Stoichiometry Effects on the Chemical Ordering and Superconducting Properties in TiZrTaNbNx Refractory High Entropy Nitrides

High‐entropy materials, an exciting new class of structural materials involving five or more elements, are emerging as unexplored ground for superconductors. Here, the effects of nitrogen stoichiometry are investigated on local chemical structure of TiZrNbTa‐based thin films by various X‐ray‐based techniques. Lattice distortion and short‐range order of a set of TiZrNbTaNx samples, including bond lengths of different atomic pairs and coordination numbers of substituting atoms are quantitatively studied. The maximum superconducting transition temperature Tc is found at 10 K for a near‐stoichiometric (TiZrNbTa)N1.08 film, which is >8 K measured for a metallic TiZrNbTa film. The underlying electronic structure and chemical bonding in these high entropy nitrides thus influence the superconducting macroscopic properties.


Introduction
Nonstoichiometry occurs widely in ceramics, including transition-metal-based carbides, nitrides, and oxides. [1]For sputter-deposited nitride thin films, the stoichiometry of nitrogen can be easily tuned by controlling the deposition condition DOI: 10.1002/andp.202300470such as reactive gas partial pressure, thus allowing for the tailoring of their physical, chemical, and mechanical properties.Superconducting properties of nitride compounds with a NaCltype structure, such as TiN x , [2] NbN x , [3] VN x , [4] MoN x , [5] ZrN x , [6] and TiNbN, [7] have been widely studied by adjusting the nitrogen stoichiometry.These transitionmetal nitrides have been found in a broad range of device applications, including single phonon detectors, [8,9] accelerator cavities, [10] superconducting resonators, [11] and qubit computers. [12]ecently, refractory high-entropy nitrides (HEN) alloys, as a family member of high-entropy alloys, have attracted increasing attention. [13]Refractory HEN films contain nitrogen and four or more components from earlytransition metals in approximately equimolar proportion.These compositions lead to a combination of mechanical, electrical, and chemical properties that are interesting for a range of potential applications as thin films.Furthermore, there is an increasing interest to explore new high-entropy superconducting thin films, [14][15][16][17] after the first observation of superconductivity in Ti-Zr-Nb-Ta-Hf with a superconducting transition temperature of T c = 7.3 K. [18] However, no superconducting highentropy nitride has been reported.Therefore, the effect of nitrogen stoichiometry on the superconducting properties needs to be explored.
In the present work, a series of TiZrNbTa-based films with different nitrogen contents were synthesized using reactive magnetron co-sputtering.The microstructure and bonding information of TiNbZrTaN x films are determined using X-ray diffraction, X-ray photoelectron spectroscopy, and X-ray absorption spectroscopy.The obtained electronic structure information, complemented by XRD, allows for the determination of the hybridization strength, chemical coordination, and bond lengths around the probed element.A sharp superconducting transition behavior was found in TiNbZrTaN x films for x = 0 and x = 1.08 while understoichiometric films were found to exhibit either semiconductor or insulating properties.The investigation of the correlation between nitrogen stoichiometry and the local electronic-and crystal structure contributes to the understanding of the macroscopic properties and enables to design of high-entropy superconductors.

