Influence of Annealing on Thin Film/Substrate Interface and Vacuum Ultraviolet Photoconductivity of Neodymium Fluoride Thin Films

High photon energy vacuum ultraviolet radiation (VUV, 100−200 nm wavelength) is challenging to detect. It easily degrades conventional silicon and semiconductor photodetectors. Fluoride photodetectors can be the answer, but the correlation between fabrication parameters and photodetector performance is not known. Here, the effect of annealing is investigated on the characteristics of neodymium trifluoride thin film/quartz substrate interface and NdF3 photoconductivity within the VUV. Thin films are deposited on unheated and heated (600 °C) substrates with post‐deposition annealing. Dark current of films on unheated substrates decreases by as much as 1/10 as resistance increases from 1 −12 TΩ after annealing. Dark current of films on heated substrates increases even after annealing, resulting in similar photo and dark currents of ≈303.7 nA and poor detectors. Fluorine diffuses from the film to the substrate during deposition, exacerbated by substrate heating but not by annealing. Fluorine diffusion degrades crystallinity near the interface, increasing the dark current. Fluorine diffusion is absent when MgF2 is used as the heated substrate. Unannealed NdF3/MgF2 detector on 600 °C‐heated substrate and 600 °C‐annealed NdF3/SiO2 detector on unheated substrate exhibit similar resistances of ≈14 TΩ. Considering the film/substrate interface and annealing is crucial when developing VUV photodetectors.


Introduction
Vacuum ultraviolet (VUV) radiation spans the wavelength range from 100 to 200 nm. [1]Its high photon energy renders VUV radiation indispensable in a wide range of applications, including semiconductor lithography, sterilization of surfaces, optical cleaning, surface modification, and spectroscopy, to name a few.[4][5][6][7][8][9] Development of VUV detectors is crucially important.[12] Since VUV light is strongly absorbed by oxygen in air, requiring propagation through vacuum, development and application of VUV photodetectors can be very challenging.Conventional silicon-based detectors, such as silicon photodiodes, which were originally designed for visible spectrum applications, can be used in the VUV region.However, these detectors have distinct disadvantages due to their inherent wide band response from X-ray to near-infrared and because they are easily degraded by high-energy radiation. [13,14][17][18][19][20][21][22][23][24][25][26][27][28][29][30][31] Despite the improved radiation hardness of WBG semiconductor-based detectors compared to silicon-based detectors, their bandgaps still lie in the UV region, and some degradation is therefore expected, especially during long-term operation.In contrast, wide bandgap fluoride-based detectors have recently emerged.Fluorides have extremely wide bandgaps of at least 6 eV, which perfectly coincide with the VUV wavelength region.Their extremely wide bandgap and ionic nature render fluorides more resistant to degradation from prolonged exposure to high-energy VUV radiation. [32][35][36][37] Unlike silicon-based detectors developed by relatively mature production technologies and WBG semiconductor-based detectors, which are widely investigated in current literature, the development of fluoride photodetectors is in its infancy and thus not much is known regarding the correlation between fabrication parameters and photodetector performance.In this work, we investigated the influence of annealing on the interface between neodymium trifluoride (NdF 3 ) thin film and quartz (SiO 2 ) substrate.NdF 3 is a wide bandgap fluoride that crystallizes in the hexagonal P6_3 cm space group.It is highly transparent down to ≈160 nm.[39] There have been reports on the growth of NdF 3 films on MgF 2 substrates using molecular beam deposition and on the growth of NdF 3 crystals from melt using the micropulling down (μ-PD) method. [40,41]In semiconductor device fabrication, annealing is the process of heating the semiconductor to a predetermined temperature, keeping it at that temperature for a fixed time, and then cooling to room temperature.Annealing is typically done to relieve stress, activate ion-implanted dopants, and reduce structural defects.During annealing, the fabricated NdF 3 film on SiO 2 substrate is heated to a temperature of 200, 400, and 600 °C inside a furnace for 3 h and then subsequently cooled to room temperature.The observed effects are then related to the photoconductivity of NdF 3 thin film-based VUV photodetector.In this study, we deposited NdF 3 on an inexpensive and practical SiO 2 substrate, and by studying the interface between fluoride and oxide, we obtained important knowledge about the heterojunction between fluoride and oxide.

