1. Introduction

The search for new and novel semiconductor materials as terahertz (THz) surface emitters is a vital aspect of THz research [1]. In particular, wide-bandgap semiconductors are of interest due to their inherent transparency in the near-infrared (NIR) and visible regions. Highly transparent materials over a broad spectral region are crucial components in THz integrated optics design [2]. It is also worth noting that wide bandgap semiconductors likewise promise to be viable THz photoconductive antennas (PCA’s) as they can be operated at higher bias fields for improved signal-to-noise ratio [3]. Their major drawback, however, is that they require intense short-wavelength excitation sources and these types of excitation sources are not in abundance at this time primarily due to prohibitive costs.

Indium oxide (In₂O₃) is a wide-bandgap n-type metal oxide semiconductor with high optical transparency in the visible region [4–7]. The applications of In₂O₃ range from transparent electronics, to light-emitting diodes (LEDs), photodetectors, solar cells, flat panel displays, thin film transistors, and even high sensitivity detection of various gases and biomedical applications [4–9]. Recently, the actual energy bandgap of this semiconductor has been extensively debated on due to the emergence of growth techniques that are capable of producing high quality defect-less single-crystal In₂O₃ films that yield varying bandgap values. Oxygen plasma assisted MBE-grown In₂O₃ films on Yttria-stabilized Zirconia (YSZ) substrates have recorded a bandgap of about 2.9 eV (427 nm), much lower than the well known value of 3.75eV- 4eV [10,11]. This direct energy gap is suggested to be the only bandgap corresponding to cubic In₂O₃ in contrast to previous reports [10,12]. The surface electronic investigation on cubic In₂O₃ also reveals preference of electron accumulation. It has been found that charge accumulates along the (001) and (111) planes of its lattice, hence having electronic characteristic similar to InN and InAs [13,14]. As these two materials are considered to be intense semiconductor surface THz emitters, these previous reports suggest the viability of In₂O₃ as a THz emission source, as well [10,11]. Thermal oxidation techniques have previously demonstrated the growth of In₂O₃ films [4]. Moreover, Sieber, et. al., reported the growth of polycrystalline In₂O₃ on magnesium oxide (MgO) [15] despite a large lattice mismatch. The In₂O₃ lattice constant along the a-axis is around 10.12 Å while MgO has 4.21 Å.

In this work, we experimentally demonstrate THz emission from photo-excited In₂O₃ films on single crystal MgO (100) prepared by thermal oxidation. The MgO substrate is so chosen due of its excellent transmission properties both in the optical and THz-frequency regimes, high-melting temperature, and low electrical conductivity [16]. Results show that even at an excitation photon energy which is lower than the bandgap, the In₂O₃ films exhibited THz emission. Additionally, the dependence of the THz emission on the oxidation temperature and excitation fluence was also studied in an attempt to explain the underlying THz radiation mechanism.

2. Experiment

The In₂O₃ thin films were formed via thermal oxidation of Indium films deposited on single crystal MgO (001) substrates. The MgO substrates were primarily cleaned, rinsed and blow-dried. Indium metal with 99.999% purity was then thermally evaporated on the substrates at a rate of 6Å/s in a 3x10⁻⁵-Torr vacuum environment. A quartz crystal sensor indicated the indium film thickness to be ~1500Å. The samples were then oxidized in ambient air at three different temperatures (350°C, 450°C, and 550°C). Initial thin film characterization was carried out via scanning electron microscopy (SEM) and X-ray diffraction (XRD). The THz emission was measured using standard THz-time domain spectroscopy (THz-TDS) methods. The optical excitation was provided for by a p-polarized mode-locked Ti:Sapphire laser delivering ~80 femtosecond pulses at a central wavelength of 800 nm with a repetition rate of 82 MHz. The samples were excited at a 45^ο incidence with an average pump power of 150 mW, loosely focused to a beam spot size of ~1 mm. The THz radiation was collected in the specular reflection direction and focused to the detector using appropriate off-axis paraboloid mirrors. The THz temporal waveforms were detected by a Hamamatsu LT-GaAs dipole-type photoconductive antenna, switched by a variable time-delayed 20 mW optical pulse.

