Insights into annealing-induced ohmic contact ...

3 downloads 588 Views 2MB Size Report
Feb 12, 2015 - S Chandramohan1, Beo Deul Ryu1, Tae Hoon Seo2, Hyunsoo Kim1,. Eun-Kyung .... analyser and a silicon photo detector, respectively. 3.
Journal of Physics D: Applied Physics J. Phys. D: Appl. Phys. 48 (2015) 095102 (7pp)

doi:10.1088/0022-3727/48/9/095102

Insights into annealing-induced ohmic contact formation at graphene/p-GaN interface with a NiOx contact layer S Chandramohan1, Beo Deul Ryu1, Tae Hoon Seo2, Hyunsoo Kim1, Eun-Kyung Suh1 and Chang-Hee Hong1 1

  Semiconductor Physics Research Center, School of Semiconductor and Chemical Engineering, Chonbuk National University, Jeonju 561-756, South Korea 2   Innovative Materials Research Center, Korea Institute of Science and Technology, Jeonbuk 565-905, South Korea E-mail: [email protected] and [email protected] Received 5 December 2014, revised 7 January 2015 Accepted for publication 8 January 2015 Published 12 February 2015 Abstract

The effect of rapid thermal annealing on the graphene/p-GaN interface with a thin NiOx contact layer in a GaN light-emitting diode (LED) was investigated. Raman and ultraviolet photoemission measurements revealed charge transfer in graphene on a NiOx contact layer. Rapid thermal annealing at temperatures above 350 °C induced the formation of metallic Ni with a simultaneous increase of Ni3+ ions within the NiOx layer covered by the graphene. As a result, the graphene electrode on a NiOx/p-GaN surface of a blue GaN LED formed a lowresistance ohmic contact with a specific contact resistance of 5.3  ×  10–4 ohm cm2. The ensuing LED chip offered a low forward voltage, comparable to that of the Ni/Au counterpart. This is attributed to a combination of distinct phenomena arising due to a work function modulation in graphene, increase in the carrier concentrations at the near surface region incited during the formation of NiOx, and thermal reduction induced modifications within the NiOx contact layer. At elevated annealing temperature, however, oxidation of graphene led to poor current spreading and thus a low optical output, but both these constraints were eliminated by using a few layer graphene. Keywords: graphene electrode, p-type GaN, NiOx, rapid thermal annealing, ohmic contact, light-emitting diode S Online supplementary data available from stacks.iop.org/JPhysD/48/095102/mmedia (Some figures may appear in colour only in the online journal)

1. Introduction

a low resistance ohmic contact to Mg-doped GaN top contact layer. This is essentially due to deficient p-type doping caused by the deep nature of the Mg acceptors, which prevents one from achieving high hole concentrations needed to make a tunnelling ohmic contact [3, 4]. The p-type GaN layer left with high resistivity needs a transparent conductive layer in order to enable hole injection uniform in the entire region of the device. For this purpose, both indium tin oxide (ITO) and Ni/Au have been predominantly used as a transparent conductive electrode (TCE) in the fabrication of GaN based LEDs [4].

Gallium nitride (GaN) based blue light-emitting diodes (LEDs) have come a long way in achieving high-brightness for commercial success and currently they are used in many applications such as solid-state white lighting, full-color displays, horticulture, architectural lighting, and so on [1]. Current spreading electrode in these devices plays a decisive role in developing chips for high-power applications [2]. For instance, in the case of devices with lateral current injection geometry, there is some complexity in achieving 0022-3727/15/095102+7$33.00

1

© 2015 IOP Publishing Ltd  Printed in the UK

S Chandramohan et al

J. Phys. D: Appl. Phys. 48 (2015) 095102

The optical transmittance measurements were carried out by using a JASCO V-570 ultraviolet − visible − near-infrared (UV − visible − NIR) spectrophotometer. Combined x-ray and ultraviolet photoelectron spectroscopy (XPS/UPS) measurements were performed on an AXIS-NOVA, Kratos spectrometer with a monochromized Al Kα x-ray source (1486.6 eV). Differential scanning calorimetry (DSC) was performed on both NiOx and graphene/NiOx samples using SDT Q600 V 20.9 Build 20 (Universal V4.5A TA instruments) thermal analyzer under N2 ambient. Raman spectra of graphene were obtained by using a 514 nm line of an argon ion laser. The resistivity measurements were carried out using a Hall Effect measurement system at room temperature with a magnetic field of 0.556 Tesla. The surface topography was probed by atomic force microscopy (AFM) in tapping mode. Current– voltage (I–V) and light output measurements were carried out at wafer scale in a probe station equipped with a parameter analyser and a silicon photo detector, respectively.

