Influence of environmental conditions and surface treatments on the photoluminescence properties of GaN nanowires and nanofins

To demonstrate the complexity of the PL response of GaN NWs in air, the dynamic behavior of the GaN PL NBE of three differently pre-treated samples is depicted in figure 2(a). The samples have been stored under ambient conditions for 14 months (aged), were HCl-etched for 5 min prior to the measurement or oxidized at a high temperature in the MBE chamber. After switching on the laser at 0 min, the intensity drops by 33% for the aged sample and increases slowly afterwards, the HCl-etched NWs lose over 70% of the initial PL intensity and stay at the low level while the oxidation triggers an enhancement by a factor of 2.5 compared to the initial value in the course of several min. For oxidized and HCl-etched NWs the change of PL intensity is reversible in ambient air, i.e. it recovers to its initial value in the dark (figure S3/figure 5(c)). Thus, the respective PL intensity change cannot be caused by a permanent oxidation induced by the UV light or a structural degradation of the material but could be induced by the physical or chemical binding of atoms or molecules from the ambient to the NWs or a slow and persistent charging effect of surface states, which leads to a changed distribution of photo-excited electrons and holes. Analogous to the PL intensity quenching for HCl-etched NWs [18], the enhancement observed for oxidized NWs is dependent on the applied excitation energy. However, while the quenching can be observed already at intensities levels as low as 1 W cm⁻², the enhancement is not significant below several hundred W cm⁻². To quantify this, we define the enhancement factor (EF), which is the ratio between the PL intensity after a certain time and the initial intensity. Figure 2(b) displays EFs after 10 and 15 min for varying excitation intensities. For excitation power densities lower than about 9 kW cm⁻², the EF increases with increasing intensity. This indicates that in this regime the relevant processes are limited by the availability of photo-excited charge carriers. For very high excitation power densities exceeding 10 kW cm⁻², a saturation of the EF is observed. As for both 10 and 15 min exposure time the saturation is observed at similar intensities and the EF is higher after 15 min, the absolute PL intensity limit is not reached yet, but the enhancement is still going on. Therefore, in the saturation regime the speed of the PL enhancement has to be limited by the speed of adsorption or desorption of molecules from the environment to or from the NW surface. In the following, we will focus on oxidized and HCl-etched NWs as the "aged" condition is not the result of a reproducible surface treatment.

