Final Report Summary - SINOPLE (Surface engineered InGaN heterostructures on N-polar and nonpolar GaN-substrates for green light emitters)
The goal of this project was to develop the potential of molecular beam epitaxy on nearly dislocation free GaN single crystals for semiconductor lasers in the green spectral range (520-550nm). The active structure was intended to consist of In-rich InGaN layers. The goal was to push the internal quantum efficiencies of green emitting InGaN devices beyond 30% and to obtain stimulated emission beyond 500 nm. Our approach is based on (i) engineering of the active structure of the device to reduce the effect of piezoelectric fields on carrier-recombination (ii) exploring molecular beam epitaxy on non-polar, semipolar and N-polar surfaces to obtain maximum In incorporation and by (iii) improving the structural perfection of the active layers by surface engineering. The project combined modern ab-initio based thermodynamics and surface kinetic calculations, with state of the art characterisation techniques. The project took full advantage of the know-how acquired at TopGaN in the growth of UV lasers by MBE and by using unique dislocation free substrates. The progress in substrates made at TopGaN is the backbone of this project since it enabled growth on any defined surface orientation required. Within the five workpackages the following results were obtained:
Substrates. We optimized dislocation free substrates with polar (Ga-polar, N-polar), a-plane, m-plane and semipolar substrates with defined miscut. Both, hydride vapor phase and high-nitrogen-pressure solution growth has been used for growth. Dislocation densities of these substrates are in the range 10^6 cm^-2. In the second period ammonothermal grown substrates have been used in addition. These crystal have superior structural properties (dislocation densities below 10^4cm^-2) and sizes up to 2 inch. A chemo-mechanical surface polishing process has been developed for these surfaces and step-flow growth has been achieved for all surface orientations under optimized growth condtions and for proper miscut.
Epitaxial Growth: InGaN on various substrate orientations by MBE. Studies on epitaxial growth by MBE on these substrates focused on the influence of the Ga- and N-rich conditions and temperature on the growth of N-polar –c(0001) GaN and compared the In incorporation in InGaN QW structures of N- and Ga-polarities under metal-rich conditions. It has been found that it is possible to grow atomically flat N-polar GaN layer under Ga-rich conditions in contrast to the N-rich growth conditions where roughening is observed. The In content of these structures is larger in N-polar InGaN than in Ga-polar InGaN due to a lower InN decomposition at low Ga/N ratio. Major improvements in the growth of N-polar (In, Ga)N structures have been made in the second period of the project. A new growth window under N-excess have been established for GaN and (In, Ga)N, that yield very smooth surfaces, and interfaces of the wells in a reproducible way. Moreover the In content that can be obtained with these new sets of parameters is as high as 20%, and do not require very large N-flux. Finally, these new parameters also allow to simplify the whole growth processes that were carried out at the same temperature without having to carefully monitor the excess of metal.
InGaN on ZnO. As an alternative way to grow green Laser diodes we developed growth of InGaN on ZnO. This approach has the advantage, that dislocation free substrates are available and that InGaN is lattice matched InGaN emitting in the green spectral range, i.e. strain induced piezoelectric fields hampering carrier injection could be reduced. We could show that atomically abrupt interfaces can be achieved under optimized growth conditions. However our studies showed also that Ga reacts strongly with ZnO, especially at elevated growth temperatures and may induce inversion domain boundaries. To reach our final goal, i.e to grow lattice matched InGaN on ZnO it is necessary to use a thin pseudomorphic InN layer, that prevents reaction of the Ga with ZnO. While pseudomorphic InN has been obtained on both polarities a homopolar latticed matched InGaN layer is still work in progress.
Characterization: A new record in precision in In fluctuation analysis by TEM. A new method has been developed to study In fluctuations with unprecedented accuracy. The work has been performed in close collaboration with MPIE who provided simulated unit cells. We could show that the local determination of lattice parameters from HRTEM image can significantly be improved by evaluation of an image series as compared to a single image. We improved the precision for the c-lattice parameter determination in the GaN layer to a STD of 1.2 pm. This allows In detect In concentrations as low as to 1 % with atomic lateral resolution. This procedure does not alter the analyzed image data by any filtering and is computational fast allowing online evaluation. The analysis of the composition of an alloy by HRTEM at the unit cell scale requires taking the in-plane lattice distortions into account, which are introduced by local composition fluctuations. This makes the out-of-plane lattice parameter dependent on strain and composition. Measuring only the out-of-plane lattice parameter is therefore not sufficient to quantify the local composition and yields error-prone results. Also, it suggests false ordering phenomena in a perfectly random alloy, i.e. formation of chains or platelets. Based on the achievable precision in measurement of unit cells were able to overcome this problem by combining the in- and out-of plane lattice parameters measured in the image. This allows separating the strain from the composition at the unit cell scale. This method enables a correct identification of the disorder and the local composition determination with much better accuracy. The feasibility of our approach was demonstrated by means of an In0.20Ga0.80N quantum well. Our method is not restricted to quantum wells but is also applicable to other strained systems like dots or wire.
