Influence of Carbon on Microstructure and Mechanical Properties of Magnetron Sputtered TaW Coatings

(1)

In ﬂuence of carbon on microstructure and mechanical properties of magnetron sputtered TaW coatings

S. Fritze

^a,

⁎ ^{, M. Hans}

^b

, L. Riekehr

^a

, B. Osinger

^a

, E. Lewin

^a

, J.M. Schneider

^b

, U. Jansson

^a

aDepartment of Chemistry-Ångström, Uppsala University, SE-751 21 Uppsala, Sweden

bMaterials Chemistry, RWTH Aachen University, Aachen, Germany

H I G H L I G H T S

• Novel TaW-C thin ﬁlms with bcc structure have been deposited by magnetron sputtering.

• Atom probe tomography revealed that carbon is evenly distributed in the bcc structure.

• The properties of these ﬁlms can be further enhanced by the addition of ~5 at.%

carbon forming a supersaturated solid solution.

• The carbon supersaturated bcc solid solution is as hard as ceramics.

• Nanoscale segregations to the grain boundaries upon annealing hinder grain growth.

G R A P H I C A L A B S T R A C T

a b s t r a c t a r t i c l e i n f o

Article history:

Received 4 May 2020

Received in revised form 6 July 2020 Accepted 17 August 2020 Available online 21 August 2020

Keywords:

Nanocrystalline refractory metals Tantalum tungsten alloy Carbon supersaturation Magnetron sputtering

(Ta,W) and (Ta,W):Cfilms with ~5 at.% C were deposited by non-reactive magnetron sputtering. They crystallised in a bcc structure with a columnar microstructure. The solid solubility of C in (Ta,W) alloys is very low, which suggests that the (Ta,W):Cfilms are supersaturated with respect to carbon. This was confirmed by diffraction and atom probe tomography (APT) showing that carbon is in the as-deposited (Ta,W):Cfilms homogeneously distributed in the structure without carbide formation or carbon segregation. Annealing at 900 °C for 2 h showed no significant column coarsening but an increased defect density at the column boundaries in the (Ta,W):Cfilms. The films were still supersaturated with respect to carbon but APT showed a partial segregation of carbon presumably to defect-rich column boundaries after annealing. The (Ta,W)films exhibited a hardness of ~12–13 GPa. Alloying with carbon increased the hardness to ~17 GPa. The hardness increased to ~19 GPa for the annealed (Ta,W):Cfilms. This annealing-induced hardness increase was explained by C segregation to the more defect-rich column boundaries, which restricts dislocation movements. (Ta,W):C coatings may be a potential alternative to ceramic coatings, worth exploring further by small scale mechanical testing to investigate if these materials are ductile.

creativecommons.org/licenses/by/4.0/).

1. Introduction

Refractory metals and alloys combine attractive properties such as high melting points [1], high density [2], high hardness [3], and high strength at elevated temperatures [4]. An excellent example is the tantalum‑tungsten (Ta-W) system, which has been extensively studied

⁎ Corresponding author.

E-mail address:stefan.fritze@kemi.uu.se(S. Fritze).

https://doi.org/10.1016/j.matdes.2020.109070

Contents lists available atScienceDirect

Materials and Design

j o u r n a l h o m e p a g e :w w w . e l s e v i e r . c o m / l o c a t e / m a t d e s

(2)

by computational and experimental materials science. Ta-W alloys have interesting chemical and physical properties. For example, Xu et al. [5]

have demonstrated that no radiation-induced clustering takes place during ion bombardment and these alloys could therefore be promising candidates for applications in fusion reactors. Furthermore, alloying W with Ta can signiﬁcantly improve the mechanical properties, resulting in higher hardness [5,6] and increasing ductility [7].

Microstructure design is an efﬁcient tool to control the mechanical properties and obtain a combination of high hardness and ductility.

This can be achieved by taking advantage of several hardening mecha- nisms, such as grain refinement, solid solution hardening and precipitation hardening. Recent reports have shown that bcc metals and alloys have much higher Hall-Petch coefficients (kHp) compared to cubic closed packed metals [8,9]. Consequently, grain refinement and formation of a material with nanocrystalline microstructure can be a fruitful pathway to significantly increase the hardness in bcc alloys. This has been demonstrated for magnetron-sputteredfilms of W and Ta. For example, Zhang et al. [9] deposited nanocrystalline Tafilms with a grain size of 77 nm that had an order of magnitude higher hardness (11.6 GPa) than coarse bulk Ta (~1 GPa). Furthermore, sputter- deposited nanocrystalline Ta-Wfilms with a grain size of 36 nm exhibited a hardness of almost 15 GPa [10].

A series of recent papers has demonstrated that grain refinement of bcc metals during magnetron sputtering can be obtained by the addition of a p-element such as C or N in small amounts without the formation of additional phases such as carbides or nitrides [11–13]. The sputtering process occurs far from equilibrium and it is easy to deposit films that are supersaturated with respect to the p-element. For example, Greczynski et al. showed that the addition of 5 at.% N in bcc Cr induced a non-columnar microstructure givingfilms with a combination of ceramic hardness and metallic toughness [11]. Furthermore, we have recently demonstrated that the addition of C to magnetron sputtered TaW-rich high entropyfilms also increases the hardness and improves the crack resistance [14]. The mechanism behind the grain refinement induced by the p-elements can be explained by their low solubility in bcc metals. For example, based on the phase diagram, we expect a low solubility of carbon (<1 at.% at temperatures below 1000 °C) in Ta or W [15,16]. Therefore, during sputtering, the carbon (or other p-elements) is expected to segregate to the surfaces of the growing grains and induce nucleation of new grains, resulting in a reduced grain size compared to a pure metalfilm sputtered under identi- cal conditions. Previous studies on metastable carbon supersaturated solid solutions have focused on the material properties in the as- deposited state [12,13]. The mechanical properties of metastable solid solutions can further be enhanced by thermal treatment leading to an age-hardening effect which is well-studied for ceramicfilms [17]. Dur- ing annealing, bulk diffusion is favoured, and precipitation of additional phases, such as amorphous carbon or crystalline carbides, can occur at grain boundaries and further increase the hardness [18]. Consequently, the addition of carbon to a bcc transition metal can significantly influence the mechanical properties by a combination of effects including grain refinement, solid solution hardening and precipitation hardening.