Results and Discussion
NdF 3 thin films were deposited on SiO 2 substrates by pulsed laser deposition (PLD) for two substrate conditions.In the first condition, the temperature of the SiO 2 substrate was maintained at room temperature (R.T.) during deposition (unheated substrate).For the second condition, the temperature of the substrate was maintained at 600 °C during deposition (heated substrate).After deposition, the thin films from both substrate conditions were Table 1.Post-deposition annealing temperatures for NdF 3 thin films deposited on heated and unheated substrates.The films were annealed in vacuum, except for one set of as-grown films without post-deposition annealing that was kept unannealed for reference.annealed in vacuum for 3 h at 200, 400, and 600 °C.As-grown films without post-deposition annealing were also kept and characterized for reference.Altogether, eight NdF 3 thin films were prepared, as shown in Table 1.
Figure 1a shows the XRD spectra of the annealed films deposited on the unheated substrate.Fitting the peaks of the XRD spectrum with pseudo -Voigt functions clearly shows the presence of four peaks (Figure 1d).The peaks at 2 = 24.7 o (blue curve), 25.4 o (cyan curve), and 28.3 o (yellow-green curve) can be indexed to NdF 3 (002), (110), and (111), respectively, with a preferred orientation in the (111) direction.There is a peak at 2 = 26.8o (magenta curve) that is clearly distinguishable from the background.This peak can be indexed to NdF 2 (111).We note that neither Nd 2 O 3 nor NdF 3 exhibit diffraction peaks at this position.The NdF 2 peak is broad due to its unstable nature.Similar fluorine removal has been reported in CeF 3 fabricated by PLD whereby CeF 3 broke down to CeF 2 . [34]As shown in Figure 1b, the full width at half maximum (FWHM) of the peaks decreased after annealing.The peak area ratio, calculated from the area of the NdF 2 (111) peak divided by all peak areas, decreased as the annealing temperature was increased up to 400 °C, as shown in Figure 1c, indicating that the amount of NdF 2 could be decreased and the crystallinity of NdF 3 could be increased by postdeposition annealing in vacuum.[44][45] However, when the sample was annealed at 600 °C, the peak area ratio increased.Since the peak area ratio is calculated from the area of the NdF 2 peak divided by all peak areas, this could be due to both the increase in the intensity of the NdF 2 peak (as is evident in Figure 1a) and the decrease in the FWHM of the NdF 3 peak (as is evident in Figure 1b).The physical properties of pure NdF 2 alone, without the presence of other NdF 3 peaks, have not yet been elucidated because of its unstable nature.We can only speculate that annealing above 400 °C improves the crystallinity of NdF 2 resulting   to the observed increase in the intensity of the NdF 2 peak in Figure 1a.
The current-voltage (I-V) characteristics of the NdF 3 thin film photodetectors deposited on the unheated SiO 2 substrate with post-deposition annealing are shown in Figure 2a.The current during VUV illumination (photocurrent or I photocurrent ) was obtained by subtracting the current measured without VUV illumination (dark current or I dark current ) from the measured current.All the films were photoconductive, resulting in a significantly higher photocurrent during VUV illumination compared to the dark current.Annealing significantly decreased the dark current compared to the as-grown film, by as much as 1/10 of the original value when annealed at 600 °C.In contrast, the photocurrent was increased by a factor of 0.7.Consequently, the signal-tonoise (SN) ratio of the annealed detectors increased by a factor of ≈7 compared to their unannealed counterparts for a bias voltage of 10 V, as shown in Figure 2b.The SN ratio is defined as the ratio of I photocurrent to I dark current .The resistance of dark current also increased by a factor of ten.These results demonstrated that post-deposition annealing improved the photoconductivity of NdF 3 thin film detectors.
Figure 3 shows the I-V characteristics of the NdF 3 thin film detectors deposited on the heated SiO 2 substrate with post-deposition annealing.Relatively high dark current values were obtained from the detectors, as shown in Figure 3a.The dark current appeared to decrease when the film was annealed at 200 °C.However, a further increase in the annealing temperature resulted in an increase in the dark current, especially when a 400 °C annealing temperature was used.Nonetheless, photoconductivity could not be measured for higher annealing temperatures.Hence, focusing on the 200 °C annealing temperature, which provided the lowest dark current measurement, Figure 3b-d depicts the I-V characteristics of the thin film detector.The photocurrent was very similar to the dark current, indicating that the resistance of the films was very low when the substrate was heated during deposition.
Furthermore, post-deposition annealing failed to improve the photoconductivity of the films.This might be explained by the fact that the thin film/substrate interface exhibited low crystallinity when the substrate was heated during deposition.As a result, the thin film near the interface had low resistance.As the schematic diagram in Figure 4 shows, use of a high bias voltage across the electrodes meant that the electric force encountered this area of low resistance, thereby leading to photocurrent values that were similar to the dark current ones.
To verify this and elucidate the observed negative effect of heating the substrate during deposition, the interface between the NdF 3 film and SiO 2 substrate was evaluated using TEM with EDS. Figure 5 shows the TEM image of the interface when the film was deposited (a) on an unheated substrate without postdeposition annealing, (b) on a heated substrate without postdeposition annealing, and (c) on an unheated substrate with postdeposition annealing at 600 °C.The interface between the annealed film and the unheated substrate revealed a clean separation between the two (Figure 5c).As discussed previously (Figure 2), the detector from this film exhibited the highest SN ratio and the highest resistance to dark current.The EDS mapping in Figures 6, 7 illustrates fluorine diffusion from the film into the substrate when the substrate was heated during thin film deposition (Figure 7b).Oxygen is also observed in the thin film, indicating that oxygen also diffused from the quartz substrate to the fluoride film at the interface possibly forming a thin layer of oxyfluoride.The formation of this thin layer of oxyfluoride caused the interface between the substrate and the film to appear blurred in the TEM image.
A comparison of fluorine composition in all three samples suggested that post-deposition annealing did not appear to induce any fluorine diffusion into the substrate, as shown in Figure 8a,b.This further indicates that diffusion of fluorine from the film to the substrate had occurred during the deposition process and that the temperature of the substrate had played a critical role in the severity of this diffusion.A small amount of fluorine was also detected in the substrate of the unheated substrate with or without post-deposition annealing, suggesting that the heat produced by PLD ablation could also contribute to the observed diffusion.However, fluorine diffusion was exacerbated when heating the substrate at 600 °C.As illustrated in Figure 8c, diffusion from the film to the substrate occurred within 40 to 50 nm across the interface. [46]Nd and Si diffusion was not observed.This is   attributed to the formation of a strong Nd-Si bond at the thin film/substrate interface, [46] which also caused F to be released.
The XRD spectra of the as-grown films that were deposited on either heated or unheated substrates are shown in Figure 9a.The peaks predominantly arise from NdF 3 and partly from NdF 2 .As shown in Figure 9b, these peaks became narrower (smaller FWHM) when the substrate was heated during the deposition process.However, Figure 9c shows that the ratio of the NdF 2 peak area relative to the area of all peaks was increased, indicating an increase in NdF 2 .The increase in NdF 2 is ascribed to the diffusion of fluorine to the substrate as observed in the TEM-EDS evaluation.
Since diffusion occurs at a depth of 40 to 50 nm, we fabricated thinner NdF 3 films to further investigate what happens at the film/substrate interface.The deposition conditions were kept the same as in the previous thicker films, but the deposition time was decreased to 30 min (instead of 2 h).The films were once again deposited on unheated and heated (600 °C)  substrates.Figure 10a shows the XRD spectra of the as-grown (unannealed) films.The XRD peaks appear at two positions similar to the thicker films, and both films consist of NdF 3 and NdF 2 .Figure 10b demonstrates that the FWHM of the thinner films increased when the substrate was heated, as opposed to thicker films where the FWHM decreased.Moreover, Figure 10c shows the ratio of the NdF 2 peak area relative to the area of all peaks was found to be increased, confirming the increase in NdF 2 when the substrate is heated.These results verified that heating the substrate during deposition led to decreased crystallinity at the film/substrate interface and increased diffusion of fluorine from the film to the substrate.Accordingly, the resistance at the interface also decreased, causing the dark current in the detector to increase to the point where it was similar to the photocurrent, as shown in Figure 3 and discussed previously.
Figure 11 shows the SEM images of the film deposited on the unheated and 600 °C-heated substrates.Droplets are observed, but there is no visible difference in the morphology of the films when the substrate was heated compared to when it was at room temperature.
Further investigation was carried out to evaluate the role of the substrate.For this purpose, the SiO 2 substrate was replaced by a magnesium fluoride (MgF 2 ) crystal.A fluoride crystal was selected as the substrate because aside from its wide bandgap, we also wanted to investigate the diffusion of fluorine from the film to the substrate when the substrate was heated during the deposition process.The NdF 3 thin film was deposited on the MgF 2 substrate using the same deposition conditions, including the deposition time of 2 h, employed with the SiO 2 substrate.The MgF 2 substrate was also heated to 600 °C during deposition and the film was not annealed after deposition (as-grown).Figure 12 shows a comparison of the dark current and photocurrent from the annealed NdF 3 film deposited on the unheated SiO 2 substrate (NdF 3 /SiO 2 ) and the unannealed (as-grown) NdF 3 film deposited on the heated MgF 2 substrate (NdF 3 /MgF 2 ).Interestingly, the film deposited on the unheated SiO 2 substrate with post-deposition annealing in vacuum at 600 °C exhibited similar dark current values as the as-grown film deposited on the MgF 2 substrate heated at 600 °C during deposition.This indicates that both detectors had similar dark current resistance, aris-ing from similar crystallinity.Furthermore, these results underline that fluorine was not diffused from the film to the substrate.

