In Situ Observations of Phase Transitions in Metastable Nickel (Carbide)/Carbon Nanocomposites

Nanocomposite thin films comprised of metastable metal carbides in a carbon matrix have a wide variety of applications ranging from hard coatings to magnetics and energy storage and conversion. While their deposition using nonequilibrium techniques is established, the understanding of the dynamic evolution of such metastable nanocomposites under thermal equilibrium conditions at elevated temperatures during processing and during device operation remains limited. Here, we investigate sputter-deposited nanocomposites of metastable nickel carbide (Ni3C) nanocrystals in an amorphous carbon (a-C) matrix during thermal postdeposition processing via complementary in situ X-ray diffractometry, in situ Raman spectroscopy, and in situ X-ray photoelectron spectroscopy. At low annealing temperatures (300 °C) we observe isothermal Ni3C decomposition into face-centered-cubic Ni and amorphous carbon, however, without changes to the initial finely structured nanocomposite morphology. Only for higher temperatures (400–800 °C) Ni-catalyzed isothermal graphitization of the amorphous carbon matrix sets in, which we link to bulk-diffusion-mediated phase separation of the nanocomposite into coarser Ni and graphite grains. Upon natural cooling, only minimal precipitation of additional carbon from the Ni is observed, showing that even for highly carbon saturated systems precipitation upon cooling can be kinetically quenched. Our findings demonstrate that phase transformations of the filler and morphology modifications of the nanocomposite can be decoupled, which is advantageous from a manufacturing perspective. Our in situ study also identifies the high carbon content of the Ni filler crystallites at all stages of processing as the key hallmark feature of such metal–carbon nanocomposites that governs their entire thermal evolution. In a wider context, we also discuss our findings with regard to the much debated potential role of metastable Ni3C as a catalyst phase in graphene and carbon nanotube growth.

Substrates for Nickel(-carbide)/carbon nanocomposite thin films sputter deposition were r-plane cut sapphire crystals for in-situ experiments. Reference films were also deposited onto native oxide covered Si(100) wafer pieces, which were however not further annealed due to the potential for Si diffusion. 2 Following deposition, samples were transported and stored in ambient air.

Ex-situ characterization and ex-situ annealing
Scanning electron microscopy (SEM) was undertaken with a Zeiss Sigma VP at 5 kV using an Everhart-Thornley detector. Film composition was determined to ~70 atom-% C and ~30 atom-% Ni using energy dispersive X-ray spectroscopy (EDX, Zeiss Supra 55 VP at 20 kV with Oxford Instruments EDX detector). Transmission electron microscopy (TEM, 200 kV) employed a JEOL200FX, a FEI Technai TF20 FEGTEM and a Philips CM200, for which cross-sectional samples were prepared by focused ion beam milling. 3 A combination of bright field (BF) TEM and selected area electron diffraction (SAED) data was acquired. SAED data was further analyzed using PASAD software 4 which was also used to extract radially integrated SAED profiles. Ex-situ Raman spectroscopy measurements employed a custom-built setup with a laser excitation wavelength of 532 nm. The employed laser power was checked to not modify the sample during measurements. All point-localized ex-situ measurements were checked at least on three macroscopically separated spots across samples to ensure homogeneity of samples and representativeness of results. Ex-situ and in-situ annealed samples were also crosschecked with ex-situ XRD either in the in-situ XRD setup described below or in a Bruker D8 (CuKα).
Ex-situ annealing of samples was undertaken in a custom-built vacuum chamber (base pressure 10 -6 mbar) using a resistive boron nitride coated heater. Temperature was controlled using a combination of thermocouples and pyrometric measurements. After annealing natural cooling of the sample (~100 °C/min initial cooling rate) was employed. Estimated uncertainty for the quoted temperatures for the ex-situ anneals is ±10 °C. After ex-situ annealing, samples were transported and stored in ambient air.

In-situ measurements
In-situ XRD: In-situ X-ray diffraction (XRD) was measured at the European Synchrotron Research Facility (beamline BM20/ROBL) using a X-ray wavelength of 0.1078 nm in a previously described setup. 5-7 A cold-wall vacuum chamber (base pressure 10 -5 mbar) is mounted onto a high-precision 6-circle goniometer. A resistive heater (Boralectric) was used for global sample heating where built-in and sample-surface-clamped thermocouples were used to control the sample temperature. We employed a grazing incidence X-ray diffraction geometry with an incident angle of 2°, thus largely suppressing strong reflections from the single crystalline sapphire substrates. The estimated information depth is ~200 nm. 8 Following annealing, samples were left to cool naturally (~100 °C/min initial cooling rate). Estimated uncertainty for the quoted temperatures is ±40 °C. We note that all diffractograms show a step in intensity at ~18° which is related to the arrangement of the detector and the X-ray entrance/exit slits into the reaction chamber. For XRD data analysis the following Inorganic Crystal Structure Database (ICSD) entries were used fcc Ni: 646089, Ni3C: 17005, graphite: 53781.

