Optical polarization properties of (11–22) semi-polar InGaN LEDs with a wide spectral range

Electroluminescence polarization measurements have been performed on a series of semi-polar InGaN light emitting diodes (LEDs) grown on semi-polar (11–22) templates with a high crystal quality. The emission wavelengths of these LEDs cover a wide spectral region from 443 to 555 nm. A systematic study has been carried out in order to investigate the influence of both indium content and injection current on polarization properties, where a clear polarization switching at approximately 470 nm has been observed. The shortest wavelength LED (443 nm) exhibits a positive 0.15 polarization degree, while the longest wavelength LED (555 nm) shows a negative −0.33 polarization degree. All the longer wavelength LEDs with an emission wavelength above 470 nm exhibit negative polarization degrees, and they further demonstrate that the dependence of polarization degree on injection current enhances with increasing emission wavelength. Moreover, the absolute value of the polarization degree decreases with increasing injection current. In contrast, the polarization degree of the 443 nm blue LED remains constant with changing injection current. This discrepancy can be attributed to a significant difference in the density of states (DOS) of the valence subbands.

www.nature.com/scientificreports www.nature.com/scientificreports/ loss due to the utilization of a polarizing filter 10 . Therefore, if semi-polar or non-polar LEDs are instead used in backlighting, such a polarizing filter becomes unnecessary and at least ~30% wasted energy consumption is saved.
In this work, a systematic study has been conducted on a series of semi-polar InGaN LEDs with a wide range of indium content (covering emission wavelengths from 443 to 555 nm) grown on (11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22) semi-polar GaN templates with a high crystalline quality. Emission polarization has been investigated as a function of both indium content and injection current.
All the LED structures are similar except for in their indium content (indium content from 0.15 to 0.3 for each case). Each LED structure consists of a 1 µm Si-doped n-type GaN layer, a 4 nm single InGaN quantum well sandwiched between two 9 nm thick un-doped GaN barriers and a final 150 nm p-type GaN capping layer (refer to Fig. 1S in Supplementary Information for scanning electron microscopy (SEM) and TEM characteristics and to Fig. 2S in Supplementary Information for a schematic diagram for the detailed LED structures). By means of a standard photolithography technique and subsequent dry-etching processes, LEDs with a standard 330 ×330 μm 2 mesa size have then been fabricated. Figure 1a schematically illustrates a primed coordinate system showing that the ′ x -′ y plane represent the (11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22) GaN plane, where the two existing orthogonal directions are labeled by ′ x and ′ y . ′ x represents the [−1-123] direction that is parallel to the projection of the c-axis ′ c ( ), where ′ y shows the [1-100] direction perpendicular to this c-axis (⊥c), and ′ z represents the growth direction. Owing to anisotropic strain the valence subbands of the (11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22) GaN split into | ′> y and | ′> x , where | ′> y and | ′> x are the first and the second valence subbands, associated with the respective emission from dipoles along the [1-100] and [−1-123] directions. However, with increasing indium content these two otherwise separate valence subbands approach each other and eventually exchange positions. As a result, the highest valence subband becomes | ′> x with a dipole parallel to [−1-123], while the second valence subband then becomes | ′> y with a dipole parallel to . Figure 1b provides a schematic illustration depicting this change before (i) and after (ii) polarization switching.
A polarization degree, denoted ρ, is defined by the ratio of the integrated polarized emission intensities along one direction relative to another direction as expressed below: [1100] [1123] A positive polarization ratio (ρ > 0) means that the dominant emission component is polarized along the [1-100] direction, while a negative polarization (ρ < 0) demonstrates that the dominant polarization component is along the [−1-123] direction.
