Fluorine doping: a feasible solution to enhancing the conductivity of high-resistance wide bandgap Mg0.51Zn0.49O active components

N-type doping of high-resistance wide bandgap semiconductors, wurtzite high-Mg-content MgxZn1–xO for instance, has always been a fundamental application-motivated research issue. Herein, we report a solution to enhancing the conductivity of high-resistance Mg0.51Zn0.49O active components, which has been reliably achieved by fluorine doping via radio-frequency plasma assisted molecular beam epitaxial growth. Fluorine dopants were demonstrated to be effective donors in Mg0.51Zn0.49O single crystal film having a solar-blind 4.43 eV bandgap, with an average concentration of 1.0 × 1019 F/cm3.The dramatically increased carrier concentration (2.85 × 1017 cm−3 vs ~1014 cm−3) and decreased resistivity (129 Ω · cm vs ~106 Ω cm) indicate that the electrical properties of semi-insulating Mg0.51Zn0.49O film can be delicately regulated by F doping. Interestingly, two donor levels (17 meV and 74 meV) associated with F were revealed by temperature-dependent Hall measurements. A Schottky type metal-semiconductor-metal ultraviolet photodetector manifests a remarkably enhanced photocurrent, two orders of magnitude higher than that of the undoped counterpart. The responsivity is greatly enhanced from 0.34 mA/W to 52 mA/W under 10 V bias. The detectivity increases from 1.89 × 109 cm Hz1/2/W to 3.58 × 1010 cm Hz1/2/W under 10 V bias at room temperature.These results exhibit F doping serves as a promising pathway for improving the performance of high-Mg-content MgxZn1-xO-based devices.

Another crucial issue restricting the practical use of high-Mg-content W-Mg x Zn 1-x O is its notably high resistance [9][10][11] . For example, the single-crystal W-Mg 0.55 Zn 0.45 O film 7 exhibits such high resistivity that the solar-blind UV PDs 2,4 fabricated on this film demonstrate a ~5 nA photocurrent when biased at 150 V and under 254 nm UV light illumination 2 , which is insufficient for practical requirements. Tuning the conductivity is therefore specifically necessary for high-Mg-content W-Mg x Zn 1-x O films and related devices. By intentionally introducing point defects-Zn interstitial, Liu et al. 12 reported their interesting result on the largely decreased resistivity of MgZnO as low as 0.053 Ω · cm. Meanwhile, heterovalent cation dopants-Ga 3+ and Al 3+ for instance-have been added into W-Mg x Zn 1-x O to create electron carriers. However, the effectiveness of these dopants as donors appears to decrease drastically as the Mg content (x) in Mg x Zn 1-x O increases 13,14 , similar to the case of Si in Al x Ga 1-x N 15 . Even more worse, such dopants like Ga might cause phase segregation in low-Mg-content W-Mg x Zn 1-x O (x ≤ 0.2) films 16 . Recently, Guo et al. 17 reported a quaternary alloy of Zn 0.9 Mg 0.1 OF 0.03 to be potentially applied as transparent electrodes. There has been no report on tuning the electrical properties of high-Mg-content W-MgZnO with bandgap in solar-blind range yet. Therefore, it is worth exploring some new methods for effective n-type doping in W-Mg x Zn 1-x O (x > 0.4) films in order to promote the corresponding device performance.
In this work, via comparative studies of doping with different cations and anion, a route was developed to replace O atoms with F for tuning the electrical properties of single-crystalline W-Mg 0.51 Zn 0.49 O films having a solar-blind 4.43 eV bandgap. The commercially available ZnF 2 powder (99.995%, Alfa Aesar) chosen as the source was firstly purified and solidified in order to exclude the possibility of unwanted impurities incorporation and meet the strict requirements by radio-frequency plasma-assisted molecular beam epitaxy (rf-MBE) growth process. Our observations solidly evidence that the incorporation of F does improve the n-type conduction behavior of high-resistance W-MgZnO deep UV components. Accordingly, the UV PD fabricated with Mg 0.51 Zn 0.49 O:F epitaxial film demonstrated an enhanced photocurrent, photoresponsivity and detectivity, one or two orders of magnitude higher than that of the device fabricated on the undoped film.