The aim of this study is to investigate the influence of carbon on the microstructure and mechanical properties of magnetron sputtered Ta-Wfilms using a combination of characterisation techniques such as X-ray diffraction (XRD), transmission electron microscopy (TEM), atom probe tomography (APT) and nanoindentation. We have depos- itedfilms without and with ~5 at.% C, which is above the solubility limit and representative for a supersaturated solid solution. A special focus of this study is on the age-hardening potential of the supersatu- ratedfilms and therefore, the most detailed analysis is carried out on the annealed samples. An annealing temperature of 900 °C was selected based on the published binary phase diagrams for W-C [19] and Ta-C [20]. As it is reasonable to assume that significantly higher temperatures (above 1250 °C) are required to form W2C precipitates we probe here if amorphous carbon or Ta2C precipitates are formed.

2. Experimental

An ultra-high vacuum (UHV) DC-magnetron sputtering system (base pressure < 10⁻⁸Pa) was used to synthesise thefilms. All films were deposited by non-reactive DC-magnetron sputtering employing a seg- mented 2-in. Ta/W (1:1) and a 2-in. C target for the carbon-containing thinfilms. All targets had a claimed purity of 99.9%. The current on the Ta/W target was 180 mA for one set of samples. For the second set of samples, the current on the Ta/W target was 165 mA, while the current on the C target was kept at 40 mA. Theα-Al2O3(00l) (10x10mm²) substrates were heated to 300 °C for 60 min prior to the deposition and a DC bias voltage of−100 V was applied to the substrate table during deposition.

An Ar⁺plasma was ignited at 0.6 Pa and the Ar gasﬂow rate was 42 sccm. The deposition rate was approximately ~200 nm/h in all experiments. An UHV furnace was used for the annealing study. The samples were placed in anα-Al2O₃crucible and the furnace was three times ﬂushed with Ar and evacuated to a pressure below 2·10⁻⁶Pa. The specimens were heated with 5 K/min to 900 °C, held at 900 °C for 120 min and then cooled down again in the furnace with a rate of ~10 K/min.

Symmetrical XRD scans were performed using a Bruker D8 diffractometer with Cu Kα radiation (θ/2θ scans). The diffractometer was equipped with a Johansson monochromator (Cu Kα1) on the primary side and a Lynx-eye detector on the secondary side. A Siemens D5000 diffractometer with parallel beam geometry was used for the grazing incidence (GI) XRD measurement with an incidence angle of 2°.

A Zeiss Merlin SEM operated at 5 kV with an in-lens detector was used to study theﬁlm surface morphology.

Electron transparent cross-section transmission electron microscopy (TEM) samples were prepared from as-depositedﬁlms using a FEI Strata DB235 focused ion beam/scanning electron microscope (FIB/

SEM). A probe corrected FEI Titan Themis instrument operated at 200 kV in scanning transmission electron microscopy (STEM) mode was used for STEM investigations. The Ta/W ratio in thefilms was measured by energy dispersive spectroscopy (EDS) in the TEM, and elemental maps were recorded with a SuperX EDS system. The EDS quantification was done using Cliff-Lorimer method and standard k-factors (theoreti- cally derived) as implemented in the Esprit 1.9 software by Bruker. For μProbe STEM, the electron beam in the TEM was set up to a nearly parallel beam condition and spot size of approximately 2 nm, after which a STEM image was recorded using the high-angle annular dark-field (HAADF) detector. The beam was then positioned at different areas of interest in the μProbe STEM image to acquire nano-beam diffraction (NBD) patterns with high lateral resolution. Additionally, selected area electron diffraction (SAED) patterns of thefilms were recorded, using a 40 μm SAED ap- erture covering a sample area with a radius of circa 570 nm, and therefore most of thefilm from substrate to surface. CrysTBox [21–23] and JEMS [24] were used to evaluate the diffraction patterns.

Time-of-Flight Energy Elastic Recoil Detection Analysis (ToF-E- ERDA) measurements, using 36 MeV¹²⁷I⁸⁺ions as primary projectiles, were carried out at Tandem Laboratory (Uppsala University) to quantify the carbon content in theﬁlms [25].

The local chemical composition at the nanometre scale was investigated spatially-resolved by atom probe tomography (APT) utilising a CAMECA LEAP 4000X HR. Field evaporation was controlled by thermal pulsing using an ultraviolet laser with 40 pJ laser pulse energy and 125 kHz laser pulse frequency. The base temperature was 60 K and the average detection rate was set to 5 ions per 1000 laser pulses. Data analysis was carried out with the IVAS 3.8.0 software. Chemical compositions were quantiﬁed by full areal integration of each peak and background subtraction. Needle-shape specimens with radii <20 nm at the tip apex were prepared by FIB techniques in a FEI Helios Nanolab 660 dual-beam microscope according to a standard protocol [26].

The mechanical properties (i.e. hardness and reduced elastic modulus) were determined with a CSM Instruments Ultra Nano Hardness Tester (UNHT) equipped with a Berkovich diamond tip. Load- displacement curves were recorded on 20 different spots with 70 nm

(3)

displacement and loading and unloading rates of 5 mN/min. The hardness of selectedﬁlms was measured at several indentation depths to ex- clude substrate and surface effects, and the variation of indentation depths between 40 and 70 nm does not affect the hardness data significantly. The hardness (H) and reduced elastic modulus (Er), were evalu- ated from at least 10 load-displacement curves as described by Oliver and Pharr [27].