Zoom In Zoom Out Reset image size

The experiments discussed above have been conducted in ambient air and at room temperature with freely exposed samples. As ambient air is a mixture of several gases and contains a relatively high amount of humidity, a more precisely controlled experimental environment is needed to understand the interactions at the NW surface. Figure 3 depicts the normalized saturated PL intensity under continuous UV illumination with an excitation intensity of 18 kW cm⁻² for HCl-etched and oxidized NWs under varied atmospheres and for different temperatures. The PL intensity measured in dry N₂ is used as the reference value for normalization. In this experiment, a fixed spot on the sample was continuously exposed to the laser illumination. The numbers depicted next to the dashed arrows refer to the transition times, i.e. the time it takes until the PL intensity saturates at a constant level after changes of the gas ambient. In dry atmospheres and at low temperature (T_eff = 46 °C), HCl-etched and oxidized NWs show qualitatively the same behavior, but the amplitude of the PL intensity reduction is more pronounced for the HCl-etched sample when the atmosphere is changed from N₂ to SA. After switching back to N₂, the initial intensities are almost recovered, but with a doubled transition time of 2 min for the HCl-etched NWs. When humidity is added to the N₂ flow, the intensities are only slightly changed at low temperature with a small increase for the oxidized and a comparable decrease for the HCl-etched NWs. By switching to humid SA, the PL intensity of the oxidized GaN NWs decreases to 60% of the dry N₂ level and to 15% for the HCl-etched NWs within 1 min. As these values are lower than in wet N₂ and dry SA separately, the quenching properties of oxygen and H₂O at low temperatures appear not to be in competition to each other, but the simultaneous presence of both decreases the PL intensity to the lowest of all observed levels. Possible explanations are the adsorption of H₂O and oxygen at different sites at the NW surface or the formation of a complex containing more than one molecule, which enables highly efficient non-radiative recombination of charge carriers. This configuration is also the most stable one under continuous illumination. At low temperature, the PL intensity stays constant in N₂ for the oxidized NWs as long as the humidity is maintained. For the HCl-etched NWs, the PL intensity recovers only slowly in humid N₂. When the flow-chamber is purged with dry N₂ as a final step, both samples increase their PL emission back to the initial level within the measurement error, which underlines the reversible nature of the observed effects. The PL response of HCl-etched NWs to atmospheric variations changes drastically when the temperature is increased by external heating of the sample to T_eff = 194 °C. Changing from dry N₂ to SA, which causes an intensity reduction at low temperature, results in a PL enhancement by a factor of over 5.5 at 194 °C. The intensity change is more persistent than its low temperature counterpart and a full relaxation to the initial value in N₂ has not been observed. Adding humidity to the N₂ atmosphere results in a doubling of the PL intensity, which is further drastically increased when the oxygen is added. Interestingly, the PL intensity in dry and humid SA are very close to each other at high temperature, while they differ significantly at low temperatures, which indicates that the effect induced by H₂O is replaced by the influence of the additional O₂ at high temperatures. By a subsequent exposure to first humid and then dry N₂, the respective PL levels could not be fully restored but remained at a higher level. Similarly, for oxidized NWs the intensity decrease at low temperature is inverted into an increase by a factor of about 5 within 3 min at high temperature when the atmosphere is changed from dry N₂ to SA. A change back to dry N₂ reduces the intensity to 1.5 times the initial value, a complete recovery was not observed. Adding humidity to the N₂ has only a minor effect on the PL intensity of the oxidized NWs. Both in SA and N₂ atmospheres the values obtained for dry and humid environments are similar. Switching back to first wet and than dry N₂ reduces the luminescence to about 3 and 1.5 times the initial value, respectively. The time needed to reach the respective saturation is comparable for etched and oxidized NWs at high temperature with the exception of the transition from dry N₂ to SA, where the oxidized NWs react more slowly (4 compared to 8 min).

Zoom In Zoom Out Reset image size

In their recently published work, Maier et al observe and discuss in detail an ambivalent nature of the influence of H₂O adsorbates on the PL intensity of (In)GaN NWs, which can have both, a quenching and an enhancing effect [32]. Our observations confirm these results for HCl-etched GaN NWs as their PL is (slightly) reduced at low temperature by adding humidity to a N₂ atmosphere, while at high temperatures, this effect is reversed to an enhancement. We observe a similar ambivalent behavior also for SA. The strong PL intensity decrease observed at low temperature turns into an even more pronounced increase at high temperature. Similar as proposed for H₂O adsorbates, [32]. we conclude that also oxygen alone is able to form two fundamentally different interacting states at the NW surface. To date, we lack the tools for a direct observation of the exact nature of these states, however, we can hypothesize that the observed change is related either to (I) the adsorption of different molecular species of oxygen, i.e. O, O₂, ${{\rm{O}}}_{2}^{-}$ (superoxide) or O₃, (II) a modification of the m-plane GaN surface due to the high temperature and illumination intensity, which adds new available adsorption sites or (III) the adsorption of the same oxygen species but with altered binding strength, e.g. chemisorption instead of physisorption.

Decisive for the validation of hypothesis I is that whether or not a specific oxygen species adsorbs should be mainly dependent on the availability of this species in the environment of the NW. The additional thermal energy added through sample heating is below 15 meV, therefore, the purely thermal influence on the relative occurrences of the oxygen species concentration is low. Further, the presence of O₃ is most probable limited by the intensity of the incident UV light, as the used wavelength of 266 nm does not create ozone, but efficiently degenerates ozone via photolysis [33]. As the laser power is constant throughout the experiment, it is not likely that the O₃ concentration changes significantly. Therefore, we think that the adsorption of O₃ from the environment is unlikely to be the major reason for our observations. However, adsorbed ${{\rm{O}}}_{2}^{-}$ at the NW surface might influence the PL intensity as it is known to catalyze chemical reactions with the surroundings due to its strong reducing character [34]. The availability of ${{\rm{O}}}_{2}^{-}$ depends on the fact whether/how efficient GaN NWs can reduce molecular O₂, which remains a subject of further investigations.

