Investigation of work function and chemical composition of thin films of borides and nitrides

Thin films of various borides, nitrides, and barium fluorides were tentatively deposited by pulsed laser deposition and by magnetron sputtering in order to develop the components of thermionic‐photovoltaic devices for the high‐temperature thermal to electrical conversion by solid state. To improve the device performance, the materials characterized by a low work function were selected. In the present work, the chemical composition and work function of obtained films were investigated by X‐ray photoelectron spectroscopy and ultraviolet photoelectron spectroscopy techniques. The values of work function were determined from the cut‐off in the He I valence band spectra. Different films were compared and estimated on the basis of X‐ray photoelectron spectroscopy and ultraviolet photoelectron spectroscopy results.


| INTRODUCTION
The rapid development of the new technologies and global world economy is remarkably accompanied by exponential increase of the energy demand. In 2010, it was predicted the growth by 56% till 2040. 1 However, the conventional energy sources based on fossil fuel resources are limited and strongly dangerous for our environment. 2 For this reason, many research activities in the last decades have been addressed to the renewable energy technologies and eco-friendly energy sources, identified as a good alternative. In particular, the solar radiation is commonly considered as an abundant, cheap, clean, and sustainable energy source. However, the performances of solar energy converters to date are still less competitive than the conventional power generators. One of the best solution to improve their performances could be the development of renewable energy sources based on hybrid systems, offering advantages in terms of cost, reliability, and efficiency. 3,4 In this context, the solid-state devices like thermionicphotovoltaic (TIPV) converters have been proposed for the conversion of concentrated solar energy. In general, thermionic converters are experiencing an increasing interest if integrated with solar concentrations. 5 TIPV converters consist of 3 main elements: electron emitter (cathode), collector (anode), and thermophotovoltaic (TPV) cell. 6 The main characteristics of the emitter are the enhanced capability to emit electrons and the selective thermal emittance, which should be able to satisfy the specific requirements of the PV cell in terms of radiation absorption. Basically, an ideal cathode must have a low work function ϕ c , in order to emit a large amount of electrons, but it must be higher than that of anode ϕ a in order to allow the collection of electrons guaranteeing a correct output voltage.
In the present work, thin films of various borides and nitrides were deposited by pulsed laser deposition (PLD), a versatile technique for the controlled deposition of nanostructured films of numerous materials on various substrates, 7 with an aim to improve the cathode and anode performances in TIPV devices. The thermionic electron emission properties of lanthanum hexaboride (LaB 6 ) and cerium hexaboride (CeB 6 ) films are well established. For this reason, they were considered as good candidates in terms of low work function (ϕ = 2.5-2.6 eV) and high melting point (T m > 2000°C). 8,9 Another valid alternative could be represented by nitride films, in particular, the amorphous carbon nitride (CN x ) and hydrogenated aluminium nitride (AlN:H), because their work function can be even lower. 10,11 In the TIPV, the cell could play itself the role of thermoionic anode. GaAs was selected for the development of the TPV cell, but it is characterized by ϕ = 4.6 eV for p-type; thus, it could not provide the necessary condition ϕ c > ϕ a . Therefore, its work function was tentatively reduced with an ultra-thin film of BaF 2 , deposited by RF magnetron sputtering. The aim of the present work was the exploration by using X-ray photoelectron spectroscopy (XPS) and ultraviolet photoelectron spectroscopy (UPS) techniques, how the surface chemical composition and work function depend on different growth parameters. The work function ϕ was calculated by using the formula: ϕ = hν (He I) − E cutoff , where E cutoff is the energy of the cut-off level in the valence band spectrum excited by He I line. 12 In order to avoid the low kinetic energy cut-off caused by the spectrometer, the work function ϕ was measured following the method proposed by Martinez et al, 12 developed to determine the influence of a negative bias (V) on the potential barrier at insulator-vacuum surface. This method consists in the acquisition of UPS spectra applying a series of negative bias voltages V. Because the photoemission threshold linearly depends on the square root of the absolute value of negative bias applied to the sample, as described by the classical Schottky model for a potential barrier, the work function value was calculated from the plot of E cutoff = f(│V│ 1/2 ).  Table 1. The thin films of fluorides (samples BF1, BF2, and BF3) were deposited by RF magnetron sputtering. The chamber was evacuated to 1.5 × 10 −6 mbar and back-filled with Ar flux to 1.7 × 10 −2 mbar.
The RF power input to the BaF 2 target was kept constant at 100 W, with a negative substrate bias of 375 V with respect to ground. The substrate holder rotated during the deposition at a speed of 60 rpm, whereas the deposition time varied from 100 to 300 seconds.