Results and Discussion
The chemical compositions of (TiZrNbTa)N x films grown on Si(100) were determined by Tof-ERDA and EDS.The nitrogen content increased from 0 to 29.7, 34.0, 36.2, 38.5, 39.2, and 52.0 at.% corresponding to variations in the nitrogen flow ratio rates f N between 0% and 38.5%.Here, the films are referred to by their nitrogen content relative to the metal content x = [N/([Ti]+[Zr]+[Nb]+[Ta])] with the general formula Me 1 N x .In this series, the x value varies from 0 to 1.08, as listed in Table S1 (Supporting Information).Within the metallic film (x = 0), the presence of two distinct bcc phases was observed, likely stemming from the segregation into TiZr-rich and NbTa-rich domains. [19]hen adding nitrogen in the films, the four metal fractions were kept close to the equimolar content, except for the x = 1.08 film with a composition of Ti:17.7%,Zr: 10.9%, Nb: 11.0%, and Ta 8.4%.The slight change in elemental ratio with different nitrogen content may result from the different poisoning behavior during the deposition. [20]igure 1 shows XRD patterns of the (TiZrNbTa)N x films, where x = 0, 0.42, 0.51, 0.56, 0.62, 0.64, and 1.08, deposited on Si(100) and quartz substrates.A metal-to-nitride phase transition is known to occur for refractory metals when accommodating nitrogen atoms in their interstitial sites. [21]The metallic TiZrNbTa film exhibits several broad diffraction peaks, which can be assigned to two different bcc phases Im 3m.For the film with x = 1.08, the diffraction peak at 41.3°can be well attributed to a (200) reflection as a typical refractory-metal nitride phase, that is, the NaCl-type of fcc phase.The films with x between 0 (metallic) and 1.08 (slightly overstochiometric in nitrogen) exhibit a preferredorientation crystalline structure.The main diffraction peak shifts from 38.4°to 41.3°with increasing x.This shift toward higher angles indicates a volume shrinkage while the N atoms are incorporated in the interstitial sites of the rock-salt structure.More-over, when comparing the two structures, the 110 reflection of bcc is parallel to the 200 reflections of fcc.Thus, the preferential orientation is maintained even when the nitrogen content gradually increases in the understoichiometric films.However, it is challenging to attribute this broad peak to either a bcc(110) or a fcc(200) reflection.
Figure 2 shows high-resolution XPS spectra recorded at the Ti 2p, Zr 3d, Nb 3d, Ta 4f, N 1s, and O 1s core-level regions for all (TiNbZrTa)N x films (0 ≤ x ≤ 0.91).As observed, there is a clear shift of all metallic peaks toward positions typical for binary nitrides at higher binding energies when x increases.This is accompanied by an increase in the relative intensity of the N 1s signal.The N 1s binding energy is in the range 397.5-397.7 eV, which is close to the binding energy (E b ) values for stoichiometric binary nitrides, [22] that is, 397.2 eV for TiN, 397.3 eV for ZrN, 397.5 eV for TaN, and 397.6 eV for NbN.The metal core level BEs for the TiNbZrTa metallic film and (TiNbZrTa)N 1.08 near-stoichiometric nitride film well agree with the reference values for the corresponding pure metals, [23] and binary nitrides. [22]A similar chemical shift trend of XPS peaks was reported in a previous article on a similar material system, where the nitrogen content x, changed from 0.25 to 0.59. [24]n the present work, the nitrogen stoichiometry is varied in a wider range from 0 to 1.08, while the nitrogen-containing films remain in a dense and columnar growth structure (Figure S1, Supporting Information) with the use of substrate biasing during deposition.The oxygen contents measured by ToF-ERDA are <0.5 at.% for all the nitrogen-containing samples, whereas the TiNbZrTa metallic film contains ≈1.0 at.% oxygen, which is the observed as oxide peaks in the Zr 3d spectrum at the top of Figure 2b.This may be due to natural post-oxidation that occurs in air (See the details in Figure S3, Supporting Information).The Ta 4f spectrum from the overstoichiometric film shows typical signs of sputter-induced damage in the form of additional low-energy structure assigned to the N-deficient top layer.[22] Figure 3 shows normalized XANES data of (TiZrNbTa)N x films measured at the Ti 1s-edge, Zr 1s-edge, Nb 1s-edge and Ta 2p 3/2 -edge.The XANES from the TiN, ZrN, NbN, TaN binary nitride samples were measured in fluorescence mode while the Ti, Zr, Nb metal foils were collected in transmission mode.For comparison, XANES was calculated using FEFF9 for the model structures for each atom (gray lines) and the average (red lines) spectra are also shown.In general, good agreement with experiment is found.Due to the electric dipole selection rules, excitation of the Ti, Zr, and Nb 1s electrons only occur to orbitals with p-character.On the contrary, the Ta 2p 3/2 electrons only dipole allowed to unoccupied orbitals of d-character.[25] The pre-edge features in the Ti, Zr, and Nb 1s XANES consist of electric dipole excitation of the 1s electron into the p-d hybridized orbitals close to the Fermi edge (E F ).The main peak, positioned near the vertical grey line, corresponds to 1s electron transitions into empty 4p orbitals, while the pre-peak features reflect the empty orbitals that have both p-and d-character that originates from the Me 1s → Me 3d -N 2p bonding.
