Exploring the thermal behavior and diffusive functionality of structural defects and phase boundaries in near-stoichiometric chromium diborides by in situ scanning transmission electron microscopy Open Access

Transition material diborides (TMB₂) are a unique class of boron-based materials that proved to be extremely hard, thermally stable, and chemically inert.^1–7 This exclusive combination of functional properties makes them suitable material candidates for applications in extreme thermal and chemical environments. Thus, the foreseen TMB₂ applications are found in aerospace, automotive, energy production, electronics, and machining industries.^8–11 

While boride materials exhibit high structural diversity, the hexagonal AlB₂ crystal structure (P6/mmm) prevails for TMB₂ in which close-packed layers of the TM atoms are interleaved by layers of boron atoms.^12,13 The ability to maintain phase stability at extreme conditions originates from the ceramic and metallic bonding nature within the TMB₂ structure. The large part of TMB₂ phases is a line compound, which forms with a precise TM-to-B ratio of 1-to-2.¹⁴ Maintaining this ratio is challenging during the synthesis of protective coatings and thin films, which are commonly deposited using various physical vapor deposition (PVD) approaches. Typically, overstoichiometric materials (with a high B-content) are deposited with the incorporation of secondary B-rich phases.^6,15–23 However, recent developments permit precise TM-to-B control and realization of under- or near-stoichiometric TMB₂.^24–36 It is essential to explore and understand structural characteristics and properties of films as the varying stoichiometry provides each film with a unique combination of properties.

In the light of recently explored understoichiometric TMB_2−x films, which have significantly widened the compositional and elemental parameter space,² it is desirable to obtain fundamental knowledge on thermal induced changes in such films at elevated temperatures and at the atomic level. Previously, it was reported that under thermal treatment, TMB₂ films undergo transformations, e.g., age hardening, resulting from recrystallization, column coarsening, and planar defect annihilation/creation.^15,37–39 Furthermore, given that the performance of advanced structural components depends on thermal stabilities of the interfacing materials, e.g., ceramic/metal, and associated diffusion phenomena, TMB₂ films are being proposed as candidates for diffusion barrier applications. In microelectronic applications, TiB₂ films were examined as diffusion barriers in ultralarge-scale integrated circuits due to their stability and low resistivity.^40–42 The other example of the application is for multilayer engineering, where CrB₂ was applied as a diffusion barrier in Mo/Si multilayers.⁴³ 

In this study, we employ in situ annealing inside a scanning transmission electron microscope to explore the thermal stability of structural defects and metal contact diffusion pathways in near-stoichiometric, high quality, epitaxial CrB₂ films. By atomic-level identification, the persistence of structural defects is explored up to 1100 °C and identified as dominant pathways for metal diffusion. We show that controlling stoichiometries and structural qualities are the critical factors for enabling TMB₂ films as diffusion barriers.

Near-stoichiometric, high crystal quality, epitaxial chromium diboride films were grown using direct-current magnetron sputtering (DCMS) on sapphire [α-Al₂O₃ (0001)] substrates. Growth was done in an Ar discharge from a chromium diboride target (Plansee) with a base pressure of ∼2.6 × 10⁻⁷ Torr in the deposition system. The substrates were heated to 900 °C during the film deposition process. The film stoichiometry (B/Cr ratio) was varied by changing p_Ar= 5 or 20 mTorr with an Ar flow of 10.3 and 47.3 SCCM, respectively. This allowed synthesis of understoichiometric CrB_1.90±0.10 (p_Ar= 20 mTorr) and overstoichiometric CrB_{2.08 ± 0.10} (p_Ar= 5 mTorr) films. Compositions of the as-grown films were determined by Rutherford backscattering spectroscopy. A detailed description of the growth conditions, composition determination, crystal quality, and atomic structure evaluation of as-grown films can be found elsewhere.⁴⁴ 

TEM sample preparation of the as-grown films was based on a mechanical and focused ion beam (FIB) lift-out sample routine described elsewhere.⁴⁵ For the in situ annealing inside scanning transmission electron microscope (STEM) experiments, the electron transparent plan-view samples (Fig. S1a)⁵⁶ were attached to the MEMS heating chip (Fig. S1b)⁵⁶ using platinum (Pt) as the contact material.

The in situ annealing experiments were performed in the Linköping double C_s corrected FEI Titan³ 60–300, using a MEMS-based double tilt heating holder (DENS solutions). In situ annealing was accomplished by increasing the temperature from ambient conditions to 1000 °C in steps of 200 °C and holding for 15 min after each step. Once the temperature reached 1000 °C, the sample was annealed for 30 min. In a final annealing step, the temperature was further increased to 1100 °C and held for an additional 30 min, with a total annealing time of 2 h, per sample.

During in situ experiments, high angle annular dark field STEM (HAADF-STEM) images were continuously acquired. Atomically resolved HAADF-STEM imaging was achieved by using a 21.5 mrad convergence semi-angle, which facilitated a sub-Ångström resolution electron probe with ∼60 pA beam current while utilizing an annular detection range between 46 and 200 mrad.

