**Vertical Integration of Nitride Laser Diodes and Light Emitting Diodes by Tunnel Junctions**

**Marcin Siekacz \*, Grzegorz Muziol , Henryk Turski, Mateusz Hajdel , Mikolaj Zak, ˙ Mikolaj Chlipała, Marta Sawicka , Krzesimir Nowakowski-Szkudlarek, Anna Feduniewicz-Zmuda, Julita Smalc-Koziorowska, Szymon Sta ´nczyk ˙ and Czeslaw Skierbiszewski**

Institute of High Pressure Physics, Polish Academy of Sciences, Sokolowska 29/37, 01-142 Warsaw, Poland; gmuziol@unipress.waw.pl (G.M.); henryk@unipress.waw.pl (H.T.); hajdel@unipress.waw.pl (M.H.); mzak@unipress.waw.pl (M.Z.); mik@unipress.waw.pl (M.C.); sawicka@unipress.waw.pl (M.S.); ˙ krzesimir.szkudlarek@unipress.waw.pl (K.N.-S.); ania\_f@unipress.waw.pl (A.F.-Z.); ˙ julita@unipress.waw.pl (J.S.-K.); szymons@unipress.waw.pl (S.S.); czeslaw@unipress.waw.pl (C.S.) **\*** Correspondence: msiekacz@unipress.waw.pl; Tel.: +48-22-876-0324

Received: 13 July 2020; Accepted: 6 September 2020; Published: 10 September 2020

**Abstract:** We demonstrate the applications of tunnel junctions (TJs) for new concepts of monolithic nitride-based multicolor light emitting diode (LED) and laser diode (LD) stacks. The presented structures were grown by plasma-assisted molecular beam epitaxy (PAMBE) on GaN bulk crystals. We demonstrate a stack of four LDs operated at pulse mode with emission wavelength of 453 nm. The output power of 1.1 W and high slope efficiency of 2.3 W/A is achieved for devices without dielectric mirrors. Atomically flat surface after the epitaxy of four LD stack and low dislocation density is measured as a result of proper TJ design with optimized doping level. The strain compensation design with InGaN waveguides and AlGaN claddings is shown to be crucial to avoid cracking and lattice relaxation of the 5 µm thick structure. Vertical connection of n-LDs allows for cascade emission of photons and increases the quantum efficiency n-times. The two-color (blue and green) LEDs are demonstrated. Application of TJs simplifies device processing, reducing the need for applications of *p*-type contact. The key factor enabling demonstration of such devices is hydrogen-free PAMBE technology, in which activation of buried *p*-type layers is not necessary.

**Keywords:** molecular beam epitaxy; nitrides; laser diode; tunnel junction

#### **1. Introduction**

The main breakthrough in III-N optoelectronic devices was related with the development of the *p*-type Mg doped layers [1]. The relatively poor *p*-type conductivity and fabrication of low resistance ohmic *p*-type contacts are still among the most challenging issues in nitrides. Recently, increased attention has been dedicated to the interband tunnel junctions (TJs) [2] for the efficient conductivity conversion from *p*-type to *n*-type in III-nitride devices [3–7]. Application of tunnel junctions (TJs) offers more freedom in device design—e.g., it eliminates the need for *p*-type contact deposition [5,8–10]. However, the utilization of TJs in wide band semiconductors is a counterintuitive approach because it is well known that the tunneling probability through *p*–*n* junction decreases exponentially with the energy gap. The additional complication, which slowed down the progress of the nitride TJs' development, was related with the *p*-type doping procedure used by metal-organic vapor phase epitaxy (MOVPE), the dominant technology for the nitride optoelectronics devices. In MOVPE it is difficult to activate the *p*-type conductivity in the (In)GaN:Mg layers that are buried below *n*-type layers due to the fact that diffusion of hydrogen is completely blocked through *n*-type layers [6,11–13].

The issue with the activation of the Mg doped *p*-type layers is not present for the hydrogen-free plasma-assisted molecular beam epitaxy (PAMBE) technology. In PAMBE, hydrogen is not incorporated into GaN:Mg layers, therefore there is no passivation of Mg dopant and no need for post-growth annealing. For this reason, PAMBE seems to be much better suited than MOVPE for practical realization of the vertical devices with buried *p*-type layers [14]. Recently, making use of PAMBE, it was shown that TJs' resistance for wide bandgap semiconductors can be significantly reduced by making use of the piezoelectric fields in the region of the junction [3,7]. The use of piezoelectric fields and heavy *p*- and *n*-type doping levels allowed us to reduce the TJs' resistance to a level appropriate for a demonstration of the continuous wave operation of nitride laser diodes (LDs) [14].

Efficient TJs enable the realization of different vertical designs of optoelectronic devices. Namely: (1) stacks of LDs or (2) multicolor light emitting diodes (LEDs). The multicolor LEDs can pave a way for efficient low energy consumption matrix displays. The stack of LDs is very attractive for many applications where pulse mode operation is required, such as gas sensing, printing and environment pollution control, or light detection and ranging (LIDAR) in cartography, automotive and industrial systems [15]. It can be very cheap and viable alternative to the arrays of LD bars. It provides much simpler coupling of the light coming from the stack of LDs with external optics than from arrays of LDs, since the spatial separation between vertical devices is two orders of magnitude smaller than for arrays of LD bars. The simultaneous operation of a cascade of *n* LDs increases the slope efficiency (SE) of the full device *n*-times, which makes high-power lasing conditions accessible for smaller currents. In addition, the level of catastrophic optical damage (COD) is *n*-times higher in comparison with a single LD. In spite of the increasing interest, there is only one report on a stack of two III-N LDs grown by MOCVD, which shows very weak evidence of simultaneous laser action from both active regions [16]. This is probably due to difficulties with Mg acceptors activation in buried *p*-type layers. Making use of the PAMBE technology we already demonstrated that it is possible to grow monolithically a stack of two LDs operating at two different wavelengths [17].

In this work we will go a step further and investigate a stack of four LDs operating at the same frequency interconnected by TJs. We demonstrate that by making use of the strain compensation concept, it is possible to grow a 5 µm nitride structure containing thick InGaN and AlGaN layers without generation of lattice relaxation. In addition, we demonstrate the potential of PAMBE for growth of vertically integrated multicolor LEDs.

#### **2. Materials and Methods**

#### *2.1. Laser Diodes*

The epitaxial structure of the LD stack presented in this work was grown entirely by PAMBE on bulk (0001) GaN crystal with the miscut of 0.5◦ . The substrate was a commercially available Ammono-GaN crystal with low threading dislocation density (TDD) about 10<sup>5</sup> cm−<sup>2</sup> [18]. The structure consists of four LD segments interconnected by TJs, as shown schematically in Figure 1a,b. At the top of each LD structure the TJ is placed, which makes it ready for the growth of a subsequent LD. The TJs are located far away (approximately 500 nm) from the waveguides of the LDs to avoid generation of additional optical losses due to heavy *p*-type doping. For such a design, the calculated optical mode overlap with the TJ is extremely low in the order of 10−<sup>7</sup> . Assuming even a high absorption loss of α = 4000 cm−<sup>1</sup> , the resulting optical loss should be at the level of 0.01 cm−<sup>1</sup> , which is negligible.

Details of the epitaxial structure of one LD segment with a TJ on top are presented in Figure 1b. At the bottom of each LD there is a 400 nm Al0.05Ga0.95N:Si cladding (except the most bottom laser diode, LD4, in which the cladding thickness was 700 nm). The LD segment consists of a 220 nm In0.04Ga0.96N waveguide and a 25 nm In0.17Ga0.83N single quantum well (SQW) [19]. Above the waveguide a 20 nm Al0.14Ga0.86N:Mg electron blocking layer (EBL) is placed. EBL is doped with Mg at the level of 2 <sup>×</sup> <sup>10</sup><sup>19</sup> cm−<sup>3</sup> , followed by upper Al0.05Ga0.95N:Mg cladding (Mg doping level is <sup>1</sup> <sup>×</sup> <sup>10</sup><sup>18</sup> cm−<sup>3</sup> ). The TJ region consists of 60 nm In0.02Ga0.98N:Mg, a 10 nm In0.17Ga0.83N quantum well

(QW), and 20 nm In0.02Ga0.98N:Si. The In0.02Ga0.98N barriers are doped with Mg and Si at the levels of 2 × 10 19 cm−<sup>3</sup> and 4 <sup>×</sup> <sup>10</sup> 19 cm−<sup>3</sup> , respectively. First, 5 nm of the QW is heavily doped with Mg at the level of 1.6 × 10 20 cm−<sup>3</sup> while the following 5 nm of the QW is *n*-type doped at the level of 1.8 × 10 20 cm−<sup>3</sup> . The Mg and Si doping profiles in TJs were optimized to achieve atomically flat surface without defects, which is essential for the growth of the subsequent devices on top within the stack.

**Figure 1.** (**a**) Schematic image of the processing design of a stack of four laser diodes (LDs) grown by plasma-assisted molecular beam epitaxy (PAMBE); (**b**) detailed layer structure of a single LD segment; (**c**) scanning electron microscopy (SEM) image of laser mesa viewed at 45 ◦ at cleaved laser facet; (**d**) band diagram of stack of four LDs.

The growth conditions of the four LD stack were the same as for a single LD and they are described in detail elsewhere [20]. The growth temperatures were 730 ◦C for GaN and 650 ◦C for InGaN layers, respectively. The temperature on the grown surface was monitored (and adjusted if necessary) by laser reflectometry described in detail in [21]. We would like to mention here that the entire structure

—

−

−

−

was grown on planar GaN substrate without special patterning applied to reduce strain caused by thick AlGaN claddings [22]. The growth of thick AlGaN claddings usually leads to the cracking of the structures. Application of InGaN waveguides surrounded by AlGaN claddings results in strain compensation and allows us to grow such thick structures without lattice relaxation. It is important to stress here, that such an approach allows us also to achieve flat wafers, which is important for device yield during the processing of lasers.

The LDs were processed as a ridge waveguide with the dimensions of 15 µm × 1000 µm. The 4.4 µm deep mesa was formed by reactive-ion etching (RIE) through active regions of the 3 LDs and reaching almost the EBL of the fourth LD (see Figure 1c). The mesa was covered by SiO<sup>2</sup> dielectric. To ensure that the metal contacts will not make a short cut on the mesa sidewalls, the top of the mesa region was also covered partially by SiO<sup>2</sup> (see Figure 1c). The overlap of the dielectric on the mesa surface reduces the size of the metal contact. The metal contact width is about 8 µm, while mesa size is 15 µm. It could be a challenge for standard LDs with *p*-type contact, in which low conductivity of *p*-type material restricts lateral carrier distribution to distances below 1 µm. This could lead to non-uniform current spreading through such processed LDs. However, in our design, the electrons can easily travel from metallization to the mesa edges (*n*-type contact: Ti/Al/Ni/Au) because the current spreading layer is *n*-type. The processed devices were cleaved and tested without dielectric mirror coatings.

The band diagram for the studied LD stack is shown in Figure 1d. Arrows indicate the electron recombination in the QWs and tunneling though the TJs.

#### *2.2. Light Emitting Diodes*

The schematic diagram of the stack of 2 LEDs operated at different wavelengths for color mixing is presented in Figure 2. The schematic working idea shown in Figure 2a explains that the power supply of the studied device can be applied either to the whole structure or separately to each of the LEDs. The structure was grown on conductive commercial GaN substrate (Saint Gobain Lumilog) with TDDs in the range from 5 <sup>×</sup> <sup>10</sup><sup>6</sup> to 1 <sup>×</sup> <sup>10</sup><sup>7</sup> cm−<sup>2</sup> . Layer sequence is presented in Figure 2b. After 30 nm In0.08Ga0.92N layer, two 2.8 nm In0.23Ga0.77N QWs were grown with 20 nm In0.08Ga0.92N barriers, followed by 20 nm Al0.15Ga0.85N:Mg EBL, 100 nm GaN:Mg and the first TJ. Above the first TJ (TJ1), the 100 nm In0.02Ga0.98N doped with Si at the level 5 <sup>×</sup> <sup>10</sup><sup>18</sup> cm−<sup>3</sup> was located. Then, the second LED was grown—with In content of 17% inside two 2.8 nm InGaN QWs. On top of the second LED, the second TJ (TJ2) was located, followed by 100 nm GaN doped Si at the level of 5 <sup>×</sup> <sup>10</sup><sup>18</sup> cm−<sup>3</sup> . The upper GaN:Si and 100 nm In0.02Ga0.98N:Si located between the bottom (green) and top (blue) LEDs were used for efficient lateral current spreading. The TJ design (thickness of the layers) are the same as for the stack of LDs described before. The differences are in the n- and *p*-type doping of the 10 nm TJ QW region (for TJ details, see Figure 1b). For TJ1, we used moderate doping levels (like for LDs described above) to provide high crystal quality, while for TJ2 the doping was increased: <sup>N</sup>Si <sup>=</sup> <sup>4</sup> <sup>×</sup> <sup>10</sup><sup>21</sup> cm−<sup>3</sup> , NMg <sup>=</sup> <sup>5</sup> <sup>×</sup> <sup>10</sup><sup>20</sup> cm−<sup>3</sup> .