3. Results

A set of (Ta,W)ﬁlms without and with ~5 at.% carbon was deposited. The composition of both, as-deposited and annealedﬁlms, was determined by APT and by combining EDS and ERDA results (C content from ERDA, Ta/W ratio from TEM EDS). It is well known that the C concentration of carbides is often underestimated by laser-assisted APT due to the detector dead time [28]. However, the comparison of ERDA and APT composition data reveals good agreement of average C contents which are 6 and 5 (as-deposited) as well as 7 and 6 at.%

(after annealing at 900 °C), respectively (Table 1). The mass spectra of the as-deposited and annealed (Ta,W):Cﬁlms are provided in the supplementary information SI-Fig. 1and the details of peak assign- ment can be found in SI-Table 1. Besides Ta, W and C, impurities of O, Ar, Ga and Re (<1 at.%) can be identiﬁed. O and Ar were incorpo- rated during the growth process and Ga was used for specimen fab- rication. The Re contribution originates most likely from the sputtering target. The maximum deviation in composition measured by different techniques is 2 at.% and the results are summarised in Table 1.

Fig. 1shows the normalisedθ/2θ and GI diffraction patterns of the as- deposited and the annealedfilms. All peaks of the as-deposited (Ta,W) film (Fig. 1a) match a bcc (A2) structure. The bcc structure is retained for the (Ta,W):Cfilm with ~5 at.% carbon (Fig. 1b) and no carbide peaks can be observed.Fig. 1(c) and (d) show diffractograms from the (Ta,W) and the (Ta,W):Cfilms after annealing at 900 °C. Both films exhibit a bcc structure and no secondary phases such as carbides are observed after annealing. Theθ/2θ scans inFig. 1show only one strong (110) peak indicating that allfilms have a very strong <110> preferred orientation. The presence of several peaks in the GI scans suggests that the preferred orientation is not perfect and that also other crystallite directions exist in the sample. The relatively strong signal from the (211) peak, however, most likely originates from planes in the textured grains as the incidence, planar, and diffraction angles can match up for the carbon- containing samples due to less aligned columns (which can be seen in Fig. 3). The relative intensities of the peaks in the GI scans are unaffected by annealing. To ensure that only data from the bulk of thefilms are used, the unit cell parameters were determined from the position of the (110) peak in theθ/2θ scans. The unit cell parameters of the as-deposited (Ta, W) and (Ta,W):Cfilms were determined to a = 3.24 Å and a = 3.28 Å, respectively. The expanded unit cell of the (Ta,W):Cfilm indicates C incor- poration at the interstitial sites of the bcc lattice. After annealing the unit cell parameters of the (Ta,W)900 and the (Ta,W):C900films were determined to a = 3.24 Å and a = 3.27 Å, respectively.

Fig. 2shows an SEM top-view image of the (Ta,W):C900film. The observed microstructure is representative of all fourfilms. The top surface is full of ridges with an orientation relationship, indicating an in- plane crystallographic ordering. Although no strict relationships can be observed, most of the ridges grow with about 110–130° orientation relative to each other which is indicated inFig. 2(b)– (d). The ridge thickness is around 12 ± 5 nm and the length is 50 ± 10 nm, both for the (Ta,W) and the (Ta,W):Cfilms.

The STEM bright-field cross-sections of the (Ta,W) and the (Ta,W):C prior and post annealing are presented inFig. 3. Thefilm thickness varies from ~520 nm for the (Ta,W)films to ~600 nm for the (Ta,W):C films. All films exhibit a columnar-like structure extending from the substrate to the top surface. The combination of SEM and TEM suggest a slab-like morphology of the columnar grains. TEM cross-section and XRD analysis, narrow full width at half maximum of the (110) peaks in theθ/2θ scans, shows that the majority columns consists of one or very few grains.Fig. 3also shows wave-like contrast variations within the columns, indicating that thefilms have a high dislocation density.

The column width, measured from the TEM cross-sections, of all four films is ~20 nm (seeTable 2) which is in agreement with the SEM top-view. No indications of column refinement or additional phases can be seen in the carbon-containingfilms. No significant coarsening of the columns can be observed when comparing the as-deposited with the annealedfilms.

Nanoindentation showed that thefilms in general have a high hardness. The (Ta,W)film exhibited a hardness of 12.1 ± 0.3 GPa with a slight increase to 13.1 ± 0.5 GPa after annealing. The as-deposited (Ta,W):C film was significantly harder (16.8 ± 0.3 GPa) and a further increase in hardness to 19.1 ± 0.6 GPa was observed after annealing. Allfilms exhibited a reduced elastic modulus of about 292 ± 10 GPa.

Due to enhanced mechanical performance, the annealed samples were analysed in more detail.Fig. 4(a) and (b) show SAED and NBD patterns from the (Ta,W)900 and the (Ta,W):C900ﬁlms, respectively.

All reflection spots in the SAED patterns of both films match the bcc structure in agreement with the XRD results inFig. 1. A more detailed analysis of the NBD patterns shows differences in column orientations between the twofilms. For the (Ta,W)900 film, the NBD patterns only show the presence of the <110> out-of-plane orientation, which is in agreement with theθ/2θ scan inFig. 1(c). The NBD patterns reveal [111] and [001] as distinct in-plane orientations. The angles between these directions are 109.5° or 126.3°, which is in the range of the observed angles in the SEM-top-view. In contrast, the NBD patterns of the (Ta,W):C900film show that three out-of-plane orientations ((1-10), (3-2-1) and (3-10)) are present, whereof the strongest agrees with the preferred <110> out-of-plane orientation observed in theθ/2θ XRD scan inFig. 1(d). The (3-2-1) and (3-10) orientations are so rare that they were only detectable by NBD and are not observed in the SAED.

The SAED pattern of the (Ta,W):C900 shows [110] and [113] as the strongest in-plane orientations. Additionally, the [111], [001], [112], [230] and [137] in-plane orientations were determined.

Fig. 5(a) shows a HAADF STEM micrograph of a column boundary between the columnar grains (GB) in the (Ta,W)900film. In the micrograph at higher magnification inFig. 5(b), the zone axis (ZA) of the left column (coloured in red) is identified as [001] and the ZA of the right column (coloured in blue) is identified as the [111], which correlates with the zone axes found with NBD. The GB is rather defect-free with a common (110) lattice plane and the orientation changes smoothly from one crystal orientation to the other. EDS maps and an extracted line scan over one GB are presented inFig. 5(d) and (e). The maps show that W is segregated to the GB and the line scan reveals that the W content increases by ~2 at.%, while Ta is depleted by ~2 at.%. EDS analysis of in totalfive GBs show either enrichment in W or no elemental segregation. It should be mentioned that such chemical enrichments are not only present at the GB but also within the columns.

