shows representative PL spectra measured from the GaP NW (the dotted line, black online) and the GaP/GaNP core/shell NW samples (the solid line, red online) at 5 K using the 325-nm line of a solid state laser as an excitation source. The PL emission from the GaP NW is rather weak and is dominated by a series of relatively sharp lines within the 2.05 to 2.32 eV spectral range due to the recombination of excitons bound to various residual impurities. Some of the PL lines are very similar to the previously reported emissions due to the recombination of excitons bound to isoelectronic centers involving N impurity, e.g., from an isoelectronic BGa
center and its phonon replica
]. Though the studied GaP NWs are intentionally undoped, the formation of the N-related centers may be caused by contamination of the growth chamber. Further studies aiming to clarify the exact origin of these emissions are currently in progress.
The PL spectra are significantly modified in the GaP/GaNP core/shell NW. First of all, the sharp excitonic lines are replaced by a broad PL band with a rather asymmetric lineshape that peaks at around 2.06 eV (Figure
). This emission originates from radiative recombination of excitons trapped at various N-related localized states
] in the GaNP shell. Secondly, a significant increase of the integrated PL intensity (by about 20 times) is observed which is largely related to the N-induced transition from the indirect bandgap in GaP to a direct bandgap in the GaNP alloy
]. The observed high efficiency of the radiative recombination in the GaP/GaNP core/shell NW implies that this material system could be potentially promising for applications as efficient nano-sized light emitters.
For practical device applications, it is essential that the high efficiency of radiative recombination is sustained up to RT. Therefore, recombination processes in the studied structures were further examined by employing temperature-dependent PL measurements. In the case of GaP NWs, temperature increase was found to cause a dramatic quenching of the PL intensity so that it falls below the detection limit of the measurement system at measurement temperatures T
exceeding 150 K. For the GaP/GaNP NW, on the other hand, the PL emission was found to be rather intense at RT even from an individual NW, though significantly weaker than that at 5 K. Moreover, thermal quenching is found to be more severe for the high energy PL components which lead to an apparent red shift of the PL maximum position at high T
. To get further insights into the mechanisms responsible for the observed thermal quenching, we have analyzed Arrhenius plots of the PL intensity at different detection energies (Edet
) as shown in Figure
a. The analysis was performed for constant detection energies since (a) the temperature-induced shift of the bandgap energy is significantly suppressed in GaNP alloys
], and (b) spectral positions of the excitons bound to various deep-level N-related centers do not one-to-one follow the temperature-induced shift of the bandgap energy. This approximation defines error bars of the deduced values as specified below. All experimental data (shown by the symbols in Figure
) can be fitted bywhere I
) is the temperature-dependent PL intensity, I
(0) is its value at 4 K, E1
are the activation energies for two different thermal quenching processes, and k
is the Boltzman constant (the results of the fitting are shown by the solid lines in Figure
a). The first activation process that occurs within the 30 to 100 K temperature range is characterized by the activation energy E1
ranging between 40 (at Edet
= 2.17 eV) and 60 meV (at Edet
= 2.06 eV). The contribution of this process is most pronounced for high energy PL components that correspond to the radiative recombination at the N-related localized states with their energy levels close to the GaNP band edge. The quenching of the high energy PL components is accompanied by a slight increase in the PL intensity at low Edet
. Therefore, this process can be attributed to the thermal ionization of the N-related localized states. Such ionization is expected to start from the N-states that are shallower in energy. The thermally activated excitons can then be recaptured by the deeper N states, consistent with our experimental observations. We note that the determined values of E1
do not one-to-one correspond to the ‘apparent’ depth of the involved localized states deduced simply from the distance between Edet
and the bandgap energy of the GaNP. This is, however, not surprising since such correspondence is only expected for the no-phonon excitonic transitions whereas recombination of excitons at strongly localized states (such as the monitored N states) is usually dominated by phonon-assisted transitions due to strong coupling with phonons.