After the in situ annealing experiments, with the sample cooled to ambient temperature, spectroscopic information was collected from the annealed structure. Electron energy loss spectroscopy (EELS) spectrum images (SIs) of 40 × 40 pixels were acquired for 5 min using 0.4 nA beam current, 0.25 eV/channel energy dispersion, 0.2 s pixel dwell time, and a 55 mrad collection semi-angle of the employed Gatan Quantum ERS postcolumn imaging filter. Elemental B and Cr distribution maps were extracted from the spectrum images using power law background subtraction and selecting distinctive edge integration windows for B-K (188–208 eV) and Cr-L₂₃ (575–590 eV). Energy-dispersive x-ray (EDX) analysis provided spectrum images of 256 × 256 pixels, which were recorded in 4 min using 0.4 nA beam current in the cumulative mode and a dwell time of 120 μs per pixel, employing the embedded high solid angle Super-X EDX detector. Elemental Pt-L_α distribution maps were extracted from the STEM-EDX spectrum images.

Figure 1 shows a series of atomically resolved HAADF-STEM images acquired from the understoichiometric CrB_1.90 film, viewed along [0001]. The images show the as-grown sample [Figs. 1(a) and 1(d)], after in situ annealing at 1000 °C [Figs. 1(b) and 1(e)], and at 1100 °C [Figs. 1(c) and 1(f)]. Note that the HAADF-STEM images were recorded from different regions of the same TEM lamella due to the absence of distinctive reference features imposed by the high-density defect network in the CrB_1.90 film and the substantial thermal sample drift upon each annealing step.

The as-grown sample exhibits many planar defects of different structural arrangements. For simplicity, these are jointly referred to as antiphase boundaries (APBs) that reside on the { $1 ¯$100} prismatic planes of CrB₂ [Figs. 1(a) and 1(d)], in agreement with previous observations of understoichiometric TiB₂, Zr_0.70Ta_0.30B_1.50, and CrB₂.^44,46–48 Compared to the as-grown sample, annealing at 1000 °C does not notably affect the structure of the film [Figs. 1(b) and 1(e)]. However, at 1100 °C, bright atomic columns appear [Fig. 1(c)], with a preferential location at the interconnections of the APBs (Fig. 1).

The signal intensity of the HAADF-STEM images is directly proportional to the atomic number (∼Z^1.7) and sample thickness.⁴⁹ This suggests that the bright columns appear due to the agglomeration of heavy atoms (high Z). The positioning and elemental nature of the formed bright columns are further explored in Fig. 2.

The interconnection shown in Fig. 2(a) is the combination of three distinctly different types of APBs (APB-1, APB-2i, and APB-2-ii), while the interconnection in Fig. 2(b) is composed of two identical APBs (APB-2i) as described in detail in Refs. 44, 46, and 47. Burgers circuit analysis around these interconnections [as shown in Fig. 2(b)] proves the presence of edge dislocations, in line with previous findings.⁴⁶ This infers that the core of the edge dislocations accommodates heavier element atoms that have diffused into the core during annealing. The chemical nature of the bright columns was investigated by STEM-EDX elemental mapping, which reveals them to be Pt as shown in Figs. 2(c) and 2(d). Pt is suggested to originate from the contact material between the TEM lamella and the MEMS chip (Fig. S1a).⁵⁶ Presumably, Pt is thermally activated at the highest temperature whereupon it migrates across the sample surface and into the accommodating dislocation cores.

Figure 2(e) shows an individual CrB₂ grain, which is misoriented with respect to the surrounding matrix, exhibiting an extended grain boundary, which is also locally decorated with Pt columns. As the grain boundary frequently displays open (unfilled) cores (dark columns), this difference suggests that the core must exhibit a sufficiently large opening for the Pt atoms to diffuse through. The similar observation can be made from Fig. 1(c), not all dislocation cores are filled.

Figure 3 shows a series of atomically resolved HAADF-STEM images of the overstoichiometric CrB_2.08 film, viewed along [0001]. The images show the as-grown sample [Figs. 3(a) and 3(d)], in situ annealed at 1000 °C [Figs. 3(b) and 3(e)] and at 1100 °C [Figs. 3(c) and 3(f)].

The as-grown film exhibits regions with dark contrast, corresponding to B-rich (and Cr-deficient) inclusions surrounded by the CrB₂ lattice as exemplified in Figs. 3(a) and 3(d). Excess boron segregation and formation of a B-rich amorphous tissue phase are well-documented phenomena in overstoichiometric TiB₂.⁶ In contrast to Fig. 1, the HAADF-STEM images were recorded from the same region on the TEM lamella due to the unique shape of the B-rich inclusion, which made it possible to locate the same region between temperature ramps. In addition to the B-rich inclusions, a single type of APBs can be observed in the film (APB-1) as shown in Fig. 3(a). Upon in situ annealing at 1000 °C, the interface between the B-rich phase and ambient CrB₂ becomes more clearly defined [Figs. 3(b) and 3(e)]. The applied heat induces a rearrangement of the interface, which results in the formation of facets on the ${1 1 ¯00}$ prismatic planes, as clearly resolved in these plan-view images. This presumably occurs to reduce the interface area and, therefore, the interface energy of the structure. Upon in situ annealing at 1100 °C, the facets become even more pronounced. Moreover, high-contrast lines become visible [Figs. 3(c) and 3(f)], which have a preferential location at the interface between the B-rich inclusions and the CrB₂ facets. The lines are typically arranged as an atomically thick layer, though thicker layers are also observed. It should also be noted that the APBs observed in the structure remain stable throughout the experiment, as in the understoichiometric film.