The XANES data of the TiZrTaNb film at Ti-1s, Zr-1s, and Nb-1s exhibit similar shape features, including the pre-edge as those of the corresponding metal foils, indicating their metallic nature.A shift to high energy can be observed as a function of x for all four spectra indicating higher oxidation states for the probed metal element.The spectra of (TiZrTaNb) 0.62 and (TiZrTaNb) 1.08 contain similar features as the corresponding ones from the binary nitrides, indicating that these two films are both nitrides, while the understoichiometric x = 0.62 sample is in a weaker nitridic state.
The local atomic structure of the (TiZrTaNb)N x (x = 0, 0.62, 1.08) film was further characterized by EXAFS, which is highly sensitive to local elemental electronic environment.
Figure 4 shows radial distribution functions (RDFs) at the Ti-1s, Zr-1s, Nb-1s, and Ta-2p 3/2 absorption edges from the (TiZrTaNb)N x films, where x = 0, 0.62, 1.08 and reference materials, respectively.It also includes RDFs of corresponding binary nitride reference films-TiN, ZrN, NbN, and TaN-all characterized by a NaCl-type fcc structure (Figure S4, Supporting Information).The RDFs were obtained from the EXAFS oscillations using Fourier transformation of the k 2 -weighted structure factors (k) by the standard EXAFS procedure. [26]For the metal foils, there is only one distinct sub-peak corresponding to Ti-Ti (3.047 Å), Zr-Zr (3.415 Å), and Nb-Nb (3.106 Å) pairs.For the TiZrTaNb metallic film, the first coordination shells show much shorter distances than for the pure metals, that is, 2.720, 2.794,    2 a, see Table S2, Supporting Information).This indicates a homogeneous atomic distribution in the metallic film due to the short-range collective rearrangements of these four metals with different atomic radii.In contrast, the pronounced two sub-peaks at shorter distance in the spectra of the nitrogencontaining films are close to the corresponding sub-peaks (Me─N and Me─Me pairs, Me is either Ti, Zr, Nb, or Ta) in the spectra of the binary nitride references signifying that the first shell of the Me-centred clusters includes both Me─N and Me─Me′ bonds.The interatomic distances include the nearest neighbors in the first coordination shell and the next to the nearest neighbors in the second coordination shell, as shown in Figure 5. Comparing two nitride films (x = 0.62 and 1.08), the difference in Ti─N and Ta─N bond distance is more significant, which indicates a local disorder that arises from the vacancies at N or the metal sites around the absorber atom, in particular Ti and Ta.
Figure 6a shows the temperature-dependence of normalized resistivities for the (TiZrNbTa)N x film.In general, the resistivity ratio /(300 K) for the nonstoichiometric films is near the vicinity of unity.Only the -T curve of the TiZrNbTa metallic film shows metallic characteristics (d/dT>0) before the onset of the superconducting transition.Three of these films display a superconducting transition behavior.A maximum T c of 10 K was achieved for the (TiZrNbTa)N 1.08 film with a nearly stoichiometric nitrogen content.The T c for the x = 0 film is ≈8 K, which is equal to the values of the TiZrNbTa film produced by a compound target [15] and the non-equiatomic Ti 15 Zr 15 Nb 35 Ta 35 bulk synthesized by arc melting. [27]It meets the general understanding that T c of early transition metal nitrides are higher than their pure elements.