In-situ Raman spectroscopy:
We used a previously described in-situ Raman setup in which the Raman probe laser is concurrently used for sample heating. [9][10][11] Laser-induced annealing was undertaken in a vacuum of ~10 -5 mbar. A laser (continuous wave [cw], 532 nm) is focused on the front side of the sample (front-illumination) to a 1 μm spot size (full-width-at-half-max [fwhm], measured using a knife-edge) with a 50× long-working-distance microscope objective through a viewport (for which it is optically compensated). For in-situ annealing experiments, the laser power was increased stepwise while measuring timeresolved Raman spectra (acquisition time 0.5 s) on a constant spot on the sample. We find that for laser powers up to 15 mW we do not observe any modification of the sample, thus allowing us to non-destructively probe the samples at these low power levels. With increasing laser power we then observe an evolution in the Raman spectra, indicating a rise in temperature. We find that for each step-wise power increase (up to 75 mW) the Raman background signal commonly stabilizes within the first few spectra, indicating that on the small heated spot equilibrium is quickly reached. As the laser both probes and heats the sample, the measured intensity in the raw data is a complex convolution of Raman scattering from temperature dependent phase contributions as well as the incident laser intensity. To eliminate the effect of incident laser intensity, normalization of the Raman spectra was employed, where the measured raw Raman intensity was divided by the applied laser power. Only such normalized in-situ Raman data is plotted herein. Following step-wise annealing, samples were left to cool naturally. For details of temperature estimation from laser-induced heating see below.
The information depth of Raman spectroscopy for such nanocomposite films is typically estimated at ~100 nm. 12 For analysis of Raman data (Figure 4d) we fitted single Lorentzians to the D, G and 2D regions. To estimate the lower bound of the inplane-ordering size of the nanocrystalline graphite we used the ratio ( ) ( ) ⁄ of the intensities of the D and G peak following ref. 13 via the equation: where the proportionality constant ≈ 4.4 nm.

In-situ XPS:
In-situ X-ray photoelectron spectroscopy (XPS) was performed at the ISISS beamline of the FHI/MPG located at the BESSY II synchrotron facility in Berlin, Germany. 14 The spectra were collected in normal emission in vacuum (10 -7 mbar) and with a probe size of ~ 100 µm x 1 mm. The samples were heated from the back using an external IRlaser (cw, 808 nm), where temperature was applied homogeneously to the sample via a SiC spacer. The temperature was controlled via a K-type thermocouple in direct contact with the sample surface. Subsequent to annealing samples were left to cool naturally (~100 °C/min initial cooling rate). Estimated uncertainty for the quoted temperatures is ±40 °C. The Ni2p and C1s scans were acquired at X-ray incident energies of 1350 eV and 725 eV, respectively, thus yielding kinetic energies of 490 eV and 440 eV, respectively. This results in an electron mean free path of λ ~1.2 nm, thus giving an estimated total information depth (3 × λ) of ~3.6 nm, 15 although we note that precise determination of information depths in electronically heterogeneous composites (such as those comprised of Ni and C, as studied here) is challenging. We emphasize however that, compared to Raman spectroscopy and XRD with information depths in the 100-200 nm range, our XPS signal is by relative amounts much more (sub-)surface sensitive. The estimated total information depth of ~3.6 nm also suggests that effects from surface oxidation and adventitious carbon contamination from sample storage in ambient air are minimized. 16 For XPS analysis, the photoelectron binding energy (BE) is referenced to the Fermi edge, and the spectra are normalized to the incident photon flux. Background correction was performed by using a Shirley background. 17 The spectra were fitted following the Levenberg-Marquardt algorithm to minimize the χ 2 . Peak shapes were modelled following Blume et al. 18 The accuracy of the fitted peak positions is ~0.05 eV.