The HAADF STEM analysis of the (Ta,W):C900ﬁlm is shown inFig. 6.

A detailed analysis ofFig. 6(a) reveals that the (Ta,W):C900 exhibits a Table 1

Chemical composition (from APT, and combined EDS and ERDA measurements) of as-deposited and annealedﬁlms, the latter labelled with the sufﬁx 900 denoting the annealing temperature.

Sample Composition [at.%] APT Composition [at.%] EDS and ERDA

Ta W C Ta W C

(Ta,W) 55 45 0 55 45 0

(Ta,W):C 52 43 5 53 41 6

(Ta,W)900 N/Aâ N/Aâ N/Aâ 55 45 0

(Ta,W):C900 50 44 6 50 43 7

aNot analysed.

(4)

more complex nanostructure than the (Ta,W)900ﬁlm, with both coherent GBs and GBs with a signiﬁcantly higher defect density. The images depicted inFig. 6(b)– (f) are extracted from the STEM image in Fig. 6(a). A coherent GB between the [110] orientation (coloured in pur- ple) and the [113] orientation (coloured in green) is depicted in Fig. 6(c). Again, both ZA share a common (110) lattice plane and

orientation changes smoothly.Fig. 6(d)– (f) shows the excerpts taken from (a) after Fourierﬁltering for the (110) lattice plane to better visualise crystal defects.Fig. 6(d) shows an anti-phase boundary. Dislo- cations are visualised inFig. 6(e) and (f).

The local chemical composition at the nanometre scale was investigated with APT. Three dimensional reconstructions from as-deposited Fig. 1.θ/2θ and GI scans of the (Ta,W) and the (Ta,W):C ﬁlms prior and post annealing. The green triangles mark the peak positions of the bcc (A2) phase.

Fig. 2. SEM top-view of the (Ta,W):C900ﬁlm.

(5)

(Ta,W) are presented inFig. 7(a) and the elements appear to be homogeneously distributed. However, a concentration proﬁle obtained perpen- dicularly to the growth direction clearly shows chemical composition modulations with Ta concentrations varying between 50 and 59 at.%.

Three dimensional reconstructions from the as-deposited and annealed (Ta,W):Cﬁlms are presented inFig. 8(a) and (b), respectively.

Ta, W and C appear to be homogeneously distributed in the as-deposited state. There are no indications of carbides or carbon clustering. Conse- quently, we can conclude that the (Ta,W) with 5 at.% C forms a solid solution in the bcc structureFig. 8(a). In contrast, after annealing at 900 °C

regions with C concentrations≥12 at.% are revealed by isoconcentration surfaces inFig. 8(b). The temperature-induced formation of C-rich regions is further emphasised by the frequency distribution analysis inFig. 8 (c) since the measured C distribution (red diamond markers) clearly devi- ates from the random, binomial distribution (red solid line). Fig. 8 (d) shows a proximity histogram of the C-rich regions indicated inFig. 8 (b) and it is evident that C increases locally from approximately 5 up to 18 at.%. The Pearson correlation coefﬁcient (μ) represents a measure of randomness [29] and values for the constitutional elements from all APT experiments are summarised inTable 3.

4. Discussion

The results show that single-phase (Ta,W)ﬁlms (55 at.% Ta) with a nanocrystalline columnar microstructure can be synthesised with magnetron sputtering. XRD reveals that they crystallise in bcc structure with a lattice parameter of 3.24 Å. This is in good agreement with an estimated cell parameter of 3.235 Å using Vegard's rule [30]. The experimental Ta-W phase diagram shows complete miscibility in form of a solid solution with bcc structure [31]. However, Turchi et al. have, based on ab initio and CALPHAD calculations, predicted the existence Fig. 3. STEM brightﬁeld cross-section overview of (a) (Ta,W), (b) (Ta,W):C, (c) (Ta,W)900, and (d) (Ta,W):C900. The top regions of the FIB lamellas are thinner than the bottom regions resulting in a contrast variation.

Table 2

Column width [d] and column height [h] estimated by cross-section TEM and ridge thickness [t] and ridge length [l] estimated by top-view SEM.

Sample Cross-section TEM Top-view SEM

d [nm] h [nm] t [nm] l [nm]

(Ta,W) 21 ± 6 ~520 12 ± 5 50 ± 10

(Ta,W):C 16 ± 4 ~600 12 ± 5 50 ± 10

(Ta,W)900 19 ± 4 ~520 12 ± 5 50 ± 10

(Ta,W):C900 17 ± 3 ~600 12 ± 5 50 ± 10

(6)

Fig. 4. (a) SAED and NBD of the (Ta,W)900ﬁlm and (b) SAED and NBD of the (Ta,W):C900 ﬁlm.

(7)

Fig. 5. (a) HAADF STEM of the (Ta,W)900ﬁlm, (b) zoom in on the GB, (c) HAADF STEM image of EDS area, (d) EDS maps, and (e) EDS line scan. The arrow in (a) indicates the growth direction, which is valid for all parts of theﬁgure.

Fig. 6. (a) HAADF STEM of the (Ta,W):C900ﬁlm, (b) Fourier transform (c) coherent GB, (d) anti-phase boundary, (d) & (f) dislocations. The arrow in indicates the growth direction, which is valid for all parts of theﬁgure.