Results
The samples were synthesized on sapphire (0001) substrates by rf-MBE with a base pressure of ~10 −10 mbar. Reflection high-energy electron diffraction (RHEED) was utilized in situ to monitor the whole epitaxial growth process. On oxygen-terminated α -Al 2 O 3 (0001) surface [shown in Fig. 1(a)], the ultrathin MgO (111) layer provides a good template for subsequent W-MgZnO epitaxy. It should be noted that sharp and streaky RHEED patterns of the highly strained MgO ultrathin layer overlap those of sapphire [ Fig. 1 To confirm the single-crystalline wurtzite structure of the F-doped Mg 0.51 Zn 0.49 O layer, X-ray diffraction (XRD) θ -2θ and φ -scans were performed. Figure 1(f) shows the XRD θ -2θ curve of the F-doped sample. The peak (41.68°) is attributed to the diffraction from sapphire (006). Diffraction from W-MgZnO:F (002) planes locates at 35.04 o obviously shifting to a much larger angle in contrast to that of pure ZnO (34.46 o ), implying a high Mg content incorporated in the film. Importantly, the appearance of only the (002) related peak without any sign of cubic MgZnO:F confirms the single wurtzite phase, consistently with the in situ RHEED findings. The inset in Fig. 1(f) shows an enlarged image of the MgZnO (002) peak, confirming the constant Mg content in the doped and undoped layers. A slight asymmetry in the W-MgZnO:F (002) peak is attributed to the contribution from the low-Mg-content buffer layer underneath. In addition, Fig. 1(g) shows the φ -scan of the MgZnO:F (101) plane, which was carried out at χ = 60.87 o [the angle between (002) and (101) planes in a hexagonal system]. Six narrow and sharp peaks with equal 60 o intervals can be clearly observed, indicating the common sixfold symmetry of the single wurtzite crystal structure, consistent with the 60° symmetry observed in RHEED patterns.
For comparison, an intrinsic Mg 0.51 Zn 0.49 O film, Ga-doped and Al-doped alloy films were also synthesized with the same growth conditions. However, we face the difficulty in obtaining single-crystalline Mg x Zn 1-x O alloy films by Ga or Al doping when increasing the Mg content x above a critical level, as other researcher encountered 14,16 . Following the same process as before, the undoped high-Mg-content MgZnO was synthesized, as illustrated in Fig. 2(a,c). After Ga doping, the RHEED patterns show the trend of phase segregation, indicated by the slightly twisted (02) and (02) reciprocal spots [ Fig. 2(b)]. As Fig. 2(e) shows, the peaks (34.96°, 34.98° and 34.89°) indicate high Mg content incorporated in these films. However, the appearance of additional peaks implies the occurrence of multiple phases after doping. Indeed, it is a challenge to dope such high-Mg-content alloy films via Ga or Al dopants. Introducing cation dopants (Ga or Al) might decrease the Mg solubility in ZnO due to the more competitive bonding between cations (Zn, Mg, Ga or Al) and anions (O). Moreover, the Ga-or Al-doped films all show huge resistance. It is therefore worth evaluating the effect of F doping on tuning the electrical properties and the device performance.  Room-temperature reflectance spectroscopy was applied to determine the band gap of the sample. As indicated by an arrow in Fig. 3 Fig. 4(a). Due to the charge accumulation into the alloy film, SIMS signals, including these for Zn and F, manifest a monotonically decreasing trend, as illustrated in the inset of Fig. 4(a). Moreover, the data obtained in the range of 110-150 nm decrease a lot, close to the SIMS detection minimum limit, resulting in the huge fluctuation after normalization based on the Zn intensity [ Fig. 4(a)]. It should be noted that the undoped MgZnO layer is too isolated to be measured by SIMS under the same conditions. Thus, the F-involved region is about 140 nm thick, in a good accordance with the designed thickness of the F-doped layer (~150 nm). Disregarding to the SIMS uncertainty in the vicinity of the surface as well as fluctuations in the vicinity to the inner interface, the average doping level is estimated to be 1.0 × 10 19 F/cm 3 [ Fig. 4(a)].
