Simulation of the XPS depth profile of pristine sampleThe structure of the pristine sample has been determined previously16,19. It is: 10.5 nm C/24.5 nm W/9 nm C/2 nm SiO2/Si substrate. The C/W and W/C interfaces are sharp and clean (no intermixing, adlayer etc. are found) and the layers are amorphous, while the Si substrate is crystalline16,19.First we carried out the simulation of the XPS depth profile of this sample with well-known structure to validate our proposed protocol, and to show the relative weight of the various distortions that might appear during XPS depth profiling. The as recorded XPS depth profile obtained on the pristine sample is shown in Fig. 3. The conditions of the ion bombardment used for the depth profiling were: 500 eV Ar+ ions with angle of incidence of 45°, and the sample was not rotated during the ion sputtering. The XPS peaks have been identified by means of their binding energies.Figure 3The as measured XPS depth profiles obtained on a pristine sample with structure of 10.5 nm C/24.5 nm W/9.0 nm C/2 nm SiO2/Si substrate. W, Si and C sign the XP intensities emitted from elemental W, Si, and graphite, while C in W2C stands for the C line which is emitted from W2C compound. SiO2 and SiC sign the photoelectron intensities of Si emitted from SiO2 and SiC compounds.At first glimpse the result is rather disappointing; the as recorded XPS depth profile show strongly intermixed layers separated by wide interfaces and demonstrates the presence of a new components (expected based on Ref.13) of W2C and SiC, thus its resemblance to the initial (known) structure is very moderate. First we checked the applicability of the routine evaluation procedure22; the result is shown in Fig. 4.Figure 4The result of a routine evaluation22 of the XPS depth profile shown in Fig. 3.According to this evaluation the sample (initial condition) depth profiled contains strongly intermixed carbon and tungsten regions which are connected by wide transitions. Besides the elemental components tungsten carbide (W2C) and silicon carbide (SiC) also appear. The proposed structure is gravely different from the initial structure of the sample, thus, the routine evaluation process for layer system having thickness comparable with those of the actual IMFPs cannot be used.In the proposed evaluation procedure, we considered that the as recorded XPS depth profile showed strong intermixing, wide interfaces, appearance of new compound which might arise partly because of the sputter removal process applied during depth profiling and partly because of the non-negligible IMFP of the analyzed photoelectrons. The two processes can be evaluated independently.First we consider the distortion of XPS depth profile only due to the finite IMFP of the photoelectrons, by simulating a “damage free” depth profile by simply peeling the initial sample.The comparison of the as measured depth profile with that simulated during the peeling process is far for being straightforward. The problem is that during the measurement the XPS intensities are measured as a function of sputtering time (s), while in the case of peeling given thickness of layer is removed that is the XPS intensities are calculated as a function of removed thickness which is regularly called as depth (nm). If the sputtering yield is known the sputtering time can be converted to number of atoms from this, if the density is also known, the removed thickness can be calculated. Since the sputtering yields of C, W and Si are rather different being 0.83, 1.05 and 1.36 (using Ar+, 0.5 keV, and 45° angle of incidence) the conversion of sputtering time to depth might result in errors. For this reason, only the interface regions where the horizontal axes can be fitted with reasonable accuracy are shown in Fig. 5, instead of the entire profile.Figure 5The comparison of the as measured depth profiles (symbols) with that of simulated ones (lines) in the case of peeling in the range of (a) C/W interface, (b) W/C; C/SiO2 and Si interfaces.The agreement between the simulated depth profiles of the main components, obtained in the case of peeling, and that of the measured depth profiles is promising. The deviations between the corresponding curves are focused to the narrow region of the interface. That is, simply the relatively high IMFPs of the photoelectrons result in the features which seem to be intermixing. Note that the “intermixing” is much larger at the C/W interface than that at the W/C interface. This can also be easily explained; the IMFP of the photoelectrons being 3.8 nm W 4f in C and 1.7 C1s in W, respectively. Consequently, the “penetration” of W to C (C/W interface) is much higher than the “penetration” of C to W (W/C interface). Thus, the correction for IMFP is absolutely important.A C/W/C/SiO2/Si system was peeled thus W2C (which appeared in the measured spectrum) evidently did not appear. To understand its appearance first of all let emphasize that the energies of W photoelectrons emitted from pure W and W2C are different, thus we can measure them separately but their IMFPs are the same. The very same is true for the C as well. It follows if the W2C is present in the sample the shape of the C in W2C profile should be similar to that of metallic W (the IMFP values are rather close). Considering Fig. 3 it is not the case; the W and C in W2C profiles in the first C layer are broad and sharp, respectively. It means that the W2C was produced during the sputter removal process. We note that this observation agrees with the XTEM results which did not show W2C layer on the interface16,19.It is known that the medium energy ion–solid interaction in the case of C/W multilayers results in intermixing and compound formation17,20. We have also shown, rather surprisingly, that ion bombardment with energy as low as 1 keV also results in tungsten carbide formation16. Thus, it also follows that the sputter removal