3. Results and discussion

As shown in Fig. 1(a) , the SEM images of the three In₂O₃/MgO films, exhibited grain structures. At a higher magnification of x1200 (inset), these grains are shown to have similar size distributions and shapes, regardless of the oxidation temperature. A more quantitative discussion is offered by the XRD data in Fig. 1(b). The most intense peak is the (002) peak associated with the MgO substrate. The other observed features shows XRD peaks at 30.26°, 35.14°, 43.98° and 77.46° corresponding to (222), (400), (422), and (820) planes (from JCPDS #06-416). The peaks are indexed against a cubic In₂O₃ unit cell with lattice parameter a = 10.1 Å. This observation is consistent with the previous studies done on the thermal oxidation of Indium films on silicon and glass substrates [4,17]. In addition to verifying the presence of In₂O₃, the XRD results also reveal that the polycrystalline characteristics of the thin films are similar, regardless of the oxidation temperature. It must be noted, however, that the samples oxidized at 350°C and 450°C confirms incomplete oxidation owing to the presence of a peak assigned to indium metal. Even as the 550°C-oxidized sample did not show this peak, it cannot be ascertained that a complete oxidation of the indium metal was achieved.

 

Fig. 1 (a) SEM images of the In₂O₃/MgO films that were oxidized at different temperatures. The films are characterized by grain-like structures. (b) XRD data showing the polycrystalline nature of the films. The most intense peak is attributed to the MgO substrate while the rest are assigned to cubic-In₂O₃ structure. Results show that incomplete oxidation occurred due to the presence of a diffraction peak assigned to Indium metal.

Download Full Size | PDF

The THz-TDS data plots for sub-bandgap optical excitation (800 nm, 600 nJ/cm² fluence) of the three samples are shown in Fig. 2 . To confirm that the THz transients did not originate from the MgO, a bare substrate was also tested yielding no THz emission. The 450°C film exhibited the most intense THz emission, followed by the In₂O₃/MgO film oxidized at 550°C. The 350°C film has a slightly weaker THz emission compared with the 550°C sample. The corresponding Fourier-transform amplitude spectra of the TDS waveforms (inset) revealed spectrally symmetric emission having a central frequency at 1 THz, with frequency components extending up to ~2 THz. The signal amplitude of the THz transients was approximately 10 times less than that of SI-GaAs and is two orders weaker than the THz emission from p-InAs. A recent survey of semiconductor surface THz emitters reported that p-InAs is currently the most intense semiconductor surface THz emitter [18] and thus, the observed THz emission from the In₂O₃ samples are very weak. However, the surveyed high-quality low-bandgap semiconductors were grown using well-established growth techniques, and were excited at energies well above the bandgap; hardly comparable with the conditions in this current study. With a signal-to-noise ratio dynamic range of approximately 1 order of magnitude in the amplitude spectra, the observed THz emission is sufficiently intense to warrant interest and further investigation. The THz data suggest a correlation between THz radiation and the presence of the cubic (222)-oriented c-In₂O₃ phase because XRD results showed that the 450°C-oxidized sample, having the highest THz emission, also exhibited the strongest intensity for this particular peak assignment. The presence of this phase appears to favor slightly more intense THz emission but this assertion needs further experimental confirmation and will not be pursued in detail in this work. On the other hand, it is important to have a better understanding of the underlying THz radiation mechanism albeit for sub-bandgap optical excitation.

 

Fig. 2 THz-TDS plots for the In₂O₃/MgO films oxidized at 350°C, 450°C, 550°C. The sample oxidized at 450°C exhibited the most intense THz emission. The inset shows the corresponding Fourier-transform spectra of the TDS data. The THz emission is centered at ~1 THz, and having frequency components of up to 2 THz. A signal-to-noise ratio dynamic range of more than 1 order of magnitude for the 450°C-oxidized sample illustrates implies intense THz emission.