Figure 2c presents a high-angle annular dark-field scanning transmission electron microscopy (HAADF STEM) image of the studied two-color LED structure. High indium content layers, such as InGaN QWs and TJs, are brighter than GaN and AlGaN layers. Sharp interfaces and no extended defects indicate high quality growth. The devices (350 <sup>×</sup> <sup>350</sup> <sup>µ</sup>m<sup>2</sup> mesa size) were separated by deep dry etching by reactive ion etching (RIE), down to the substrate. The schematic picture of the LED after processing will be presented in the following part of the paper.

The band diagram for the studied structure is shown in Figure 2d. Again, similarly to the LD stack band diagram shown in Figure 1d, the arrows indicate the electron recombination in QWs and tunneling though the TJs.

− −

**Figure 2.** (**a**) Schematic diagram and (**b**) layer sequence of two-color light emitting diode (LED) structure. (**c**) High-angle annular dark-field scanning transmission electron microscopy (HAADF STEM) image presenting the studied two-color LED structure. (**d**) Band structure of two-color LED.

#### **3. Results and Discussion**

#### *3.1. Laser Diodes*

μm The LDs were operated with 200 ns long pulses and a repetition rate of 1 kHz. The light–current (L–I) characteristics of the cascade of four LDs are shown in Figure 3a. Three lasing thresholds had been observed: the first one at a current density of 3.8 kA/cm<sup>2</sup> with a slope efficiency of 0.8 W/A; the second one occurred at 5.9 kA/cm<sup>2</sup> and the observed slope efficiency was equal to 1.5 W/A; the third one was at 6.4 kA/cm<sup>2</sup> and the slope efficiency increased to 2.3 W/A. The multiplications of the slope efficiency indicate that the same electrons (and holes) are used three times to generate light—once in each of the three LDs.

– – To verify the observation of lasing from the LDs' stack, we measured near-field patterns collected using a Gaussian beam telescope setup [23]. Strong filamentation is observed as expected for wide-ridge LDs [24]. At j = 3.8 kA/cm<sup>2</sup> , there is only a single near-field pattern for LD1 visible (see Figure 3b). Above the second threshold a second near-field pattern appears below. For current densities higher than 6.4 kA/cm<sup>2</sup> , a third peak in the near-field pattern is observed. Further increase of the current density does not change this pattern. This experiment shows that the LDs start the lasing action in the following order: LD1, LD2 and LD3. Furthermore, we observed that LD4 was not lasing. We suspect that the reason why LD4 was not lasing is that the mesa was too shallow. The distance from the surface to the EBL of LD4 is 4.8 µm, while the etching depth measured by SEM (see Figure 1c) was 4.4 µm. Therefore, one can expect very different waveguiding properties for LD4 in comparison to LD1–LD3, which were etched through the whole structure. For LD1–LD3, the optical mode is confined laterally by SiO<sup>2</sup> deposited on the sides of the mesa, while for LD4 the optical mode can spread laterally to unpumped regions (see Figure 1c). This geometry increases the optical confinement factor for LD1–LD3

and increases the internal optical losses for LD4. Therefore, we expect a substantial increase in the lasing threshold current density for LD4. –

– –

'

—

The maximum optical power obtained for the studied structure was 1.1 W per laser facet and can be further increased using dielectric coatings. Application of this design for *n*-LDs interconnected by (*n*−1) TJs will allow us to increase SE *n*-times. This construction paves a way to achieving III-nitride high-power pulse laser diode stacks for LIDAR applications. −

– – – **Figure 3.** (**a**) The light–current characteristics of the stack of four LDs structure grown by PAMBE. Three lasing thresholds are observed. Total slope efficiency follows the number of lasing devices multiplied by slope efficiency of the first LD; (**b**) the near-field patterns collected by Gaussian beam telescope setup for the lasing regions (1)–(4) denoted in the light–current characteristics.

' — <sup>−</sup> — − The important part in the design of the LDs' stack presented in this work is related with optimization of the TJ design and epitaxy. The TJ should have low resistance and the growth should not introduce additional structural defects. In this work, we applied TJ design which consists of InGaN QW to increase the current tunneling probability [3,7]. The *p*- and *n*-type doping levels were optimized to provide low serial resistance of a TJ. However, the *p*-type and *n*-type doping are limited by inherent physical properties or by structural deterioration. For the *p*-type Mg doping, above the doping level 2—5 × 10 19 cm−3—the self-compensation process is observed, which reduces efficiency of magnesium as an acceptor [25,26]. Contrary to *p*-type, the *n*-type doping is still efficient for very high Si doping levels. Indeed, the increase in the Si doping reduces the tunnel junction resistance, however, for concentrations at the order of 5 × 10 20 cm−<sup>3</sup> a deterioration of the surface morphology is observed. The mechanism of this process is probably related to the Si masking effect when the surface is exposed to the very high Si flux. The surface morphology of our four LDs' stack is shown in Figure 4a. The *n*-type Si doping in TJ equals to 1.8 × 10 20 cm−<sup>3</sup> . For higher *n*-type doping—above 5 × 10 20 cm−3—surface roughening is observed, and many dislocations are generated, as shown in Figure 4b. The TJ presented in Figure 4b, was used for our first demonstration of TJ LDs [14]. Heavy Si doping allowed us to achieve a very low resistance TJ; however, high defect density and rough surface morphology would not allow for the growth of subsequent LDs. Note that the surface root mean square roughness measured by atomic force microscope (AFM) at 5 <sup>×</sup> <sup>5</sup> <sup>µ</sup>m<sup>2</sup> scans is 0.29 nm for the four LDs' stack with optimized TJs and 2.34 nm for the single LD with the highly doped TJ on top of it, see Figure 4a,b, respectively.

The defect density of the four LDs' stack was studied. We used defect-selective etching (DSE) in molten KOH–NaOH eutectics at 450 ◦C for 15 min. After DSE, etch pit density (EPD) was evaluated using optical microscope images. For a given image, the etch pits are counted per known area and –

— <sup>−</sup> —

'

'

thus their density is easily calculated. An example of the optical microscope image of the four LDs' stack surface after DSE is shown in Figure 4c, in which the pits (some are marked by arrows) can be observed. In Figure 4c, 24 pits on the area of 13.4 µm<sup>2</sup> result in EPD of 1.8 × 10 6 cm−<sup>2</sup> . Note that some pits have a flat bottom that may indicate dislocations generated during epitaxy of the four LDs' stack. The etch pits corresponding to the dislocations which originate from the substrate are denoted with black arrows and their density matches approximately the TDD of the GaN substrates used. The overall low defect density revealed by the DSE shows that the stacking of LDs by TJ using PAMBE is a promising technology for high-power pulse laser diodes [27]. ' − '

'

−

− − ' − **Figure 4.** Surface morphology of (**a**) stack of 4 LDs with Si doping in TJ of 1.8 × 10 20 cm−<sup>3</sup> ; (**b**) single LD with heavy Si doping in TJ above 5 × 10 20 cm−<sup>3</sup> . (**c**) Optical microscope image of the studied 4 LDs' epitaxial structure after defect-selective etching. Etch pit density is 1.8 × 10 6 cm−<sup>2</sup> . Some of etch pits are flat-bottomed, indicated by pink arrows.

#### *3.2. Multicolor Light Emitting Diodes*

' — We analyzed the properties of the device that consisted of two standard LED structures emitting at true blue and green range, interconnected and capped with heavily doped In0.02Ga0.98N/In0.17G0.83aN/In0.02Ga0.98N TJs. The device structure is presented in Figure 2. The electroluminescence taken at different currents is presented in Figure 5 and two emission lines centered at 466 nm and 530 nm are visible. For the bottom and top of the LEDs' stack we used *n*-type Ti/Al/Ni/Au contacts. Two-color LED operates at reasonably low voltages of about 8 V for currents around 150 mA. It is important to point out that the light is extracted through the upper LED region which is not covered by metallization. This is characteristic for TJ LEDs—since electrons are more mobile than holes and TJ enables efficient horizontal current spreading. As an example, in Figure 6 we demonstrate the electroluminescence of our TJ LED devices with different shapes of upper contact metallization which defines the output light pattern.

Further development of such device, when adding the third LED, emitting red color, could lead to a phosphorous-free white LED. We would like to stress here that vertical interconnection by TJs also simplifies the processing of individually addressed LED devices. This could be interesting for the future design of multicolor LED matrix displays. In Figure 7a, a schematic diagram of the etched structure is presented. To obtain middle contact we partially etched the structure down about 400 nm and deposited *n*-type contact (Ti/Al/Ni/Au) to In0.02Ga0.98N:Si doped layer (for structure details see Figure 2). It is a great advantage, since there is no need to fabricate *p*-type contacts on the etched surface of the device, which is a challenging issue. Using this configuration, we can bias each LED individually. In Figure 7b, the Current–Voltage (I–V) characteristics of the upper (blue), bottom (green) and both LEDs are presented. We observe a higher turn-on voltage for the green diode (i.e., TJ1 + green LED) than for the blue diode (TJ2 + blue LED). This effect is probable due to the higher voltage drop on TJ1 in comparison to TJ2. In addition, for the green diode, the increase in turn-on voltage can be caused

by stronger piezoelectric field related to higher indium content in green QWs. As was mentioned before, application of TJs simplifies preparation of the contacts on the etched surface. Here, we applied *n*-type contact between green and blue LEDs. The slope of the I–V curve above the turn-on voltage is a measure of the serial resistance of the device. The green LED and both green and blue LEDs have a similar slope. The increased serial resistance for blue LED is due to lateral resistance of the *n*-type layer located below the blue LED. The electrons must travel from the bottom contact of the blue LED for several microns through In0.02Ga0.98N:Si layer doped at moderate values (see Figures 2b and 7a). We can eliminate this effect by increased *n*-type doping of this layer and/or by smaller spacing between mesa and the bottom contact.

– – **Figure 5.** (**a**) The optical spectra of 2 LEDs connected by tunnel junction (TJ) for different driving currents. (**b**) The Current–Voltage (I–V) characteristics for the stack of 2 LEDs; inset to this figure is the same plot in semi-log scale and an indication of very low current leakage in reverse direction. – –

μm μm – **Figure 6.** (**a**) The optical microscope image of operating multicolor TJ LEDs with full mesa size of <sup>350</sup> <sup>×</sup> <sup>350</sup> <sup>µ</sup>m<sup>2</sup> , surrounded by other devices. The devices have various geometry of upper metal contact; metallization blocks the light generated in the active region as schematically depicted in the insert and therefore the top contact pads are visible as black. When an LED chip is biased though a needle placed on top, the bright light emission is visible. (**b**–**e**) Operation of LEDs with different metallization patterns.

– –

– –

–

–

– **Figure 7.** (**a**) Diagram of two LEDs after dry etching and contact deposition. (**b**) I–V and (**c**) spectral characteristics (together with true-color electroluminescence pictures) of blue, green and both LEDs.