process (Ar+ energy is 0.5 keV) might be blamed for the appearance of the W2C. Accepting this we will simulate the sputter removal process applying the TRIDYN code (using the experimental sputtering conditions of Ar+, 500 eV 45°) which accounts for the intermixing and the model for compound formation20. The initial structure for the simulation is same as before. The results for the interface regions are shown in Fig. 6a,b.Figure 6The measured and simulated XPS depth profiles on pristine sample. (a) Initial structure. Depth profiles (b) the C/W interface, (c) W/C interface; (symbols) measurement, (lines) simulation.Figure 6a shows the initial ideal structure of the pristine sample based on the XTEM image. This structure is the input for the simulation. In the case of the C/W interface (see Fig. 6b) the agreement between the measured and simulated profiles is excellent; it clearly shows the drastically different shapes of C lines originate from the graphite layer and that from the W2C produced by the sputter removal process. In case of the W/C interface (Fig. 6c) the agreement of the depth profiles of the elements are again excellent, but in the case of the sputtering induced W2C there is some deviation; the as measured profile of the W2C is broader the expected (as previously in Fig. 5a). This happens most likely because of the roughening of the interface due to the unfavorable sputtering conditions. Anyhow this result exhibits that both the distortions due to the IMFP and the sputter removal processes are correctly described by our models in most part, and shows how the initial structure of the sample results in the XPS depth profile shown in Fig. 3.This protocol will be applied for the simulation of the XPS depth profiles obtained from samples of unknown thin layer systems.Medium energy irradiated samplesIn this section we will report on the application of our trial and error protocol for “real” problem. The pristine sample has been irradiated by medium energy ions of various kinds and energies to produce tungsten carbide rich nano-layers at room temperature, which can be a used as corrosion protection and/or high wear resistance layers. It is evident that their structure, chemistry is to be revealed before further applications.It has been shown above that even at Ar+ bombardment of 0.5 keV, used for depth profiling, intermixing and compound formation happens (producing the same W2C, which is produced by the medium energy irradiation) which distorts the measured depth profile, and which should be corrected.Artifact production mainly occurs if there are pure C/pure W and/or pure W/pure C interfaces, so it is only disturbing at the beginning of the tungsten carbide production. It is favorable that the artifact production is well visible in case of XPS depth profiling. This is due to the relatively high IMFP of the excited photoelectrons. If the to be depth profiled sample contains tungsten carbide and we measure the C in W2C bound photoelectron (C1s) signal then due to the high IMFP (it is in pure C and W, 3.3 nm 1.7 nm, resp.) in the case of C/W interface a very broad, while in the case of W/C interface a broad peak will appear. (Note that the penetration depth of an electron is roughly 3* IMFP.) On the other hand, if the ion bombardment produces the observed tungsten carbide the corresponding C1s peak, then this peak will have a sharper appearance since the projected range of the 0.5 keV Ar+ ions is in the range of 1.6 ± 0.5 nm. Based on the difference of the shapes one can distinguish between the artifact and the already present tungsten carbides. Anyhow our evaluation protocol account for this problem, which will be demonstrated for the case of the sample irradiated by 0.1E16 Xe+/cm2 of 120 keV.Making the initial guess of the irradiated sample one should remember that the medium energy irradiation produces tungsten carbide. This carbide growth from the interfaces by means of a quasi-diffusional process and thus, their initial structure will be approximated by erfc functions.Sample irradiated by 0.1E16 Xe+/cm2 of 120 keVPrevious studies20 and simulations show that this slight irradiation causes a modest tungsten carbide production and mainly in the region of the W/C interface. The as measured XPS depth profile of sample irradiated by 0.1E16 Xe+/cm2 of 120 keV is shown in Fig. 7.Figure 7The as measured XPS depth profile obtained on sample irradiated by 0.1E16 Xe+/cm2 of 120 keV. The C 1s peak (283.6 eV) which is in carbide form signed as C in W2C is multiplied by 10 for better visibility.The as measured depth profile similarly to the previous one shows two so called intermixed regions and broad transition regions. The carbon region of the XPS spectrum could be decomposed into two peaks as graphitic C (284.8 eV) and C in W2C carbide (283.6 eV) accordingly we have two carbon profiles one of them is the C in W2C signed as C in W2C, while the other is the graphitic C signed as C. Both the ion irradiation and ion bombardment used for depth profiling produce W2C, the latter is an artefact. It is evident that for determining the W2C produced by ion irradiation the artefact W2C is to be removed. This is far not easy procedure. It should be mentioned that C in W2C profile shows two distinct regions; a sharp feature at the C/W interface and a much broader one at the W/C interface. As we have learnt previously the sharp feature is most likely due to the artefact, while the broad transition produced by an already present W2C and the “intermixing” is due to the high IMFP. On the other hand, in the simulation we can easily distinguish between the two carbides; this will be shown in the following. Based on the two distinct shapes observed in the as measured profile the initial guess for the structure of the irradiated sample (which is used in the calculation of the simulated depth profiles) is the following: C/W interface is free of