Recently, the two-dimensional material graphene is considered as a potential alternative to commonly used ITO or Ni/ Au due to its outstanding properties together with low production costs. Graphene combined with ultrathin metal films and/ or metal nanowires/nanodots have been demonstrated to perform better than bare graphene as a TCE in GaN-based LEDs [5–10]. In the hybrid electrode configuration, the metal compensates the high sheet resistance of the graphene, enhances the adhesion of graphene with the underlying epilayer, and acts as a contact layer for facilitating ohmic contact formation. In the search of finding an appropriate contact layer, different groups proposed different materials which include a thin NiOx, ITO and silver nanowire networks [8, 10, 11]. Among these, the non-stoichiometric NiOx has many advantages such as high transparency to visible light, p-type conductivity, and high work function as high as 5 eV [12, 13]. Furthermore, it is well known that the crystalline NiOx at the metal/p-GaN interface reduces the barrier height and provokes hole injection [4, 14]. Indeed, it has been shown that the graphene electrode performance can be significantly improved in GaN LEDs by using NiOx at the graphene/p-GaN interface [8, 15]. However, the ohmic contact formation mechanism in graphene/p-GaN junction using NiOx contact layer is not well understood, especially its apparent dependence on the annealing process. Herein, the effects of rapid thermal annealing (RTA) of graphene formed on NiOx/p-GaN are systematically evaluated to aid understanding of the origin of ohmic contact formation. The results showed new insights into the physics at the surface and interface of graphene on metal oxides and nitrides under thermal treatment. The problematics discussed in this paper may be of interest for the field of III-nitrides LED development in general, e.g., for the case of AlN in both its three-dimensional and even two-dimensional electronic material variant [16, 17].

3. Results It is common practice to anneal the metal contacts to p-GaN in an oxygen containing ambient (either in only O2 or N2+O2) at elevated temperatures because annealing in such reactive ambient gives rise to enhanced carrier concentration near the surface by the removal of hydrogen that bonds with Mg or N. This tactic has been widely used for the formation of low resistance ohmic contact to p-GaN [4, 14, 18]. In this work, thermal oxidization of Ni is therefore preferred for the formation of NiOx. Prior to device fabrication, the characteristics of the NiOx film as a function of annealing ambient, temperature, and time are evaluated to find out an optimum oxidization condition. It is observed that the transmittance of the Ni films increases with increasing annealing temperature and complete oxidization occurs at temperatures ≥500 °C (online supplementary figure  S1)(stacks.iop.org/JPhysD/48/095102/mmedia). Moreover, the transmittance of the films annealed for different durations from 90 to 300 s showed no significant variations indicating that an annealing time of 90 s or above would be sufficient for complete oxidization of the Ni film. This is further supported by the XPS results (online supplementary figure S2) (stacks.iop.org/JPhysD/48/095102/mmedia), which obviously showed the absence of metallic Ni with similar Ni/O atomic ratio for the films annealed in either ambients for 90 and 300 s. It should be noted that both NiO (Ni2+) and Ni2O3 (Ni3+) phases coexisted in all the films analogous to previous reports [8, 19, 20]. Furthermore, the transmittance of the film is found to be a strong function of Ni film thickness and the films annealed under N2+O2 ambient have slightly enhanced transmittance. In consideration of the above results, a thin Ni film oxidized in N2+O2 ambient for 180 s is employed in our devices. The characteristics of the graphene/NiOx bilayer stack are also evaluated as a function of annealing temperature. Figure 1 shows the optical transmittance of the graphene/NiOx bilayer stacks subjected to RTA at different temperatures. The transmittance of the graphene/NiOx stack is ~95% at 450 nm and shows a systematic decrease with increasing annealing temperature. For instance, the transmittance values at 450 nm