Conclusion
In summary, the effect of annealing on the interface between NdF 3 thin film and SiO 2 substrate and hence on the photoconductivity of NdF 3 thin film VUV photodetector, was investigated.Thin films were deposited under two substrate conditions: unheated substrate at R.T. and heated substrate at 600 °C.The deposited films were then annealed in vacuum and their structure and photoconductivity were compared to those of the unannealed (as-grown) films.The films deposited on the unheated substrate exhibited an increase in dark current resistance and hence a decrease in dark current after annealing.Consequently, our findings revealed a sevenfold increase in the SN ratio of the detectors following post-deposition annealing.Resistance to dark current of the films deposited on the 600 °C-heated substrate decreased, resulting in similar values to those of the photocurrent at high bias voltages and suggesting poor photoconductivity.Investigation of the NdF 3 /SiO 2 interface revealed that fluorine diffused from the film to the substrate during deposition.Diffusion can be attributed to the energy deposited by PLD ablation; however, diffusion is exacerbated when heating the substrate during deposition.Post-deposition annealing did not seem to significantly contribute to fluorine diffusion.Diffusion of fluorine was found to degrade NdF 3 crystallinity near the interface, explaining the decrease in dark current resistance and increase in dark current.Deposition of NdF 3 on a MgF 2 substrate heated at 600 °C did not result in fluorine diffusion.Consequently, the SN ratio of the NdF 3 /MgF 2 detector that was deposited on a 600 °Cheated substrate without post-deposition annealing was similar to that of the NdF 3 /SiO 2 detector that was deposited on an unheated substrate with post-deposition annealing at 600 °C.Our results highlight the importance of selecting an appropriate substrate and the role of annealing in improving photoconductivity, including the timing for that annealing should be performed.Equally, our results could pave the way to improved photoconductivity of VUV photodetectors through annealing and careful choice of substrate.