In order to assess the effect of fluorine incorporation on the tuning of electrical properties, the Mg 0.51 Zn 0.49 O:F film was characterized by the temperature-dependent Hall measurement (TDH) using the van der Pauw technique in a magnetic field of 10 kG and a temperature range of 20-300 K. Figure 4  shows the carrier concentration (n) as the function of the reciprocal temperature, revealing two linear regions, labeled as I and II, respectively. In both regions, the electrical conductivity increases with increasing temperature. In order to determine the donor concentration (N D ) and the activation energy (E D ), we fitted these data by the least-squares method, assuming the charge neutrality equation for n-type semiconductor containing predominant donors and compensated acceptors: where N A , N c, g, k B and T denote the compensating acceptor concentration, the effective density of states in the conduction band, the donor degeneracy factor (~2), Boltzmann constant and absolute temperature, respectively. The impact from the semi-insulating MgZnO beneath the doped layer was reasonably neglected. The impact of the enhanced conductivity on the device performance was evaluated via fabrication of two Schottky type interdigital planar metal-semiconductor-metal (MSM) UV PDs with the undoped and F-doped Mg 0.51 Zn 0.49 O, respectively. Ti (10 nm)/Au (50 nm) was deposited to form finger electrodes with 5 μ m width, 300 μ m length, and 5 μ m gap, as illustrated in the inset of Fig. 5(a). The well-defined symmetrical rectifying behavior in Fig. 5(b,d) indicates the back-to-back Schottky contacts of the non-alloyed Ti/ Au on high-Mg-content films. The dark current of F-doped Mg 0.51 Zn 0.49 O UV PD is increased by more than two orders of magnitude compared to that of the undoped one [ Fig. 5(b)]. For a Schottky contact in an ideal case, if E 00 ≈ k B T, the thermionic-field emission (TFE) dominates the electronic transport process, which is a combination of thermionic emission (TE) and field emission (FE). E 00 , k B and T denote the characteristic energy, Boltzmann constant and absolute temperature, respectively. E 00 is defined as: Where q, ħ, N, m e * and ε s denote the elementary charge, the reduced Planck constant, the carrier concentration, the effective electron mass and the dielectric permittivity, respectively. The carrier concentration N is ~10 14 cm −3 and 2.85 × 10 17 cm −3 for the undoped and F-doped films, respectively. The dielectric permittivity in Mg 0.51 Zn 0.49 O is taken as 9.20ε o by assuming a linear increase from ε ZnO = 8.75ε o 22 to ε MgO = 9.64ε o 23 with the Mg-content x in Mg x Zn 1-x O. As a result, E 00 is 0.11 meV for the undoped film, which is smaller than the thermal energy k B T at room temperature (26 meV). However, E 00 for the doped one increased to a much larger value (6.0 meV) and could be to some extent comparable to the thermal energy. Therefore, the I-V curve is roughly determined by TE model and TFE model for undoped and doped cases, respectively, as illustrated in Fig. 5(c). F doping could narrow the potential barrier and result in larger tunneling currents 24 . Under 254 nm light illumination, the photocurrent of F-doped Mg 0.51 Zn 0.49 O UV PD increased by two orders of magnitude compared to that of the undoped one [ Fig. 5(d)]. Thus, the photoresponsivity of the device is dramatically enhanced, although sacrificing some contrast ratio.