2.  Experimental details The InGaN/GaN LED wafer used in this study was grown by metal-organic chemical vapour deposition on a c-plane sapphire. The peak emission wavelength of the sample was around 450 nm. The wafer was pre-etched by a combination of lithography and dry etching processes for forming mesa structure LED chips (1  ×  1 mm2). All the samples were cleaned with standard organic solvents followed-by surface treatment with buffered oxide etch solution. The device fabrication consisted of the following steps. First, a thin Ni film of 2 nm was deposited onto the p-GaN surface by electron beam evaporation and subsequently oxidized in a rapid thermal processing system under oxygen-containing ambient. For the optimization of graphene/NiOx bilayer stack (the terminology ‘bilayer stack’ refers to graphene on NiOx), Ni films of different thicknesses were deposited onto glass substrates. Monolayer graphene films were instantly transferred onto the glass and epiwafer substrates with and without NiOx via a wet-transfer procedure as described elsewhere [15]. Post-graphene-transfer annealing of the samples was performed at 350, 450, and 550 °C for 300 s in ambient N2. Finally, Ti/Au metal fingers were formed onto the exposed n-GaN and graphene electrode as metal pads for current injection in making of devices. 2

S Chandramohan et al

J. Phys. D: Appl. Phys. 48 (2015) 095102

[21]. In order to make sure of the reliability of the above result, a similar contact structure is fabricated with a different quality epi-wafer which eventually offered a reasonable ρc of 7.6  ×  10–4 ohm cm2. This result corroborates the forward I–V characteristics of the devices. 4. Discussion In order to understand the reason why the forward voltage and the contact characteristics improves with the annealing temperature, the structure and interface electronic properties of graphene on the p-GaN substrate with NiOx contact layer are evaluated as a function of annealing temperature. Metal oxide and substrate induced Fermi level shift in graphene is a well-known phenomenon, by which the work function of graphene can be modified [22–25]. Raman spectroscopy is an authoritative technique to study the charge transfer induced doping in carbon-based materials. Figure  4(a) shows the micro Raman spectra of the graphene sheet on a p-GaN with and without NiOx contact layer. In the Raman spectrum of graphene on p-GaN, the characteristic G and 2D peaks are observed at 1588.3 cm−1 and 2687 cm−1, respectively. We observed an up-shift and stiffening of the G band for graphene on NiOx/p-GaN and this effect is further enhanced upon thermal annealing. For instance, the G band position is measured to be 1591.8 cm−1 and 1596.5 cm−1, respectively, for the graphene on NiOx/p-GaN substrate before and after annealing at 550 °C. In a similar fashion, the 2D peak position also upshifted with respect to the addition of NiOx contact layer and annealing. This shows a clear signature of charge transfer induced p-doping of graphene [23, 26–28]. The doping of graphene is further verified by evaluating the ratio of the intensity of the 2D peak to the intensity of the G peak i.e., I(2D)/I(G), which shows a significant decrease from 1.6 (on p-GaN) down to 1.15 (on NiOx/pGaN) and 0.87 (on NiOx/pGaN with subsequent annealing at 550 °C). The stiffening of the G band is usually ascribed to disorder and/ or doping effects [29]. Even though the positional change and stiffening of the Raman G band, and the decrease of the I(2D)/I(G), in unison can be strongly related to doping in graphene [28–31], the present results cannot be related only to doping effect because the intensity grows up around the D band position for the graphene on NiOx contact layer, as confirmed from the Raman spectra with normalized intensity scale (online supplementary figure  S3) (stacks.iop.org/ JPhysD/48/095102/mmedia). However, it should be noted that no further change in the D band intensity is observed after the annealing process. Ultraviolet photoemission measurements were performed to discriminate between doping and annealing effects in graphene. Figure 4(b) shows the UPS spectra of representative graphene samples. The work function (Φ) is estimated from the empirical relation Φ = hυ − (EF − Ecutoff), where hυ, EF, and Ecutoff are the photon energy of the excitation light (21.22 eV), the Fermi level edge, and the measured secondary electron cutoff, respectively [6]. The surface work function of graphene on NiOx is calculated to be 4.98 eV and 4.67 eV, respectively, before and after annealing at 450 °C. These values are higher than the work

