z = 3 to 1 for n = 1 to 3.³⁸⁾ As the average Z value for V_III–nSi complexes is likely above unity, a complex can cause much stronger scattering than a single charged Si donor. This can explain the mobility degradation as the V_III–nSi concentrations increase, as shown in Fig. 4(b). Additionally, when the 2.4 eV–band intensity ratio is above a critical level, the carrier concentration abruptly dropped, as shown in Fig. 4(a). This can be explained by the majority of electrons being captured by the V_III–nSi acceptors, where the concentration of V_III–nSi complexes was likely closer to that of shallow donors. Consequently, the resistivity showed two trends: a gradual decrease due to mobility degradation and the abrupt drop due to compensation with the increase of the 2.4 eV–band intensity ratio. As described above, the suppression of V_III–nSi complexes evidently plays a key role in the resistivity control of n-AlGaN layers.

Zoom In Zoom Out Reset image size

In summary, we achieved the lowest resistance in n-AlGaN with an AlN mole fraction of 62%. The resistivity had a strong correlation with the formation of V_III–nSi complexes, which emerged as a 2.4 eV band in a CL spectrum. The key growth parameters to suppress the formation of V_III–nSi defects were clarified. We found that the increase in V_III–nSi concentration leads to mobility degradation due to ionized impurity scattering, even in the case of a V_III–nSi concentration significantly lower than that of the Si donors. Consequently, the suppression of the V_III–nSi complexes by control of the growth parameters is significant in a high Si doping regime for reducing the resistivity of Al-rich n-AlGaN layers. Because the dominant mechanism of resistivity increase likely depends on the AlN mole fraction, further investigations at different mole fractions are needed for applications to wider fields.