www.acsnano.org
Quasiliquid Layer Promotes Hexagonal Boron Nitride (h-BN) Single-Domain Growth: h‑BN on Pt(110) Dominik Steiner,† Florian Mittendorfer,‡ and Erminald Bertel*,† †
Institute of Physical Chemistry, Universität Innsbruck, 6020 Innsbruck, Austria Institute of Applied Physics and Center for Computational Materials Science, Vienna University of Technology, 1040 Vienna, Austria
Downloaded via UNIV OF SOUTHERN INDIANA on July 18, 2019 at 21:38:05 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
‡
S Supporting Information *
ABSTRACT: Hexagonal boron nitride (h-BN) monolayers were grown on Pt(110) using borazine as a precursor molecule. The resulting surface structure was studied by scanning tunneling microscopy, low-energy electron diffraction, and density functional theory calculations. Borazine fragments reduce the roughening temperature of pristine Pt(110) (Tr = 1090 K); consequently, growth below T = 1100 K results in a serrated hBN/Pt(110) surface with small terraces, defects, and domain boundaries. Surprisingly, h-BN deposition at T > 1100 K yields large terraces covered by a carpet-like single-domain h-BN monolayer. Despite the incommensurability and different symmetry, the epitaxial growth is almost perfect. The key to this counterintuitive behavior is the “soft” Pt(110) surface responding to the h-BN overlayer in two ways: First, the (1 × 2)missing-row (m.r.) reconstruction is converted into a (1 × n)-m.r. reconstruction with a regular alternation of n = 5 and 6, yielding a superperiodicity of the Moiré pattern. Second, the remaining rows experience significant relaxations. Some Pt surface atoms are mobile underneath the h-BN monolayer, even at room temperature. Under growth conditions, the top metal layer is disordered and highly mobile, rendering the h-BN growth comparable to that on liquid gold. Such a mechanism may be of general relevance for the epitaxial growth of 2D materials. Because epitaxial deposition of Pt(110) on various substrates has been demonstrated, the present system appears scalable, and its regular 1D grooves render it a promising template for nanowire arrays. KEYWORDS: hexagonal boron nitride, Pt(110), 2D materials, epitaxy, chemical vapor deposition, scanning tunneling microscopy, density functional theory mismatch, for example, Rh(111),1 Ni(111),8,9 Pd(111),10,11 Pt(111),12 and several others,13 to fcc(110)14−16 and bcc(110)17 substrates. For a review, see Pakdel et al.,6 Laskowski et al.,18 and Auwärter.19 Because of the predominantly electrostatic h-BN−metal interaction resulting in weak directional bonding, one observes in most cases the nucleation of various (Novaco−McTague) rotational domains. This gives rise to inhomogeneous films with defects and domain boundaries compromising the construction of strictly periodic and unidirectional templates. Large-scale single-orientation hBN growth was recently achieved by Cun et al.4 on singlecrystalline Rh(111) and by Lee et al.,5 who, instead of exploiting epitaxial relations, used liquid gold as a perfectly
H
exagonal boron nitride (h-BN) monolayers, also sometimes dubbed “white graphene”, are the focus of intense research interest nowadays. As a material with a wide band gap of ∼6 eV and being isostructural with graphene, h-BN can serve as a dielectric layer in graphenebased 2D electronic devices. High thermal conductivity, a low friction coefficient, and chemical inertness are other prominent features of monolayer h-BN. It may therefore serve as a template for cluster deposition,1 as a catalyst and catalyst support,2,3 as a protective layer,4,5 and in several other applications.6 Most recently, it was discovered that h-BN monolayers are also efficient single-photon sources.7 Hence it is of pre-eminent interest to understand the growth and to control the structure and properties of h-BN monolayers. The growth of h-BN monolayers has been investigated on a large array of metal surfaces ranging from the natural choice of fcc(111) and hcp(0001) surfaces with only a minor lattice © 2019 American Chemical Society
Received: March 27, 2019 Accepted: June 7, 2019 Published: June 7, 2019 7083
DOI: 10.1021/acsnano.9b02377 ACS Nano 2019, 13, 7083−7090
Article
Cite This: ACS Nano 2019, 13, 7083−7090
Article
ACS Nano