Figure 1.  Optical transmittance spectra of graphene/NiOx bilayer

stacks annealed at different temperatures.

are estimated to be 93.6%, 88.6% and 87%, respectively, for the films annealed at 350 °C, 450 °C, and 550 °C. Figure 2 shows the XPS Ni 2p3/2 core-level spectra of a graphene/NiOx bilayer stack before and after RTA. The spectra were fitted by using a XPS peak fitting program (XPSPEAK4.1) with Shirley background. As shown in the figure, the Ni 2p3/2 spectrum could be well fitted by using two peaks at a binding energy position of 854 eV and 855.8 eV, respectively. The binding energy difference of 1.8 eV between the two peaks supports the assignment of NiO (Ni2+) and Ni2O3 (Ni3+) phases as the prevailing states in our films [20]. Upon thermal annealing, the line shape of the spectrum changes and a new peak emerges at a binding energy position of 852.9 eV, which is ascribed to the formation of metallic Ni (Ni0) in the film [13]. Moreover, a positive shift in the binding energy position of about 0.34 eV is observed for the Ni2+ and Ni3+ peaks. The speculation that the new peak might be a sign of formation of NiC is excluded as the DSC spectrum indicated obviously no endothermic or exothermic peaks (figure 2(b)). Therefore, the observed decrease in the optical transmittance is attributed to the formation of metallic Ni in analogous to previous experiment [15]. Figure 3(a) shows the I–V curves of GaN LEDs fabricated with graphene electrodes by incorporating similar NiOx contact layer. One can notice a systematic decrease in the forward voltage with increasing annealing temperature. The forward voltages at an injection current of 20 mA for the pristine device and devices annealed at 350 °C, 450 °C, and 550 °C are estimated to be 5.4 V, 4.4 V, 3.9 V, and 3.5 V, respectively. Thus, it is apparent that the electrode performance is significantly enhanced after being annealed at elevated temperature. Figure  3(b) shows the I–V curves measured on a graphene/ NiOx/p-GaN circular transmission line model contact structure annealed at 550 °C for different gap spacing. It is obvious from the figure  that the graphene electrode forms an ohmic contact to the p-GaN by the addition of the NiOx contact layer. Moreover, the ρc is estimated to be 5.3  ×  10–4 ohm cm2, which is comparable to that of the state-of-the art Ni/Au electrode 3

S Chandramohan et al

J. Phys. D: Appl. Phys. 48 (2015) 095102

Figure 2. (a) XPS Ni 2p3/2 spectra of graphene/NiOx bilayer stack before and after annealing at 450 °C. (b) DSC heat flow traces of NiOx

and graphene/NiOx films heated in N2 ambient.

of the carboxyl functional groups by depolymerization of the PMMA and a reduction in the p-doping of graphene [31, 32]. However, the Raman spectrum corresponding to the annealed sample shows a progressive blue shift in the peak positions, which is incongruous to the UPS results. This anomaly can be associated to the competition between the different doping effects by the simultaneous removal of the PMMA and the oxidation of graphene, as reported by Ahn et al [32]. Because the decomposition of the carboxyl functional groups at high annealing temperatures and associated oxidation of graphene would also give rise to a shift in the Raman peaks. Based on the above considerations, the relatively high stiffening of the G and 2D peaks and increased intensity of the G peak after thermal annealing could be a sign of oxidation of graphene. Because of the overlapping of the D band position with the substrate phonon modes, we were unable to figure out how much broadening is contributed by doping and disorder effects. The effective resistivity of the graphene/p-GaN and graphene/NiOx/p-GaN stacks is estimated to be 2.55   ×   10–1  ohm cm and 8.2   ×   10–2 ohm cm, respectively. A further decrease in the resistivity of the trilayer stack (hereafter refers to graphene/NiOx/p-GaN, unless otherwise stated) down to 5.6  ×  10–2 ohm cm is observed after being annealed at 550 °C. It is known that the formation of Ni2+ vacancies and the increase of Ni3+ ions both could increase the hole concentration in the non-stoichiometric NiOx film [12, 13]. To verify this supposition in our contact geometry and to evaluate the effect of contact layer, the resistivity of the trilayer stack subjected to RTA at 350 °C is measured with and without graphene. One can notice that the resistivity of the p-GaN with NiOx contact layer after the removal of the graphene is lower than that of the initial NiOx/p-GaN layer, as shown in figure 4(c). The above results indicate that the contact layer itself plays a significant role in decreasing the resistivity of the p-GaN. Indeed, Ni is known to react with the surface Ga oxides during annealing in oxygen ambient where it could act as a cleaning agent [18]. This is also found true based on our experience during the Hall measurements. For instance, we were able to