Regarding hypothesis II, to the best of our knowledge, no information on the changes in surface structure or reconstruction in the relevant temperature range are available for m-plane GaN. However, for both polarities of c-plane GaN, phase transitions of the surface reconstruction occur at about 250 °C in UHV [35]. It is possible that for m-planes and under atmospheric pressure, similar phase transitions take place in the temperature range measured. Therefore, hypothesis II can neither be excluded nor proven due to the lack of reliable experimental data.

In favor of hypothesis III is the higher stability of the PL intensity change induced by dry SA at high temperatures: at low temperature, the recovery to the initial PL intensity in dry N₂ takes 2 min whereas at high temperature the initial level is not re-gained even after 4 min. As for elevated temperatures the available energy for deconstruction of a surface layer is higher, the more persistent PL intensity change indicates that the high temperature surface modification is significantly more stable, hence, that the binding of the adsorbed species on a molecular level is stronger.

So far, the quenching and enhancing effects of the air constituents have been discussed in terms of relative intensities normalized to the PL intensity observed in dry N₂ atmosphere. Further insight can be gained by focusing on absolute intensities. The aforementioned ambivalent role of O₂, which can act as both, a quenching and enhancing agent, is again illustrated in figure 4(a), which depicts the absolute PL intensities at three temperatures as a function of the O₂ content in a background of dry N₂. Already at concentrations of about 0.3%, the O₂ content starts to dominate the PL response at low and high temperature, causing a rapid quenching/enhancing, respectively, after which the PL intensity stays roughly at the same level for higher O₂ contents. At an intermediate temperature of T_eff = 97 °C, the PL intensity is almost unaltered by the O₂ content. As already shown in figure 3, the observed PL intensity change is not completely reversible, which indicates a permanent attachment of O₂-related adsorbates. Interestingly, the intensity variation in N₂ across the temperature range measured is significantly higher than in the O₂ atmosphere. Figure 4(b) shows the absolute PL intensities of the same spot of the HCl-etched sample in dry SA and N₂ at different temperatures. In dry N₂ atmosphere, the PL intensity decrease follows an Arrhenius-like behavior over the measured temperature range. In contrast, the PL intensity in SA is almost constant up to a temperature of about 400 K and decreases at higher temperatures. By assuming that the observed quenching is mainly caused by an increase of non-radiative recombination and that the internal quantum efficiency approaches unity at 0 K, the PL decreasing parts of the experimental data can be modeled by an Arrhenius-type fit. The resulting activation energies (E_a ) are similar, with 240 meV in N₂ and 200 meV in SA. This indicates that the thermal quenching of the PL intensity above about 400 K is induced by the same processes in both atmospheres. For GaN, constant PL intensity despite increasing temperature, as observed in SA between 300 and 400 K, is an uncommon behavior and suggests, that an additional, competing effect has to be considered. As discussed later in more detail we suggest that in the case of GaN NWs in SA atmosphere this second effect is the passivation of surface states by oxygen, which leads to a reduction of non-radiative recombination.

Zoom In Zoom Out Reset image size

Under steady state illumination the environmental changes are decisive for the changes of the PL intensity. In contrast, the PL intensity evolution immediately after the UV excitation onset probes the light-induced changes of the surface properties in a given ambient. The PL response of differently treated GaN NWs in air (figure 2) already demonstrated the impact of surface treatments on the environmental sensitivity. In order to understand the underlying mechanisms, similar measurements in controlled atmospheres have been performed. Figure 5(a) depicts the time-resolved PL intensity of oxidized and HCl-etched GaN NWs immediately after starting the UV excitation in dry and humid N₂. The PL intensity of the oxidized GaN NWs in dry N₂ increases within 1 min to 140% of the initial value, whereas in humid N₂ it reaches only 120%. The HCl-etched GaN NWs show almost no change in the PL intensity in dry N₂, but a pronounced quenching of the PL signal to 50% of the initial intensity in humid N₂. The equivalent measurements in dry and humid SA (figure 5(b)) show a very similar PL response as in pure N₂ with the increasing PL signal of the oxidized NWs and a constant/quenching PL evolution of the HCl-etched NWs. Our results show that the humidity level of the ambient and the GaN NW pre-treatment are the decisive factors for the time-dependent PL intensity during the first minutes of UV illumination and that the O₂ content in the ambient only influences the PL evolution to a minor extent. This is particularly interesting as under prolonged illumination (steady state) humidity was only secondary for the change of the PL intensity but rather the O₂ content in N₂ was the decisive factor. Further, the fact that HCl-etched GaN NWs show almost no PL intensity evolution in dry N₂ and SA proves that the adsorption of quenching O₂ is not UV-induced but already takes place in the dark.

