An Accurate Quantitative X-ray Photoelectron Spectroscopy Study of Pure and Homogeneous ZrN Thin Films Deposited Using BPDMS

X-ray photoelectron spectroscopy (XPS) technique is widely used for evaluating the chemical state of the near surface elements revealed on a material, through the determination of the chemical shift occurring for peak positions in relation to the species surrounding the atoms under investigation. Generally, the analysis performed is only qualitative and in better cases semiquantitative, but in this way it does not express its powerful potentiality. The possibility of extracting quantitative information from surface materials, with an accuracy of ±1%, gives XPS analysis great potential. Of course, the intention to perform quantitative analyses should be clear to the analysts before operating and setting the experimental acquisition parameters. Calibration procedures and deep knowledge of the apparatus ensure a high degree of reliability, repeatability, and reproducibility of data at any time. In the present work a quite recent PHI VersaProbeII 5000 multi-equipped spectrometer was employed for XPS spectrum acquisitions. A monochromatic Alkα X-ray source with energy 1486.6 eV, power 12.5 W and 50 µm lateral resolution was employed for the acquisitions of all data sets. The polychromatic photoelectrons coming from the samples’ surface were collected using a hemispherical analyzer set in FAT mode, whose operating parameters were chosen to optimize the acquisition of high-resolution spectra. In some particular cases, depth profiling along the whole thickness of the specimens under investigation could prove a fruitful method for overcoming the limits of the resolution depth, typical of XPS. Therefore, the performances of a sputter monoatomic Ar+ ion gun and cluster Ar+₂₅₀₀ ion gun with maximum tension voltage of 20 kV were compared before acquiring depth profiling. The monoatomic Ar+ ion gun was chosen for sputtering all samples, taking a film of Ta₂O₃ as reference sample for calibrating sputtering time and thickness of eroded material according to working conditions. In the literature it is underlined that gas cluster sputtering guns are widely accepted for depth profiling of polymers samples, avoiding chemical damage or crosslinking but surprisingly in inorganic samples, and in particular on the commercial standard Ta₂O₅ (BCR-261T), preliminary tests showed that (i) the lower penetration depth does not lead to improved depth resolution when Ta₂O₅ is profiled, and (ii) even though cluster guns present a less preferential sputtering of the lighter elements in the first sputtering cycles, the use of 6 kV Ar+1000 ions onto Ta₂O₅ showed a gradual decrease in the O concentration. This progressive loss of oxygen atoms as a function of depth was further enhanced when the experimental conditions changed (higher tension and/or introducing sample rotation). However, an exhaustive comment for the choice of the mono ion gun rather than the cluster gun, for sputtering purposes, will be object of the section entitled ”Discussions”. The different ZrN films, deposited using the DPDMS method on conductive substrates, generally did not show any charging shift in spectra, despite sample ZrN D that displayed a 2.1 eV shift to lower binding energies suddenly correcting after spectra acquisition with respect to the lines of a ZrN-standard present in our laboratory database. With respect to the past study [16], the authors changed the fit function from SGL (30) to an “asymmetric Lorentzian” LA (a,b,c) included in the built-in functions of the CasaXps v.2.3.24PR1 software. LA (a,b,c) represents a kind of superset of Voigt function, with three variables and the possibility to include tail asymmetries on both sides in each synthetic peak considered. Nominally the “a” variable regulates the shape factor and must be properly chosen in combination with “b” and “c”. From an operative point of view “a” regulates the tail behavior on the right side of the synthetic peak (i.e., at high BE), while” b” varies in the interval [1, 200] and influences both the tail factor on the left side and the FWHMs values. If properly tuned until an optimized residual standard deviation value is obtained (generally in the range [1, 3]) for numerical residual standard deviation-RSD),“b” gives good output results for the peak fits improving the quality of elaborated data and giving a lower margin of errors. Finally, “c” is an integer between 0 and 499, relative to the width of the synthetic peak it belongs to, but also influences the tail factor “a” and “b” variables together. It assumes widely different values during data elaboration. An accurate quantification of XPS data also depends on the ad hoc choice of the range of background subtraction. It absolutely remains a matter of primary importance, otherwise generating unreproducible results. The Shirley method has been chosen and the energy slot for correspondent core regions was assigned and fixed for background subtraction. In Table 2 the start/end points are reported for Zr 3d and N 1s core regions for 3 representative samples of the three groups of films, at surface and after sputtering (level 2). It is worthwhile to point out that the “start” and “end” points of level 2 remain identical for all the levels of depth profiling under the surface for both N 1s and Zr 3d regions. In Table 2 some experimental values for Shirley subtraction method are reported.

Peak area corrections were operated via tabulated e-RSFs (i.e., empirical sensitivity factors reported in the PHI handbook of the apparatus), and the correction factors depend also on the specific transmission function of the analyzer. It was activated by ticking the “intensity correction” box and filling the empty field with the value −0.6. The theorical constrictions, such as the peak ratio in Zr 3d doublet fixed at 0.67 and the FWHM upper limit fixed at 2 in the satellite doublet, were introduced directly on the synthetic peaks in the “Narrow quantification window” of the software. A standard sputter Ar+ ion gun (model EX05) was employed for depth profiles using the bending geometry and shallow-condition angles (60–70° referred to the perpendicular axis to the surface of the sample).

This sample presents, unlike all the others, two important features: (i) the ZrN compound can easily be revealed already on the as-received sample (Figure 1a level one), and (ii) a little amount of oxygen is trapped in the whole thickness of the sample (Figure 2 O 1s for level two, but similar spectra are available for the subsequent levels). In Figure 1a, the nitride peak position for N 1s stands at 397.30 ± 0.02 eV with a FWHM of 1.34 eV. After fit, Zr 3d showed instead two doublet-components, one centered at (179.5, 182.02) eV and associated to ZrN, the other at (181.99, 184.45) eV, representing the satellite’s structure related to the zirconium nitride (3d_5/2, 3d_3/2 ) synthetic peaks (refer to Figure 1b). A perfect correspondence between the experimental Δ = 2.43 ± 0.02 eV and the literature value has been found for the ZrN compound, where Δ represents the distance between 3d _3/2 and 3d_5/2 lines. The stoichiometry of the film at level two was calculated considering the ratio between N 1s total area and Zr 3d total area, but without the satellite’s contribution. It resulted as being equal to 0.83 for level two, under stoichiometric in nitrogen when the N 1s peak position falls at 397.3 eV and 0.87 for level four when N 1s is peaked at 397.4 eV. Moreover, the Zr 3d of level four has a right-shift (as visible in Figure 1b right).

A small amount of oxygen contamination [18], in this case, influences only the satellite contribution forming zirconium oxide. The peak fit with the LA (a,b,c) fit function has been pictured in Figure 3 (left and right) for level two of Zr 3d and N1s, respectively of sample ZrN A. Analogous situation was found analyzing the twin sample ZrN F.