With the introduction of nitrogen, possible superconductivity is first observed for films of x = 0.64, which exhibits a two-step transition toward to superconducting state, and a zero-resistivity superconducting state can only be expected for temperatures below 2 K.The values of T c for these three films with x = 0, 0.62, 1.08 also deviated in a reasonable range compared to the T c 's for the binary counterparts, that is, TiN: 6.0 K, [2] ZrN: 10.4 K, [28] NbN: 17 K, [3] and TaN 10.8 K. [29] However, T c values of these TiNbZrTaN x films may not directly correlate with a simple linear combination of the T c values of individual binary nitrides due to the intricate nature of the interactions from composition, crystal structure, electron-phonon interactions, and the presence of nitrogen.In TiN x , the superconducting transition temperature T c , reduced with the decrease of x from 1.0 to 0.55. [2]The nitrogen-containing films with x < 0.64, exhibit an insulating behavior or semiconducting behaviors, that is, a negative temperature derivative of the resistivity (d/dT<0), which is seen for the x = 0.56 and 0.62 films.This result gives evidence that metallic to insulating system can be tuned through adjustment of nitrogen stoichiometry in the (TiZrNbTa)N x material system.
To investigate the difference in superconductivity between HEA and HEN films, we also performed systematical magnetotransport measurements for TiZrNbTa and (TiZrNbTa)N 1.08 films with external magnetic fields up to 9 T perpendicular to the film surface.Figure 6b-d shows the field-dependent normal to superconducting transitions for these two films, where both show a sharp drop of resistivity in the vicinity of the transition.Despite a lower T c in TiZrNbTa, the suppression of superconductivity by external magnetic fields is less pronounced than that in the (TiZrNbTa)N 1.08 film.By defining the T c as the temperature where the resistivity drops to half of the normal state resistivity, we summarize the field dependence of the T c in Figure S5 (Supporting Information).By fitting the B c2 (T) dependence with the Werthamer-Helfand-Hohenberg (WHH) theory in the "dirty" limit [30,31] (Figure S5, Supporting Information), the critical fields B WHH c2 (0) for x = 0 and 1.08 are 13.7 and 13.5 T, respectively.The shape of the transition of the x = 0 film remains similar with the ones of the non-equiatomic Ti 15 Zr 15 Nb 35 Ta 35 bulk. [27]

Conclusion
The effects of nitrogen stoichiometry on local chemical structure of TiZrNbTa-based nitride films were investigated by various X-ray-based techniques.For a metallic film x = 0, the structure exhibited that of bcc, whereas nitrogen containing films (x > 0) exhibited a preferred-orientation with a NaCl-type structure.A gradual change of chemistry and structure was observed between the bcc to the fcc phase transition as confirmed by the XPS and XANES spectra.A maximum T c , ≈10 K was obtained for a (TiZrNbTa)N 1.08 film with a nearly stoichiometric nitrogen content while a T c of ≈8 K was observed for x = 0. Understoichiometric films (0< x <0.62) were found to be either insulating or semiconducting.The bond distances of the different atomic pairs including metal-nitrogen, and metal-metal indicate a chemical order in nitrogen-containing films.Understanding how stoichiometry influences the formation of secondary phases or defects is crucial for overcoming potential challenges that might hinder superconducting properties.The tailored stoichiometry of TiZrTaNbN x opens possibilities for various applications, including high-performance superconducting wires, magnets, and electronic devices.

Experimental Section
Materials Synthesis: The nitrogen-containing TiZrNbTa-based films were deposited on Si(100) and amorphous quartz substrates using magnetron sputtering in an ultrahigh vacuum chamber [32] (base pressure < 10 −7 Pa) with four elemental 50.8 mm diameter targets.The distance between the targets and the substrate holder was ≈140 mm.The substrate holder was maintained at 400 °C under a constant rotation of 20 rpm and a substrate bias of −100 V was applied by a pulsed DC power supply during depositions.The discharge power applied to each target was adjusted individually to achieve an equimolar composition (details are given in Table S1, Supporting Information).The total gas flow of the Ar/N 2 mixture was kept at 65 sccm corresponding to a total pressure of 0.47 Pa (3.5 mTorr) where the nitrogen ratio of f N = N 2 /(Ar+N 2 ) were set to 0% and 38.5% for the metallic and nitride film, respectively.To vary the nitrogen content in the TiZrNbTa-based films, f N was adjusted from 0%, 3.85%, 4.00%, 4.15%, 4.30%, 4.50%, and 38.5%, respectively.The Si(100) and quartz substrates with a size of 10 × 10 mm 2 were cleaned sequentially with acetone and ethanol in an ultrasonic bath for 10 min, and finally blow-dried with nitrogen gas.For each deposition, the substrates were preheated for at least 30 min to obtain a homogeneous temperature.The deposition time was 15 min for all the metallic films, and 20 min for all the nitrogen-containing films.