Zoom In Zoom Out Reset image size

The light-induced ad- and desorption respectively the chemical interaction of the water molecules with the GaN surface are influenced by the pre-treatment of the NWs, e.g. oxidized Ga- and N-polar GaN surfaces show a better wetting behavior than as-grown or HCl-etched surfaces [36]. Additionally, the electronic properties of both polar and non-polar surfaces are influenced by pre-treatments like HCl-etching or O₂ plasma exposure, since this modifies the surface states and creates distinct surface dipoles due to the presence of adsorbates remaining after the treatments. This behavior is found to be similar for the GaN (0001) and $(1\bar{1}00)$ surfaces, which are the dominant surface facets in the case of GaN NWs [21]. Particularly, each pre-treatment might cause a specific rate of non-radiative recombination due to the different SBB and adsorbates. Upon beginning UV illumination in dry atmospheres, the change of one or several of these properties lower the proportion of non-radiative recombination in the case of oxidized NWs, whereas for HCl-etched NWs the respective properties related to PL intensity are unaffected. In all cases, H₂O acts as an additional source of non-radiative recombination, which reduces the PL enhancement for oxidized NWs and leads to a quenching of the PL intensity for HCl-etched NWs.

The reversibility of the PL responses was investigated in a recovery study, the results of which are shown in figure 5(c). The PL signal of HCl-etched GaN NWs was quenched to 30%–40% of the initial PL intensity in dry SA by exposure to the excitation light with an intensity of 21 kW cm⁻² in humid SA. Then, the laser was blocked at 0 min. The laser was switched on for 1 s to probe the PL intensity recovery in the dark and immediately switched off again. In humid SA, almost no change in the PL intensity is observed except a slight quenching possibly due to the additional UV illumination of the laser during the short measurements. After 15 min, the measurement chamber was purged with dry N₂ and dry SA, respectively. The PL intensity recovers to its initial value (7300 cts) in dry SA within 45–60 min. In dry N₂, this recovery to 100% takes longer (100 min), followed by a further PL enhancement to about 130% of the initial PL intensity (10 500 cts) within the subsequent 50 min. This overshoot reflects the higher PL intensity at temperatures close to RT in N₂ compared to SA (figure 3). For the oxidized GaN NWs, a recovery from the PL enhancement is observed when measured in a comparable way in ambient air (figure S3, SI). As already mentioned, the complete recovery of the PL intensity shows that no permanent change of the GaN NW surface is responsible for the PL intensity evolution. However, the fact that the PL signal remains quenched in humid SA but a complete recovery of the PL intensity is observable in dry atmospheres shows that H₂O from the ambient is not only responsible for the PL quenching after illumination onset but it also hinders the recovery. In the literature, changes in the PL intensity after illumination onset have been attributed to O₂ chemisorption on the GaN surfaces, which results in a Fermi level pinning reducing the quantum efficiency of the PL [37–39]. However, our results show that the ad- and desorption of H₂O is the decisive factor for the PL intensity evolution after illumination onset, which is also in accordance with the observations by Hetzl et al [18].

While the PL experiments presented so far clarify the quantitative and qualitative evolution of the NBE PL intensity of GaN NWs in different atmospheres, complementary electrical measurements of GaN NFs can provide further insight into the underlying mechanisms. To this end, the current characteristics of GaN NFs on AlN substrates were investigated in different atmospheres. Figure 6(a) shows the electrical current through GaN NFs under illumination with 266 nm laser light and the simultaneously measured NBE PL in varied dry atmospheres at room temperature. Under illumination, the total current (dark and photocurrent) decreases to 88% and recovers to 95%. At the same time, the PL decreases to about 85% in SA and reaches its initial value 10 min after switching back to N₂. The corresponding dark current at room temperature (figure 6(b), blue curve) is reduced only slightly to about 99% of its initial value after replacing N₂ with SA. It recovers partially when switching back to N₂. The evolution of the PL intensities after illumination onset (figures 5(a)/(b)) showed that the changes of the GaN NW luminescence due to dry oxygen is not light-induced but mainly happens already in the dark. Consequently, the oxygen-related mechanism which causes a reduction of the dark current will similarly reduce the current under illumination. However, the amplitude of the current reduction under illumination is higher, indicating a stronger sensitivity of the photocurrent. A possible explanation lies in the location of photo-excitation within the GaN NF, which mainly takes place in the surface region, while the majority of the dark current contribution is through the conductive channel in the center of the NF. Thus, an increase of the band bending due to oxygen adsorption has a stronger impact on the photocurrent contribution of the overall current, which makes the current under illumination more sensitive. The simultaneously measured PL decreases in SA, too. As speed and amplitude of the PL decrease and current reduction under illumination correlate strongly, it is likely that both effects have the same origin, i.e. changes of the SBB.