#### **4. Conclusions**

− − We demonstrated a stack of four nitride LDs interconnected by TJs grown by PAMBE. We show that it is possible to grow four LDs with a structure almost 5 µm thick without lattice relaxation. Quality of epitaxy and design of TJ are reflected by low dislocation density at the level of low 10 6 cm−<sup>2</sup> . The lasing wavelength is 453 nm. We show that three LDs are operating simultaneously, and the slope efficiency is increased three times. The first LD started to operate at 3.8 kA/cm<sup>2</sup> with SE equal to 0.8 W/A. When all three LDs were operating, the SE increased to 2.3 W/A. This result, together with the near-field pictures, is proof for simultaneous lasing of three LDs. Application of this design for *n*-LDs interconnected by (*n* − 1) TJ will allow us to increase SE *n*-times. The presented design is a viable alternative to achieving III-nitride high-power pulse laser diodes for many applications such as gas sensing or LIDARs.

We also investigate the stack of multicolor LEDs interconnected by TJs for white color, phosphorus-free LEDs and for LED array displays applications. The use of TJs simplifies the electrical connections to buried LED structures, eliminating the need of *p*-type contacts application.

investigation, M.S. (Marcin Siekacz), M.H., M.Ż., K.N. Ż.; resources, K.N. **Author Contributions:** Conceptualization, M.S. (Marcin Siekacz), G.M. and C.S.; methodology, G.M.; investigation, M.S. (Marcin Siekacz), M.H., M.Z., K.N.-S., S.S., M.C., J.S.-K. and A.F.- ˙ Z.; resources, K.N.-S. and A.F.- ˙ Z.; C.S. and ˙ M.S. (Marcin Siekacz); writing—original draft preparation, M.S. (Marcin Siekacz), C.S. and M.S. (Marta Sawicka); writing—review and editing, M.S. (Marcin Siekacz), G.M., H.T., M.H., M.C., M.S. (Marta Sawicka) and C.S.; visualization, M.S. (Marta Sawicka), M.H., M.Z.; supervision, C.S. All authors have read and agreed to the ˙ published version of the manuscript.

**Funding:** This work was supported partially by TEAM-TECH POIR.04.04.00-00-210C/16-00, POWROTY POIR.04.04.00-00-4463/17-00 and HOMING POIR.04.04.00-00-5D5B/18-00 projects of the Foundation for Polish Science co-financed by the European Union under the European Regional Development Fund and the Polish National Centre for Research and Development Grants LIDER/29/0185/L-7/15/NCBR/2016 and

LIDER/35/0127/L-9/17/NCBR/2018 and National Science Center Poland within grants nos. 2019/35/D/ST5/02950, 2019/35/D/ST3/03008, 2019/35/N/ST7/04182 and 2019/35/N/ST7/02968.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

## **Influence of Si Substrate Preparation Procedure on Polarity of Self-Assembled GaN Nanowires on Si(111): Kelvin Probe Force Microscopy Studies**

#### **Marta Sobanska 1,\* , Núria Garro <sup>2</sup> , Kamil Klosek <sup>1</sup> , Ana Cros <sup>2</sup> and Zbigniew R. Zytkiewicz <sup>1</sup>**


Received: 26 October 2020; Accepted: 11 November 2020; Published: 13 November 2020 -

**Abstract:** The growth of GaN nanowires having a polar, wurtzite structure on nonpolar Si substrates raises the issue of GaN nanowire polarity. Depending on the growth procedure, coexistence of nanowires with different polarities inside one ensemble has been reported. Since polarity affects the optical and electronic properties of nanowires, reliable methods for its control are needed. In this work, we use Kelvin probe force microscopy to assess the polarity of GaN nanowires grown by plasma-assisted Molecular Beam Epitaxy on Si(111) substrates. We show that uniformity of the polarity of GaN nanowires critically depends on substrate processing prior to the growth. Nearly 18% of nanowires with reversed polarity (i.e., Ga-polar) were found on the HF-etched substrates with hydrogen surface passivation. Alternative Si substrate treatment steps (RCA etching, Ga-triggered deoxidation) were tested. However, the best results, i.e., purely N-polar ensemble of nanowires, were obtained on Si wafers thermally deoxidized in the growth chamber at ~1000 ◦C. Interestingly, no mixed polarity was found for GaN nanowires grown under similar conditions on Si(111) substrates with a thin AlO<sup>y</sup> buffer layer. Our results show that reversal of nanowires' polarity can be prevented by growing them on a chemically uniform substrate surface, in our case on clean, in situ formed SiN<sup>x</sup> or ex situ deposited AlO<sup>y</sup> buffers.

**Keywords:** gallium nitride nanowires; polarity; Kelvin probe force microscopy

#### **1. Introduction**

GaN nanowires (NWs) are found as promising building blocks for a future generation of electronic and optoelectronics devices. In general, NWs are almost strain-free objects without lattice misfit defects propagating into the crystalline structure even if grown on highly lattice-mismatched substrates. Therefore, these nanostructures facilitate, for instance, the integration of GaN-based devices with Si electronics. In addition, complicated heterostructures can be ideally grown in the form of NWs with a crystallographic quality not achievable in the case of comparable planar heterostructures. However, the growth of wurtzite GaN nanowires on nonpolar Si substrates raises the issue of GaN NW polarity, which is known to have a critical impact on the structural properties of GaN, such as the incorporation of dopants [1–4], surface reactivity [5] and thermal stability [6], as well as on the nucleation and growth of GaN NWs [7,8]. Polarity also causes the onset of a spontaneous built-in electric field in nitride heterostructures, which, together with a piezoelectric contribution due to strain-related electric polarization, accounts for polarization-induced doping [9] and the formation of two-dimensional electron gas in high-electron mobility transistor structures [10]. Overall, polarity significantly affects

the performance of GaN-based devices, so it must be carefully controlled and kept uniform over a large surface area.

While in planar films one polarization domain may overgrow the other and result in a single film polarity [11], this is very unlikely in NWs due to their low lateral growth rate. It is well established already that GaN NWs grow exclusively under N-rich conditions with the c-axis parallel to the growth direction [12–18]. There is no consensus, however, about NW polarity. Although the majority of self-assembled GaN NWs grown on Si substrates are found to be N-polar, the coexistence of NWs with different polarities inside the NW ensemble has been observed [9,11,19–22]. The formation of mixed-polarity NWs is not fully understood and the parameters influencing this behavior, which include the interface chemistry and the growth procedure, are still under debate. However, it is widely reported that the polarity of self-assembled NWs is determined at the GaN nucleation stage and can be reversed from N- to Ga-polar by the high local surface concentration of impurities like Mg, Si, Ti or O [20,23–30]. These findings clearly indicate that proper substrate preparation is decisive for the achievement of homogeneous polarity. In particular, in the case of growth on Si surfaces, the processing procedure and cleanness of Si further determine the chemical and morphological uniformity of the silicon nitride film created during a Si nitridation step prior to GaN growth. As will be shown, these are critical steps for the successful formation of GaN NW ensembles with uniform N polarity.

In this work, we tested the procedures commonly used to prepare Si(111) substrates prior to plasma-assisted molecular beam epitaxial (PAMBE) growth of GaN NWs regarding their impact on the polarity of ensembles of self-assembled GaN nanowires. From a wide range of techniques used to determine the polarity of NWs (see [31] for a review), Kelvin probe force microscopy (KPFM) was chosen. By contrast to the techniques based on electron microscopy which require complicated sample preparation, are time-demanding and may lead to sample damage, KPFM allows the polarity assessment of a statistically significant number of single NWs over micrometer large surface areas with nanometer resolution and without the need of any special sample preparation [32]. Our studies show that the uniformity of the polarity of GaN NWs on Si(111) strongly depends on the procedure used for the substrate processing prior to NW growth. Interestingly, no mixed polarity was found for GaN NWs grown under similar conditions on Si(111) substrates covered by a thin amorphous AlO<sup>y</sup> buffer layer [33,34]. This shows the crucial role the chemistry at the GaN/Si(111) interface plays for the determination of GaN NWs polarity.

#### **2. Experiment**

The samples used in this study were grown by PAMBE using a solid-source effusion Ga cell and the radio frequency Addon nitrogen plasma source controlled by an optical sensor of plasma light emission [35]. All samples were grown on n-type low-resistivity (1–30 Ωcm) silicon (111) substrates. Five of them, later referred to as samples A–E, were grown on bare silicon and they differ only by the procedure used for Si substrate cleaning before epitaxial growth. Details of these procedures are listed in Table 1. Irrespective of substrate preparation, all wafers were transferred in air to the PAMBE system and outgassed in the loading chamber at ~150 ◦C for 1 h and then at ~400 ◦C in the preparation chamber for 2 h to remove any volatile contamination before the transfer to the growth chamber. The substrate temperature was calibrated prior to growth on bare Si(111) substrate by observation of the 7 × 7 to 1 × 1 reflection high-energy electron diffraction (RHEED) pattern transition at 830 ◦C [36,37]. The growth procedure started by exposing silicon substrates to an active nitrogen flux for 15 min at 750 ◦C. The aim was to create a thin SiN<sup>x</sup> film on the surface as explained in detail in our previous report [17]. Finally, the Ga source was opened to induce incubation of three-dimensional GaN islands which subsequently transformed to NWs [38–41]. In addition, sample F was grown on Si(111) deoxidized by the HF dip, similar to sample A, and then transferred in air to the atomic layer deposition (ALD) system for the deposition of a 15 nm thick amorphous AlO<sup>y</sup> buffer layer at 85 ◦C [42]. Then, the substrate was loaded to the PAMBE system, degassed in loading and preparation chambers as for the rest of the samples and transferred to the growth chamber. No substrate nitridation was used for that sample and, after

reaching the growth temperature and ignition of the N source, the Ga and N shutters were opened simultaneously to start the growth. All samples reported in this work were grown at the temperature of 750 ◦C and under nitrogen-rich conditions (N/Ga flux ratio of 1.8) to promote NW growth.



Frequency-modulation KPFM was used to nondestructively determine the polarity of individual NWs within the assembly. The technique is based on the local measurement of the contact potential difference (CPD) between the NW top facet and the atomic force microscopy (AFM) tip [43,44]. Assuming similar electron affinities in N- and Ga-polar GaN, the measured CPD is used to reveal differences in the work function between polar faces [31], which allows their identification [45]. Calibration of the tip contact potential was performed by measuring the CPD on the N- and the Ga-polar facets of a GaN bulk sample grown by hydride vapor-phase epitaxy. For all Pt-coated tips used, the resulting CPD was typically comprised in the interval of 0.20–0.35 V for the Ga-polar face and 0.75–0.90 V for the N-polar face, the difference between N- and Ga-polar faces measured with the same tip being constant (0.55 ± 0.05 V). Complementary images of KPFM, namely topography and contact potential difference, were analyzed [32]. They allow measuring the polarity of individual NWs over an area of tens of µm<sup>2</sup> and provide good statistics on the polarity of the ensemble.

#### **3. Results and Discussion**

Figure 1a shows the AFM topographic view image of sample A. The NWs have a diameter of ~200 nm, but this is convoluted with the tip, which has a diameter of ~20 nm. Figure 1b shows a CPD map of the same area. Some darker NWs, marked by blue arrows, are clearly visible on the map. Figure 1c,d present the CPD value and NW height profiles along the green line, respectively. As seen in Figure 1b,c, the mean CPD value for the majority of the NWs is ~1000 mV, which is compatible with their N polarity. On the contrary, for dark NWs marked by blue arrows, the CPD values drop to ~350 mV. Those values could be attributed to a reverse polarity in Ga-polar NWs. As shown in the last column of Table 1, nearly 18% of 125 NWs checked in total in sample A exhibited reversed polarity.

**Figure 1.** (**a**) AFM topographic view image (1 × 1 µm μ 2 , color scale from 0 to 370 nm) and (**b**) the corresponding contact potential difference (CPD) map (color scale from 0 to 1.5 V) of sample A. Blue arrows in (**b**) mark Ga-polar NWs. The CPD value and NW height profiles along the green line, (**c**) and (**d**), respectively. (**e**) shows the 3D superposition of the topography (xyz axis) and CPD value (color scale). Blue marks Ga-polar and yellow N-polar NWs.