| Surface characterization
The surface analyses were carried out by using an ESCALAB 250Xi The UPS measurements were carried out at a pressure of approximately 2 × 10 −8 and 5 × 10 −9 mbar for He I and II, respectively. The spectra were collected at 2 eV pass energy. In order to check the contribution of the surface contamination, the spectra were also registered after a short surface cleaning (30 seconds) by ion sputtering.
The work function ϕ of the samples was calculated by measuring the characteristic cut-off energy in the He I spectrum, as shown in Figure 1. A negative bias of 2, 5, 10, and 15 V was applied to shift the spectra from the spectrometer threshold. The E cutoff values measured at different bias voltages were plotted as E cutoff vs |V| 1/2 ( Figure 1), as it was suggested in Martinez et al. 12 From these plots were extrapolated the values of the work function ϕ at zero bias.
Spectroscopic data were acquired and processed by the Avantage v.5 software. Shirley background subtraction and mixed Lorentzian/ Gaussian peak shape (30%) were used for the peak fitting. La 3d, Ce 3d, and B 1s spectra are shown in Figure  can partially derive from a preferential sputtering due to the significant mass difference between La and B atoms. 15 A large content of oxygen in the bulk of all films (Figure 3) was most probably due to the favorable tendency for La oxidation. 16 As it can be expected, operating at a lower pressure (sample LB3), the oxygen content slightly decreases, but at the

| Nitrides
The chemical composition of nitride films reported in Table 1 displays   the presence of Al, N for oxynitrides. 18,19 The calculated atomic ratios of Al/N (Table 1) evidenced an excess of Al that was partially oxidized. The situation was more or less the same for CN sample, where was registered a

| Barium fluoride
The films of BaF 2 were tentatively deposited onto GaAs substrate in order to reduce its work function. The XPS analysis of the BaF samples revealed the presence of C, O, Ga, and As, beside of Ba and F ( Table 1).
The presence of C was related to the carbon contamination, which was completely removed after 30 seconds of ion sputtering. The presence of Ga 3d and As 3d signals indicates that the deposited films were very thin. In fact, their thickness was estimated to be approximately of 0.5, 0.8, and 2.0 nm for the samples BF1, BF2, and BF3, respectively. This result was correlated with increasing deposition time. The Ba 3d 5/2 peak ( Figure 4) was positioned at BE = 780.3 eV, which can be assigned contemporaneously to fluoride 22 or oxide, 23 whereas a single F 1s peak was positioned at BE = 684.7 eV assigned to fluoride. 21 However, except in the thinnest film, the atomic ratio of Ba/F was not stoichiometric, indicating that also Ba was partially oxidized. The work function of the BF samples, calculated from UPS (Table 1) to have an efficient energy conversion, the emitter work function must be in the range of 2.5-2.7 eV. Therefore, the obtained work function value is sufficiently compliant with the request specifications. Conversely, a work function than 2.0 eV is expected for the TIPV anode.
The barium fluoride films, although reduced the work function of the GaAs substrate (TPV cell) by 1.3-1.4 eV, are far from the requested values. Most probably, it depends from the high content of incorporated oxygen. Therefore, in the near future, it will try to act on deposition parameters to reduce the oxygen partial pressure as well as the thickness of the films, preferably monolayers.