Zoom In Zoom Out Reset image size

Generally, changes in PL intensity of NSs under constant excitation originate in changes of the ratio of radiative to non-radiative charge carrier recombination. This is either caused by the creation or passivation of non-radiative recombination channels, e.g. unsaturated surface states, or by changing the amount of radiative recombination through spatial separation of electrons and holes, e.g. due to SBB. In contrast, the electrical current through NFs in the dark is influenced by atmospheric changes only via the SBB, which alters the size of the conductive channel within the NS. In the present case, the conductivity of NFs is not sensitive to passivation effects of surface states as long as the band bending remains unaltered. Similar to PL, the photocurrent is influenced by both SBB and the non-radiative recombination rate. However, not necessarily with the same sensitivity.

For confirmation, we measured the macroscopic surface photovoltage (SPV) and CPD, which is a measure for the position of the Fermi level at the surface, of bulk HCl-etched m-plane GaN (figure 7(a)). The CPD value of the GaN surface in the dark is about 0.1 V higher in SA compared to N₂, reflecting an increased SBB in SA. Under illumination with 6 mW cm⁻² of 340 nm LED light, the CPD follows the same trend with a more pronounced response to oxygen (0.2 V CPD difference). The recovery to the initial CPD values in N₂ is faster under illumination, which might indicate a light-induced process accelerating the desorption of oxygen. The SPV was about 0.3 V in N₂ and almost 0.2 V in SA. In both atmospheres, the maximum intensity of the UV-LED used was not sufficient to achieve flat band conditions (figure S2, SI). The changes of the SPV and the current under illumination both show a double-exponential decay with similar time constants of 11 s and 430 s, respectively 4 s and 440 s, which indicates that the same oxygen-induced processes are responsible for the reduction of current under illumination and SPV. Figure 7(b) summarizes the CPD and SPV results in a schematic drawing of the GaN NS SBB in SA and N₂ atmosphere. In the dark, the SBB in SA is higher than in N₂. This difference increases upon above band gap illumination as the SBB reduction due to the light is more pronounced in SA compared to N₂. This matches well with the increased sensitivity of the current to changing the atmosphere under illumination compared to in the dark (figure 6): the higher reduction of the SBB under illumination leads to a stronger shrinking of the conductive channel within the NF, which increases the electrical resistivity to a greater extend. The picture changes, when the NFs are heated to 375 K. The dark current (figure 6(b), red curve) decreases to 90% in SA and recovers only slightly in N₂. While the current under illumination decreases to about 80%, the PL increases to about 150% (figure 6(c)). Both effects recover majorly within 10 min after switching back to N₂. The qualitatively different behavior of the electrical current and PL prove that beside the increase in SBB a second, contrary effect exists, which reduces non-radiative recombination. We suggest that the adsorption of O₂ on the heated GaN NF surface passivates non-radiative recombination paths, which increases radiative recombination. The overshoot of the PL intensity after switching back to N₂ indicates that the passivation is temporary and reacts to atmospheric changes more slowly than the change in band bending: in the first minutes after the SA is replaced in the measurement chamber, the band bending relaxes relatively quickly, which increases the PL and current. In combination with the more stable passivation, the already lowered SBB results in a peaking PL intensity.