(8)

of an ordered B2 (CsCl type) phase for a wide composition range (~20–80 Ta at.%) at low temperatures [32]. The predicted B2 to bcc (A2) phase transition temperature is ~730 °C [33]. In contrast,ﬁrst-principles calculations using ab initio and cluster expansion methods by Muzyk et al. [7]

and Blum and Zunger [34] with a large number of bcc-like structures pre- dict other ordered phases. They conclude that more complex structures with >50 at.% W are more stable configurations as they exhibit more neg- ative enthalpies of formation. DFT calculations also show that the mechanical properties of a Ta-W alloy should be influenced by the formation of such ordered structures [7]. Ordered structures give rise to superstructure reflections in the diffractograms. However, Ta and W have very similar X-ray scattering factors which are expected to result in weak ordering peaks, which may be challenging to detect with lab- scale XRD equipment. Complementary studies with neutron scattering, where the scattering factors differ sufficiently, are therefore needed to give more details on the presence of ordered phases [35]. An interesting observation is that the APT results inFig. 7(a) from the as-deposited (Ta,W)film show a modulation in the composition with regions slightly

enriched in W and others enriched in Ta. This chemical modulation suggests that an elemental segregation and chemical ordering occur in Ta- W alloys.

Upon annealing to 900 °C, the (Ta,W)film retains a bcc structure without significant column coarsening. An interesting observation is that the GBs between the columnar grains are more or less coherent without many defects (Fig. 5). However, the EDS analysis suggests that some GBs are enriched in W. Murdoch et al. have calculated the GB segregation enthalpy for nanocrystalline (Ta,W) alloys and found a tendency for Ta segregation to the GBs [36]. This is in contradiction to our experimentalfindings and suggests that a solely enthalpy-based approach used in ref. [36] cannot explain the GB segregation. The TEM EDS mapping data inFig. 5(d) of the (Ta,W) 900film also show a modulation in the composition with some areas slightly enriched in W and others enriched in Ta. A possible explana- tion for the long-range variations in the composition may be found in the fact that thefilms are slightly enriched in Ta (55 at.%), while first principles calculations suggest that the most stable composition is at about 40 at.% Ta.

We propose that this may be a contributing factor for the segregation of W to the GBs and the formation of what seems to be a long-range composi- tional variation in more Ta- and more W-rich areas in the as-deposited and the annealedﬁlm. Further studies are needed to clarify the chemical ordering effects in (Ta,W) alloyﬁlms but this is outside the scope of the present study.

The hardness of the as-deposited (Ta,W)ﬁlm is 12.1 ± 0.3 GPa and around three times higher than for coarse-grained W bulk samples [3].

Hardness values of 10.9 GPa [37] and 14.1 GPa [38] have been previously reported for strongly <110> textured Ta and W thinfilms. These high hardness values are explained by the nanocrystalline structure and the fact that bcc metals have high Hall-Petch coefficients (kHPfor pure Ta is 760 MPaμm^1/2and for W 1000 MPaμm^1/2[39]) leading to a high hardening potential with reduced grain size [8]. Furthermore, Wang et al. have shown that also in binary refractory thinfilms the grain size hardening effect is significantly larger than the solid solution hardening effect [10].

Consequently, the high hardness of the as-deposited (Ta,W)film can be attributed to the nanocrystalline microstructure. The hardness of the (Ta,W)film increases slightly to 13.1 GPa after annealing to 900 °C. The high hardness of the (Ta,W)900film is attributed to the absence of column coarsening upon annealing. As can be seen inFig. 5, the GBs between the columns are coherent, which may suggest that a grain size effect should be small in thesefilms. However, the SEM top-view image (Fig. 2) shows that the columns in the (Ta,W)films exhibit specific in- plane orientations with approximately ~120° angles between each other which is similar to the microstructure in a TaW-rich high entropy alloy [14]. It is likely that the slip systems are not perfectly aligned over the GBs and therefore slip over the GBs is difficult.

A major aim of this study was to investigate the effect of C on the microstructure and mechanical properties of the (Ta,W)films. Carbon has maximum solubility of 0.3 at.% in W [15] and 2.7 at.% in Ta [16] close to the melting point and therefore a carbon content of ~5 at.% is far above the solubility limit at room temperature. However, the XRD and APT data show a solid solution of C without the formation of carbides. The addition of ~5 at.% carbon into the bcc structure should lead to a lattice expansion of ~0.5% [40] which is in fair agreement with the lattice dilatation of the as-depositedfilm observed with XRD inFig. 1. The APT results confirmed that carbon is evenly distributed in the bcc structure and no C-rich regions are observed in the as-depositedfilm. Furthermore, no segregation of Ta and W can be observed (Pearson coefficients close to zero) in the carbon-containingfilm. An unexpected observation is that addition of carbon has no observable effect on column width or introduction of an equiaxed microstructure. This is in contrast to the results by Greczynski et al. [11], where the addition of N to bcc Crfilms led to the formation of equiaxed grains and Shinde et al., where the addition of carbon to a CrNbTaTiW alloy led to a complete amorphisation [41].

DFT calculations on bcc metals with carbon suggest that carbon in solid solution is located in the octahedral sites of the bcc structure [42,43]. The sizes of these sites are small and should lead to a Fig. 7. APT data from the as-deposited (Ta,W) coating, showing the local chemical

composition at the nanometre scale. (a) Reconstructions of Ta and W atomic positions for the as-deposited thinﬁlm. The cylinder in (a) indicates the region for the chemical composition proﬁle which is shown in (b).

(9)

thermodynamically unfavourable material, with a significant driving force to push carbon out of the solid solution during annealing. TEM of the (Ta,W):C900film shows a much more complex GB structure. The NBD patterns inFig. 4exhibit more in-plane orientations between the columns for the carbon-containingfilms compared to the (Ta,W) films. Fur- thermore, many defects (anti-phase boundaries and dislocations) are observed along the GBs in the high-resolution images inFig. 6. The APT analysis shows that Ta and W are evenly distributed in the thinfilm but that a significant C segregation has occurred with clustering of carbon atoms, most likely along the defect-rich GBs. The driving force for this

clustering at the GBs is the energetically more favourable position of carbon at a defect site compared to an octahedral position in a perfect bcc structure.

The addition of carbon has a significant influence on the hardness. The as-deposited (Ta,W):Cfilm has a hardness of 16.8 ± 0.3 GPa which is about 4.7 GPa harder than for (Ta,W)film. Several factors can contribute to this increase in hardness. Firstly, the GBs in the carbon-containingfilms are more complex with a larger number of defects. There are also more in- plane orientations that can contribute to GBs which are less favourable for dislocation movements. Furthermore, magnetron sputtered thinfilms usually contain a high number of dislocations (10⁹/cm²) [44]. Ma et al.

