Light-driven permanent transition from insulator to conductor in Ga2O3

The transition from insulator to conductor can be achieved in some materials but requires modification of both the arrangement of atoms and their electronic configurations. This is often achieved by doping. Here we reveal a mechanism the lattice may adopt to induce such a transition. We show that limited exposure to sub-bandgap light caused a permanent transition from an insulator state to a conductor state in the insulating oxide Ga2O3 with 9-orders of magnitude increase in electronic conduction. Photoexcitation modifies the charge state of an O-vacancy and the redistribution of the localized electrons, leading to a massive structural distortion in the lattice that is shown to be the underlying mechanism. It modifies density of states and introduces new stable states with shallower energy levels, leading to this intriguing behavior. We suggest that this mechanism may occur in other wide bandgap metal oxides leading to drastic modification in their electronic properties.


Introduction
When light is impinged on a semiconductor material, charge carriers -electrons and holes -may be generated resulting in an enhancement in conductivity. If the energy of the incident photons is greater than the band gap of a semiconductor, it excites an electron from the valance band to the conduction band, a phenomenon called intrinsic photoconductivity. On the other hand, if the energy of the photon is less than the band gap, it may excite electrons from defect levels to the conduction band enhancing conductivity. This is referred as extrinsic photoconductivity 1 . In either case, if the conductivity persists after turning off the photo excitation, it is known as persistent photoconductivity. In this situation, when the electron-hole pairs are generated, there must be microscopic or macroscopic potential barriers that separate charge carriers and reduce the probability of recombination between them, resulting in enhanced conductivity for a longer period of time 2 .
Persistent photoconductivity at room temperature has been primarily reported in hetero-structures of semiconductors and in a few bulk materials 2,3 . In this work we report an extrinsic persistent photoconductivity behavior in bulk Ga 2 O 3 and a surprising permanent transition from the insulator state to the conductor state upon exposure to sub-band gap light for a limited period of time. First, Ga 2 O 3 bulk crystals exhibited massive persistent photoconductivity upon exposure to sub-bandgap light of much lower energy than the band gap. Then, the induced meta-stable states created by light became stable simply by increasing the photoexcitation time, leading to a permanent transition from the insulator state to the conductor state. To our knowledge, such behavior has never been reported in any material system and could have vast implications on the material properties and applications.
Gallium oxide (Ga 2 O 3 ) is the widest band gap transparent (up to UV-C range) semiconducting oxide known so far  . Its ultra-wide band gap (~ 4.5-4.9 eV) 6 may lead to unusual electronic phenomena. Due to this wide band gap, UV-C transparency, and excellent thermal and chemical stability, it has numerous potential applications in power and high voltage devices, Schottky diodes, field effect transistors, gas sensors, phosphors and electroluminescent devices, UV photo detectors and more [7][8][9][10][11]24 . Ga 2 O 3 exhibits polymorphism, denoted by α, β, γ, δ, and 12 , with β-Ga 2 O 3 being the most stable phase from room temperature to its melting point 13 . As the most stable form, β-Ga 2 O 3 is also the most studied polymorph. It crystalizes into a monoclinic structure with space group C2/m and lattice parameters a = 12.2140 Å, b = 3.03719 Å, c = 5.7819 Å and β= 103.83° 13 . It contains both octahedral and tetrahedral cation sites in equal numbers. As Ga 2 O 3 has a wide band gap, it is an insulator at room temperature, but electron conduction has been reported when synthesized under reducing condition 5 . This conductivity is often attributed to oxygen vacancies, though recent theoretical calculations showed that oxygen vacancies are deep states and cannot provide conduction electrons 26 . It has been also proposed that silicon, which is a major impurity in Ga 2 O 3 , might be the cause of electron conductivity 12 . The effective hole condition in Ga 2 O 3 has not been reported; theoretical calculations show that the valance band is flat, indicating larger effective mass for holes, making p-type conductivity difficult 6 .