Figure 3. (a) I–V characteristics of the LEDs with graphene and graphene/NiOx hybrid electrodes at different annealing temperatures. (b) I–V characteristics of graphene/NiOx/p-GaN contact for different gap spacing.

function of the pristine graphene (4.56 eV). This result substantiates the charge transfer induced doping of graphene on NiOx, as discussed earlier based on the Raman measurements. However, a reduction in the work function of 0.2 eV is observed for the annealed sample, compared to non-annealed counterpart. The comparatively higher work function of the graphene before annealing is believed likely due to the presence of residual polymethyl methacrylate (PMMA) on the surface (online supplementary figure  S4) (stacks.iop.org/JPhysD/48/095102/ mmedia), because it has been shown that the residual PMMA is a source of hole doping in transferred graphene [31–33]. On the other hand, the surface of the graphene subjected to annealing is rather clean with a low surface roughness, as evidenced from the AFM images (online supplementary figure S4) (stacks.iop. org/JPhysD/48/095102/mmedia). This result is consistent with previous reports which have indicated that thermal annealing at temperatures above 300 °C could lead to effective removal 4

S Chandramohan et al

J. Phys. D: Appl. Phys. 48 (2015) 095102

Figure 4. (a) Raman spectra of graphene on a p-GaN substrate with and without a NiOx contact layer. (b) UPS spectra acquired on

graphene and graphene/NiOx. (c) Variations in resistivity of p-GaN with respect to contact layer and annealing conditions. (d–f) AFM surface topography of a p-GaN, NiOx/p-GaN and graphene/NiOx/p-GaN annealed at 550 °C, respectively.

measure the resistivity of the NiOx/p-GaN sample consistently with comparable values in each measurement, whereas measurements on a p-GaN were unreliable. It is also observed that the resistivity of the trilayer stack decreases with increasing annealing temperature, signifying a substantial modification in the near surface region at the graphene/NiOx/p-GaN interface. Furthermore, the bulk carrier concentration in the respective trilayer stack is found to increase from 2.2  ×  1017 cm−3 to 5.9  ×  1017 cm−3 following the annealing process. To give a further insight, the morphology of the samples is evaluated. Figures 4(d)–)f) represent the surface topography of the p-GaN, NiOx/pGaN, and graphene/NiOx/pGaN annealed at 550 °C, respectively. The bright particles detected in the later sample (figure 4(f)) are most likely the metallic nickel, because the XPS analysis revealed the formation of Ni0 in annealed samples. In addition, the spatial distribution of the particles is neither random nor uniform, but has a contour similar to that of the line defects (grain boundaries and wrinkles) in graphene. This implies that during the annealing process, the reduction of NiOx is prominent under the graphene line defects. The mechanism of formation of Ni0 can be understood by revisiting figure 2(a). The peak area analysis indicates an increase in the Ni3+/Ni2+ ratio for the annealed sample. This result suggests that NiOx is simultaneously reduced and oxidized at elevated temperatures leading to the formation of Ni0 and increase in the Ni3+ ions [19, 20]. The reduction might have been caused by the excess oxygen in the form of surface hydroxide introduced