[45] have recently shown that such a high dislocation density accounts for only ~0.1 GPa of the total hardness of magnetron sputtered Wfilms but their presence can lead to a solid solution hardening effect due to formation of so called Cottrell atmospheres where the interstitial carbon atoms are preferentially bonded to dislocations. Finally, the very high Hall-Petch coefficients in bcc metals can lead to a significant increase in hardness also for a very small change in average grain size. The precise estimation of the Hall-Petch coefficients for the presented Ta–W–C films is for two reasons not possible. (I) The grain shape and therefore the grain size of the presentedfilms is highly anisotropic (ridge length/thickness ratio ~4) making it difficult to estimate a grain size which can be used Fig. 8. APT data from the (Ta,W):Cfilms, showing the local chemical composition at the nanometre scale. a) Reconstruction of Ta, W and C atomic positions for the as-deposited thin film and b) after annealing at 900 °C. C-rich regions are indicated by isoconcentration surfaces with≥12 at.% (red regions). c) Frequency distribution analysis for the 900 °C annealed thin film and the measured distribution of Ta, W and C (scatter plot) is compared to a random, binomial distribution (line plot). Pearson correlation coefficients are represented by μ values.

d) Proximity histogram for the 900 °C annealed thinﬁlm of regions with ≥12 at.% C as indicated by the isoconcentration surfaces in b).

Table 3

Pearson correlation coefﬁcients (μ) for (Ta,W) and (Ta,W):C thin ﬁlms in the as-deposited state as well as after annealing at 900 °C.

(Ta,W) (Ta,W):C

As dep. 900 °C^a As dep. 900 °C

μTa 0.06 N/A 0.05 0.05

μW 0.06 N/A 0.03 0.06

μC – – 0.04 0.39

aNot analysed.

(10)

to calculate the kHP. (II) The carbon containingfilms cannot be synthesised as bulk material (single crystal), since they are supersaturated solid solutions, which would be necessary to determine H₀in the Hall- Petch relationship. However, the high hardness values of the presented films suggest that the kHPdo not differ significantly from the values for pure Ta and W as described in ref. [39]. Although we cannot directly observe a column width reduction in the carbon-containingfilms, it cannot be excluded that a small decrease in column width (within the error mar- gins) contributes to the hardness increase. After annealing, we observe a small, but significant, additional hardening of the film to 19.1 ± 0.6 GPa.

We attribute this to the formation of C-rich clusters in the GBs, indicated by APT inFig. 8. A similar behaviour has been observed in ferritic steels with C as interstitial [46] where the yield strength of the steel was strongly affected by the presence of interstitial carbon. The Hall-Petch co- efﬁcient was found to increase from 150 MPaμm^½in low carbon steel to 600 MPaμm^½in steels with more than 0.005 wt% C with the same grain size. A plausible mechanism for this is that carbon affects the dislocation emissions at the GBs [46].

5. Conclusions

In this work, we have investigated (Ta,W) thinfilms without and with ~5 at.% carbon in the as-deposited state and upon annealing to 900 °C. The XRD and TEM results show that the (Ta,W)film crystallises in a single-phase bcc structure and the high hardness of 12 ± 0.3 GPa for the (Ta,W)film is a result of the nanocrystalline microstructure. The bcc structure of thefilm is retained upon annealing to 900 °C and no softening upon annealing is observed. The (Ta,W):Cfilm also exhibits a single- phase bcc structure. Additionally, the APT data show that carbon is evenly distributed in the structure which confirms the formation of a supersaturated solid solution. The (Ta,W):Cfilm is with 16.8 ± 0.3 GPa more than 4 GPa harder than the (Ta,W)film which is explained by the more complex microstructure. The (Ta,W):C900film is 19.1 ± 0.6 GPa hard and the increased hardness upon heat treatment may be caused by the segregation of carbon to the GBs which gives an additional age- hardening effect. In conclusion, a carbon supersaturated solid solution can be a potential alternative material to ceramic coatings, worth exploring further by small scale mechanical testing.

Data availability

The data that support theﬁndings of this study are available from the corresponding author upon reasonable request.

Authors contribution

Stefan Fritze, Erik Lewin, and Ulf Jansson were responsible for the concept and experimental design. Stefan Fritze and Barbara Osinger deposited the samples and performed all experiments except of APT and TEM. Lars Riekehr performed the TEM analysis. Marcus Hans performed the APT analysis. Jochen M. Schneider, Erik Lewin, and Ulf Jansson su- pervised the project. All authors interpreted the data and wrote the manuscript.

Declaration of Competing Interest

The authors declare that they have no known competingﬁnancial interests or personal relationships that could have appeared to inﬂu- ence the work reported in this paper.

Acknowledgments

The authors acknowledge the funding of The Swedish Research Coun- cil for funding under Grant Nos. 2018-04834. The operation of the tandem accelerator at Uppsala University is supported by the Swedish Research

Council, VR-RFI (contracts #821-2012-5144 & #2017-00646_9), and the Swedish Foundation for Strategic Research (SSF, contract RIF14-0053).

References

[1] G.H. Aylward, T.J.V. Findlay, SI Chemical Data, 6th ed. John Wiley & Sons, 2013.

[2] O.N. Senkov, G.B. Wilks, D.B. Miracle, C.P. Chuang, P.K. Liaw, Refractory high-entropy alloys, Intermetallics 18 (2010) 1758–1765,https://doi.org/10.1016/j.intermet.

2010.05.014.

[3] A. Leitner, V. Maier-Kiener, D. Kiener, Extraction ofﬂow behavior and Hall-Petch parameters using a nanoindentation multiple sharp tip approach, Adv. Eng. Mater. 19 (2017) 1–9,https://doi.org/10.1002/adem.201600669.

[4] O.N. Senkov, G.B. Wilks, J.M. Scott, D.B. Miracle, Mechanical properties of Nb25Mo2525Ta25W25 and V20Nb20Mo20Ta20W20, Intermetallics 19 (2011) 698–706,https://doi.org/10.1016/j.intermet.2011.01.004.