Zoom In Zoom Out Reset image size

Recently, several groups have reported that oxygen can bind in different configurations to GaN m-plane [40, 41]. Although the underlying mechanism is not yet completely clarified, density functional theory calculations and experimental observations indicate that dissociative adsorption of oxygen and bridge adsorption of O₂ can coexist. Due to the differing binding configuration, the respective binding energy is different, with the dissociative absorption being the more stable variant. From the perspective of our results, we speculate that at room temperature bridge adsorption is dominant, while the share of dissociative adsorption increases when the sample is heated during the exposure to O₂. Following this argumentation, bridge adsorption of O₂ would mainly increase the SBB, while dissociative adsorption would passivate non-radiative surface recombination paths. However, at present this remains only a speculation, which could motivate future work.

(Al)GaN-based high-electron-mobility transistors (HEMTs) and field-effect transistors are known to be sensitive to H₂O, various organic solvents like alcohols, acetone or propanol and the pH-value in aqueous solutions [42–45]. However, there are only few reports which deal with the optical readout of group III-nitride-based sensors in liquids. E.g. it has been shown that the luminescence of (In)GaN NWs can be used to monitor salt concentrations in H₂O or the pH-value [1, 18]. The PL response of HCl-etched GaN NWs in various liquids is shown in figure 8(a) on a logarithmic scale. The already discussed PL quenching in ambient air is shown for comparison. In all liquid environments, a PL enhancement is observed. In DI H₂O, the PL intensity doubles after 6min of exposure. When the NWs are surrounded by ethanol or acetonitrile, the PL intensity almost quadruples. In cyclohexane the PL intensity even exceeds 26 times of its initial value. The adsorption of molecules from an electrolyte at a semiconductor surface can lead to a charge transfer between the adsorbed species and the semiconductor, causing formation of a Helmholtz layer at the semiconductor surface. This in turn influences surface properties of the semiconductor like its SBB and, thereby, the PL intensity [46]. Further, alcohols like methanol are known to act as hole-scavengers. These influence the recombination mechanisms in GaN NWs e.g. they prevent them from oxidation and decomposition in H₂O, [19] which might influence the PL signal as well. Candidates for the significant deviation in the magnitude of PL enhancement in the liquids are (I) their molecular dipole moment, (II) the molecule size, (III) differences in the potential for several redox-reactions and (IV) the oxygen solubility. (I) It is known from (Al)GaN-based HEMTs that the source-drain current is sensitive to liquids with different molecular polarization strengths in the vicinity of the polar GaN-gate as they compensate polarization-induced charges in the heterostructure [43, 44]. Even though GaN NWs expose mainly the non-polar m-plane facets, comparable processes might influence the surface properties of the NWs and their PL intensity. Cyclohexane is an unpolar molecule, whereas ethanol and H₂O have comparable dipole moments of 1.69 D and 1.85 D, respectively, and acetonitrile has the highest moment of 3.92 D. However, the amplitude of the PL enhancement for the different liquids does not reflect this ordering as a similar evolution of the PL intensity is observed for ethanol and acetonitrile. Thus, although a secondary role of the dipole moment is possible, it cannot be the main reason for the observed PL intensity evolution. (II) The molecule size might influence the built-up speed and capacitance of the emerging Helmholtz layer manipulating the SBB. H₂O is the smallest molecule, ethanol and acetonitrile are approx. twice as large and cyclohexane is significantly larger. This matches the observed ordering of the PL enhancement. We would expect that the built-up time for Helmholtz layers is significantly shorter than the seconds to minutes of the here observed effects. However, further investigations are needed to clarify the role of the Helmholtz layer on the recombination mechanisms of the NWs. (III) The redox energy levels for illumination induced reactions, e.g. water oxidation or superoxide formation are not necessarily the same in different electrolytes. Some of the emerging products of these reactions may alter the GaN surface, leading to passivation of surface states. For us, a definite conclusion over this possible explanation for the observed effects is difficult as no controlled potential was applied to the NWs in the here discussed experiments. Therefore, this approach has to remain open for further investigations. (IV) The mole fraction solubility in units of 10⁻⁴ of O₂ in the respective liquids is approx. 0.2 for H₂O, 6 for ethanol, 8 for acetonitrile and 12 for cyclohexane at room temperature and ambient pressure [47, 48]. As this reflects the overall PL intensity enhancement trend, dissolved oxygen may play a role in the PL response of GaN NWs in liquids, too. Adsorbed oxygen at the NW surface, which acts as a PL quencher at room temperature in gaseous surroundings (see above), dissolves more easily in e.g. cyclohexane compared to H₂O, which might cause a much more pronounced PL enhancement.