during the wet transfer process or through the reactions with the polymer residues on the graphene line defects. When the oxygen occupies the interstitial position, Ni2+ vacancies can be created and their subsequent ionization forms more Ni3+ ions and holes, as evidenced from the significant increase in the contribution of Ni3+ ions in both Ni 2p and O1s spectra (online supplementary figure S5) (stacks.iop.org/JPhysD/48/095102/ mmedia). The increasing of the effective carrier concentration in the trilayer stack therefore suggests the formation of nonstoichiometric p-type NiOx at the interface. The current spreading ability and the optical output power are also investigated in an attempt to understand the transmittance and sheet resistance tradeoff for the graphene electrodes on NiOx contact layer as a function of annealing temperature. The light output versus current characteristics of the LED chips are compared in figure 5. For the LED with graphene on NiOx annealed at 350 °C, the light output is enhanced over that of a device made with Ni/ Au translucent electrode. But, annealing at temperatures above 450 °C results in a decrease in the light output at injection currents above 20 mA, despite an improvement in the forward voltage. Considering the high transmittance of the graphene/ NiOx bilayer stack even after annealing as depicted in figure 1, the low light output is a direct consequence of the poor sheet resistance of the graphene. This can be understood by looking at the light emission images of the respective devices. An obvious current crowding occurs under the metal pad of the 5

S Chandramohan et al

J. Phys. D: Appl. Phys. 48 (2015) 095102

Figure 5.  Comparison of the light output characteristics of different LED chips investigated. The insets are the photographs of the LED chips during the light emission at 1 mA injection current.

(a) A systematic decrease in the optical transmittance of the graphene/NiOx bilayer stack with increasing annealing temperature from 350–550 °C is a consequence of the metallic Ni formation. (b) The observation of Ni particles at graphene line defects with a simultaneous increase of Ni3+ ions suggests that the reduction of NiOx is likely caused due to oxygen indiffusion, the source of which could be the buried water molecules or the polymer residues. (c) The electronic properties of graphene are also modulated with respect to the nature of the substrate and annealing conditions. The work function of graphene on NiOx is enhanced due to charge transfer and defect activation is dominated at elevated annealing temperatures above 450 °C. (d) A systematic decrease in the resistivity and increase in the carrier concentration of graphene/NiOx/p-GaN stack with increasing annealing temperature suggests that the observed low resistance ohmic contact formation and associated reduction in the forward voltage are primarily governed by the local variations in the physical properties of the NiOx contact layer. (e) Stiffening of the Raman bands with increased G band intensity, a reduction in the surface work function, and a low optical output with an obvious current crowding observed for the graphene electrode proves oxidation of graphene at 550 °C. However, the effect of oxidation may be prevailed over by using few layer graphene.

device subjected to annealing at 550 °C. We attribute this effect to the increase in the sheet resistance of graphene at this particular annealing temperature, as the annealed graphene/NiOx bilayer stack on a non-conducting substrate showed a three-fold increase in the sheet resistance. Thus, it is evident that oxidation of graphene likely occurs, as already discussed on the basis of stiffening of the Raman bands. However, we did not observe any similar current crowding when a multi- or few-layer graphene was used. As a proof of concept, the performance of the device made with a three layer graphene formed by layerby-layer transfer and annealed at 550 °C is also compared, as shown in figure 5(b). In this case, the light output is enhanced compared to its Ni/Au counterpart, and there is no obvious current crowding occurs. Considering the fact that a high sheet resistance of graphene could bring in severe current crowding to the device, the alleviation of current crowding using a fewor multi-layer graphene suggests that the effect of oxidation is outweighed by comparatively low sheet resistance of few- and multi-layer graphene. It is worthy of mentioning in this connection that the depolymerization of residual polymer may not necessarily increase the sheet resistance of graphene, as graphene subjected to similar annealing conditions on substrates without NiOx has shown improved sheet resistance. This means that the extent of oxidation may likely be different in graphene with different layer numbers such that a monolayer graphene is more vulnerable to oxidation on NiOx at high temperatures. Thus, by controlling the sheet resistance of graphene and optimizing the annealing conditions it is possible to form a low resistance stable ohmic contact to p-GaN with the use of NiOx contact layer and reliable device performance can be achieved.