Comment on estimation of temperatures
The most accurate temperature estimate of ±10 °C during vacuum annealing was achieved for the ex-situ anneals via careful cross-calibration using a combination of thermocouples clamped to the sample surface and the substrate holder as well as pyrometric measurements. The in-situ XRD and in-situ XPS measurement have larger estimated uncertainties in temperature of ±40 °C which are due to less controlled thermal contact of sample and thermocouples in the in-situ XRD and in-situ XPS setups.
The in-situ Raman measurements do not provide reliable direct temperature measurements of the local temperature at the 1 μm laser spot. (At higher laser powers >75 mW the Raman spectra develop a contribution from thermal blackbody radiation, 9 which is however insufficient in intensity to reliably estimate local temperatures from it.) Therefore, in order to calibrate estimated temperatures at the laser spot as a function of applied in-situ Raman laser power, we employ cross-calibration against ex-situ annealed samples. To this end, we compare the observed degree of carbon ordering (G peak width, I(D)/I(G), I(2D)/I(G)) from in-situ Raman annealed sample spots after cooling (i.e. after reducing laser power to 5 mW) against ex-situ Raman measurements on controlled ex-situ annealed samples after cooling. Thereby, in-situ Raman measurements were estimated to locally yield ~300 °C at 35 mW and ~700 °C at 85 mW applied laser power, respectively. We note that this temperature estimation for the in-situ Raman measurements is also in good agreement with previous studies which showed a linear increase of the local laser-heating-induced temperature rise with respect to the applied laser power. 11,19 Since this temperature estimation is a very indirect method involving fits to broad Raman spectral features and differences in annealing kinetics, we conservatively estimate a comparably large uncertainty of the quoted insitu Raman temperatures of ±50 °C.
We note that the quoted (thermocouple-derived) in-situ XRD and XPS annealing temperatures were similarly corroborated, comparing Raman spectral features (G peak width, I(D)/I(G), I(2D)/I(G)) from ex-situ Raman measurements of the globally annealed in-situ XRD and XPS samples (i.e. after cooling) against ex-situ Raman measurements of ex-situ annealed reference samples (after cooling).

Comment on Ni3C phase assignment
There is a long standing debate in the literature on the notorious difficulty to reliably distinguish carbon-containing Ni3C (space group: R3 c; ICSD 17005) from the largely isostructural carbon-free hexagonally closed packed (hcp) Ni phase (space group: P6 /mmc). [20][21][22][23] This difficulty arises from the fact that both phases consist of a hcp Ni (sub-)lattice, which in the case of Ni3C comprises an additional ordered sublattice of interstitial carbon. 20 The additional ordered interstitial carbon however only slightly changes the resulting diffraction patterns which is why SAED or XRD routinely exhibit difficulties in identifying either phase with certainty. This picture is further complicated since recent reports 20,21 suggested the formation of carbon-containing hcp Ni (termed hcp NiCy) which differs with respect to Ni3C only in terms of decreased interstitial carbon ordering i.e. hcp NiCy is a disordered form of Ni3C. In contrast, previous work showed that even trace carbon contamination in hcp Ni formation processes will lead to the formation of Ni3C. 20 Low angle diffraction measurements have been recently suggested to allow unambiguous identification of Ni3C based on very low intensity superlattice reflections. [20][21][22] The sensitivity of our time-resolved in-situ XRD measurements under our processing conditions is however not high enough to either confirm or exclude the presence of these superlattice reflections. We however note that in the SAED measurement of the as deposited nanocomposite films (Figure 1b) we find indications of weak diffraction spots which could be assigned as the (104) superlattice reflection of Ni3C. Presence of these reflections unambiguously proves the existence of Ni3C and excludes hcp Ni. [20][21][22] The presence of carbon-free hcp Ni in our as deposited films can be further excluded with confidence based on the following arguments: 1. It has been previously shown that even trace carbon contamination in hcp Ni formation processes will lead to the formation of Ni3C 20 or at least its disordered form hcp NiCy. 21 Given the overall carbon content in our films of ~70 atom-% C, carbon-free formation of hcp Ni is therefore excluded under our conditions. 2. We observe a very low binding energy signature (~283.0 eV) in the carbon C1s signature in XPS for the as deposited films and for anneals up to 300 °C. This low binding energy component has in previous literature 24 been assigned to crystalline Ni3C. Also in our measurements its presence correlates directly with the presence of the suspected Ni3C SAED and XRD patterns. Since this XPS signature is measured on the carbon C1s core level, its presence clearly excludes carbon-free hcp Ni, which due to lack of carbon would not show any corresponding C1s signal. Combined our observations confirm the presence of a Nickel-carbide in our as deposited films. The possible weak (104) diffraction spots in the SAED pattern of the as deposited films give a good indication that this Nickel-carbide is indeed Ni3C.
We note however that our general usage of the terms Nickel-carbide and Ni3C in this manuscript not only includes fully ordered Ni3C but may also include disordered forms of Ni3C approaching the recently suggested 20,21 hcp NiCy form (i.e. hcp Ni sublattice including somewhat disordered interstitial carbon).