Download Full Size | PDF

Initially, a two-photon absorption process was suspected. A two-photon photoconductive THz emission in ZnSe has been previously reported [19]. However, our further experimental verifications showed that photocarrier generation due to this nonlinear process is unlikely. The 2nd harmonic (400 nm at 40 mW) of the 800 nm wavelength emission of the Ti:Sapphire laser was used to excite the 450°C sample in order to investigate above-bandgap optical excitation. Its THz emission was then compared with the emission for the fundamental-line excitation at the same power. The results are shown in the inset of Fig. 3 , wherein the above-bandgap pump yielded lower THz emission. In general, the probability two-photon absorption is orders of magnitude lower than the single photon case and the comparable THz emission intensities in the inset do not support a dominant two-photon scenario. Moreover, even though the number of photons in the short-wavelength excitation case is only half of the fundamental line pumping, a two-photon process should still have exhibited less efficient photo-carrier generation (and THz emission). The difference in the THz-TDS waveforms in the two excitation conditions, however, suggests that the origin of photocarrier generation that leads to THz emission are different. The above-bandgap excitation case exhibited an almost single-cycle THz transient while the THz emission due to sub-bandgap absorption was characterized by several cycles. This is an indication of differences in the photocarrier dynamics between 400 nm and 800 nm excitation cases that could be an interesting topic for succeeding studies.

 

Fig. 3 Excitation fluence dependence of the THz emission from the 450°C-oxidized sample. The slope of the log-log plot indicates that the THz radiation mechanism is not a nonlinear optical process. The inset compares the THz emission from below-bandgap and above-bandgap excitation showing that the 800 nm wavelength pump is more efficient (the traces have been offset for ease of comparison); thereby ruling out a two-photon absorption process as the origin of the photo-carriers causing THz emission.

Download Full Size | PDF

Excitation fluence dependence measurements were then undertaken to further study the origin of the THz emission from the In₂O₃ films. The log-log plot of the fluence dependence in Fig. 3 did not exhibit a slope-2 dependence which would otherwise characterize a two-photon process [20]. The linear fluence dependence indicates that the origin of the THz emission could not be mainly attributed to two-photon absorption; although this possibility may not be completely excluded. Previous reports have demonstrated sub-bandgap excited THz emission in GaAs owing to the inverse Franz-Keldysh effect which is due to the third-order nonlinear susceptibility [21]. Similar results on the observation of THz emission via the second order nonlinear effect called optical rectification (OR) mechanism in sub-bandgap excited 6H-SiC [22] and in LiNbO₃, LiTaO₃, and dimethyl amino 4-N-methylstilbazolium tosylate (DAST) electro-optic crystals [23]. The OR THz emission mechanism does not necessitate photocarrier generation or above-bandgap excitation (as with the inverse Franz-Keldysh effect). However, these are nonlinear optical processes which are strictly dependent on the orientation of the crystal axis relative to the polarization of the optical excitation. In this regard, the OR mechanism is excluded as a possible THz radiation mechanism for the In₂O₃ samples in this study. The polycrystalline In₂O₃ film, comprising of different crystalline phases, is not favorable for OR-related THz emission [1]. More importantly, the azimuthal angle dependence of the THz emission was investigated and no distinct angular dependence was observed. This agrees well with the assertion that an OR-related nonlinear THz mechanism is improbable as the In₂O₃ films are polycrystalline.