Zoom In Zoom Out Reset image size

The ambivalent role of H₂O around room temperature (PL quenching in gaseous atmospheres and enhancement in a liquid environment) of the HCl-etched GaN NWs might be explained by the work of Shen et al [49]. They found that a full dissociation of H₂O molecules occurs at the GaN m-planes, leading to a localized and high concentrations of disordered H₃O⁺ and OH⁻ molecules at the GaN surface, which separate and align at the GaN NW surface under UV illumination. In contrast, when the NWs are in a H₂O bath this effect vanishes as the ion concentration is thermodynamically determined by the auto-protolysis of the ambient reservoir. As the ion concentration in the vicinity of the NWs is a decisive factor for the PL intensity evolution this has a major impact [18]. The different magnitude of the PL response in various liquid environments opens a possibility to use GaN NWs to optically detect impurity molecules for, e.g. H₂O quality control applications.

In figure 8(b), the PL intensity evolution of GaN NWs with several oxide shells after the onset of illumination under ambient conditions in air is depicted. All MBE-coated samples (ZnO, Ga_x O_y ) show an intensity increase, however, the amplitude differs significantly. While the PL signal of the GaN NBE of the oxidized NWs more than doubled within 2 min, for the thickest Ga_x O_y shell it increases only by 6%. As a general trend, the enhancing effect inflicted by the atmospheric exposure is smaller the thicker the surrounding shell is. A confirmation that the observed PL intensity increase results from an increased emission related to GaN and not to ZnO is given in figure S4(a) in the supporting information. For the ALD-grown Al₂O₃ shell, a quenching to 24% of the initial intensity is observed. Although the comparison of absolute PL intensities of different samples has to be treated with care, some clear trends are observed: the absolute PL intensities (table S5, SI) of the GaN-ZnO core–shell NWs decrease by a factor of almost 25 from the thinnest to the thickest shell due to absorption of the excitation laser in the ZnO. Further, the sample with Al₂O₃ shell exhibits the highest PL signal and is at the same time the only one exhibiting a decreasing PL intensity over illumination time. We speculate that in the case of an Al₂O₃ shell the electronic interface configuration is already in a beneficial state with high radiative recombination efficiency and cannot be further improved upon UV laser exposure. For the other types of oxide coatings, this may initially be not the case leaving room for UV-induced improvement. The qualitatively different behavior of MBE- and ALD-grown core–shell NWs can be possibly explained by differences in the growth procedure. Prior to the shell growth via MBE, the GaN NWs were oxidized in the MBE chamber due to process requirements. During ALD growth of amorphous Al₂O₃ no comparable oxidation of the GaN surface took place on the previously HCl-etched NWs. Additionally, an accumulation of positive charges on the Al₂O₃ side of the heterointerface to GaN has been reported, although their origin remains disputed [50]. These charges can annihilate the band bending of the GaN m-plane [51]. Under illumination, the positive interface charge could be slowly compensated by photo-excited electrons resulting in an upward band bending. From an electronic point of view, this would resemble the adsorption of anions on bare GaN NWs, which leads to a quenching of the PL intensity and has been discussed in detail by Hetzl et al [18]. Further, the spotty appearance of the MBE-grown oxide shells suggests that the shells could be not completely tight and allow some atmospheric molecules to still reach the GaN NW surface. For the tight 50 nm Al₂O₃ shell, a diffusion of molecules/ions from the environment to the GaN-Al₂O₃ interface can be excluded [52]. However, as ALD on GaN for passivation can be challenging, it cannot be excluded that the recipe could influence the PL evolution as factors like the GaN pre-treatment, deposition temperature and post-deposition annealing treatments influence the GaN-Al₂O₃ interface as well as the Al₂O₃ itself [53–56]. Note that the illumination-induced PL intensity changes for the core–shell NWs were reversible similar to the uncoated NWs described above (figure S4(b), SI), which excludes permanent changes in the GaN-oxide interface like changed oxidation states as possible explanation of the observed effects. In sum, thick ZnO and especially Ga₂O₃ shells are well-suited to serve as an efficient passivation the GaN NW surface and stabilisator of the PL intensity, which could be exploited in NW-based (opto-)electronic devices.