the domain boundary was not resolved. The kink could result from the offset of two equivalent features in the h-BN across the domain boundary, for example, a B row above a Pt closepacked atom row (see the structural proposal by Achilli et al.16 in their supporting information). However, later we will demonstrate that the dark lines are caused by missing rows in the Pt substrate. Depending on the tunneling parameters and the tip condition, the Moiré showed stripes that are associated with the close-packed Pt rows below the h-BN monolayer (Figure 1c). Apparently, the surface was so rough and serrated under these conditions that the h-BN film occasionally fractured and partly detached from the surface. Figure 1d shows such a flake on which the h-BN hexagons could be resolved. Because its position and orientation were slightly different in the forward and backward scans, we concluded that it was not in intimate contact with the Pt substrate. By and large, these experimental findings are in excellent agreement with those reported by Achilli et al.,16 excluding a significant influence of the precursor molecule on the film structure. However, our interpretation of the contrast in terms of a structural model is quite different because the more perfect growth conditions at higher temperature allowed a deeper insight into the actual geometry of the system. h-BN Growth at T > 1100 K. Counterintuitively, an increase of the h-BN deposition temperature to 1120 K, that is, beyond the roughening temperature of pristine Pt(110), resulted in a much better ordered surface with large, flat terraces of up to ∼100 × 400 nm2 extension. Here the Moiré pattern could be very well resolved, as shown in Figure 2. Interestingly, on this well-ordered surface, the Moiré pattern
isotropic substrate. The relation of the h-BN monolayer to the substrate lattice after cooling was not explored. h-BN growth on Pt(110) has been studied recently by Achilli et al.16 They observed a rather strongly corrugated surface and a correspondingly defect-rich h-BN overlayer. The roughness of the Pt(110) surface is unexpected given a growth temperature Tgrowth = 1000 K significantly below the Pt(110) roughing temperature (1090 K< Trough < 1120 K).20−22 It therefore appears to be worthwhile to investigate whether the roughness was induced by the growth process and if there is a way to avoid it. Here we show that the growth behavior is surprisingly different from other fcc surfaces, even from Ni(110)14 or Pd(110).15 We demonstrate uniform single-domain h-BN growth on large terraces of the (1 × 1)-Pt(110) surface or, more precisely, on a (1 × n)-missing-row (m.r.) surface, where n is periodically modulated. Details of the preparation protocol and the resulting peculiar structure suggest a growth mechanism that is similar to the one observed by Lee et al.5 on liquid gold. As a consequence, an alternative choice of substrate for 2D epitaxy is suggested.
RESULTS AND DISCUSSION h-BN Growth at T < 1100 K. To gauge the effect of the precursor molecules, we dosed borazine instead of amino− borane used by Achilli et al.16 The first exposures to borazine were carried out at a sample temperature of 1020 K. Whereas the Pt crystal showed well-developed large, atomically flat terraces after cleaning, borazine exposure invariably resulted in a rough surface with terraces rarely exceeding 20 nm in the [001] direction and 80 nm in the [11̅0] direction. On such terraces appeared the characteristic Moiré pattern of an h-BN layer on Pt(110) with B−N−B zig-zags oriented along the close-packed row [11̅0] direction of Pt(110). The (1 × 2)-m.r. reconstruction of clean Pt(110) was lifted beneath the h-BN monolayer, as indicated by the small-scale stripes visible, for instance, in Figure 1c. Occasionally, a slight offset in the Moiré pattern indicated the presence of antiphase or mirror domains, as shown in Figure 1b. The kinks in the dark lines separating the apparent hills in the Moiré pattern signaled the location of the associated domain boundary, but the atomic structure of
Figure 1. h-BN monolayer grown at 1020 K on Pt(110): (a) Moiré pattern. (b) Domain boundary signaled by shifted unit cells (black grids) and kinks (white circles). (c) Pt close-packed rows resolved within the Moiré pattern. (d) h-BN hexagons resolved on a detached flake. Tunneling parameters It, Ubias (w.r.t. tip): (a) 200 pA, 1 V; (b) 200 pA, 0.5 V; (c) 200 pA, 0.3 V; (d) 200 pA, 1.5 V.