Theory and Simulation: In surface segregation within mixed In/Ga bilayers on Ga-polar and N-polar c-plane In0.25Ga0.75N surfaces were investigated within the framework of DFT, with calculations performed for the case of pseudomorphic growth on a GaN substrate. In all cases, In was found to preferentially occupy the uppermost surface of the bilayer. The results indicate that only a single Ga adlayer (one Ga atom per 1×1 surface unit cell) is enough to completely block the incorporation of In. This provides evidence that metal-rich growth conditions alone are not sufficient to ensure incorporation of In, but that the In/Ga ratio plays a decisive role in this regard. This is in agreement and explains a previous plasma-assisted MBE observation indicating that In content is limited by the (Al+Ga)/N ratio for the growth of AlInGaN alloys as well as recent MME investigations showing that a modulation of the metal dose can be advantageous to eliminating indium segregation. The dissociation of In from Ga-polar and N-polar surfaces were also investigated by determining energy barriers directly using DFT, with a larger energy barrier found for the latter. This was explained in terms of the greater number of In-N bonds necessary to break for the N-polar surface, and is consistent with the enhanced In surface segregation found for the Ga-polar surface as compared to the N-polar surface as calculated previously. In order to investigate the interplay between the adatom kinetics on the flat m-plane surface and the Ehrlich-Schwoebel barriers for diffusion across low index step edges we have developed a kinetic Monte Carlo (kMC) approach. Our calculations indicate that despite the stronger binding coefficients at the c-steps the island was advancing by elongating along the [11-20] directions. Hence, the strong anisotropy of the diffusion barriers dominates over the strong anisotropy of the binding energies.
Devices: Within the project an optimized design has been developed that is based on high In–content InGaN waveguide having a higher refractive index than GaN to effectively enhance the optical confinement. The design has been verified numerically and successfully applied to the growth of c(0001)-GaN laser diodes by plasma-assisted molecular beam epitaxy (PAMBE) working in cw operation. Beside m(1010)-based blue LDs, this design appears also suitable to the growth of blue and green LDs on -c(0001) and semi-polar (2021) planes due to their high In incorporation efficiency. Work on device processing and on the demonstration of the operation of high-power c(0001) InGaN laser diode ‘‘mini-arrays’’ showed that the most promising high-power laser diode design turned out to be a three stripe solution. In addition to the high power operation, the three stripe device has good spectral characteristics accompanied by very reasonable differential efficiency making it a good candidate for ultra-high optical power systems like laser projectors.
Based on our theoretical and experimental results, we produced lasers in the UV, blue and green spectral range. We found that the high nitrogen flux that could be achieved in the new MBE system at TopGaN allows growing higher quality InGaN in plasma enhanced MBE. A record for electrically pumped nitride MBE Laser diodes with wavelength emission at 482 nm has been achieved. An electrically driven ultraviolet semipolar (202 ̅1) laser diodes were fabricated. The growth under metal-rich growth conditions provided smooth surface morphology and high structural quality. The devices were processed with the ridge waveguide along the [1 ̅21 ̅0] direction. Lasing was observed at room temperature in pulse mode at 388.2 nm. We presented LDs with mirrors successfully fabricated FIB etching giving as good slope efficiency and threshold current as devices with cleaved mirrors. Continuous wave laser diodes on low dislocation GaN were demonstrated with a wavelength of 450 nm, a power of 60 mW and a lifetime exceeding 5 000 h on substrates with low dislocation densities (<1x104 cm-2), showing that MBE could compete with classical MOCVD, but may have advantages with the flexibility in growth temperature, the range of chemical potentials and the possibility to grow on a variety of substrates (e.g. ZnO) that are not accessible for MOCVD. An optically pumped laser with a wavelength of 501 nm has been achieved on conventional c-plane, which is still behind the ultimate target of 530nm-550 nm. The new approach to grow at high N-flux may pave the the way to this goal within the near future.