Dipping of the as-received wafer in diluted (~5%) HF acid is the most commonly used procedure for Si substrate preparation before GaN NWs growth by PAMBE [46–50]. This is due to the simplicity of the technique. Moreover, the high-temperature substrate heater required by other methods is not needed. There are a number of studies showing that etching of silicon in a solution of HF removes native silicon oxide and results in hydrogen-terminated and locally ordered surfaces. [51–54]. The hydrogen surface passivation provides protection against oxygen during the wafer transfer to the ultra-high vacuum (UHV) system, where it is removed by annealing at moderate temperatures. Unfortunately, it degrades with exposure to air and moisture. Therefore, the efficiency of surface protection critically depends on the handling of the etched sample. In particular, the time of surface exposure to the laboratory air before transfer to the UHV chamber must be made as short as possible. Moreover, the use of a glove box with dry inert atmosphere connected to the load lock chamber to prevent exposure of the freshly etched sample to oxygen is recommended [55].

Mixed polarity of GaN NWs grown by PAMBE on HF-treated Si substrates is widely observed. Concordel et al. reported the amount of Ga-polar NWs to be below a small percent [20], while the KPFM studies by Minj et al. revealed ~5% of GaN NWs with Ga instead of N polarity [32]. Although these values are much lower than those found in our study, direct comparison is not straight forward since most publications do not provide a precise description of how the freshly etched substrates have been handled.

In the case of sample A, prior to the NW growth, the Si(111) substrate was etched in 5% aqueous HF solution for 1 min, followed by a short deionized water bath and drying with nitrogen. Then, it was transferred in air to the MBE system. The freshly etched substrate was exposed to the air for ~10 min until pumping of the load lock chamber started. After initial thermal treatment in the load chamber as described in Section 2 above, the hydrogen passivation was removed by substrate

annealing in the growth chamber at a temperature of ~700 ◦C. This resulted in a sharp 7 × 7 RHEED pattern characteristic of a clean, oxide-free Si(111) surface. Despite that, presumably some oxide islands were left on the surface. As proposed by Borysiuk et al. [19], these islands might locally protect the Si(001) substrate against the creation of a SiN<sup>x</sup> amorphous layer during the substrate exposure to the nitrogen flux. Thus, after the nitridation stage, the surface was covered by a silicon nitride amorphous layer with some spots of silicon oxide where GaN nucleated, first as zinc-blende (ZB) GaN pyramids and then transformed into wurzite Ga-polar GaN NWs [19].

The hypothesis that on the HF-treated Si substrate the islands of residual oxide were responsible for inducing Ga-polar growth of GaN NWs is strongly supported by results of recent X-ray diffraction measurements [56]. Due to the high intensity of the synchrotron radiation beam used and the application of grazing incidence geometry, the presence of tiny ZB-GaN pyramids that are the seeds for Ga-polar NWs could be detected on the HF-treated Si(001) substrate. For that study, we used a GaN NWs sample grown on HF-treated Si(001). We then compared the results with those obtained for NWs grown on a similar substrate, but for which, after H passivation desorption and substrate nitridation, the so-called gallium-induced surface cleaning [57–60] was used. The procedure consisted of the deposition of a few monolayers of Ga at a low temperature of 500 ◦C in the absence of active nitrogen, followed by gallium desorption at 800 ◦C. This step was repeated three times, after which the substrate was nitridated again. It is well established that Ga-induced cleaning leads to the creation of volatile Ga2O on the substrate that is removed from the surface during heating [61,62]. We anticipated that if after the first nitridation step the substrate was cleaned by gallium flux, the residual silicon oxide islands should be removed, so the second nitridation step could complete the SiN<sup>x</sup> film on places initially covered by the oxide. As a result, the concentration of ZB-GaN pyramids should be significantly reduced. This was indeed observed, together with a corresponding reduction in the number of Ga-polar NWs [56].

In silicon manufacturing, the standard way of treating Si wafers before high-temperature processing steps is the well-known RCA clean developed by the Radio Corporation of America in 1965 [63]. In this cleaning procedure, the native oxide on silicon is dissolved and a new oxide layer forms. Such oxide regeneration is an important factor in the removal of particles and chemical impurities from the surface. The thin volatile oxide created on the wafer may be removed at ~800 ◦C under UHV if a pure silicon surface is needed for epitaxial growth [13,38,64–67]. The Si substrate for sample B was prepared by its exposure for 10 min to a mixture of water-diluted hydrogen peroxide and ammonium hydroxide at 80 ◦C (SC-1 bath), followed by a short immersion in a 1:20 solution of aqueous HF (oxide removal) and a final etching in a mixture of water-diluted hydrogen peroxide and hydrochloric acid at 80 ◦C for 10 min (SC-2 bath) [68]. After the deionized water rinse and drying in nitrogen flow, the substrate was annealed in the PAMBE growth chamber at ~950 ◦C for 10 min to desorb the oxide film. Next, standard substrate nitridation and GaN NW growth were performed. KPFM studies of the as-prepared sample B revealed that 4% of the 50 NWs checked exhibited reversed polarity. This result is similar to that reported by Eftychis et al., who used KOH etching to assess the polarity of GaN NWs grown by PAMBE on RCA-cleaned Si(111) substrates [65].

Interestingly, if after desorption of the oxide film formed by the RCA clean two Ga-triggered deoxidation steps were additionally applied (sample C), the number of Ga-polar NWs reduced only slightly to ~3% (180 NWs analyzed in total). This indicates that oxides were only a fraction of the impurities responsible for the reversed polarity, while the majority of them most probably originated from a residual contamination of chemicals, water, glassware or handling tools.

Next, sample D was prepared, for which, instead of using the RCA clean, the native oxide was thermally desorbed from an as-received wafer in the PAMBE growth chamber at ~1000 ◦C for 10 min. Then, two Ga-induced surface deoxidation steps were applied before substrate nitridation as for sample C. KPFM studies showed that none of the 225 NWs tested in sample D exhibited Ga polarity. In order to elucidate whether this was due to the thermal oxide desorption itself or the Ga-triggered

deoxidation, sample E was studied for which the substrate was prepared as for sample D but without the Ga-induced surface cleaning steps.

Figure 2a,b show, respectively, the AFM topographic view image and the CPD map of the same area of sample E. The profile in Figure 2c shows a uniform CPD distribution with the mean value of ~850 mV, which is compatible with the nitrogen polarity of the NWs. No dark spots corresponding to Ga-polar NWs, as those marked with arrows in Figure 1b, are found on the CPD map of sample E (120 NWs analyzed in total). This evidences that thermal native oxide desorption at high temperature alone provides a clean Si surface and that additional Ga-triggered deoxidation steps, as those used for sample D, are not necessary. However, it is worthy noticing that results of the substrate cleaning procedure may depend on the particular conditions available in various laboratories. For instance, Carnevalle et al. reported ~10% of Ga-polar NWs grown by PAMBE on Si(111) substrates thermally deoxidized in the growth chamber at a temperature of ~1000 ◦C [9], i.e., under conditions similar to those used for sample E.

μ **Figure 2.** (**a**) AFM topographic view image (1 <sup>×</sup> <sup>1</sup> <sup>µ</sup>m<sup>2</sup> , color scale from 0 to 100 nm) and (**b**) the corresponding CPD map (color scale from 0 to 1.5 V) of sample E. The CPD value and NW height profiles along the green line, (**c**) and (**d**), respectively. (**e**) shows the 3D superposition of the topography (xyz axis) and CPD value (color scale).

In summary, from the silicon substrate cleaning procedures tested in this work, thermal native oxide desorption at high temperature inside the growth chamber provides the best uniformity of polarity in the GaN NWs ensemble. Obviously, the cleaner the surface of the silicon substrate before its nitridation, the cleaner and more chemically uniform the silicon nitride layer on which NWs nucleate. If any contamination is left on the Si surface, it disturbs the SiN<sup>x</sup> nucleation layer and may lead to the formation of NWs with reversed polarity. We underline that each of the substrate cleaning techniques presented above resulted in a sharp 7 × 7 RHEED pattern characteristic of a clean, oxide-free Si(111) surface prior to the switching on the nitrogen source. Results of our KPFM studies show that this is not sufficient to ensure uniform N polarity in the whole NW ensemble.

Another strategy for the formation of polarity-uniform GaN NWs arrays by PAMBE is to grow them on a thin uniform buffer layer deposited ex situ on the silicon substrate. If such a buffer is

chemically stable and conformally buries residual impurities on the substrate, it should prevent NW polarity reversal from N to Ga.

Recently, there is an increasing interest in the application of amorphous AlO<sup>y</sup> films deposited by ALD as nucleation layers for the PAMBE growth of GaN nanostructures. Such buffers effectively induce selective area formation of GaN NWs on sapphire [33] and GaN [69,70]. As shown in previous studies [34,41,71,72], AlO<sup>y</sup> buffer layers significantly enhance the nucleation rate of GaN with respect to nitridated Si without a loss of structural and optical quality [34]. Additionally, AlO<sup>y</sup> buffers prevent diffusion of silicon from the substrate [34], facilitating the growth of GaN nanostructures at high temperatures without incorporating any impurities [73], thus potentially leading to exceptional optical properties. However, the polarity of GaN NWs grown by PAMBE on Si(111) substrate with a thin AlO<sup>y</sup> buffer layer deposited by ALD has never been tested. In this study, we used KPFM to fill this gap.

Figure 3a shows the AFM topographic view image of sample F. The NWs have a diameter of ~90 nm, but as before, this is convoluted with the AFM tip of ~20 nm diameter. The CPD values shown in the map in Figure 3b and in the CPD line profile in Figure 3c are quite uniform with the mean value of ~750 mV, which is compatible with the N polarity of the NWs. In some points of the map, the CPD value slightly decreases to ~450 mV, but these points do not correspond to the top of the NWs and are assigned to NWs' sidewalls. This allows us to conclude that no mixed polarity with a certainty above 99.8% (more than 400 NWs analyzed) was observed for GaN NWs grown on Si(111) substrates covered by a thin amorphous AlO<sup>y</sup> buffer layer.

μ **Figure 3.** (**a**) AFM topographic view image (1 <sup>×</sup> <sup>1</sup> <sup>µ</sup>m<sup>2</sup> , color scale from 0 to 140 nm) and (**b**) the corresponding CPD map (color scale from 0 to 1.5 V) of sample F. The CPD value and NW height profiles along the green line, (**c**) and (**d**), respectively. (**e**) shows the 3D superposition of the topography (xyz axis) and CPD value (color scale).

Finally, it is worth mentioning that the silicon substrate for sample F was cleaned by a HF dip in the same procedure as used for sample A, and then transferred in air to the ALD system. It took a few hours before the deposition of the buffer layer started. Due to the instability of the H-passivated substrate surface, it certainly got locally oxidized during the long time of unprotected storage. Nevertheless, no reversed NW polarity was found in sample F. This indicates that the islands of oxide that led to the appearance of Ga-polar NWs in sample A were efficiently covered by the buffer and their presence under the amorphous AlO<sup>y</sup> layer had no impact on GaN nucleation.

#### **4. Summary and Conclusions**

In this work, the most common procedures used to process Si substrates prior to epitaxial growth are tested to find their impact on polarity uniformity inside the ensemble of self-assembled GaN NWs grown by PAMBE. Since local contamination of the silicon substrate surface eventually disturbs the uniformity of the SiN<sup>x</sup> film formed during the nitridation step preceding the GaN nucleation, this might lead to the formation of NWs with reversed Ga polarity.

Kelvin probe force microscopy was used to determine the polarity of GaN NWs. The technique allows the polarity assessment of a statistically significant number of single NWs on the wafer over micrometer large surface areas with nanometer resolution and without the need of any special sample preparation. Complementary images of KPFM, namely topography and contact potential difference, were analyzed.

We showed that the uniformity of the polarity within the ensemble of GaN NWs on Si(111) strongly depends on the procedure used for substrate cleaning. As high as 18% of NWs with reversed polarity (i.e., Ga-polar) were detected by KPFM if the Si substrate was etched in diluted HF and then annealed in the growth chamber to remove hydrogen passivation prior to the substrate nitridation. We ascribe this behavior to the low stability of the hydrogen passivation layer. Apparently, it did not sufficiently protect the freshly etched substrate during its transfer to the UHV system and some islands of oxide were formed, which induced the growth of Ga-polar GaN NWs by the mechanism reported earlier [19]. Such conclusion is strongly supported by previous studies showing that the mixed polarity of GaN NWs on HF-treated Si can be eliminated by additional Ga-triggered deoxidation of the substrate performed just before its nitridation inside the growth chamber [56].