Figure 2. h-BN monolayer grown at 1120 K: Moiré pattern, showing a 5−6−5−6 superperiod in the [001] direction and height profiles measured at the indicated positions. The width of the smaller stripes corresponds to exactly five Pt close-packed row distances a (a = 392 pm), that for the wider ones to 6a. It = 200 pA, Ubias = 1 V. 7084
DOI: 10.1021/acsnano.9b02377 ACS Nano 2019, 13, 7083−7090
Article
ACS Nano
Figure 3. (a) STM image illustrating the long-range 5−6−5−6 superperiodicity of the h-BN/Pt(110) Moiré pattern. It = 200 pA, Ubias = 2 V. (b) LEED image recorded at U = 136 V. The dotted rectangle corresponds to the (1 × 1) unit cell of the pristine surface. A profile along the white line is presented in the Supporting Information (Figure S3).
was often modulated with some superperiod, namely, 5−6−5− 6, 6−5−5−6, or 5−6−6−5, in terms of the Pt lattice constant (a = 392 pm). This is illustrated in the profiles shown in Figure 2 for a 5−6−5−6 modulation, which appears to be the most common periodicity. Figure 3a shows a large-scale scanning tunneling microscopy (STM) image illustrating the regular periodicity over a distance of 40 nm. Figure 3b displays an associated low-energy electron diffraction (LEED) pattern. Note that the observed periodicities in the [001] direction are consistent with the c(8 × 10) and c(8 × 12) unit cells identified by Achilli et al.16 Surprisingly, the 11-fold periodicity, to be expected from a 5−6−5−6 ordered structure, is not clearly evident in the LEED pattern. Motivated by Figure 3a, we show a calculated diffraction pattern of a 1D array of scattering centers in the Supporting Information (Figure S4). The different periodicity of the form factor and the structure factor combined with the finite width of the diffraction peaks due to the finite coherence length cause a slight shift of the observed diffraction maxima, consistent with what is observed in the actual LEED pattern. In addition, we note that the 5− 6−5−6 periodicity is well developed only on large terraces, whereas LEED averages over a broad terrace distribution. With appropriate tunneling parameters, it was possible to resolve a smaller-scale stripe pattern, which is apparently caused by the underlying close-packed atom rows of the Pt(110) surface (Figure 4). However, the average apparent distance between the rows is 428 pm and thus is somewhat larger than the Pt lattice constant a. In contrast, the dark serpentine lines separating the broader stripes in the Moiré pattern of Figure 4 have an average distance of exactly six Pt lattice constants. Finally, we observe that the Moiré pattern on the upper terrace (top left corner of Figure 4) is shifted relative to the pattern on the lower terrace by exactly half a Pt lattice constant in the [001] direction. This indicates the existence of a continuous carpet-like h-BN film extending across the step because the shift corresponds precisely to the offset of the Pt close-packed rows in the upper terrace. The step height is 138 pm, as it is for steps on the pristine Pt(110) surface. A similar
Figure 4. h-BN monolayer grown at 1120 K: Moiré pattern, showing a 5−6−5−6−6 superstructure in the [001] direction and a height profile measured at the indicated position. It = 200 pA, Ubias = −2 V. Black and white inset: DFT-based STM simulation (Tersoff−Hamann approach, see the Methods section).
carpet-like morphology is observed for graphene on Pt(110).23 As far as we can judge from inspecting different regions by STM, the size of a domain is limited only by imperfections of the crystal surface, that is, areas with densely accumulated steps (scratches) where we cannot resolve the film structure and hence cannot exclude the existence of domain boundaries. The step pattern appearing in Figure 4 is typical for the hBN/Pt(110) system and indicates a dramatic change in the 7085
DOI: 10.1021/acsnano.9b02377 ACS Nano 2019, 13, 7083−7090
Article
ACS Nano
closely resemble the results of Achilli et al.,16 our calculations show pronounced structural effects as well as an increased adsorption energy on the latter structure. We expect a similar result for the adsorption on the (1 × 6)-m.r.-Pt(110) substrate. The optimized structure for the adsorption on the (1 × 5)-m.r. surface (Figure 6) clearly shows that the interaction is
ratio of step formation energies for steps running in the [001] and [11̅0] directions, respectively. On the clean (1 × 2)-m.r. reconstructed Pt(110), this ratio exceeds one order of magnitude,24 and the large anisotropy gives rise to the socalled fish-scale pattern on slightly miscut Pt(110) crystals.25,26 After being covered by the h-BN layer, steps in both orthogonal directions appear to be almost equivalent in energy; however, steps running along the [11̅0] direction occasionally pair up underneath the h-BN film, thus forming double steps of 277 pm height, as seen in Figure S1. This figure also shows a comparison with the rough surface obtained at deposition temperatures below 1100 K. In this case, single and double steps appear in stochastic succession, as expected for a surface just above the roughing temperature. Apparently this morphology is preserved upon cooling by the defective h-BN film on top, whereas at T > 1100 K, the film is uniform and allows the surface underneath to anneal to the long-range ordered state. A final observation relates to the mobility of Pt surface atoms at room temperature. Figure 5 shows two images of the
Figure 6. (a) Side view and (b) top view illustrating the geometry of the h-BN/(1 × 5)-m.r.-Pt(110) slab according to the DFT calculations. The model is based on a c(10 × 10) unit cell, as marked by the lozenge. Pt surface atoms are shown in light gray, Pt bulk atoms in dark gray, N atoms in light blue, and B atoms in green. (c) Moiré pattern of the structure in panel b. For clearer visibility, Pt atoms here are colored in black. (d) Enlarged section of panel b. The structure models are visualized using VESTA.27
Figure 5. (a) Kinks in the dark lines separating Moiré stripes. (b) After a time lap of ∼20 min, the kinks have changed their relative position. (c) Profiles along lines 1 and 2 in panel a. Note that the distance between the missing rows is almost identical to the Pt lattice constant, whereas the adjoining rows have significantly relaxed toward the gap. It = 2.8 nA, Ubias = 0.6 V.