Arrhenius plots of the PL intensity measured at different detection energies from the GaP/GaNP NWs (a) and GaNP epilayer (b).
The second thermal quenching process is characterized by the activation energy E2 of approximately 180 ± 20 meV, which is the same for all detection energies. This process becomes dominant at T > 100 K and leads to an overall quenching of the PL intensity irrespective of detection energies. We therefore ascribe it to thermal activation of competing non-radiative recombination which depletes photo-created free carriers and, consequently, causes a decrease in the PL intensity. It is interesting to note that the competing NRR process remains active even when the excitation photon energy (Eexc) is tuned to 1.96 eV, which is below the GaNP bandgap. Indeed, Arrenius plots of the PL intensity measured at Edet = 1.73 eV under Eexc = 2.33 eV (the open circles in Figure
a) and Eexc = 1.96 eV (the dots in Figure
a), i.e., under above and below bandgap excitation, respectively, yield the same activation energy E2. In addition, the PL thermal quenching under below bandgap excitation seems to be even more severe than that recorded under above bandgap excitation. At first glance, this is somewhat surprising as the 1.96 eV photons could not directly create free electron–hole pairs and will be absorbed at N-related localized states. However, fast thermal activation of the photo-created carriers from these localized states to band states will again lead to their capture by the NRR centers and therefore quenching of the PL intensity. Moreover, the contribution of the NRR processes is known to decrease at high densities of the photo-created carriers due to partial saturation of the NRR centers which results in a shift of the onset of the PL thermal quenching to higher temperatures. In our case, such regime is likely realized for the above bandgap excitation. This is because of (a) significantly (about 1,000 times) lower excitation power used under below bandgap excitation (restricted by the available excitation source) and (b) a high absorption coefficient for the band-to-band transitions.
The revealed non-radiative recombination processes may occur at surfaces, the GaNP/GaP interface or within bulk regions of GaNP shell. The former two processes are expected to be enhanced in low-dimensional structures with a high surface-to-volume ratio whereas the last process will likely dominate in bulk (or epilayer) samples. Therefore, to further evaluate the origin of the revealed NRR in the studied NW structures, we also investigated the thermal behavior of the PL emission from a reference GaNP epilayer. It is found that thermal quenching of the PL emission in the epilayer can be modeled, within the experimental accuracy, by the same activation energies as those deduced for the NW structure. This is obvious from Figure
b where an Arrhenius plot of the PL intensity measured at Edet
= 2.12 eV under Eexc
= 2.33 eV from the epilayer is shown. However, the contribution of the second activation process (defined by the pre-factor C2
in Equation 1) is found to be larger in the case of the GaNP/GaP NWs. This suggests that the formation of the responsible defects is facilitated in the lower dimensional NWs and that the defects could be at least partly located either at the surface of the GaNP shell or at the GaNP/GaP hetero-interface, consistent with the results of
The activation of the NRR recombination processes at elevated temperatures is also confirmed by the performed time-resolved PL measurements. Typical decay curves of the integrated PL intensity at 5 K and RT are shown in Figure
. At 5 K, the PL decay is found to be rather slow, i.e., with the decay time τ
of the dominant decay component longer than 60 ns (the exact value of τ
could not be determined from the available data due to the high repetition frequency of the laser pulses). Such slow decay is likely dominated by the radiative lifetime τr
as it is of the same order of magnitude as previously determined for the radiative transitions within the N-related localized states in the GaNP epilayers
]. A temperature increase above 100 K causes significant shortening of the PL decay, down to several ns at RT (see the inset in Figure
). The measured decay time contains contributions from both radiative and NRR processes so that
denotes the non-radiative decay time. Therefore, the observed dramatic shortening of the measured decay time at elevated temperature implies thermal activation of non-radiative carrier recombination, consistent with the results of cw-PL measurements (Figure
Decays of the integrated PL intensity measured from the GaP/GaNP NWs at 5 K and RT.