Our project has shown that the combined effort of theory, characterization and epitaxial growth is a promising approach to improve real devices. This effort is far beyond the possibilities of an SME like TopGaN. On the other hand this project produced a insight into fundamental growth phenomena and basic phyiscs of II-Nitriedes which is major importance for the academic partners. .
Web site of the project: www.sinople-iapp.eu
Substrates. We optimized dislocation free substrates with polar (Ga-polar, N-polar), a-plane, m-plane and semipolar substrates with defined miscut. Both, hydride vapor phase and high-nitrogen-pressure solution growth has been used for growth. Dislocation densities of these substrates are in the range 10^6 cm^-2. In the second period ammonothermal grown substrates have been used in addition. These crystal have superior structural properties (dislocation densities below 10^4cm^-2) and sizes up to 2 inch. A chemo-mechanical surface polishing process has been developed for these surfaces and step-flow growth has been achieved for all surface orientations under optimized growth condtions and for proper miscut.
Epitaxial Growth: InGaN on various substrate orientations by MBE. Studies on epitaxial growth by MBE on these substrates focused on the influence of the Ga- and N-rich conditions and temperature on the growth of N-polar –c(0001) GaN and compared the In incorporation in InGaN QW structures of N- and Ga-polarities under metal-rich conditions. It has been found that it is possible to grow atomically flat N-polar GaN layer under Ga-rich conditions in contrast to the N-rich growth conditions where roughening is observed. The In content of these structures is larger in N-polar InGaN than in Ga-polar InGaN due to a lower InN decomposition at low Ga/N ratio. Major improvements in the growth of N-polar (In, Ga)N structures have been made in the second period of the project. A new growth window under N-excess have been established for GaN and (In, Ga)N, that yield very smooth surfaces, and interfaces of the wells in a reproducible way. Moreover the In content that can be obtained with these new sets of parameters is as high as 20%, and do not require very large N-flux. Finally, these new parameters also allow to simplify the whole growth processes that were carried out at the same temperature without having to carefully monitor the excess of metal.
InGaN on ZnO. As an alternative way to grow green Laser diodes we developed growth of InGaN on ZnO. This approach has the advantage, that dislocation free substrates are available and that InGaN is lattice matched InGaN emitting in the green spectral range, i.e. strain induced piezoelectric fields hampering carrier injection could be reduced. We could show that atomically abrupt interfaces can be achieved under optimized growth conditions. However our studies showed also that Ga reacts strongly with ZnO, especially at elevated growth temperatures and may induce inversion domain boundaries. To reach our final goal, i.e to grow lattice matched InGaN on ZnO it is necessary to use a thin pseudomorphic InN layer, that prevents reaction of the Ga with ZnO. While pseudomorphic InN has been obtained on both polarities a homopolar latticed matched InGaN layer is still work in progress.
Characterization: A new record in precision in In fluctuation analysis by TEM. A new method has been developed to study In fluctuations with unprecedented accuracy. The work has been performed in close collaboration with MPIE who provided simulated unit cells. We could show that the local determination of lattice parameters from HRTEM image can significantly be improved by evaluation of an image series as compared to a single image. We improved the precision for the c-lattice parameter determination in the GaN layer to a STD of 1.2 pm. This allows In detect In concentrations as low as to 1 % with atomic lateral resolution. This procedure does not alter the analyzed image data by any filtering and is computational fast allowing online evaluation. The analysis of the composition of an alloy by HRTEM at the unit cell scale requires taking the in-plane lattice distortions into account, which are introduced by local composition fluctuations. This makes the out-of-plane lattice parameter dependent on strain and composition. Measuring only the out-of-plane lattice parameter is therefore not sufficient to quantify the local composition and yields error-prone results. Also, it suggests false ordering phenomena in a perfectly random alloy, i.e. formation of chains or platelets. Based on the achievable precision in measurement of unit cells were able to overcome this problem by combining the in- and out-of plane lattice parameters measured in the image. This allows separating the strain from the composition at the unit cell scale. This method enables a correct identification of the disorder and the local composition determination with much better accuracy. The feasibility of our approach was demonstrated by means of an In0.20Ga0.80N quantum well. Our method is not restricted to quantum wells but is also applicable to other strained systems like dots or wire.