Much better homogeneity of NW polarity was obtained on the Si substrate cleaned by the RCA procedure in which the native oxide was dissolved and a new oxide layer was formed. Next, this oxide was desorbed by annealing in the PAMBE growth chamber at ~950 ◦C for 10 min. KPFM studies of an NW ensemble grown on such substrate revealed that around 4% of NWs had reversed polarity. This number decreased to less than 3% if after oxide desorption, but before the nitridation step, the substrate was additionally cleaned by Ga-triggered deoxidation. In the latter case, the polarity flip could be due to surface contamination by residual pollution of chemicals, water, glassware or handling tools, instead of local surface oxidation.

The best results, i.e., purely N-polar ensemble of NWs, were obtained on epi-ready Si wafers thermally deoxidized in the growth chamber at ~1000 ◦C just prior to their nitridation. Additional Ga-induced surface cleaning steps were not needed in that case. Our studies indicate that high-temperature silicon oxide desorption under UHV produces the cleanest Si surface, resulting in the formation of a uniform SiN<sup>x</sup> layer that prevents an NW polarity reversal from N to Ga. This requires, however, that a high-temperature substrate heater is available in the PAMBE system, which is not always the case.

It is worth mentioning that each substrate cleaning technique tested in this work resulted in a sharp 7 × 7 RHEED pattern characteristic of a clean Si(111) surface prior to switching the nitrogen source on. Results of our KPFM studies show that this is not sufficient to ensure uniform N polarity in the whole NW ensemble.

Finally, no mixed polarity with a certainty above 99.8% was found for GaN NWs grown under similar conditions on Si(111) substrates covered by a thin amorphous AlO<sup>y</sup> buffer layer. It is well known that the ALD technique produces compact, pinhole-free layers conformally covering the surface. Our results indicate that an ALD-deposited AlO<sup>y</sup> buffer efficiently buries oxides that might eventually form during a few hours' storage in air of the HF-dipped Si wafer before buffer deposition in the ALD system. Therefore, possible contamination of the substrate surface under the amorphous AlO<sup>y</sup> layer

had no impact on GaN polarity. In summary, our results show that reversal of GaN nanowires' polarity can be efficiently prevented by growing them on a chemically uniform substrate surface, in our case on clean, in situ formed SiN<sup>x</sup> or ex situ deposited AlO<sup>y</sup> buffers.

**Author Contributions:** Conceptualization, M.S., N.G., K.K., A.C., and Z.R.Z.; methodology, M.S., N.G., A.C., and Z.R.Z.; software, N.G. and A.C.; investigation, M.S., N.G., and A.C.; resources, M.S. and K.K.; writing—original draft preparation, M.S. and Z.R.Z.; writing—review and editing, M.S., N.G., A.C., K.K., and Z.R.Z.; supervision, A.C. and Z.R.Z.; funding acquisition, A.C., N.G., and Z.R.Z. All authors have read and agreed to the published version of the manuscript.

**Funding:** M.S. and Z.R.Z. acknowledge partial support from the Polish National Science Centre grant 2016/23/B/ST7/03745 and the Polish National Centre for Research and Development project PBS1/A3/1/2012 Pol-HEMT. N.G. and A.C. are grateful for support from the Generalitat Valenciana (Spain), grant PROMETEO2018/123 EFIMAT and PID2019-104272RB-C53, co-financed by the Spanish MICINN and FEDER funds.

**Acknowledgments:** The authors thank S. Gieraltowska for ALD deposition of AlO<sup>y</sup> buffer layers and G. Tchutchulashvili for his help with chemical treatment of Si wafers.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


73. Corfdir, P.; Hauswald, C.; Zettler, J.K.; Flissikowski, T.; Lähnemann, J.; Fernández-Garrido, S.; Geelhaar, L.; Grahn, H.T.; Brandt, O. Stacking faults as quantum wells in nanowires: Density of states, oscillator strength, and radiative efficiency. *Phys. Rev. B* **2014**, *90*, 195309. [CrossRef]

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Influence of Growth Polarity Switching on the Optical and Electrical Properties of GaN/AlGaN Nanowire LEDs**

**Anna Reszka 1,\*, Krzysztof P. Korona <sup>2</sup> , Stanislav Tiagulskyi <sup>3</sup> , Henryk Turski <sup>4</sup> , Uwe Jahn <sup>5</sup> , Slawomir Kret <sup>1</sup> , Rafał Bozek ˙ 2 , Marta Sobanska <sup>1</sup> , Zbigniew R. Zytkiewicz <sup>1</sup> and Bogdan J. Kowalski <sup>1</sup>**


**Abstract:** For the development and application of GaN-based nanowire structures, it is crucial to understand their fundamental properties. In this work, we provide the nano-scale correlation of the morphological, electrical, and optical properties of GaN/AlGaN nanowire light emitting diodes (LEDs), observed using a combination of spatially and spectrally resolved cathodoluminescence spectroscopy and imaging, electron beam-induced current microscopy, the nano-probe technique, and scanning electron microscopy. To complement the results, the photo- and electro-luminescence were also studied. The interpretation of the experimental data was supported by the results of numerical simulations of the electronic band structure. We characterized two types of nanowire LEDs grown in one process, which exhibit top facets of different shapes and, as we proved, have opposite growth polarities. We show that switching the polarity of nanowires (NWs) from the N- to Ga-face has a significant impact on their optical and electrical properties. In particular, cathodoluminescence studies revealed quantum wells emissions at about 3.5 eV, which were much brighter in Ga-polar NWs than in N-polar NWs. Moreover, the electron beam-induced current mapping proved that the p–n junctions were not active in N-polar NWs. Our results clearly indicate that intentional polarity inversion between the n- and p-type parts of NWs is a potential path towards the development of efficient nanoLED NW structures.

**Keywords:** nanowires; GaN; AlGaN; LEDs; growth polarity

#### **1. Introduction**

While light emitting diodes (LEDs) or laser diodes (LDs) made of group III nitrides have widely replaced conventional light sources in everyday life and various brands of technology [1–3], they still suffer from drawbacks such as a limited internal quantum efficiency and low light extraction efficiency. These problems are related to structural imperfections (such as a high dislocation density), strong polarization electric fields at interfaces, or inefficient doping in the planar multilayer structures, which are composed of materials with markedly different structural parameters. One solution to at least some of those problems that has been seriously considered is to replace continuous layers of semiconductors with ensembles of quasi-1D nanowires (NWs) grown perpendicularly to the substrate.

**Citation:** Reszka, A.; Korona, K.P.; Tiagulskyi, S.; Turski, H.; Jahn, U.; Kret, S.; Bozek, R.; Sobanska, M.; ˙ Zytkiewicz, Z.R.; Kowalski, B.J.; et al. Influence of Growth Polarity Switching on the Optical and Electrical Properties of GaN/AlGaN Nanowire LEDs. *Electronics* **2021**, *10*, 45. https://doi.org/10.3390/ electronics10010045

Received: 12 October 2020 Accepted: 23 December 2020 Published: 29 December 2020

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

III-N compounds and their solid solutions are highly suitable for the construction of photonic devices. Their band gap energies cover a wide range from 0.7 eV for InN, through 3.4 eV for GaN, to 6.1 eV for AlN [4]. In principle, appropriate selection of the nitride composition in the active region of an optoelectronic device should allow for the obtention of electromagnetic radiation from any part of the visible spectrum or a considerable part of the ultraviolet band. Thus, they can be used in the fabrication of lighting systems imitating the daylight spectrum for houses and offices; light source-acceleration of photosynthesis for agriculture; and health care applications. Additionally, full-color displays or high-density optical data storage systems based on nitride laser diodes can be produced. Nitride-based optoelectronic devices can also be applied in UV detection [5,6] for aerospace and automotive engineering, biology/medicine, and astronomy. All the components of these devices, such as quantum wells, barriers, and p- and n-type layers in heterojunctions, can be fabricated from proper nitrides or their alloys. This is an important advantage of group III nitrides compared to many other wide-band-gap materials. Nevertheless, technical problems associated with the complex synthesis of multilayer semiconductor structures make the reproducible fabrication of reliable devices challenging. It is of great importance to select substrates that are lattice-matched and thermally compatible, while also large and cost-effective. While ammonothermal [7,8] and hydride vapor-phase epitaxy (HVPE) [9,10] methods of growing large bulk GaN substrates have already been successfully developed, sapphire or SiC substrates are still routinely used and have shown promising results. The growth of high-quality nitride structures on silicon would make the integration of nitride-based devices with advanced Si microelectronics possible. This would provide a great advancement in optoelectronic device technology.

The lattice mismatch between the substrate and the nitride layer results in dislocations, relieving the strain at the interface but deteriorating the electrical and optical properties of the structure. Various methods, such as buffer layer engineering and the epitaxial lateral overgrowth technique, have been devised to reduce the dislocation density in nitride-layered structures. Replacing 2D layers with a set of quasi 1D NWs improves the strain accommodation conditions at the markedly smaller interface. This effect was also demonstrated for nitride NWs grown on a Si substrate [11]. The appearance of stressrelieving dislocations can be avoided not only between the substrate and the nanowire but also at the interfaces inside the nanowire [12].

The quasi 1D geometry of the structure has more advantages than those already mentioned: For nano-light-emitters created as axial or radial heterostructures, it increases the light generation efficiency; reduces the light losses in the structure; enhances the efficacy of light extraction from the structure; and allows for the growth of multicolor micro-LED pixels [13–15]. Such objects, with a regular shape defined by the crystalline structure and a size comparable to the wavelength of light, can also confine electro-magnetic radiation and play the role of a resonator for such nano-devices. However, some other critical issues influencing the functionality of nanowire-based devices still exist. It is necessary to solve the problems associated with effective p-type doping (in particular, for ternaries with high In or Al contents) in order to reduce defect-related nonradiative recombination channels, including those related to the developed surface of the system. This would improve hole transport, which is limited by defects, and thus tackle the heat management issue in quasi 1D devices. Some of these issues, such as p-type doping, defect creation (e.g., defects caused by stacking faults (SFs) and polarity inversion domains) are directly related to the nanowire growth conditions.

An important feature of NWs is the possibility of fabricating them by a bottom-up growth process, without the use of technically demanding and expensive technological processes, such as photo- or electron-beam lithography, resist deposition, or etching. Instead, the spontaneous formation of NWs governed by the growth conditions is often applied. A variety of growth methods allowing for the fabrication of NWs on various substrates have been successfully tested (including chemical vapor deposition, metalorganic chemical vapor deposition (MOCVD), and molecular beam epitaxy (MBE) [16]), and numerous examples of ensembles of NWs that are of a highly crystalline quality and have well defined electronic and optical properties have been demonstrated (for a review, see [17]).

Among the techniques currently used to grow III-N low-dimensional structures, MBE has important advantages. The high purity of the source materials and ultra-highvacuum growth conditions, as well as the availability of many techniques of in-situ growth monitoring, allow for a reduction in the unintentional impurity level and improvement in the crystalline quality of the grown material. In general, the two most commonly used approaches to nanowire growth by MBE are based on spontaneous nucleation and vapor-liquid-solid growth, in which whiskers are grown beneath a catalyst droplet. The former [18] is routinely used to grow nitride NWs, since it eliminates the unwanted contamination of NWs with catalyst atoms. While the MBE method is not suitable for the mass production of devices, it is a convenient tool for studies motivated mainly by an interest in fundamental physical processes involving electrons and photons in NWs. The structure of NWs can be carefully controlled and modified in accordance with the results of microscopic, structural, and optical measurements, even if some uncertainties concerning the NW size or composition, resulting from the spontaneous nature of the growth process, should be borne in mind. However, the conclusions can lead to improvements in the architecture and technology of real devices or, at least, the construction of demonstrator devices.

Full control of the course of the growth process and properties of the resulting system is still far from being established. This leads to a scattering of the important physical properties of NWs, even those obtained on the same substrate in a single process. The inner structure of the nanorods may contain spontaneously formed defects, chemical composition fluctuations, domains with built-in strain, growth polarity domains, etc. Fingerprints of some of these defects can be detected by cathodo- and photoluminescence, for example, stacking faults [19,20] and inverted polarity domains [21,22]. On the other hand, intentionally grown axial or radial heterojunctions, quantum wells, etc., are formed in the NWs in order to study basic physical phenomena (e.g., those related to elementary excitations in semiconductors) or create new functionalities [13,23]. Since the properties fluctuate between nanowires, the parameters should not be analyzed as global quantities describing the sets of many nanowires. It is necessary to investigate the optical and electronic properties of NWs with a submicron or nanometer resolution, which allows the properties of the sub-structures of NWs, such as quantum wells, barriers, and p- or n-type segments in the heterojunctions to be revealed.