initial carbide, while on the W/C interface there is carbide (produced by the medium energy irradiation; its C content in the figures signed by C car and some pure W/C region. It should be emphasized that the total number of C and W atoms in the test layer system is the same as in the pristine sample except the slight C loss due to the high energy irradiation. This C loss is accounted for by changing the thickness of the first C layer. The best agreement between the simulated and the as measured depth profile was reached by choosing the initial concentration distributions shown in Fig. 8a.Figure 8The comparison of the as measured and simulated in-depth concentration distributions of the sample after irradiation by 0.1E16 Xe+/cm2 of 120 keV. (a) Assumed initial concentration distributions; C, W and C in W2C stands for graphite, metallic W and C in W2C bound (produced medium energy irradiation), respectively, (b) comparison of the simulated (lines) and measured (symbols) depth profiles of C (graphite), W (metal) and C in W2C bound, (c) comparison of the simulated depth profiles of C in variously produced W2C phases with the measured C in W2C Car med s, car art s and car all s stand for simulated depth profiles of carbides made by the medium energy irradiation, induced by the ions used for depth profiling (artefact) and the sum of the two previous, resp.Figure 8b demonstrates the measured (symbols) and simulated (lines) depth profiles, for better visibility the carbide profiles are multiplied by 3. It can be seen that the profiles are in good agreement, therefore one should accept that the assumed initial concentration distributions (Fig. 8a) are correct. Based on the simulation one can also conclude that the broad interfaces and seemingly “intermixed” regions are the consequence of the relatively high IMFP of the photoelectrons. On the other hand, the sharp feature of the C in W2C line at the C/W interface is an artefact (due to the ion bombardment used for depth profiling). This is detailed in Fig. 8c. The figure shows separately the simulated W2C carbides produced by the medium energy ion bombardment (car med s) and that produced by the ion bombardment used for depth profiling (car art s) which is an artefact. Car all s is the sum of the two previous. This figure also proves that the artefact signal is the highest in case of pure interfaces16.Sample irradiated by 0.25E16 Xe+/cm2 of 120 keVFigure 9 shows the proposed initial structure (Fig. 9a) and the comparison of the measured and simulated depth profiles (Fig. 9b) for the sample after 0.25E16 Xe+/cm2 of 120 keV irradiation; for better visibility the low intensity carbide profiles are shown separately (Fig. 9c).Figure 9The same as Fig. 8 but after irradiation of 0.25E16 Xe+/cm2 of 120 keV.Concerning the main components, the conclusion is the same as before. On the other hand, the simulated depth profile of the C in W2C shows nice agreement at the C/W interface while at the W/C interface the agreement is not so good. The deviation is due to the too much higher artefact value. This suggest that the assumed initial carbide distribution covers higher part of the W/C interface than that the assumed one or as in the case of the pristine sample bombardment induced roughening occurs.Sample irradiated by 1E16 Ar+/cm2 of 40 keVFigure 10 depicts the proposed initial structure (Fig. 10a) and the comparison of the measured and simulated depth profiles (Fig. 10b) for the sample after 1E16 Ar+/cm2 of 40 keV irradiation. The carbide profiles with detailed simulation results (Fig. 10c) are also provided separately.Figure 10The same as Fig. 8 but after irradiation by 1E16 Ar+/cm2 of 40 keV.At this irradiation the shape of the C in W2C bound profile is different from the previous ones having a nearly zero value in the middle of the W layer. The simulated profiles nicely agree with the corresponding as measured ones for the main components but in the case of the C in W2C bound profile the agreement is reasonable and poor at the C/W and W/C interfaces, respectively and good at the other region. The deviation at the W/C interface might be explained as previously. At the C/W interface it seems that the undisturbed C/W interface is larger by 10–20% than that assumed by the amount of the carbide produced by the medium energy ions. This problem might be connected with the sharp change of the produced carbide, see Fig. 10a, in the C matrix.In summary, we conclude using our method the initial concentration distributions of the XPS depth profiles samples could be estimated even when the as received XPS depth profiles were seriously distorted because of the high IMFP of the photoelectrons and damages introduced by the sputter removal used for the depth profiling. The results also emphasize that the routine evaluation of these measurement should not be attempted, especially in cases where carbide formation is moderate. Routine evaluation may lead to physically improper conclusions, such as intermixing of the first C layer.The very same system has been studied by AES depth profiling20, where due to the much lower IMFP values of the Auger electrons, the unexpected “intermixing” had not caused any problem. The tungsten carbide production due to the depth profiling appeared similarly; anyhow the total artefact production was much lower in the case of AES depth profiling. On the other hand, the distinction of C Auger peak emitted from WC and W2C because of the more complex Auger process is far not straightforward, thus the identification of the carbide produced by the medium energy ion bombardment is uncertain. Thus, if one need to identify the type of carbide, XPS depth profiling is to be used even if the quality of the depth profiles is poorer than that in the case of AES depth profiling. Summarizing the proposed evaluation method can be used safely for evaluating the XPS depth profiles of complex systems.