Experimental Section
NdF 3 thin films were deposited on SiO 2 substrates by pulsed laser deposition (PLD) using the fourth harmonics of a Nd:YAG laser operating at 266 nm wavelength, 5 ns pulse width, and 10 Hz repetition rate.The laser power was set at 100 mW.PLD was used to the NdF 3 films because the small difference in chemical composition between the source target and the deposited film enables to obtain high-quality thin films without using toxic fluorine when depositing fluoride thin films.It was also easy to evaporate materials with a high melting point using PLD.[35] The ceramic NdF 3 target has dimensions of 20 x 10 mm and purity of 99.9%.The distance between the target and the substrate ranged from 27 to 30 mm, with 27 mm being the case when using a new target.The deposition chamber was maintained in vacuum at 6 × 10 −6 Pa and deposition was carried out for 2 h.The structure and crystallinity of the thin films were characterized using X-ray diffractometry (XRD, Smartlab, Rigaku).The interface between the NdF 3 films and the SiO 2 substrate was evaluated using transmission electron microscopy with energy dispersive X-ray spectroscopy (TEM-EDS, JEM-2100F).The distribution of F, Nd, O, and Si on the fabricated films was determined using EDS mapping.To fabricate the NdF 3 photoconductive detector, a pair of aluminum interdigitated electrodes was deposited on the surface of each set of thin films.The patterned area of the interdigitated electrodes was 5 × 2.4 mm 2 .The gap between the electrodes was 0.2 mm and the width and length of the electrodes was 2.4 and 5 mm, respectively.Photoconductivity was characterized in the VUV region by illuminating the patterned area with a deuterium lamp (Hamamatsu Photonics K.K. L10366 series).The deuterium lamp emitted wavelengths in the range of 120-180 nm with a dominant emission peak at 160 nm and an average power of 88 μW.The current was measured using a high-resistance electrometer (8340A, ADCMT).