In this work, we report on the nano-scale correlation of the morphological, structural, electrical, and optical properties of GaN/AlGaN nanowire LEDs, as observed by a combination of scanning electron microscopy (SEM), spatially and spectrally resolved cathodoluminescence (CL), andelectron beam-induced current (EBIC), atomic force, and Kelvin probe force microscopies. GaN/AlGaN LED nanowire structures with three GaN quantum wells in the p–n junction and AlGaN barriers were grown on silicon (111) substrates, without any catalyst, using plasma-assisted molecular-beam epitaxy. The bottom n-type part of the structure was made of GaN:Si. Then, the GaN wells sandwiched between AlGaN barriers were formed. The top part of the nanowire, which acted as the p-type part of the p–n junction, was composed of AlGaN:Mg. As this part was employed with the intention of forming a quasi-planar base for the top electrical contact of the structure, the growth mode had to be adapted to the conditions, which enhanced the lateral growth, leading to an inversely tapered form of the NWs. Therefore, the upper part of the structure was formed under particularly demanding conditions, and its properties resulted from the interplay of several coinciding physical phenomena. It was grown from an AlGaN solid solution at a reduced growth temperature and under intense Mg-doping, which should give the p-type conductivity. The results of our investigations indicated that unintentional fluctuations in the growth conditions caused the growth polarity to switch from the N- to the Ga-face for a considerable part of the NWs. Such a polarity reversal causes differences in the growth and changes the direction of the built-in spontaneous electric field [24]. The EBIC signal and the CL spectral features recorded for individual NWs proved that in

the N-face NWs, the p–n junctions were not active, and the quantum-well luminescence was reduced. A comparison of the experimental data with the results of the numerical simulations of the electronic band structure of individual NWs facilitated a discussion of the physical mechanisms responsible for the activation or deactivation of the p–n junctions. We ascribed the reduced functionality of the p–n junction to an ineffective Mg-doping of the top part of the NWs grown with N-polarity. This conclusion established the growth parameters that result in the successful growth of functional nanoLEDs in NWs.

#### **2. Samples and Experimental Methods**

#### *2.1. Samples*

The GaN/AlGaN NW LEDs were grown on in-situ nitridated 3" Si (111) substrates using plasma-assisted molecular beam epitaxy (PAMBE) in a Riber Compact 21 system with elemental sources of Ga, Al, Si, and Mg. A radio frequency Addon nitrogen plasma cell, controlled by an optical sensor of plasma light emission [25], was used as the source of active nitrogen species. No catalyst was used to induce the nucleation of the NWs. The nanowire LEDs with the three GaN quantum wells in the area of the p–n junction were grown according to the following procedure. First, the 500–600 nm-long GaN nanowires doped with Si (nominal doping level: 5 × 10<sup>18</sup> cm−<sup>3</sup> ) were grown under N-rich conditions at a temperature of 790 ◦C to achieve the n-type part of the structure. Then, the quantum structure was grown. It began with AlxGa1−xN segments with a gradually increased composition (3%, 6%, 9%, 12%) and was followed by the growth of 3 GaN quantum wells (QWs) with thicknesses of 3.5 nm, which were sandwiched between the 10 nmthick Al0.15Ga0.85N barriers. The number of QWs and other parameters of the quantum structure were chosen on the basis of our previous experience, and reports are available in the literature (e.g., [26]). As such a configuration provides efficient luminescence, it was suitable for testing the functionality of our nanoLED in NWs. Finally, the p-type part of the structure consisting of 1 µm-thick Al0.2Ga0.8N doped with Mg (nominal doping level: 5 × 10<sup>19</sup> cm−<sup>3</sup> ) was grown. The growth temperature at this step was reduced to 715 ◦C to enhance the lateral growth rate and to broaden the nanowire tops, with the aim of making the further processing of the device easier.

A detailed scheme of the NW LED structure is shown in Figure 1. More details on the growth method used for the fabrication of the NW structures can be found in [27–29]. The contact with the n-type (cathode) was obtained by covering the Si substrate with indium (for electroluminescence measurements) or aluminum (for EBIC studies). The anodes were made by the evaporation of a thin, semitransparent layer of gold on the upper surfaces of the NWs. The anodes for the electroluminescence measurements had a shape characterized by circular spots with a diameter of 1 mm. The layer of gold was continuous and conductive due to the good coalescence of the tops of the NWs, which was obtained by their lateral growth.

#### *2.2. Methods*

The morphology of the LED NWs was examined by scanning electron microscopy (SEM) using a field-emission Hitachi SU-70 scanning electron microscope (Hitachi, Tokyo, Japan). The SEM machine, equipped with a Gatan MonoCL3 cathodoluminescence system, including the continuous-flow liquid helium cryo-stage and Gatan electron beam induced current (EBIC) setup, enabled studies of the local optical and electrical properties of single NWs.

High-resolution transmission electron microscopy (TEM) experiments of the LED NW structures were conducted using the FEI Titan 80–300 Cubed high-resolution TEM (HRTEM) operating at 300 keV (FEI Company, Hillsboro, ON, USA).

The photoluminescence (PL) spectra were excited by a system composed of a Ti:Sapphire laser and third harmonic generator at a wavelength of 300 nm, with a pulse frequency of 80 MHz. The time-resolved photoluminescence (TRPL) kinetics were measured using a Hamamatsu C5680 streak-camera (Hamamatsu Photonics, Shizuoka, Japan).

**Figure 1.** Scanning electron microscopy (SEM) image and scheme of the light emitting diode (LED) nanowire structure, which consists of: a GaN:Si n-type part on the bottom, an active region, which starts with AlGaN segments with an increasing Al content (3%, 6%, 9% and 12%), followed by 3 GaN quantum wells (QWs), which are sandwiched between the Al0.15Ga0.85N barriers and the p-type Al0.2Ga0.8N:Mg top.

The I–V characteristics of individual LED NWs were measured in the chamber of the Tescan Lyra3 scanning electron microscope (Tescan Orsay Holding, a.s., Brno-Kohoutovice, Czech Republic) using a Keithley 236 measurement unit. A tungsten tip of a SmarAct nano-probe served as a top ohmic contact (SmarAct GmbH, Oldenburg, Germany).

The atomic force microscopy (AFM) and Kelvin probe force microscopy (KPFM) were carried out in the tapping mode using a MultiMode AFM, with a Nanoscope IIIa controller and Nanoscope Extender (Bruker/Digital Instruments, Billerica, MA, USA).

#### **3. Results**

#### *3.1. Morphology and Structure*

Despite the fact that the NWs were grown in the same growth process, the SEM investigations revealed two kinds of NW morphologies, as shown in Figure 2.

Figure 2a shows a cross-sectional SEM image, where two types of nanowires can be easily distinguished. Some of them have a regular broadening, with flat sides, and some have an irregular "oval-like" shape in the laterally grown area. In the plane-view SEM (Figure 2b), differences are also clearly visible: some of the NWs have a flat hexagonal top with sharp edges, while the others have a less regular, rough top with rounded edges.

We note that, due to a radial temperature gradient across the wafer, two areas could be easily distinguished on the substrate surface. Area A (Figure 2b, upper panel) is a ~10 mm-wide ring near the edge of the wafer, while area B (Figure 2b, bottom panel) is a circle in the center of the wafer. Independent measurements of the temperature distribution, performed by growing planar GaN layers on such substrates, indicated a ~30 ◦C higher growth temperature at the wafer periphery than in the central B area. Interestingly, such a temperature profile led to the noticeably higher concentration of hexagonally shaped NWs in area A than in the colder part (area B) of the wafer, as shown in the SEM images.

The TEM studies in the Z-contrast scanning transmission mode (STEM) revealed that both types of NWs contain three 3.5 nm-thick QWs (inset in Figure 2a). The bottom parts and QW regions of NWs are similar in all wires, and differences are observed only in the upper p-type Al0.2Ga0.8N:Mg part of the LED structure. The NWs with the flat hexagonal tops are more uniform, while in the NWs with the "oval-like" top area, some features resembling grain boundaries are visible.

(**<sup>a</sup>**) (**<sup>b</sup>**)

**Figure 2.** Cross-section SEM and Z-contrast scanning transmission mode (STEM) (inset) images (**a**) and plane-view SEM images (**b**) of the LED NWs. Two types of NWs can be easy distinguished: those with flat sides and hexagonal tops with sharp edges, and those which are "oval-like", with less regular, rough tops and rounded edges. Due to the temperature gradient across the substrate, the ratio of these two types varied. Two types coexist in area A, while in area B, NWs with rounded tops are mainly present.

In the [0001] direction, GaN can grow either in the Ga or N direction, which are referred to as polarities. Most nitride layers are grown in the Ga polarity, while spontaneously nucleated GaN NWs usually grow in the N polarity. However, depending on the interface chemistry and the growth procedure, changes to the Ga polarity inside one NW ensemble are often observed. Since the specific polarity of NWs affects both the optical and electrical properties and result in different properties for two morphologically distinct types of NWs, it is necessary to assess the polarity of individual NWs. This was conducted by the selective etching of the sample in an aqueous solution of potassium hydroxide (KOH). The principle of this method is based on the fact that the etching behavior strongly depends on the chemical nature of the surface. The resulting etching rate, etched surface shape, and roughness are determined by the polarity, which can, therefore, be clearly identified. It was revealed in research by several groups that the KOH etching rate is much lower for Ga-polar GaN surfaces than for N-polar ones (see, for instance, [30–33]).

In our case, the samples were etched in a KOH:H2O 6.5 g/100 mL solution at a temperature of 160 ◦C for 10 min. SEM images of the samples' surfaces are presented in Figure 3. In the SEM image of the as-grown sample (Figure 3a), two kinds of NWs are visible—hexagonal- and oval-shaped, as mentioned earlier. Figure 3b presents exactly the same area of the sample, after etching in the KOH solution. It can be seen that the NWs of a hexagonal shape with flat tops have disappeared, while some of the oval-top NWs remain untouched. This experiment proved that the hexagonal-top surfaces were N-polar NWs, while the oval-top ones had Ga polarity.

**Figure 3.** SEM images in the plane-view of the same NW array, before (**a**) and after (**b**) KOH treatment. Before KOH treatment, two kinds of NWs—those with flat hexagonal tops and those with rounded edges—were visible. After etching, the hexagonal ones disappeared, while the oval-top NWs remained untouched.

#### *3.2. Macroscopic Electro- and Photoluminescence*

The semiconductor structure inside the NW was designed to work as a light emitting diode. The bottom part was n-type (Si-doped), the upper part was p-type (Mg-doped), and three QWs were placed in the region of the p–n junction in order to trap electrons and holes, which should promote radiative recombination. Figure 4 (upper panel) shows the photoluminescence spectra at different points in the sample. It can be observed that the spectra are significantly different. For example, the GaN-related G1 peak is much stronger in the A area, while area B is dominated by AlGaN emission (peak A4). To check the operation of the NW LED devices, a series of small diodes were prepared for macroscopic electroluminescence (EL) measurements in a few rows, from the edge to the center of the wafer, across the A and B areas. After application of a voltage, blue-violet-UV electroluminescence was only observed in the B area. Diodes in the A area, mostly consisting of the hexagonally shaped top facets, were not active, which agrees well with the I–V data shown in Figure 9. The electroluminescence (EL) spectra of the forward polarized ensemble of the NWs in the B area, as a function of the applied voltage, are plotted in Figure 4 (bottom panel).