[5] A. Xu, D.E.J. Armstrong, C. Beck, M.P. Moody, G.D.W. Smith, P.A.J. Bagot, S.G. Roberts, Ion-irradiation induced clustering in W-Re-Ta, W-Re and W-Ta alloys: an atom probe tomography and nanoindentation study, Acta Mater. 124 (2017) 71–78, https://doi.org/10.1016/j.actamat.2016.10.050.

[6] Y.J. Hu, M.R. Fellinger, B.G. Bulter, Y. Wang, K.A. Darling, L.J. Kecskes, D.R. Trinkle, Z.K.

Liu, Solute-induced solid-solution softening and hardening in bcc tungsten, Acta Mater. 141 (2017) 304–316,https://doi.org/10.1016/j.actamat.2017.09.019.

[7] M. Muzyk, D. Nguyen-Manh, K.J. Kurzydłowski, N.L. Baluc, S.L. Dudarev, Phase stabil- ity, point defects, and elastic properties of W-V and W-Ta alloys, Phys. Rev. B Condens. Matter Mater. Phys. 84 (2011) 1–17,https://doi.org/10.1103/PhysRevB.

84.104115.

[8] D. Wu, J. Zhang, J.C. Huang, H. Bei, T.G. Nieh, Grain-boundary strengthening in nanocrystalline chromium and the Hall-Petch coefﬁcient of body-centered cubic metals, Scr. Mater. 68 (2013) 118–121,https://doi.org/10.1016/j.scriptamat.2012.09.025.

[9] M. Zhang, B. Yang, J. Chu, T.G. Nieh, Hardness enhancement in nanocrystalline tantalum thinﬁlms, Scr. Mater. 54 (2006) 1227–1230,https://doi.org/10.1016/j.

scriptamat.2005.12.027.

[10] C.L. Wang, M. Zhang, J.P. Chu, T.G. Nieh, Structures and nanoindentation properties of nanocrystalline and amorphous Ta-W thinﬁlms, Scr. Mater. 58 (2008) 195–198, https://doi.org/10.1016/j.scriptamat.2007.09.042.

[11] G. Greczynski, J. Lu, O. Tengstrand, I. Petrov, J.E. Greene, L. Hultman, Nitrogen-doped bcc-Crﬁlms: combining ceramic hardness with metallic toughness and conductiv- ity, Scr. Mater. 122 (2016) 40–44,https://doi.org/10.1016/j.scriptamat.2016.05.011.

[12] J.P. Palmquist, Z. Czigany, M. Odén, J. Neidhart, L. Hultman, U. Jansson, Magnetron sputtered W-Cﬁlms with C60 as carbon source, Thin Solid Films 444 (2003) 29–37,https://doi.org/10.1016/S0040-6090(03)00937-4.

[13] L. Yang, C. Liu, M. Wen, X. Dai, Y. Zhang, X. Chen, K. Zhang, Small atoms as reinforced agent for both hardness and toughness of group-VIB transition metalﬁlms, J. Alloys Compd. 735 (2018) 1105–1110,https://doi.org/10.1016/j.jallcom.2017.11.208.

[14] S. Fritze, P. Malinovskis, L. Riekehr, L. von Fieandt, E. Lewin, U. Jansson, Hard and crack resistant carbon supersaturated refractory nanostructured multicomponent coatings, Sci. Rep. 8 (2018) 14508,https://doi.org/10.1038/s41598-018-32932-y.

[15] H.J. Goldschmidt, J.A. Brand, The tungsten-rich region of the system tungsten- carbon, J. Less Common Met. 5 (1963) 181–194,https://doi.org/10.1016/0022- 5088(63)90012-2.

[16] G. Hörz, K. Lindenmaier, R. Klaiss, High-temperature solid solubility limit of carbon in niobium and tantalum, J. Less Common Met. 35 (1974) 97–105,https://doi.org/

10.1016/0022-5088(74)90149-0.

[17] P.H. Mayrhofer, C. Mitterer, L. Hultman, H. Clemens, Microstructural design of hard coatings, Prog. Mater. Sci. 51 (2006) 1032–1114,https://doi.org/10.1016/j.pmatsci.

2006.02.002.

[18] U. Jansson, E. Lewin, Carbon-containing multi-component thinﬁlms, Thin Solid Films 688 (2019) 137411,https://doi.org/10.1016/j.tsf.2019.137411.

[19] P. Gustafson, A thermodynamic evaluation of the C-Fe-W system, Metall. Mater.

Trans. A. 18A (1987) 175–188,https://doi.org/10.1007/BF02825699.

[20] H. Okamoto, Phase diagrams for binary alloys, ASM Int. 1752 (2010)https://doi.org/

10.1016/S0966-9795(01)00037-1.

[21] M. Klinger, A. Jäger, Crystallographic Tool Box (CrysTBox): automated tools for transmission electron microscopists and crystallographers, J. Appl. Crystallogr. 48 (2015) 2012–2018,https://doi.org/10.1107/S1600576715017252.

[22] M. Klinger, More features, more tools, more CrysTBox, J. Appl. Crystallogr. 50 (2017) 1226–1234,https://doi.org/10.1107/S1600576717006793.

[23] M. Klinger, L. Polívka, A. Jäger, M. Tyunina, Quantitative analysis of structural inho- mogeneity in nanomaterials using transmission electron microscopy, J. Appl.

Crystallogr. 49 (2016) 762–770,https://doi.org/10.1107/S1600576716003800.

[24] P. Stadelmann, JEMS Version 4.6031U2017, 2017.

[25] Y. Zhang, H.J. Whitlow, T. Winzell, I.F. Bubb, Detection efﬁciency of time-of- ﬂight energy elastic recoil detection analysis systems, Nucl. Inst. Methods Phys. Res. A 149 (1999)https://doi.org/10.1016/S0168-583X(98)00963-X.

[26] K. Thompson, D. Lawrence, D.J. Larson, J.D. Olson, T.F. Kelly, B. Gorman, In situ site- speciﬁc specimen preparation for atom probe tomography, Ultramicroscopy 107 (2007) 131–139,https://doi.org/10.1016/j.ultramic.2006.06.008.