The fluence dependence results also show that saturation of the THz emission occurs at about 400 nJ/cm². This finding is currently being studied in more detail but the low saturation value could provide insights on the characteristics of the THz mechanism. Although room-temperature optical spectroscopy such as photoluminescence and transmittance in the NIR region did not yield any informative results, this sub-bandgap absorption near the excitation wavelength region is currently believed to be the origin of the photo-carriers responsible for THz generation. This weak absorption band could originate from still-unverified electronic states coming from defects and impurities similar to earlier reports, which are common in wide bandgap semiconductor such as In₂O₃ [24,25]. The actual origin of this absorption band could not be confirmed by deep level transient spectroscopy (DLTS) as well, due to the difficulty in fabricating electronic contacts. As the XRD data showed, incomplete oxidation, and therefore the presence of indium, prevented the preparation of devices suited for DLTS measurements. These photo-carriers from defect-related absorption are presumably driven by a strong, strain-induced electric field at the In₂O₃/MgO interface region. This strain field coming from the huge lattice mismatch (58.36%) between the In₂O₃ film and the MgO can facilitate carrier drift at the In₂O₃/MgO interface region, resulting in THz emission [15,26]. In Refs [10,11], a 5 nm-deep electron accumulation layer near its surface similar to InAs and InN, is implied as the main reason why In₂O₃ may be considered to be a contender for intense THz emission. However, the optical excitation conditions in this work suggest otherwise. The In₂O₃/MgO sample transmits most of the 800 nm-wavelength light and the sample’s estimated 1,100 nm film thickness (according to AFM measurements) is large compared with the 5 nm depth of the reported surface accumulation layer. It must be noted that the excitation laser is practically transmitted through the In₂O₃ film. Thus the THz emission due to carriers driven by a surface field in the accumulation layer should be negligible compared with the sub-bandgap absorption deep beneath the surface. As such, this situation agrees well with the conjecture that carrier drift at the strained interface region is a more probable origin of THz transients as opposed to a surface effect. Furthermore, Ref [10] provided experimental data for the effective mass and carrier mobility for In₂O₃, which are 0.35m₀ and 32 cm²/(V.s), respectively. Aside from the fact that the sub-bandgap excitation results in carriers lacking excess energy required for photo-Dember emission, these nominal effective mass and carrier mobility values further suggest that a Dember-related mechanism is not favored. Thus, a field-driven photo-carrier drift should be the more plausible THz mechanism for the In₂O₃. The low saturation fluence exhibited by the samples could not be fully accounted for at this time. However, it is possible that long carrier lifetimes resulting in carrier accumulation and therefore, saturation could be a contributing factor. More so, defect-related states are prone to scattering [27] and should contribute to increased carrier lifetimes.

The presence of these defects has been shown not to detrimentally affect the THz emission properties of the In₂O₃/MgO sample in a significant manner. The micron-scale grain structures could scatter incident NIR light but the optical quality of the sample in the THz region has not been compromised. Shown in Fig. 4 is the comparison of the THz emission from the 450°C-oxidized sample in the transmission excitation geometry for the cases where the optical excitation was made incident on the film side and on the MgO substrate side. An important observation is that the peak intensities of the two signals are very much comparable. The intensity attenuation for the case of MgO substrate-side excitation is attributed to the attenuation of the excitation pump laser after passing through the MgO substrate, possibly due to reflection and optical scattering. As such, it can be deduced that: (a) the MgO substrate is sufficiently transparent to both the excitation beam and to THz radiation and (b) the In₂O₃ film is transparent to THz radiation.

 

Fig. 4 In the transmission-type excitation geometry, the THz emission intensity does not vary much whether the sample was excited from the MgO substrate side or from the In₂O₃ film side. Even as the sample’s over-all optical transmission is relatively low due to strong scattering, its THz optical qualities are not compromised. The TDS waveforms have been shifted for ease of comparison.

Download Full Size | PDF

4. Summary

This work experimentally demonstrated the generation of THz transients in polycrystalline In₂O₃ thin films on MgO substrates, prepared by thermal oxidation. Sufficiently intense THz emission can be observed even for excitation photon energy below the In₂O₃ bandgap of 2.9 eV. A comparison of the THz emission intensity between above- and sub-bandgap excitation initially suggested that a two-photon absorption process is not mainly responsible for the THz emission. This was confirmed by excitation fluence dependence measurements. The sub-bandgap absorption is currently attributed to weak absorption from defects and impurities although this has not been confirmed. Transmission excitation geometry THz measurements showed that the samples are sufficiently transparent in the THz region. Studies on the actual utilization of In₂O₃/MgO films in wide-bandgap PCA’s is beyond the scope of this work. However, results suggest that the necessity of short-wavelength excitation sources for such future devices may be circumvented owing to the observation of sub-bandgap excited THz emission from In₂O₃.