Composition Measurement: The elemental compositions of the nitrogen-containing TiZrNbTa films were determined using Energy-Dispersive X-ray Spectrometry (EDS, Oxford Instruments X-Max) and Time-Of-Flight Elastic Recoil Detection Analysis (ToF-ERDA) measurements.ToF-ERDA measurements were performed using a 36 MeV 127 I 8+ beam at 67.5°incidence relative to the surface normal and a 45°recoil angle. [33]The data were analyzed using the simulation code Potku. [34]The contents of the light elements N and O were determined by ToF-ERDA, while the metallic contents were determined by a combination of EDS and ToF-ERDA, because of the overlap between Zr and Nb that occurs in the ToF-ERDA mapping.Given sufficient statistics, the relative accuracy in obtaining the concentration of light elements by ToF-ERDA is better than 1.0 at.% with high absolute accuracy demonstrated for nitrides. [35]tructural Characterization: The crystal structure was evaluated by Xray diffraction (XRD) −2 measurements with a PANalytical X'Pert Pro diffractometer in a Bragg-Brentano geometry using a Cu K  X-ray source.X-ray Absorption Near-Edge Spectroscopy (XANES) and Extended X-ray Absorption Fine Structure (EXAFS) experiments at room temperature were conducted for selected multicomponent films at Ti 1s-edge (4966 eV), Zr 1s-edge (17 998 eV), Nb 1s-edge (18 986 eV) and Ta 2p 3/2 -edge (9881 eV) at the BALDER beamline on the 3 GeV electron storage ring at MAX IV, Lund, Sweden. [36]A Si 111 crystal pair in the double crystal monochromator with a Pt-coated Si focusing mirror was used.The energy resolutions were set to 1, 4, 4, and 2 eV at the 1s edges of Ti, Zr, Nb, and Ta 2p 3/2 -edges of the beamline monochromator, respectively.A 7-element SDD detector was used for the data collection in fluorescence yield mode.The samples were positioned with incidence and exit angles of 45 degrees and the scans were made with 0.25 eV energy steps.For energy calibration, 6, 16, and 15 μm thick foils of Ti, Zr, Nb, respectively, were measured in transmission mode.The calibration of Ta was conducted using the 2p 3/2 edge from a 3 μm thick W foil.The spot size on the samples was optimized for sufficient spatial resolution to H × V = 50 μm × 60 μm.
The XANES and EXAFS data were processed and analyzed using the Demeter software package as a base for the Athena and Artemis software.First, a linear function was subtracted from the pre-edge region, then the edge jump was normalized using the Athena software. [37]The k 2 -weighted EXAFS oscillations (k) were fitted with least-squares refinement after extracting the Fourier-transformed data in R-space using the EXAFS equation by applying the Artemis software as a front end to FEFF calculations of the scattering paths of the model systems.
Ab-initio density functional theory simulations had been performed to calculate optimized lattice parameters of TiZrTaNbN x .The Quan-tumEspresso simulation package had been used with projector augmented-wave (PAW) method [38] and with the Perdew-Burke-Ernzerhof parametrization of exchange-correlation energy density. [39]he alloys had been modeled by special quasirandom structures (SQS) generated by ATAT [40] at x = 0, 0.5, and 1.To account for the same substitutional configurational freedom, supercells with 64 atoms were used in the metal sublattice in both, the bcc and fcc lattices.