We note that in order to obtain a sufficient current, it was necessary to apply a bias of the order of 10 V, which was due to the high resistivity of the NWs. The main decrease in the voltage was most probably in the p-type part of the NWs. Nevertheless, the change in bias caused a change in electric field in the area of QWs, which led to a shift in the emission energy from 3.15 eV up to 3.28 eV, with an increase in the bias (see Figure 4—bottom panel). The effect can be explained as state filling due to a high-density electron current or a reduction in the quantum-confined Stark effect (QCSE). In the second case, an increase in the external voltage led to a reduction in the total electric field, which means that the build-in electric field has the opposite direction to that of the external one. Figure 4 shows the photoluminescence (PL) spectra obtained by the photoexcitation of the diode without voltage and PL (plus EL), observed under a bias of 14 V. In the case of EL, electrons are provided by a cathode, while holes are injected by an anode. The only part of the structure that contains both types of carriers is an active region with the QWs. Thus, in EL, only one peak was visible. During light excitations, electrons and holes were generated throughout the structure, so we could also observe the peak A4 at 3.6 eV, which is related to AlGaN. The A4 peak decreased after the application of the voltage, which was probably caused by the increase in the electric field in the AlGaN part. This meant that the build-in electric field existing in AlGaN without application of an external voltage had the same direction as the external field applied in the forward direction. The EL was only observed at room

temperature, since at lower temperatures, the sample was not electrically conductive. However, it was possible to measure PL at lower temperatures.

**Figure 4.** Luminescence spectra at room temperature: PL at different points in the sample (upper panel) and EL at different voltages (bottom panel). It can be observed that the GaN-related G1 peak is much stronger in area A. However, in area B, it was possible to obtain EL, which was dominated by the G1 peak.

A comparison of PL at 4 and 300 K in the A and B areas can be found in Figure 5. The spectra were measured in the time-resolved mode, and the kinetics of the light emission could be determined. At room temperature (upper panels in Figure 5a,b), the AlGaN-related emission was observed in a range above 3.5 eV. It was diminished by fast relaxation (lifetime about 0.1 ns) due to the escape of carriers to GaN. The main GaN peak, G1, was at about 3.4 eV at 300 K. Its lifetime depended on the location in the sample at which it was measured, from 1 ns in area A to 0.2 ns in the B area. It was observed that the GaN peak at t = 0 was at about 3.4 eV, and its energy decreased to 3.3 eV after 2 ns. Such an effect can be caused by the QCSE in the QWs. Since the energy of this peak coincided with the electroluminescence energy, we can assume that it was emitted by the QWs. Due to a shift in the energy gap [34], the luminescence spectra at 4 K are expected to be at an energy that is about 0.1 eV higher, compared to that at room temperature. The AlGaN emission was observed in the 3.6–3.8 eV range, and the GaN emission was at 3.47 eV (G1) and about 3.43 eV (G2). As can be seen in in the bottom panels of Figure 5a,b, the emission dynamics in the A and B areas were similar. The G1 peak was observed only under a high-power excitation. The lifetimes ranged from 0.2 to 0.5 ns for the AlGaN emission and 1–2 ns for the G2 emission. The exciton lifetimes in 3D GaN were below 1 ns [35], but similarly long lifetimes were observed for QWs [26,36] and SF (which was also compared to QW [20]). Therefore, we assumed that G2 is related to QWs.

**Figure 5.** Time-resolved photoluminescence spectra in area A (**a**) and B (**b**) at T = 300 K (upper panels) and T = 4 K (bottom panels). The GaN emission can be seen in the 3.3–3.5 eV range, and that of AlGaN can be observed in the 3.6–3.8 eV range. The lifetimes at a low temperature are visibly longer. It can be noticed that the AlGaN emission decays faster than the GaN emission. In area A, GaN PL has a longer lifetime.

#### *3.3. Microscopic Optical and Electrical Properties*

#### 3.3.1. Cathodoluminescence Studies

SEM combined with CL enables a direct correlation of luminescence maps and sample morphologies at the nanoscale level. The high spatial and spectral resolution and high energy excitation of this method make it the perfect tool for studying the optical properties of nanostructures that emit light in the UV range [37–39].

NW arrays were studied by CL spectroscopy and imaging at 5 K using an acceleration voltage (AV) of 5 kV and a beam current (Ib) of 1 nA. Spectra and monochromatic CL maps were collected in the cross-sectional geometry using the photomultiplier mode. In the CL spectrum (Figure 6, central panel), four main emission bands can be distinguished. Two bands—A1: 3.83 eV; and A4: 3.63 eV—are related to AlGaN luminescence. The CL maps (Figure 6, right side) confirmed that the emission comes from the upper (p-type) part of the NWs. The presence of two bands is most probably related to the existence of inhomogeneous Al contents in AlGaN. The third emission line, G1 centered at 3.47 eV, is related to the luminescence of the QWs. The emission of QWs is clearly visible in the corresponding CL map (Figure 6, left side). The fourth one, G3 at 3.3 eV, most probably comes from the recombination of donor–acceptor pairs.

**Figure 6.** Cathodoluminescence spectrum of the NW array (central panel) and corresponding monochromatic CL maps superimposed with SEM images (sides), collected in the cross-sectional mode. Two bands—A1: 3.83 eV; and A4: 3.63 eV—are related to AlGaN luminescence. In the corresponding CL maps (right side), the emission is visible in the upper (p-type) part of the NWs. The third emission line, G1 centered at 3.47 eV, is related to the luminescence of the QWs, which proves the corresponding CL map (left side). The fourth one, G3 at 3.3 eV, most probably comes from the recombination of donor–acceptor pairs.

To study local optical properties in detail, CL spectrum line-scans at 5 K were recorded for a number of individual NWs (in a cross-sectional geometry) using an AV of 5 kV and I<sup>b</sup> of 1 nA. A CCD camera in the CL line-scan mode enabled the acquisition of a series of luminescence spectra of the individual NWs, which were excited with the electron beam point-by-point along the specified line parallel to the NW axis. The direction of the line-scans is illustrated in the SEM insets shown in Figure 7. Typical results are shown in Figure 7. The set of spectra in the upper panel corresponds to the CL line-scan of the nanowire with an oval shape, where the polarity switched to the Ga-one, while the bottom panel shows the spectra for a fully N-polar hexagonal NW.

As can be seen in the upper panel in Figure 7, the bright luminescence from the active region is observed in the Ga-face NWs. Since the QWs emission is the main source of LED radiation, we conclude that a reversal of NW polarity is crucial for LED efficiency. The N-face NWs had a strong emission related to the A4 peak at 3.65 eV, which was emitted in the upper part of the NW.

A series of spectra of the individual NWs was also collected in the top-view configuration. The CL in the spot-mode was measured at 6 K using an AV of 5 kV and I<sup>b</sup> of 1.9 nA. The electron beam parameters have been adjusted so that only the upper, p-type Al0.2Ga0.8N:Mg part of the LED NW structure was excited. A Monte Carlo simulation of the CL intensity distribution, performed using the Casino v2.48 (2.4.8.1) software [40], allowed us to estimate the size of this region. For AV = 5 kV, about 90% of the CL signal comes from a depth of 200 nm.

Observing Figures 7 and 8, one can easily see that the different peaks in the CL spectra correlate with the differences in the shape of the NWs. The AlGaN:Mg emission was expected to be above 3.5 eV. The hexagonal-shaped (N-polar) NWs emissions corresponded to the A1 and A4 peaks, while the oval-shaped (Ga-polar) NWs emitted mainly at 3.78 eV (peak A2). The lower energy of the oval-shaped NWs implies that these NWs have a lower Al content. Since the Al desorption at used growth temperatures is negligible, irrespective of the polarity,

this was probably caused by the higher Ga incorporation into AlGaN on the N-polar than on the Ga-polar surface, which is similar to the case of In during InGaN growth [41].

**Figure 7.** CL line-scans collected along the individual NWs: upper panel—scan along the oval-like NW; bottom panel—scan along the hexagonal one. The bright emission from the active region is observed in the Ga-face NWs. The N-face NWs had a strong emission related to the A4 peak at 3.65 eV, which was emitted in the upper part of the NW. The arrows on the SEM images indicate the direction of the scans.

**Figure 8.** Series of cathodoluminescence spectra collected in the spot-mode using AV = 5 kV, where only the top (p-type) part of the structure is excited. Differences between NWs with flat hexagonal tops (black curves) and irregular ones (red curves) are clearly visible. The inset shows the geometry of the measurements.

#### 3.3.2. EBIC and Nano-Probe Results

The EBIC technique was used to assess and analyze the local electrical properties of the structure. The interaction of the electron beam with the semiconductor structure results in the generation of electron–hole pairs, which are separated in the presence of an internal field near the p–n junction. As a result, an electrical current is generated, providing the EBIC signal. This technique, combined with SEM, enabled the visualization of the position and continuity of the p–n junction and estimation of the width of the depletion region [42,43].

The electrical measurements conducted on the wafer with the NWs revealed that the NW LED structures in the central B area of the wafer exhibited diode-like I–V curves (Figure 9a), which means that we could expect a current generation caused by an electron beam. The obtained EBIC contrast image is presented in Figure 9b. EBIC measurements were performed at room temperature using an AV of 5 kV and I<sup>b</sup> of 0.2 nA, without applying any bias voltage. The Au+Pd contact was deposited on the p-type NW tops using a sputter coater. Al contacts were deposited onto the exposed Si (n-type) substrate using the same method. The sample with deposited contacts was cleaved for EBIC measurements in the cross-sectional geometry. The EBIC contrast (red) superimposed on the SEM image (grey) revealed the position of the active part of the p–n junction. We observed a strong EBIC signal only in the NWs with the "oval-like" shape tops. The depletion region in these LED NWs was located on the side of the p-type part of the structure, and its width was about 300 nm. A Monte Carlo simulation of CL from GaN under such conditions shows that about 65% of the emission originates from a volume with a diameter of 10–15 nm. This also gives us an estimate of the lateral resolution of the EBIC experiment and confirms that it insignificantly interferes with the observed width of the depletion region [40]. This observation meant that the p-type part had a lower concentration of electrically active acceptor dopants, compared to the concentration of donors in the n-type part, even if the nominal concentration of the acceptors was ten times higher. This was due to the weak activation of Mg acceptors.

**Figure 9.** (**a**) Macroscopic I–V curves of the A and B areas of the wafer. Lines fitted to the curves show serial resistance deduced from the slope and turn-on voltage deduced from the crossing with the zero axis. (**b**) SEM image (grey) with the superimposed EBIC signal (red) of the LED NWs. The strong EBIC signal reveals the position of the active p–n junction. The inset shows the geometry of the contacts.

To investigate the I–V characteristics of individual LED NWs, a tungsten-tip nanoprobe was used to form the top ohmic contact. The native tungsten oxide was preliminary removed from the tip using a 30 keV focused Ga+ ion beam. Thin silver wire, glued by conductive silver paste, provided the ohmic contact to the silicon substrate. The bias voltage was applied between the needle and the grounded substrate [44,45]. The current–voltage characteristics were measured for the two different types of nanowires with different shapes and polarities. The measurements were repeated on different spots of the sample, achieving a reasonable reproducibility (see Figure 10). The sharp hexagons (black curves) exhibit nonlinear, symmetrical I–V curves, with a low turn-on voltage. The NWs with the rounded edges (red curves) exhibit nonlinear and nonsymmetrical diode-like I–V curves.

**Figure 10.** Series of I–V curves (central panel) of individual nanowires (NWs) (black curves: hexagonal top NWs; red curves: irregular NWs) and corresponding SEM images, showing the geometry of the measurements (sides). The sharp hexagons (black curves) exhibit nonlinear, symmetrical I–V curves with a low turn-on voltage, while the NWs with rounded edges (red curves) exhibit nonlinear and nonsymmetrical diode-like I–V curves.

Both the EBIC and nano-probe measurements revealed that the p–n junction works properly only in the NWs with the oval-like top segment and that only these NW structures act efficiently as LED diodes.

#### 3.3.3. Kelvin Probe Force Microscopy

The change in the NW polarity should produce significant differences in the surface potential at the top of the NW. The surface potential profile at the microscale for individual NWs can be measured by Kelvin probe force microscopy (KPFM). The KPFM allows for the determination of the polarity of single NWs over micrometer-large surface areas with a nanometer resolution, without the need for any special sample preparation. In the KPFM studies, the local-contact potential difference (VCPD) between the metallic tip and sample was measured. By scanning the sample surface in KPFM measurements, we can compare the potentials of different wires, and local changes can be determined.