Figure 4 provides further insight into the interface, the APBs, and the elemental nature of the CrB_2.08 film after the annealing experiment.

Figures 4(a) and 4(b) reveal that high-contrast regions are present at the interface between the B-rich inclusions and the CrB₂ structure. The EELS and EDX elemental analysis, shown in Figs. 4(d)–4(f), of the inclusion in Fig. 4(a), again confirms that Pt is present at the interface. Figure 4(b) shows that Pt not only diffused to the interfaces but also exhibits atomic order on the CrB₂ { $1 ¯$100} prismatic facets where it appears to add complete layers. Again, presumably initiated by the drive to reduce the interface energy. While metallic Pt crystallizes as FCC, with a spacing of ∼3.94 Å, the distance between atomic columns in this figure is measured as 3.16 Å, which would correspond to a ∼6.3 Å unit cell. The detailed structure of the Pt containing interface remains to be explored. The APBs present in the film again accommodate Pt in-diffusion at edge dislocations as shown in Fig. 4(c), in agreement with the observations from CrB_1.90 at same conditions.

The applied in situ annealing up to 1100 °C of slightly under- and overstoichiometric CrB₂ films did not induce significant structural changes to the film. Observed defects, predominantly APBs, were not affected by the applied heat. The most apparent structural change was observed in the overstoichiometric film where the phase boundaries between the B-rich tissue phase and the CrB₂ structure developed a well-defined structure, inferred to reduce the interface energy of the system. This was also shown in previous studies, albeit for significantly more overstoichiometric material, exhibiting a less ordered structure.^15,37

The most apparent change in the structure comprises the diffusion of thermally activated Pt atoms, which serves as a model element for this study. Pt was available from the FIB metal weld that migrates toward and into the boundaries between phases and into the dislocation cores formed by the APBs. The driving force for the diffusion of non-native elements into dislocation cores is inferred to also originate from a desire to reduce the defect energy.^50–55 

Similarly, the decoration of the interfaces between the diboride and the amorphous tissue phase in the overstoichiometric film is inferred to originate from the ability of Pt to arrange itself in a morphic structure on the diboride facets, which arguably reduces the total energy compared to Pt dissolving into the amorphous tissue phase.

TMB₂ are high temperature materials, of interest to withstand much higher temperatures than explored herein. Though this investigation suggests minor reformation of the structure, depending on stoichiometry, it highlights the influence of adjacent elements on the structural properties. Both dislocation cores and phase boundaries act as diffusion paths for non-native species although the boundaries evidently allow for diffusion of more material. Both artifacts should therefore be avoided. It is known from earlier investigations that these defects become increasingly available with higher off-stoichiometry. Accordingly, perfectly stoichiometric or slightly understoichiometric TMB₂ should be pursued as diffusion barriers and for high temperature applications.

Near-stoichiometric films of CrB₂ were annealed in situ, up to 1100 °C, to explore the thermal stability of structural defects depending on the composition. It was observed that line defects of both samples were unaffected by the applied heat. However, the interfaces between the boron-rich tissue phase and the CrB₂ phase became better defined because of the reduction of interface energy in the overstoichometric film. Moreover, at the highest temperature, Pt was observed to diffuse into dislocation cores and in between the diboride/amorphous boron tissue phase interface, which is inferred as a means to reduce the energy of the system. This study presents the first direct atomic scale evidence of metal diffusion in near-stoichiometric TMB₂ films via edge dislocations, interfaces, and grain boundaries. Precise control of the B/TM ratio and structural characteristics of the film is suggested to be of great importance to reduce the observed metal diffusion phenomena and implementation of TMB₂ for diffusion barrier applications.

The Knut and Alice Wallenberg Foundation is acknowledged for support of the electron microscopy laboratory in Linköping and for project funding through The Boride Frontier (No. KAW 2015.0043). The Swedish Research Council (VR) and the Swedish Foundation for Strategic Research (SSF) are acknowledged for funding a research program (No. EML16-0004) and for access to ARTEMI, the Swedish National Infrastructure in Advanced Electron Microscopy (Nos. 2021-00171 and RIF21-0026). J.P. acknowledges support from VR (No. 2021-03652) and Carl Tryggers Stiftelse (No. CTS:21 1272). Megan Dorri is acknowledged for providing samples presented in this work.

The authors have no conflicts to disclose.

Per O. Å. Persson: Funding acquisition (equal); Resources (equal); Writing – review & editing (equal). Johanna Rosen: Funding acquisition (equal); Writing – review & editing (equal). Ivan Petrov: Writing – review & editing (equal). Justinas Palisaitis: Conceptualization (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Project administration (equal); Supervision (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal).

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