Electroluminescence (EL) measurements have been carried out at room temperature in a continuous wave (cw) mode. A detailed description for the EL measurement system has been provided in the "Methods" section, and a schematic for the system has been illustrated in Fig. 3S in Supplementary Information. There is a rotating linear polarizer placed between an objective lens and spectrometer, which allows polarization dependent EL measurements to be conducted between the two orthogonal directions, namely  and [−1-123]. The 0° and 90° angles of the polarizer position correspond to the electric fields aligned along  and [−1-123] directions, respectively. Figure 2 shows the EL spectra of each of the semi-polar InGaN LEDs with a peak emission wavelength ranging from blue (443 nm) to yellow (555 nm), which are measured with the polarizer aligned at 0° and 90° under 20 mA injection current. In each case, the EL spectrum colored with a red-line was measured with the polarizer positioned at a 0° angle, corresponding to the electric field aligned along [1-100] direction, while the EL spectrum labeled with a blue-line was measured at a 90° polarizer angle, indicating the electric field aligned to the [−1-123] direction. This demonstrates that the polarized EL spectra depend on the peak emission wavelength and therefore indium content. Figure 3a shows the polarization degrees of all the semi-polar LEDs as a function of peak emission wavelength, where the polarization degree has been extracted from Fig. 2 using Eq. 1. Figure 3a shows a polarization degree ranging from 0.15 for the shortest wavelength emitter (443 nm LED) to −0.33 for the longest wavelength LED (555 nm LED). The polarization degree of the 443 nm blue LED exhibits a positive sign (ρ > 0), meaning that the intensity of the emission with the electric field polarized along the  direction is higher than that of the emission with the electric field polarized along the [−1-123] direction. Figure 3a also demonstrates that the polarization degree approaches zero at ~470 nm. For the longer wavelength LEDs above ~470 nm the polarization degree therefore switches to a negative sign (ρ < 0) which signifies that the intensity of the polarized emission along the [−1-123] direction is now higher than that of the polarized emission along the  direction. This is entirely consistent with previous studies 19, 25,26 . As the emission moves towards longer wavelength, the polarization degree becomes larger in the negative direction owing to the larger energy separation of the two topmost | ′> x and | ′> y valence subbands. Figure 3b depicts the energy separation between the two polarized emissions (along [−1-123] and  directions) for each LED as a function of peak emission wavelength. The energy separation labeled as Δ E can be defined as:  Figure 3b shows that the Δ E between the two polarized emissions of the 443 nm blue LED is positive. As the emission wavelength increases, the Δ E becomes negative and continues to further reduce with wavelength meaning that the absolute value of Δ E further increases. A comparison between Fig. 3a,b depicts that a negative polarization degree (ρ < 0) is always connected with a negative valued energy separation (Δ Ε < 0), which is in agreement with other report 19 . This is due to the transitional probability between the topmost valence subband and the conduction band being higher than that between the second topmost valence subband and the conduction band. Figure 4a depicts the polarization degrees of all the LEDs measured as a function of injection current from 5 to 100 mA, while Fig. 4b shows the Δ Ε between the two polarized emissions (along [−1-123] and [1-100] www.nature.com/scientificreports www.nature.com/scientificreports/ directions) also as a function of injection current in the same range. It is noted that the 443 nm LED exhibits a different relationship to polarization degree (ρ) and Δ Ε with injection current in comparison to the longer emission wavelength LEDs. The longer wavelength LEDs with a negative polarization degree (ρ <0) demonstrate that an absolute value of polarization degree (|ρ|) decreases with increasing injection current, while both the polarization degree and the Δ Ε of the 443 nm blue LED remains almost constant regardless of injection current. Furthermore, for each of the longer wavelength LEDs, the change in polarization degree decreases with increasing injection Figure 2. Polarized EL spectra measured at 20 mA injection current for the semi-polar (11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22) LEDs with increasing indium content and peak emission wavelength. In each case, the EL spectrum labeled in red was measured at a polarizer angle of 0° corresponding to the electric field aligned along , while the EL spectrum in blue was measured at a polarizer angle of 90° corresponding to the electric field aligned along [−1-123]. www.nature.com/scientificreports www.nature.com/scientificreports/ current in each case. In addition, this change in polarization degree also increases with increasing indium content (emission wavelength). In greater detail, the change in polarization degree from 5 to 100 mA is approximately 0.06, 0.1 and 0.14 for each of the 485, 508 and 555 nm LEDs, respectively. Figure 4S in Supplementary Information also provides very clear data.
The reduction in the absolute value of polarization degree with increasing injection current can be attributed to band filling effects. In the longer wavelength LEDs, the topmost valence subband is | ′> x with a resulting dominant emission component polarized along the [−1-123] direction, while an emission associated with the second valence subband | ′> y is polarized along the [1-100] direction. At low injection current holes mainly occupy the first valence subband | ′> x . With increasing injection current holes then begin to fill the second valence subband | ′> y states, once the more favorable states of the first subband are fully occupied. Consequently, the | ′> y valence subband related emission increases with increasing injection current, and therefore the overall polarization degree decreases.