The time dependence of photocurrent properties was studied by using periodic 254 nm and 365 nm illumination alternatively from a UV lamp [ Fig. 6(a,b)]. The 10%-90% rise and decay time are less than 0.12 s for both two PDs, which is the limit of our testing system. The sharp curve [ Fig. 6(a)] indicated F doping didn't result in extra defects, especially oxygen vacancies, which could cause persistent photocurrent (PPC) 25 . Figure 6(c) shows the photoresponsivity curve of both devices at a 10 V bias. The cutoff wavelength is 278 nm and 276 nm for undoped and F-doped MgZnO-based devices, respectively, which agrees well with the optical bandgap of these two films. Note that the peak photoreponsivity was enhanced from 0.34 mA/W to 52 mA/W after F doping. (The peak at 320 nm is induced by the exchange action of two gratings in the testing system.) The noise equivalent power (NEP) can be evaluated by NEP = (4k B T/ R dark + 2qI dark ) 1/2 ∆f 1/2 /R, where R dark , I dark , ∆f and R denote the differential resistance ( = / R dV dI) under the bias, the dark current under the bias, the electrical bandwidth and responsivity, respectively. k B ,T and q have the same meaning as above. The NEP is determined to be 4.21 × 10 −11 W/Hz 1/2 and 2.22 × 10 −12 W/Hz 1/2 for the undoped and F-doped device, respectively, under a 10 V bias at room temperature. The detectivity (D*) can be then determined by , where A is the optical active area. The detectiviy D* is 1.89 × 10 9 cm Hz 1/2 /W and 3.58 × 10 10 cm Hz 1/2 /W for undoped and F-doped devices, respectively. The above results indicated F doping would not cause severe quality degeneration of high-Mg-content  MgZnO films, but could enhance the responsivity and detectivity by 1 ~ 2 orders of magnitude. It unambiguously implies that F doping can robustly tune the conductivity of high-Mg-content W-MgZnO films in a controllable way and have a positive impact on the device performance.
To confirm the tuning effect of fluorine, the electrical properties of W-Mg x Zn 1-x O:F (0≤ x ≤ 0.3) thin films were investigated by Hall measurement. The resistivity increased from 5.32 × 10 −3 Ω · cm to 0.16 Ω · cm when increasing Mg content x from 0 to 0.3, with fluorine concentration of ~9.8 × 10 19 cm −3 (supplementary information).
In conclusion, the conductive W-Mg 0.51 Zn 0.49 O:F thin film was prepared with a fluorine concentration of ~1.0 × 10 19 cm −3 , lifting the electron concentration from ~10 14 cm −3 to 2.85 × 10 17 cm −3 . The conductivity and photoresponsivity increased by two orders of magnitude. The detectivity was enhanced from 1.89 × 10 9 cm Hz 1/2 /W to 3.58 × 10 10 cm Hz 1/2 /W. Two energy levels were revealed after fluorine incorporation, which is tentatively ascribed to built-in electric field discrepancy along the c-axis between two different configurations. The results indicate that F doping can dramatically modulate the electrical properties of high-Mg-content W-MgZnO and improve the UV device performance in solar-blind range, which is of crucial importance to promote the competitiveness of these active components. In addition, the purification and solidification process to ZnF 2 powder and their unique advantages, high-purity and solid phase for example, may offer a new approach for fluorine doping attempts in other wide bandgap oxides.

Methods
F doping into W-Mg 0.51 Zn 0.49 O film was realized with a solid anhydrous ZnF 2 source by radio-frequency plasma-assisted molecular beam epitaxy (rf-MBE). After degreasing in acetone and ethanol, a sapphire wafer was loaded into the vacuum chamber and thermally cleaned at 750 °C, followed by exposure to active oxygen radicals at 500 °C. An ultrathin unrelaxed cubic MgO buffer layer (~1 nm) was deposited at 500 °C, providing an epitaxial template for the growth of W-MgZnO film. Further, a quasi-homo Mg 0.36 Zn 0.64 O buffer layer (~20 nm), a high-Mg-content Mg 0.51 Zn 0.49 O epilayer (~80 nm) and a fluorine-doped Mg 0.51 Zn 0.49 O epilayer (~150 nm) were subsequently grown at 450 °C, respectively. Following the same growth process, intrinsic Mg 0.51 Zn 0.49 O film, Ga-doped and Al-doped alloy films were also synthesized. The K-cell temperature for ZnF 2 , Ga and Al was set at 420 °C, 500 °C and 890 °C, respectively. More growth details can be found elsewhere 7,26 .
XRD was performed using Cu Kα radiation (Empyrean System). RBS system is based on 1 MeV NEC Tandem accelerator. Additional insights into the bandgap were obtained using room-temperature reflectance spectroscopy (Cary 5000 System). SIMS measurements were performed using a Cameca IMS 7 microanalyzer. The electrical properties were characterized by TDH in Lakeshore 7604 system. Semiconductor parameter analyzer (Keithley 6487) was employed for I-V characterization. The spectra response was performed using the SpectraPro-500i (Acton Research Corporation) optical system with a 75 W Xe-arc lamp combined with a 0.5 m monochromator as light source.