same area that were continuously scanned for several minutes. Two kinks in the black, snaky lines are marked. The time lapse between Figure 5a and Figure 5b is ∼20 min. Both kinks have moved within this time span and, in particular, have changed their relative positions. Within the observation window, we noticed both forward and backward jumps, and the jump rate is estimated as about one nearest-neighbor (nn) distance jump (277 pm) per minute. The jumps are apparently not tipinduced because we never noticed a movement between forward and backward scans (time lapse of 0.5 s). Density Functional Theory Results. Density functional theory (DFT) calculations have been performed for a c(10 × 10) supercell, minimizing the computational misfit between the BN layer and the Pt(110) substrate. We studied the adsorption of the BN layer both on the unreconstructed Pt(110) surface and on the (1 × 5)-m.r.-Pt(110) surface suggested by the periodic elimination of Pt rows in the STM image. Whereas our results for the unreconstructed surface
dominated by the interaction of the N atoms (light blue) with the Pt substrate (see bonding interactions shown in the ball-and-stick representation of the side view in Figure 6), resulting in a strongly corrugated BN layer: For the N atoms located directly on top of the underlying Pt substrate, the calculations predict the smallest N−Pt distance of ∼2.30 Å (in good agreement with Achilli et al.16), whereas the maximal height is found for a N atom in a bridge position along the same Pt(110) row, resulting in a total corrugation of the layer of ∼100 pm. This is larger than the 50 pm corrugation found by the Padua group on (1 × 1)-Pt(110).16 In addition, we find that the presence of the (1 × 5)-m.r. reconstruction leads to a low-lying, rather flat stripe located above the missing Pt row in the Moiré pattern. We also found that the interaction between the BN moiré and the substrate is too weak to lift the Pt(110) missing row reconstruction if the N atoms are located in hollow positions of the substrate: A comparison with the adsorption of the BN 7086
DOI: 10.1021/acsnano.9b02377 ACS Nano 2019, 13, 7083−7090
Article
ACS Nano layer on the unreconstructed c(10 × 10) Pt(110) surface shows that the latter structure, which has the same number of B and N atoms, is slightly disfavored by 44 meV per (1 × 1)-Pt surface unit cell when the energy of the missing Pt atoms is referenced to the DFT-D3 bulk value (EB = −6.86 eV). In addition, the distance between the Pt rows in the surface layer is increased by up to 14 pm in the regions of the Moiré with the strongest interaction, with slightly smaller values for the regions with a weaker bonding. A minor difference between the present approach and that of Achilli et al.16 is the use of a different van der Waals (VdW) correction (DFT-D3 vs DFTD2 in ref 16), but the main distinction is, of course, the inclusion of the (1 × 5(6))-m.r. reconstruction of the substrate, yielding a contrast inversion above the missing row in the simulated STM images. Structure files containing the atom coordinates are available on request from the authors. Discussion: h-BN Growth at T < 1100 K. The h-BN monolayer shown in Figure 1 has been grown at 1020 K. The roughening temperature of Pt(110) is >1090 K.21,22 The considerable roughening observed during borazine deposition is due to step decoration by fragments, which lowers the step formation energy and therefore the roughening temperature. This is an effect known for many adsorbates. For instance, for bromine adsorption on Pd(110), we observed spontaneous step formation, even at room temperature.28 However, although the monolayer is grown above the roughening transition, one might expect that during the cool down the Pt(110) surface returns to its flat low-temperature morphology. This is apparently not the case. A possible explanation is the pinning of steps at defects and domain boundaries in the hBN film grown under these circumstances. Furthermore, borazine fragments not inserted into the 2D lattice and still remaining at the Pt surface or boron segregating from the Pt bulk during cool down could block the formation of large terraces. Because the surface seems to strongly fluctuate during deposition, it is not surprising that the monolayer is partly ruptured, with small lappets even detached from the substrate to such a degree that the position and orientation did not completely match in the forward and backward scans. The apparent decoupling from the substrate allowed imaging of the monolayer honeycomb structure, which elsewhere was obscured by the Pt-induced local density of states (LDOS) (Figure 1d). By and large, the observations agree with the Padua group (Achilli et al.)16 Discussion: h-BN Growth at T > 1100 K. Although counterintuitive at first sight, the results of Lee et al.5 for h-BN monolayer growth on liquid gold encouraged us to try