Finally, the present study unveils that by controlling the layer numbers of graphene and annealing conditions, the combination of graphene electrode with NiOx contact layer would be a promising alternative to conventional electrodes in the fabrication of III-nitride LEDs of superior electrical and optical performances.

5. Conclusions In summary, a systematic study has been carried out in order to understand the origin of ohmic contact formation at the graphene/p-GaN interface with a thin NiOx contact layer under RTA. RTA of graphene/NiOx bilayer stack modifies the inherent properties of the individual constituents. From the experimental results, the following conclusions are derived.

Acknowledgments This research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and 6

S Chandramohan et al

J. Phys. D: Appl. Phys. 48 (2015) 095102

Technology (2012R1A1A3011103 & 2013R1A1A2013044), and financially supported by the Ministry of Education (MOE) and NRF through the Human Resource Training Project for Regional Innovation (2013H1B8A2032197).

Defects in hexagonal-AlN sheets by first-principles calculations Eur. Phys. J. B 85 48 [18] Pettersen S V, Grande A P, Tybell T, Riechert H, Averbeck R and Grepstad J K 2007 Formation and electronic properties of oxygen annealed Au/Ni and Pt/Ni contacts to p-type GaN Semicond. Sci. Technol. 22 186–93 [19] Kim D S and Lee H C 2012 Nickel vacancy behavior in the electrical conductance of nonstoichiometric nickel oxide film J. Appl. Phys. 112 034504 [20] Yun D J and Rhee S W 2008 Deposition of NiOx thin films with radio frequency magnetron sputtering and their characteristics as a source/drain electrode for the pentacene thin film transistor J. Vac. Sci. Technol. B 26 1787 [21] Chandramohan S, Ko K B, Yang J H, Ryu B D, Katharria Y S, Kim T Y, Cho B J and Hong C H 2014 Performance evaluation of GaN light-emitting diodes using transferred graphene as current spreading layer J. Appl. Phys. 115 054503 [22] Meyer J, Kidambi P R, Bayer B C, Weijtens C, Kuhn A, Centeno A, Pesquera A, Zurutuza A, Robertson J and Hofmann S 2014 Metal oxide induced charge transfer doping and band alignment of graphene electrodes for efficient organic light emitting diodes Sci. Rep. 4 5380 [23] Hellstrom S L, Vosgueritchian M, Stoltenberg R M, Irfan I, Hammock M, Wang Y B, Jia C, Guo X, Gao Y and Bao Z 2012 Strong and stable doping of carbon nanotubes and graphene by MoOx for transparent electrodes Nano Lett. 12 3574 [24] Nistor R A, Kuroda M A, Maarouf A A and Martyna G J 2012 Doping of adsorbed graphene from defects and impurities in SiO2 substrates Phys. Rev. B 86 041409(R) [25] Zhong H, Xu K, Liu Z, Xu G, Shi L, Fan Y, Wang J, Ren G and Yang H 2014 Charge transport mechanisms of graphene/semiconductor Schottky barriers: a theoretical and experimental study J. Appl. Phys. 115 013701 [26] Pham V P, Kim K N, Jeon M H, Kim K S and Yeom G Y 2014 Cyclic chlorine trap-doping for transparent, conductive, thermally stable and damage-free graphene Nanoscale 6 15301 [27] Kim D, Lee D, Lee Y and Jeon D Y 2013 Workfunction engineering of graphene anode by bis(trifluoromethanesulfonyl)amide doping for efficient polymer light-emitting diodes Adv. Funct. Mater. 23 5049 [28] Voggu R, Das B, Rout C S and Rao C N R 2008 Effects of charge transfer interaction of graphene with electron donor and acceptor molecules examined using Raman spectroscopy and cognate techniques J. Phys.: Condens. Matter 20 472204 [29] Ferrari A C 2007 Raman spectroscopy of graphene and graphite: disorder, electron–phonon coupling, doping and nonadiabatic effects Solid State Commun. 143 47 [30] Das A et al 2008 Monitoring dopants by Raman scattering in an electrochemically top-gated graphene transistor Nat. Nanotechnol. 3 210 [31] Pirkle A, Chan J, Venugopal A, Hinojos D, Magnuson C W, McDonnell S, Colombo L, Vogel E M, Ruoff R S and Wallace R M 2011 The effect of chemical residues on the physical and electrical properties of chemical vapor deposited graphene transferred to SiO2 Appl. Phys. Lett. 99 122108 [32] Ahn Y, Kim H, Kim Y H, Yi Y and Kim S I 2013 Procedure of removing polymer residues and its influences on electronic and structural characteristics of graphene Appl. Phys. Lett. 102 091602 [33] Deng C et al 2014 Reversible charge-transfer doping in graphene due to reaction with polymer residues J. Phys. Chem. C 118 13890