[27] W.C. Oliver, G.M. Pharr, An improved technique for determining hardness and elastic modulus using load and displacement sensing indentation experiments, J. Mater.

Res. 7 (2011) 1564–1583,https://doi.org/10.1557/JMR.1992.1564.

[28] M. Thuvander, J.J. Weidow, J. Angseryd, L.K.L. Falk, F. Liu, M. Sonestedt, K. Stiller, H.O.

Andrén, Quantitative atom probe analysis of carbides, Ultramicroscopy 111 (2011) 604–608,https://doi.org/10.1016/j.ultramic.2010.12.024.

[29] M.P. Moody, L.T. Stephenson, A.V. Ceguerra, S.P. Ringer, Quantitative binomial distribution analyses of nanoscale like-solute atom clustering and segregation in atom

(11)

probe tomography data, Microsc. Res. Tech. 71 (2008) 542–550,https://doi.org/10.

1002/jemt.20582.

[30] L. Vegard, Die konstitution der mischkristalle und die raumfüllung der atome, Zeitschrift Für Phys. 5 (1921) 17–26,https://doi.org/10.1007/BF01349680.

[31] C.H. Schramm, P. Gordon, A.R. Kaufmann, The alloy systems uranium-tungsten, uranium-tantalum and tungsten-tantalum, Jom 2 (1950) 195–204,https://doi.org/

10.1007/bf03398995.

[32] P.E.A. Turchi, V. Drchal, J. Kudrnovský, C. Colinet, L. Kaufman, Z.K. Liu, Application of ab initio and CALPHAD thermodynamics to Mo-Ta-W alloys, Phys. Rev. B Condens.

Matter Mater. Phys. 71 (2005) 1–14,https://doi.org/10.1103/PhysRevB.71.094206.

[33] K. Masuda-Jindo, V. Van Hung, N.T. Hoa, P.E.A. Turchi, First principles calculations of thermodynamic and mechanical properties of high temperature bcc Ta-W and Mo- Ta alloys, J. Alloys Compd. 452 (2008) 127–132,https://doi.org/10.1016/j.jallcom.

2006.12.163.

[34] V. Blum, A. Zunger, Prediction of ordered structures in the bcc binary systems of Mo, Nb, Ta, and W fromﬁrst-principles search of approximately 3,000,000 possible con- ﬁgurations, Phys. Rev. B Condens. Matter Mater. Phys. 72 (2005) 3–6,https://doi.

org/10.1103/PhysRevB.72.020104.

[35] W. Steurer, Single-phase high-entropy alloys– a critical update, Mater. Charact. 162 (2020) 110179,https://doi.org/10.1016/j.matchar.2020.110179.

[36] H.A. Murdoch, C.A. Schuh, Estimation of grain boundary segregation enthalpy and its role in stable nanocrystalline alloy design, J. Mater. Res. 28 (2013) 2154–2163, https://doi.org/10.1557/jmr.2013.211.

[37] A.A. Navid, A.M. Hodge, Nanostructured alpha and beta tantalum formation- relationship between plasma parameters and microstructure, Mater. Sci. Eng. A 536 (2012) 49–56,https://doi.org/10.1016/j.msea.2011.12.017.

[38] L. Yang, K. Zhang, M. Wen, Z. Hou, C. Gong, X. Liu, C. Hu, X. Cui, W. Zheng, Highly hard yet toughened bcc-W coating by doping unexpectedly low B content, Sci.

Rep. 7 (2017) 1–8,https://doi.org/10.1038/s41598-017-09807-9.

[39] Z.C. Cordero, B.E. Knight, C.A. Schuh, Six decades of the Hall-Petch effect– a survey of grain-size strengthening studies on pure metals, Int. Mater. Rev. 61 (2016) 495–512,https://doi.org/10.1080/09506608.2016.1191808.

[40] Y. Pauleau, P. Gouy-Pailler, Very hard solid-solution-type tungsten-carbon coatings deposited by reactive magnetron sputtering, Mater. Lett. 13 (1992) 157–160, https://doi.org/10.1016/0167-577X(92)90129-8.

[41] D. Shinde, S. Fritze, M. Thuvander, P. Malinovskis, L. Riekehr, U. Jansson, K. Stiller, El- emental distribution in CrNbTaTiW-C high entropy alloy thinﬁlms, Microsc.

Microanal. 25 (2019) 489–500,https://doi.org/10.1017/S1431927618016264.

[42] D.C. Ford, P. Zapol, L.D. Cooley, First-principles study of carbon and vacancy structures in niobium, J. Phys. Chem. C 119 (2015) 14728–14736,https://doi.org/10.

1021/acs.jpcc.5b00372.

[43] Y.-W. You, X.-S. Kong, X.-B. Wu, C.S. Liu, Q.F. Fang, J.L. Chen, G.-N. Luo, Interaction of carbon, nitrogen and oxygen with vacancies and solutes in tungsten, RSC Adv. 5 (2015) 23261–23270,https://doi.org/10.1039/C4RA13854F.

[44] P. Petroff, T.T. Sheng, A.K. Sinha, G.A. Rozgonyi, F.B. Alexander, Microstructure, growth, resistivity, and stresses in thin tungstenﬁlms deposited by rf sputtering, J. Appl. Phys. 44 (1973)https://doi.org/10.1063/1.1662611.

[45] H. Ma, A.S. Sologubenko, M. Döbeli, K. Sanvito, A. Heusi, K. Pletscher, R. Spolenak, The curious mechanism of irradiation-induced cryogenic grain growth in tungsten thinﬁlms: a pathway to single crystals, Acta Mater. 187 (2019) 153–165,https://

doi.org/10.1016/j.actamat.2019.12.042.

[46] S. Takaki, D. Akama, N. Nakada, T. Tsuchiyama, Effect of grain boundary segregation of interstitial elements on hallpetch coefﬁcient in steels, Mater. Trans. 55 (2014) 28–34,https://doi.org/10.2320/matertrans.MA201314.