borazine deposition at an even higher temperature. At a deposition temperature of 1120 K, we actually already obtained a much smoother surface with large, uniformly covered terraces after cooling to room temperature. Because here the growth temperature exceeds the effective roughening temperature in the presence of borazine by at least 100 K, the Pt surface is expected to strongly fluctuate. The Pt surface atoms are therefore highly mobile and form a liquidlike, dynamically disordered layer. Thus the nucleation, the growth conditions, and the monolayer formation mechanism are comparable to those on liquid Au. Furthermore, at this temperature, fragments might become unstable and either be incorporated into the h-BN network or desorb. As will be seen later, the Pt atom mobility underneath the h-BN monolayer is surprisingly large. Hence the surface is able to revert to its long-range ordered morphology during cooling with no
contaminants, grain boundaries, or defects blocking the process. Next, we address the nature of the comparatively sharp, snaky dark lines separating the stripes of the Moiré pattern in Figures 3−5. The corresponding minimum in the LDOS could result from (i) a domain boundary in the h-BN film, (ii) the combined effect of the overlayer and the substrate, or (iii) a feature in the Pt surface layer. Domain boundaries in the h-BN layer can be excluded because the periodicity of the Moiré pattern is preserved across the dark lines and remains the same irrespective of the different spacings between the lines. A combined effect of the overlayer and the substrate could arise if the LDOS minimum stems from a coincidence of certain structures in the substrate and adsorbate, as suggested by Achilli et al.16 We discount this possibility on the basis of the following arguments: First, if the pattern results from the superposition of rigid h-BN on Pt(110), then the dark lines would necessarily be strictly periodic. The different observed periodicities can only result if the h-BN adlayer is elastically distorted. A switch between a 5a and 6a line separation (a being the Pt lattice constant) would imply a strain of ∼4%, corresponding to several tens of millielectronvolts energy per atom.18,29 Whereas our DFT calculations find variations in the B−N bond length of up to 3.2%, these variations are locally compensated, rendering the global strain of the h-BN layer almost negligible. Even more importantly, the presence of abrupt, mobile kinks cannot be reconciled with a continuous hBN monolayer evident from the regular Moiré pattern. The simulated image based on the DFT calculations (see Figure 4, inset) allows us to unambiguously attribute the dark lines and their occasional kinks to a feature in the substrate, namely, a missing row in the otherwise (1 × 1)-terminated Pt(110) surface. The missing row has no effect on the position of the unit cells in the Moiré pattern, in contrast with a domain boundary such as the one shown in Figure 1b. Whereas on the pristine Pt(110) surface the (1 × 2)-m.r. reconstruction is stabilized by ∼75 meV relative to a (1 × 1) unit cell,30 the present (1 × 5)-m.r. reconstruction underneath the h-BN layer is preferred by ∼44 meV relative to a h-BN/(1 × 1)-Pt(110) unit cell. Figure 4 shows that the measured corrugation of the dark lines amounts to ∼80 pm, as compared with the 138 pm depth expected for a missing-row trough. The difference is attributable in part to a smoothing effect of the continuous hBN layer. Actually, in the DFT results, both the lowest and the highest parts of the h-BN film are found above the closepacked Pt rows depending on whether the N sits on top or in a bridge site, respectively. DFT yields a total corrugation of the h-BN film of ∼100 pm, as compared with ∼80 pm extracted from the STM data in both orthogonal directions. The small-scale stripes seen in Figures 1c, 4, and 5 are obviously related to the Pt close-packed rows. Their separation exceeds the row-to-row distance on clean Pt(110) by 30−40 pm. The LDOS does not necessarily reflect the exact position of the rows, but the DFT results indeed show a maximum distance of Pt atoms in neighboring rows of ∼409 pm relative to the calculated lattice constant of 395 pm. Although this is significantly less than the value estimated from the STM measurements, it does imply a substantial relaxation of the remaining rows toward the vacant one. In addition to the relaxation, the close-packed Pt rows adjacent to the missing rows are meander-like distorted by ∼25 pm. The snake-like periodic lateral distortion is well reflected in the DFT STM simulation (inset of Figure 4). Obviously, the atom rows of the 7087
DOI: 10.1021/acsnano.9b02377 ACS Nano 2019, 13, 7083−7090
Article
ACS Nano
applications appears to be possible. This would add another alternative to the Auliq and the Rh(111)/SiO2 templates,4,5 each of which have their particular merits.