Results and Discussion
Ga 2 O 3 single crystals were illuminated by sub-bandgap light and the conductivity and carrier density were measured during illumination. Figure 1 illustrates the dependence of photoconductivity in undoped β-Ga 2 O 3 single crystals on photoexcitation energy and intensity. In Fig. 1b, the photoconductivity is plotted versus light intensity for photon energies of 3.39, 3.22, 3.1, 2.69 eV, revealing that the conductivity abruptly increases at low photon intensity and quickly saturates. In Fig. 1c, the photoconductivity and photo-induced charge carrier density are plotted as a function of photon energy. Before each measurement, the sample was heated at 400 o C for 1 hour to retain its initial dark conductivity, and excitation was carried at the same photon intensity for all energies. It is interesting to note that all sub-band gap photo-excitations led to an increase in conductivity, even at low energies of 1.9 and 1.45 eV. The maximum photoconductivity occurs at 3.1 eV, which is much smaller than 4.5/4.9 eV, the band gap energy of Ga 2 O 3 6 . The increase of conductivity with sub-band gap photoexcitation can be explained due to the excitation of electrons from a localized state in the gap to the conduction band as shown in Fig. 1a. In contrast to undoped Ga 2 O 3 , Fe-and Mg-doped samples behave differently when exposed to light. Both Fe and Mg-doped crystals showed decreased conductivity when exposed to 400 nm and 365 nm light (Fig. 2). This indicates that common impurities in Ga 2 O 3 such as Fe is not behind this photoconductivity.
After tuning off the photo-excitation, the sample illuminated by 3.1 eV shows significant persistent photoconductivity. To calculate the associated potential barrier that prevents the recapture of charge carriers by their centers after light is turned off and is thus the ultimate origin of the persistent photoconductivity 3,27 , the photoconductive sample was annealed at various temperatures (from 300°C to 390°C) for 10 minutes at each temperature inside the Hall-effect chamber. After each anneal, the sample was then cooled to room temperature and the electrical conductivity and carrier density were measured. The steps of the experimental procedure are illustrated in Fig. 3a. Figure 3b show how the conductivity and charge carrier density of the sample vary with annealing temperature and Figure 3c is a corresponding graph for the natural logarithm of charge carrier density vs reciprocal thermal energy (1/kT). Using the Arrhenius equation, n = Ae -Eth/kT , the slope of the best-fit line gives a thermal energy barrier of about 0.157±0.04 eV. In Fig. 3d, we illustrate the process of electron pumping from the localized center to the conduction band where the center relaxes to a metastable state and the subsequent process of electron recapture through a barrier energy E th . However, it should be mentioned here that subsequent prolonged and repeated photo-excitation led to stable electron conduction that does not decay even after heating up to 400 o C. The decay of the induced conductivity and its dependence on photon energy, and intensity and illumination time are discussed in detail in the following.  Figure 5b (the blue curve) shows the decay of conductivity as a function of time after the 70 hours exposure to 3.1 eV excitation. After decay, the sample was then re-exposed to 3.1 eV photons for 10 minutes. The red curve in Fig.   5b represents the subsequent decay of conductivity. Each of the decay curves in Fig.   5b exhibits two time decay constants, one relatively fast and one slow. The fast decay rate of conductivity was 0.4 Ω -1 cm -1 /min after 1 h excitation, 0.008 Ω -1 cm -1 /min after 70 h excitation and 0.004 Ω -1 cm -1 /min after repeating photoexcitation with higher intensity. The dependence of photoconductivity decay on excitation time is unusual.
However these measurements clearly demonstrate the strong dependence of decay rate of conductivity on the energy, intensity and time of photo-excitation. It is interesting that repeated exposure to light significantly impacts the decay of conductivity, leading to a more stable electron conduction.
To further investigate the conditions that cause the permanent transition from insulator to conductor state, an undoped Ga 2 O 3 sample was exposed to light several times and the conductivity was monitored. Initially the conductivity was 1.08x10 -8 Ω -1 .cm -1 . When exposed to photo-excitation of 3.1 eV, the conductivity promptly increased by almost two orders of magnitude but retained nearly the same initial value after the light was turned off. However, by repeating photo-excitation and after prolonged exposure to light, the conductivity was increased by 9 orders of magnitude and was held after turning off the light without decay, indicating a complete conversion from the insulator to conductor state. Annealing the sample at 400 o C for 1 hour in dark did not remove or decrease the conductivity. Annealing at a much higher temperature of 800 o C for 2 hours in O 2 flow was necessary to revert the sample to an insulator with a conductivity of 7.69×10 -7 Ω -1 .cm -1 . However, this annealing also completely eliminated the photoconductivity feature of the sample.  Fig. 7 shows a freeze out region for the electrons indicating that these new states are still within the bandgap.
To understand the reason behind the permanent conversion from insulator to conductor and reveal the mechanism that prevented the electrons from returning to their center after turning off light, the change in the structural properties of β-Ga 2 O 3 has been examined by first-principles electronic structure calculations. There are two different types of Ga sites present in the β-Ga 2 O 3 crystal structure. The first is Ga coordinated by four oxygen (denoted as Ga 1 ) while the second is Ga coordinated with six oxygen (denoted as Ga 2 as shown in Fig. 8 Package) 33 . The core electron behaviour and the interaction between the valence electrons and the ion are described by the projector augmented wave method (PAW) 34 . The Perdew-Burke-Ernzerhof (PBE) form of the generalized gradient approximation (GGA) has been employed as the exchange-correlation functional to obtain the optimized ground state structure 35 . The Brillouin zone has been sampled using 3x3x3 and 7x7x7 meshes of Monkhorst-Pack k-points for optimization and electronic structure calculations, respectively. The valence electrons are described by a plane waves basis set with a converged energy cut-off of 520 eV. A supercell of 160 atoms (32 formula unit) is considered in this calculation. The structure has been optimized until the calculated Hellmann-Feynman forces are smaller than 0.0001 eV Å -1 .