References [1] Nakamura S and Krames M R 2013 History of gallium– nitride-based light-emitting diodes for illumination Proc. IEEE 101 2211–20 [2] Narukawa Y, Ichikawa M, Sanga D, Sano M and Mukai T 2010 White light emitting diodes with super-high luminous efficacy J. Phys. D: Appl. Phys. 43 354002 [3] Moroc H 2008 Handbook of Nitride Semiconductors and Devices vol 2 (Weinheim: Wiley) [4] Song J O, Ha J S and Seong T Y 2010 IEEE Trans. Electron. Devices 57 42–59 [5] Lee J M, Jeong H Y, Choi K J and Park W I 2011 Metal/ graphene sheets as p-type transparent conducting electrodes in GaN light emitting diodes Appl. Phys. Lett. 99 041115 [6] Chandramohan S et al 2012 Chemically modified multilayer graphene with metal interlayer as an efficient current spreading electrode for InGaN/GaN blue light-emitting diodes J. Phys. D: Appl. Phys. 45 145101 [7] Shim J P, Seo T H, Min J H, Kang C M, Suh E K and Lee D S 2013 Thin Ni film on graphene current spreading layer for GaN-based blue and ultra-violet light-emitting diodes Appl. Phys. Lett. 102 151115 [8] Zhang Y, Li X, Wang L, Yi X, Wu D, Zhu H and Wang G 2012 Enhanced light emission of GaN-based diodes with NiOx/ graphene hybrid electrodes Nanoscale 4 5852 [9] Seo T H, Kim B K, Shin G, Lee C, Kim M J, Kim H and Suh E K 2013 Graphene-silver nanowire hybrid structure as a transparent and current spreading electrode in ultraviolet light emitting diodes Appl. Phys. Lett. 103 051105 [10] Li Z, Kang J, Liu Z, Du C, Lee X, Li X, Wang L, Yi X, Zhu H and Wang G 2013 Enhanced performance of GaN-based light-emitting diodes with graphene/Ag nanowires hybrid films AIP Adv. 3 042134 [11] Kun X, Chen X, Jun D, Yanxu Z, Weiling G, Mingming M, Lei Z and Jie S 2013 Graphene transparent electrodes grown by rapid chemical vapor deposition with ultrathin indium tin oxide contact layers for GaN light emitting diodes Appl. Phys. Lett. 102 162102 [12] Lee M K and Lai Y T 2013 Characterization of transparent conducting p-type nickel oxide films grown by liquid phase deposition on glass J. Phys. D: Appl. Phys. 46 055109 [13] Greiner M T, Helander M G, Wang Z B, Tang W M and Lu Z H 2010 Effects of processing conditions on the work function and energy-level alignment of NiO thin films J. Phys. Chem. C 114 19777 [14] Li X J et al 2014 The significant effect of the thickness of Ni film on the performance of the Ni/Au Ohmic contact to p-GaN J. Appl. Phys. 116 163708 [15] Chandramohan S et al 2013 Impact of interlayer processing conditions on the performance of GaN light-emitting diode with specific NiOx/graphene electrode ACS Appl. Mater. Interfaces 5 958–64 [16] Kakanakova-Georgieva A, Gueorguiev G K, Yakimova R and Janzén E 2004 Effect of impurity incorporation on crystallization in AlN sublimation epitaxy J. Appl. Phys. 96 5293–97 [17] de Almeida Junior E F, de Brito Mota F, de Castilho C M C, Kakanakova-Georgieva A and Gueorguiev G K 2012

7