Theory and Simulation: In surface segregation within mixed In/Ga bilayers on Ga-polar and N-polar c-plane In0.25Ga0.75N surfaces were investigated within the framework of DFT, with calculations performed for the case of pseudomorphic growth on a GaN substrate. In all cases, In was found to preferentially occupy the uppermost surface of the bilayer. The results indicate that only a single Ga adlayer (one Ga atom per 1×1 surface unit cell) is enough to completely block the incorporation of In. This provides evidence that metal-rich growth conditions alone are not sufficient to ensure incorporation of In, but that the In/Ga ratio plays a decisive role in this regard. This is in agreement and explains a previous plasma-assisted MBE observation indicating that In content is limited by the (Al+Ga)/N ratio for the growth of AlInGaN alloys as well as recent MME investigations showing that a modulation of the metal dose can be advantageous to eliminating indium segregation. The dissociation of In from Ga-polar and N-polar surfaces were also investigated by determining energy barriers directly using DFT, with a larger energy barrier found for the latter. This was explained in terms of the greater number of In-N bonds necessary to break for the N-polar surface, and is consistent with the enhanced In surface segregation found for the Ga-polar surface as compared to the N-polar surface as calculated previously. In order to investigate the interplay between the adatom kinetics on the flat m-plane surface and the Ehrlich-Schwoebel barriers for diffusion across low index step edges we have developed a kinetic Monte Carlo (kMC) approach. Our calculations indicate that despite the stronger binding coefficients at the c-steps the island was advancing by elongating along the [11-20] directions. Hence, the strong anisotropy of the diffusion barriers dominates over the strong anisotropy of the binding energies.
Devices: Within the project an optimized design has been developed that is based on high In–content InGaN waveguide having a higher refractive index than GaN to effectively enhance the optical confinement. The design has been verified numerically and successfully applied to the growth of c(0001)-GaN laser diodes by plasma-assisted molecular beam epitaxy (PAMBE) working in cw operation. Beside m(1010)-based blue LDs, this design appears also suitable to the growth of blue and green LDs on -c(0001) and semi-polar (2021) planes due to their high In incorporation efficiency. Work on device processing and on the demonstration of the operation of high-power c(0001) InGaN laser diode ‘‘mini-arrays’’ showed that the most promising high-power laser diode design turned out to be a three stripe solution. In addition to the high power operation, the three stripe device has good spectral characteristics accompanied by very reasonable differential efficiency making it a good candidate for ultra-high optical power systems like laser projectors.
Based on our theoretical and experimental results, we produced lasers in the UV, blue and green spectral range. We found that the high nitrogen flux that could be achieved in the new MBE system at TopGaN allows growing higher quality InGaN in plasma enhanced MBE. A record for electrically pumped nitride MBE Laser diodes with wavelength emission at 482 nm has been achieved. An electrically driven ultraviolet semipolar (202 ̅1) laser diodes were fabricated. The growth under metal-rich growth conditions provided smooth surface morphology and high structural quality. The devices were processed with the ridge waveguide along the [1 ̅21 ̅0] direction. Lasing was observed at room temperature in pulse mode at 388.2 nm. We presented LDs with mirrors successfully fabricated FIB etching giving as good slope efficiency and threshold current as devices with cleaved mirrors. Continuous wave laser diodes on low dislocation GaN were demonstrated with a wavelength of 450 nm, a power of 60 mW and a lifetime exceeding 5 000 h on substrates with low dislocation densities (<1x104 cm-2), showing that MBE could compete with classical MOCVD, but may have advantages with the flexibility in growth temperature, the range of chemical potentials and the possibility to grow on a variety of substrates (e.g. ZnO) that are not accessible for MOCVD. An optically pumped laser with a wavelength of 501 nm has been achieved on conventional c-plane, which is still behind the ultimate target of 530nm-550 nm. The new approach to grow at high N-flux may pave the the way to this goal within the near future.
Our project has shown that the combined effort of theory, characterization and epitaxial growth is a promising approach to improve real devices. This effort is far beyond the possibilities of an SME like TopGaN. On the other hand this project produced a insight into fundamental growth phenomena and basic phyiscs of II-Nitriedes which is major importance for the academic partners. .
Web site of the project: www.sinople-iapp.eu