Figure 1 .
Figure 1.a) XRD spectra of films deposited on an unheated SiO 2 substrate (@R.T.) with post-deposition annealing at 200, 400, and 600 °C.As-grown refers to the film without post-deposition annealing.b) FWHM of the XRD peaks as a function of annealing temperature.c) Area of NdF 2 (111) peak divided by all peak areas (peak area ratio) as a function of the annealing temperature.d) Fitting of the XRD spectrum peaks from the as-grown sample.

Figure 2 .
Figure 2. a) I-V characteristics of the NdF 3 thin film photodetectors deposited on an unheated SiO 2 substrate (@R.T.) with post-deposition annealing at 200, 400, and 600 °C.As-grown refers to the film without post-deposition annealing.b) Signal-to-noise ratio (SNR) and resistance of the photodetectors as a function of the annealing temperature.

Figure 3 .
Figure 3. a) Dark current of the NdF 3 thin film photodetectors deposited on a heated SiO 2 substrate (@600 °C) with post-deposition annealing at 200, 400, and 600 °C.As-grown refers to the film without post-deposition annealing.b) Dark current and c) photocurrent of the photodetector deposited on the heated SiO 2 substrate and annealed at 200 °C.d) Enlarged section of the photocurrent of the detector from (c) for applied bias voltage up to 30 V.

Figure 4 .
Figure 4. Schematic diagram of the mechanism for low resistance and increased dark current values measured from the NdF 3 film deposited on a heated SiO 2 substrate.

Figure 5 .
Figure 5. TEM image of the NdF 3 /SiO 2 interface when the film was deposited a) on an unheated substrate without post-deposition annealing, b) on a heated substrate without post-deposition annealing, and c) on an unheated substrate with post-deposition annealing at 600 °C.

Figure 6 .
Figure 6.TEM-EDS showing the distribution of fluorine (F), neodymium (Nd), oxygen (O), and silicon (Si) in the film deposited on a) an unheated SiO 2 substrate without post-deposition annealing, b) a heated SiO 2 substrate without post-deposition annealing, and c) an unheated SiO 2 substrate with post-deposition annealing at 600 °C.

Figure 7 .
Figure 7. Intensity as a function of the fluorine (F), neodymium (Nd), oxygen (O), and silicon (Si) position in the film deposited on a) an unheated SiO 2 substrate without post-deposition annealing, b) a heated SiO 2 substrate without post-deposition annealing, and c) an unheated SiO 2 substrate with post-deposition annealing at 600 °C.

Figure 8 .
Figure 8.Comparison of the a) raw and b) curve-fitted distribution of fluorine in the film deposited on an unheated SiO 2 substrate without postdeposition annealing (as-grown@R.T.), a heated SiO 2 substrate without post-deposition annealing (as-grown@600 °C), and c) an unheated SiO 2 substrate with post-deposition annealing at 600 °C (600 °C@R.T.).c) Schematic diagram of fluorine diffusion from the film to the SiO 2 substrate.

Figure 11 .
Figure 11.SEM images of the films deposited on a) unheated and b) 600 °C-heated substrates.

Figure 12 .
Figure 12. a) Dark and b) photocurrent of the NdF 3 thin film photodetectors deposited on an unheated SiO 2 substrate (@R.T.) with post-deposition annealing at 200, 400, and 600 °C.As-grown@R.T. refers to the film without post-deposition annealing onto an unheated SiO 2 substrate.As-grown@600 °C to the film without post-deposition annealing onto a 600 °C-heated MgF 2 substrate.