Theoretical XANES spectra were calculated from first principles within density functional theory [41,42] by using the real-space Green's function (RSGF) method as implemented in the FEFF9 program. [26,43,44]Here, the Hedin-Lundqvist exchange-correlation potential was used.The 1s-edge spectra in Ti, Zr, Nb, and the 2p 3/2 -edge spectrum in Ta were computed separately for each atom in the respective supercell structures.From this, individual atoms as well as averaged spectra were obtained in each investigated system.In order to account for final state effects, a core-hole was inserted at the central scattering atom for which the spectrum was obtained.To converge the spectral features, clusters with >940, 640, and 500 atoms respectively, were used for the supercells with N occupation set to 1, 0.5, and 0. At last, the theoretical spectra were aligned with measurements for comparison.For a brief comparison between different methods to calculate XANES, see for example, ref. [45].
XPS Analysis: The X-ray photoelectron spectroscopy (XPS) measurements were performed in an Axis Ultra DLD instrument from Kratos Analytical (UK) employing monochromatic Al K  radiation (h = 1486.6eV).The surface contamination due to the exposure of the sample to air was first removed by sputter-etching the films for 120 sec with a 4 keV Ar + ion beam incident at 70°with respect to the sample normal; the Ar + ion energy was then reduced to 0.5 keV for 600 s to minimize surface damage.The size of the sputter-etch cleaned area was 3 × 3 mm 2 , while the spectra were collected from a 0.3 × 0.7 mm 2 area centered in the middle of the etched crater and with electrons emitted along the surface normal.The binding energy (BE) scale of the spectrometer was calibrated using the ISO-certified procedure [46] to avoid problems related to a use of the C 1s peak of adventitious carbon. [47,48]The analyzer pass energy was set to 20 eV that resulted in the full width at half maximum of 0.55 eV for the Ag 3d 5/2 peak.Elemental quantification was performed using the Casa XPS software (version 2.3.16)based upon the peak areas obtained from narrow energy range scans and the elemental sensitivity factors supplied by Kratos Analytical Ltd.
Resistivity Measurements: Resistivity measurements were performed in a Quantum Design physical property measurement system (PPMS) (9 T) using a standard van der Pauw method. [49]Contacts were made using 25 μm aluminum wire bonded on the grown film using the TPT wire bonder.

Figure 1 .
Figure 1.XRD patterns for the (TiZrTaNb)N x films grown on (a) Si(100) and (b) quartz substrates as a function of the nitrogen composition x.

Figure 3 .
Figure 3. Normalized XANES spectra at the Ti 1s-edge, Zr 1s -edge, Nb 1s -edge and Ta 2p 3/2 -edge for the (TiZrTaNb)N x films, where x = 0, 0.62, 1.08, and Ti, Zr, and Nb transmission foils as well as binary TiN, ZrN, NbN and TaN films on quartz.The thickness of the vertical gray lines indicates the shift of the main peak (see Figure S2, Supporting Information).Theoretical XANES for each atom (gray lines) in the model structures and the average (red lines) spectra are shown.

Figure 4 .
Figure 4. Interatomic distance determined by XRD and EXAFS spectra of Ti, Zr, Nb, Ta atoms in the (TiZrTaNb)N x films, where x = 0, 0.62, 1.08, and the corresponding reference samples: Ti, Zr, and Nb metal foils as well as the binary TiN, ZrN, NbN, and TaN films.

Figure 5 .
Figure 5. Fourier transform profiles of the experimental EXAFS spectra of Ti 1s-edge, Zr 1s-edge, Nb 1s-edge and Ta 2p 3/2 -edge for the (TiZrTaNb)N x films, where x = 0, 0.62, 1.08, and the corresponding reference samples: Ti, Zr, and Nb metal foils as well as binary TiN, ZrN, NbN and TaN films on amorphous quartz substrates.A phase shift of ≈0.5 Å is not included in these plots.The vertical lines indicate the positions of the main peaks.

Figure 6 .
Figure 6.a) Normalized resistivities (T)/(300 K) of the (TiZrNbTa)Nx films as a function of temperature from 2 and 300 K. b-d) The magnetic field dependence of the superconducting transition from 0 to 9 T of the (TiZrTaNb) N x films, where x = 0. 0.62, and 1.08.