topmost Pt layer respond vividly to the potential energy surface (PES) provided by the h-BN overlayer. This is understandable if one considers the inherent instability of the Pt(110) surface. Finally, the diffusive motion of the kinks is easily explained in this model as a hopping of the outermost Pt atom in a truncated chain to the neighboring, oppositely truncated chain. Close-packed rows are eliminated if their position would fall onto an unfavorable locus in the PES provided by the h-BN adlayer, for instance, if a B row would coincide with a Pt closepacked row. Neighboring Pt rows may both be similarly close to such a line, as illustrated in a ball-and-stick model of the undistorted h-BN monolayer on top of a (1 × 1)-Pt(110) surface in Figure S2. It is conceivable that a kink appears in the pattern, where the missing row position switches by one lattice constant. The insertion of a missing row enables an elastic relaxation in the Pt layer, much as the solitons do in the Frenkel−Kontorova model of incommensurate overlayers.31 The periodicity may change between 5 and 6 depending on temperature and perhaps the cooling rate.
METHODS The base pressure in our ultrahigh vacuum (UHV) preparation chamber was ∼5 × 10−11 mbar. The Pt(110) single crystal serving as the substrate was cleaned by heating in oxygen (1 × 10−7 mbar, 970 K) as well as repeated cycles of Ar sputtering and annealing to ∼1120 K. A final cleaning step consisted of two cycles of O2 adsorption at T < 120 K and subsequent flashing to 950 K. For the monolayer deposition, borazine was dosed from a Peltier-cooled storage vessel through a capillary onto the crystal in UHV. Typical exposures were 20 s at 5 × 10−8 mbar (∼5 × 10−7 mbar in front of the capillary). During exposure, the crystal was kept at a deposition temperature between 950 and 1170 K via electron impact heating. After terminating the borazine exposure, the crystal was kept at the exposure temperature for ∼60 s and subsequently cooled with a cooling rate of 1 K/s to 940 K. Then, the heating was switched off. The cooling rate below 940 K was initially ∼12 K/s. Our efforts to prepare h-BN islands at reduced exposures to determine the island shapes and orientations as well as the nucleation sites have failed up to now. We either got complete coverages or completely covered terraces adjacent to completely empty terraces. At pressures p < 5 × 10−8 mbar, no film growth occurred at all. We conclude that the nucleation barrier is high, requiring a substantial supersaturation. Once nucleation took place on a terrace, the monolayer spread rapidly across the whole terrace. After the self-terminating monolayer deposition, the crystal was transferred to the variable-temperature scanning tunneling microscopy (VT-STM) chamber and examined by STM. Bias voltages in the figure captions are given with respect to the tip. STM images were background-subtracted and in some cases Gauss-filtered using Gwyddion.35 DFT calculations were performed with the Vienna Ab initio Simulations Package (VASP)36,37 using the Perdew−Burke−Ernzerhof (PBE) xc functional38 with DFT-D3 corrections39 to account for the dispersion interactions. A 2 × 2 × 1 k-point mesh was used for the integration of the Brillouin zone. The surface was modeled by a sixlayer Pt(110) slab using the optimized DFT-D3 Pt lattice constant of 3.92 Å and a c(10 × 10) superstructure to allow for the adsorption of the BN layer. The BN layer and the uppermost four layers of the substrate were relaxed until the residual forces were 1100 K, and with Tgrowth > Trough, the Pt surface is strongly disordered, rendering the top layers liquid-like. The adaptability of the Pt(110) surface allows the h-BN film to grow