Nitrides of V-group elements, including AlGaN, exhibit a strong built-in electric field caused by spontaneous polarization, together with a piezoelectric contribution due to a strain-related electric polarization along the AlGaN axis. We assume that the potential at the wafer and along the NW up to the QW region is the same for all NWs. Then, above the QW, the polarity is reversed, so the built-in field should be different by a few MV/cm. Consequently, we expected a difference in the potential at the top of the nanowire [30]. The VCPD is expected to be higher at the N-polar than the Ga-polar surfaces. This assumption is in agreement with our calculations presented below and with earlier reports on GaN epitaxial layers [46] and nanowires [47] grown according to the MBE technique.

In the topography map (Figure 11a), the hexagonal wires appear to be higher and brighter. In the KPFM contrast (Figure 11b), still larger brightness differences between the hexagonal-shaped (N-polarity) and oval-shaped (Ga-polarity) NWs are visible. This means that the N-polar wires had a much higher electrical potential than the Ga-polar ones. In Figure 11c, the height and VCPD profiles along the line marked in Figure 11a,b through four NWs are plotted as red and black curves. The black curve denotes the height profile from the AFM topography image, in which two NWs with flat hexagonal tops—NW2 and NW4—and two NWs with rough, oval tops—NW1 and NW3—are visible. The red curve is the profile of VCPD from the KPFM contrast map. It can be seen that the surface potential is not correlated with the height of the NWs and is much higher for the NWs with flat

hexagonal tops; therefore, the contrast shown in the KPFM map is not an artifact resulting from the NW topography. The results of the KPFM experiments are fully consistent with the results concerning the polarity determination, performed by KOH etching of the NWs (see Section 3.1).

− − **Figure 11.** (**a**) Atomic Force Microscopy (AFM) topography (color scale from 0−500 nm), (**b**) the Kelvin probe force microscopy (KPFM) contrast map (color scale from 0−0.2 V), and (**c**) line-scan profiles of the height (in black) and potential on the surface (in red) through four NWs.

#### **4. Discussion**

It is known that during the growth of catalyst-free GaN on silicon using the PAMBE technique, the N-rich conditions promote columnar growth, leading to the formation of NWs that are mainly nitrogen-polar [18,48–50]. In the case of our GaN/AlGaN NW LED structures, the n-type part of the structure (GaN:Si) and the active region of the structure with three QWs, which were grown under N-rich conditions, exhibit nitrogen polarity, which was confirmed for many previously grown structures. During the growth of the ptype part, to enhance the Mg incorporation and increase the NW diameter, the temperature was decreased, which promoted lateral growth. However, lowering the growth temperature increases the possibility of stacking fault formation and switching of the polarity to Ga-polar. Moreover, such conditions enabled the more efficient incorporation of Mg dopant [51]. The polarity inversion induced by Mg-doping in GaN has been observed before by many groups, but usually 2D structures were studied, and the polarity changed from Ga- to N-polar in the samples from both the MOCVD and MBE growth processes [52–56]. It was reported that doping of N-face GaN with Mg resulted in a crystalline phase transition from wurtzite to zinc blende [57,58]. However, switching from N- to Ga-polarity was also observed as a result of Mg-doping during PAMBE growth. In one case, it was achieved by exposing the surface of the GaN layer to Mg and N fluxes during growth interruption at a reduced substrate temperature, and the formation of a MgxN<sup>y</sup> compound was suggested to be responsible for inverting the crystal polarity [59]. In the second case, the polarity of GaN was inverted from the N- to Ga-face by inserting a composite AlN/AlO<sup>x</sup> interlayer structure at the inversion interface [60].

In wurtzite structures, the growth polarity strongly influences the incorporation of dopants and impurities and affects the formation of native point defects. Additionally, the doping behavior of Mg and resulting conductivity of the doped structures were found to be strongly dependent on the surface polarity of the growing GaN planes. In the PAMBE process, the incorporation rate of the most common dopant used for achieving p-type conductivity is significantly higher in Ga-polar structures, while N-polar growth inhibits effective Mg doping. A lower Mg incorporation efficiency in N-face GaN has been predicted theoretically [61] and confirmed experimentally [53,62,63].

Oxygen atoms are one of the most common impurities in the GaN growth process, especially if Al-containing alloys are used. Oxygen acts as a shallow donor [64,65] and compensates Mg acceptors. It was widely reported that oxygen incorporation is also dependent on the growth polarity. Oxygen can be incorporated much more easily into N-polar (Al,Ga,In)N films in comparison to metal polar films. The higher oxygen incorporation efficiency on the N-polar surface was associated with a facile exchange between N-surface atoms and oxygen [62,66,67].

In order to understand the origin of different electrical characteristics of the two types of nanowires, a set of band profiles was calculated using a one-dimensional Drift-Diffusion Poisson Schrodinger Solver [68]. For simulation purposes, for each III-nitride layer, we consider only the majority dopant (i.e., acceptor or donor) concentrations, defined as the difference between the majority and minority type of dopant. Keeping in mind that the oxygen (donor) background doping levels that are typically observed are in a range between 1 × 1016–5 × 10<sup>17</sup> atoms/cm<sup>3</sup> , the numbers indicated below may be significantly different to those implied by the dopant fluxes introduced during growth. In Figure 12, two band profiles obtained for unbiased structures of both types of NWs are compared. The black dotted line corresponds to the fully N-polar structure with n-type doping (2 × 10<sup>18</sup> atoms/cm<sup>3</sup> ) below the MQW and p-type AlGaN:Mg (2 × 10<sup>15</sup> atoms/cm<sup>3</sup> ) above it. Such a lightly doped structure should result in an extremely thick depletion region and could be easily perturbed by other effects, such as residual oxygen contamination. The blue dashed line in Figure 12 indicates how residual oxygen, at the relatively low concentration of 1 × 10<sup>16</sup> atoms/cm<sup>3</sup> , in the nominally p-type region changes the band profile, resulting in a lack of p–n junction in the structure. This effect could explain the lack of a rectifying characteristic in N-polar NWs, as shown in Figure 10. The last profile, represented by the red line in Figure 12, corresponds to the structure in which a p-type layer polarity inversion from N- to Ga-polar took place at the beginning, and the incorporation of 5 × 10<sup>16</sup> atoms/cm<sup>3</sup> of Mg started. Our calculations show that this corresponds to a depletion width of approximately 300 nm situated above the MQW region, which is in agreement with the results of the EBIC studies. The assumed difference between the Mg concentrations for the Ga- and N-polar structures was in accordance with the previously reported discrepancy between both polarities [53].

In fact, there is only a small difference in doping between the models, which are represented by the black and blue curves in Figure 12. However, as shown in Figure 12, this difference is sufficient to induce a change from a weak p-type to a weak n-type semiconductor. It is possible that some N-polar NWs are slightly p-type, some are slightly n-type, and some are insulating. We have observed that some of them are electrically conductive, such as those shown in Figures 9a and 10, and some are insulating. The slightly n-type NWs have their electric fields screened at the top, which is shown as the flattening of the blue dashed curve in Figure 12. This means that this part of the NW can emit CL, which is in agreement with the observation that some N-polar NWs emitted bright luminescence from their tops. It is also worth noting that the potential on the top of the Ga-face NWs is lower than the potential of the N-face NWs, which is in agreement with the KPFM measurements.

− Ф Ф **Figure 12.** Conduction and valence band profiles for three different nanowire structures, as a function of the location in the wire. The growth direction goes from left to right. The dotted black line denotes the profile for a lightly doped N-polar structure, with a 2 × 10<sup>15</sup> atoms/cm<sup>3</sup> Mg concentration. The dashed blue line denotes the profile for the case in which oxygen compensated p-type doping, resulting in n-type doping. The red profile corresponds to the structure in which polarity inversion took place right at the beginning of the p-type layers. The dotted arrow indicates the position where polarity inversion took place for the wire presented in red. From that moment, the structure was Ga-polar, with a 5 × 10<sup>16</sup> atoms/cm<sup>3</sup> Mg concentration. The vertical axis shows the electron energy E = −eΦ, where Φ is the potential.

#### **5. Summary and Conclusions**

A comprehensive dataset revealing the optical and electrical properties of individual nanoLED structures embedded in GaN/AlGaN NWs emitting EL at approximately 3.3 eV has been collected. The results were obtained by complementary SEM, TEM, CL, and EBIC spectroscopy and imaging; AFM and Kelvin probe force microscopy experiments; and I–V characteristics measurements for individual NWs using a nano-probe. Further information was acquired by studying the properties of NW ensembles using electroluminescence and time-resolved photoluminescence.

We identified two types of nanowires that differ in terms of the shape of their upper parts. KOH etching and KPFM studies show that they have different polarities. We believe that unintentional fluctuations in the conditions during the growth of the Mg-doped AlGaN part of the structure caused a reversal of the growth polarity of some NWs from the N- to Ga-face.

The interpretation of the experimental data was supported by the results of numerical simulations of the electronic band structure of individual NWs, which allowed the physical mechanisms responsible for the activation or deactivation of the p–n junctions, depending on the polarity of the last p-type part in the nanoLED structures, to be discussed.

The experimental techniques with a spatial nano-resolution allowed for the study of the correlating luminescence and electrical properties of NWs, with the growth polarity switching in the p-type AlGaN part of the NW. The CL spectra showed that the N-face NWs had emissions at the highest energy of 3.83 eV, while the Ga-face NWs energy of emissions was mainly at 3.78 eV. The lower emission energy of the Ga-polarity NWs indicated that there was a lower Al content in these NWs. This, in turn, suggested a higher Ga desorption for N-face NWs than for Ga-face NWs. The monochromatic CL maps provided clear evidence of the QW emission at about 3.5 eV at helium temperature, which is the main recombination path producing electroluminescence in nanoLEDs. We observed that only Ga-polar NWs emitted bright luminescence from QWs. This is a clear indication that

switching polarity was crucial for LED efficiency. According to our calculations of the band structure of the NWs, the change in polarity induced an electric field profile that attracted carriers to the region of QWs, which is the necessary condition for bright light emission. The EBIC mapping proved that p–n junctions were not active in N-polarity NWs, while Ga-polar NWs generated a strong current when excited in the junction area. The observed depletion region was mostly on the p-type side of the junction due to the relatively weak activation of Mg acceptors. The time-resolved photoluminescence showed that the lifetime of the QW-related emission was about 1–2 ns, which proved good quality QWs. The electroluminescence experiments showed that an ensemble of the investigated nanoLED NWs was acceptable as a source of UV radiation, with an energy of about 3.3 eV.

The collected data and results of numerical simulations indicate that the intentional polarity inversion between the n- and p-type parts of NWs is a potential path towards the development of efficient nanoLED NW structures. We believe that a more effective p-type doping of AlGaN, supporting an active p–n junction, would prevail over the disadvantage of the possible creation of structural defects in the polarity inversion region.

**Author Contributions:** Conceptualization, K.P.K., A.R., B.J.K. and Z.R.Z.; methodology, A.R., K.P.K. and B.J.K.; software, H.T.; formal analysis, A.R., K.P.K. and H.T.; investigation, A.R., K.P.K., S.T., S.K., R.B. and U.J.; resources, M.S. and Z.R.Z.; writing—original draft preparation, A.R., K.P.K., B.J.K. and H.T.; writing—review and editing, Z.R.Z., S.T., U.J. and M.S.; visualization, A.R., K.P.K. and H.T.; supervision, K.P.K., B.J.K. and Z.R.Z.; funding acquisition, B.J.K., Z.R.Z. and H.T. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was partly funded by the Polish National Science Centre (NCN), Grants Nos. UMO-2016/21/B/ST5/03378 and UMO-2016/23/B/ST7/03745, by the Polish National Centre for Research and Development project PBS1/A3/1/2012 Pol-HEMT and by the Foundation for Polish Science, co-financed by the European Union under the European Regional Development Fund Homing POIR.04.04.00-00-5D5B/18-00.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Acknowledgments:** The authors would like to thank M.Sc. Kamil Klosek for his assistance in the growth of the NW structures, M.Sc. Giorgi Tchutchulashvili for his help with the SEM imaging and KOH etching of the NW LED structures, and M.Sc. Michal Sztyber for his assistance in the electroand photo- luminescence measurements.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Electronics* Editorial Office E-mail: electronics@mdpi.com www.mdpi.com/journal/electronics

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18

www.mdpi.com ISBN 978-3-0365-1521-2