However, the blue 443 nm LED demonstrates a polarization degree that remains nearly constant with increasing injection current, which is similar to an existing report 27 . This effect is attributed to the significantly higher density of states (DOS) associated with the | ′> y subband (polarized emission along [1-100] direction) than is associated with the | ′> x subband (polarized emission along [−1-123] direction 25 . Consequently, the | ′> y subband can accommodate a significantly high density of holes, leading to a lower probability for holes to occupy the | ′> x subband before polarization switching. This naturally means that the polarization degree remains unchanged with increasing injection current.
In principle, Δ Ε is determined mainly by the difference in energy state between the | ′> y and | ′> x subbands, and thus is not sensitive to injection current. Consequently, as shown in Fig. 4b, the Δ Ε remains almost unchanged with increasing injection current.
In conclusion, a systematic study of the influence of both indium content and injection current on polarization properties has been performed on a series of semi-polar LEDs with a wide spectral range between 443 and 555 nm all grown on (11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22) semi-polar GaN templates with a high crystal quality. Detailed polarization dependent EL measurements demonstrate that the polarization degree strongly depends on the LED emission wavelength, which varies from a positive polarization degree of 0.15 at the shortest wavelength (443 nm) to a negative polarization degree of −0.33 for the longest wavelength LED. A linear fitting indicates that a polarization switching takes place at around 470 nm. Furthermore, the longer wavelength LEDs with a negative polarization degree exhibit a consistent relationship between polarization degree and injection current, while the 443 nm blue LED (before polarization switching) exhibits an insensitivity in polarization degree to injection current. Methods epitaxial growth. All the semi-polar LEDs were grown on high quality (11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22) semi-polar GaN templates on m-plane sapphire by a low-pressure metal-organic vapour phase epitaxy (MOVPE) system. The semi-polar (11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22) GaN templates were obtained by using our well-established overgrowth approach on micro-rod arrays, where the micro-rod diameter is typically 4 µm 11-13 . For the micro-rod array fabrication, a SiO 2 layer with a thickness of 500 nm was initially deposited on a standard single semi-polar (11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22) GaN layer with a thickness of ~400 nm grown on m-plane sapphire. Subsequently a standard photolithography patterning technique and then dry-etching processes were employed to etch the SiO 2 film into regularly arrayed micro-rods with a diameter of 4 µm, which serve as a secondary mask to etch GaN underneath, forming regularly arrayed GaN micro-rods. Finally, the regularly arrayed semi-polar GaN micro-rods were reloaded into the MOVPE chamber for overgrowth. The overgrown semi-polar (11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22) GaN layer with a thickness of ~4 μm exhibits a typical dislocation density of 2×10 8 cm −2 and a typical basal stacking fault density of 4 × 10 4 cm −1 11-13 . All the LED structures were further grown on the semi-polar overgrown templates, beginning with a 1 μm Si-doped n-type GaN layer, then InGaN/GaN SQW and a final 150 nm Mg-doped p-type GaN layer. The only difference in the growth between these LEDs is the growth temperature for the InGaN SQW used to control the indium content, allowing for emission wavelengths ranging from 443 to 555 nm. www.nature.com/scientificreports www.nature.com/scientificreports/ Device fabrication. By means of a standard photolithography technique and subsequent dry-etching processes, LEDs with a standard mesa size of 330 ×330 μm 2 have been fabricated. 100 nm ITO layer was used as a transparent p-type contact, while an n-type Ti/Al/Ti/Au alloy contact was then deposited on to n-type GaN. Finally, Ti/Au alloy bond-pad are deposited by thermal evaporation to form both p-type and n-type contact electrodes.

Electroluminescence (EL) measurements.
Polarization dependent EL measurements have been conducted by using an electroluminescent system equipped with A Keithley 2400 Source-meter and an objective lens (50× magnification; NA = 0.42). EL emission is collected by the objective lens, and is directed through a 50:50 beam splitter. 50% of the emission goes to a CMOS camera which is used to identify the position of a sample. The rest 50% emission is then introduced into the Shamorck 500i Czerny-Turner monochromator via a fiber collimator and finally detected by an air-cooled charge coupled device (CCD). A rotatable polarizer is placed between the objective lens and the fiber collimator, allowing for polarized EL measurements. A halogen lamp is used as a calibration light source in order to remove any polarization from the system. (refer to Fig. 3S in Supplementary Information for a schematic diagram for the EL system).