in a single domain without incorporating energetically costly domain walls. This is different from nearly all other metal substrates used for h-BN growth because their rigid surfaces do not respond to the h-BN layer, and thus the buildup of elastic energy is released by the formation of multiple domains. The h-BN growth in the present system is more comparable to the one observed on liquid gold and hints to a general strategy for the seamless growth of 2D materials on liquids or surfaces with quasi-liquidlayer termination: Instead of templates matching the film geometry under normal conditions, it may be preferable to use substrates exhibiting quasiliquid surface layers under growth conditions. Typically, both Au and Pt exhibit reconstructing, that is, unstable, surfaces that not only offer the required mobility at the growth temperature but also are able to adapt to the h-BN film structure upon cooling, as observed in the present case. In the case of the liquid gold template, the low-temperature structure of the h-BN-covered Au surface has not been investigated, although this would certainly be interesting.5 The peculiar structure of the present h-BN/Pt(110) system appears promising as a template for nanowire growth. Moreover, the epitaxial growth of Pt(110) on MgO(110) 32,33 and SrTiO3(110)34 has been demonstrated. Therefore, up-scaling to achieve seamless h-BN monolayer films for device
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsnano.9b02377. Comparison of the surface morphology after h-BN deposition above and below 1100 K, respectively (Figure S1), periodicity of an undistorted h-BN film on top of a (1 × 1)-Pt(110) surface (Figure S2), evaluation of the LEED profile shown in Figure 3b (Figure S3), and simulation of the diffraction pattern for a 1D scattering array with periodicity 5−6−5−6 (Figure S4) (PDF)
AUTHOR INFORMATION Corresponding Author
*E-mail:
[email protected]. ORCID
Florian Mittendorfer: 0000-0002-5073-9191 Erminald Bertel: 0000-0002-9006-8222 7088
DOI: 10.1021/acsnano.9b02377 ACS Nano 2019, 13, 7083−7090
Article
ACS Nano Author Contributions
(16) Achilli, S.; Cavaliere, E.; Nguyen, T. H.; Cattelan, M.; Agnoli, S. Growth and Electronic Structure of 2D Hexagonal Nanosheets on a Corrugated Rectangular Substrate. Nanotechnology 2018, 29, 485201. (17) Müller, F.; Hüfner, S.; Sachdev, H. One-Dimensional Structure of Boron Nitride on Chromium (110) − a Study of the Growth of Boron Nitride by Chemical Vapour Deposition of Borazine. Surf. Sci. 2008, 602, 3467−3476. (18) Laskowski, R.; Blaha, P.; Schwarz, K. Bonding of Hexagonal BN to Transition Metal Surfaces: An Ab Initio Density-Functional Theory Study. Phys. Rev. B: Condens. Matter Mater. Phys. 2008, 78, 045409. (19) Auwärter, W. Hexagonal Boron Nitride Monolayers on Metal Supports: Versatile Templates for Atoms, Molecules and Nanostructures. Surf. Sci. Rep. 2019, 74, 1−95. (20) Cvetko, D.; De Renzi, V.; Floreano, L.; Morgante, A.; Tommasini, F. Deconstruction and Roughening Transitions on (1 × 2)Pt(110). Solid State Commun. 1994, 91, 539−543. (21) Speller, S.; Parascandola, S.; Heiland, W. Structural Parameters of Pd(110) and Pt(110)-(1 × 2) in the Temperature Range 300−900 K Studied by Low-Energy Ion Scattering. Surf. Sci. 1997, 383, 131− 136. (22) Krzyzowski, M. A.; Zeppenfeld, P.; Romainczyk, C.; David, R.; Comsa, G.; Kuhnke, K. E.; Kern, K. Thermal Disordering of the Pt(110)-(1 × 2) Surface. Phys. Rev. B: Condens. Matter Mater. Phys. 1994, 50, 18505−18516. (23) Gao, T.; Xie, S.; Gao, Y.; Liu, M.; Chen, Y.; Zhang, Y.; Liu, Z. Growth and Atomic-Scale Characterizations of Graphene on Multifaceted Textured Pt Foils Prepared by Chemical Vapor Deposition. ACS Nano 2011, 5, 9194−9201. (24) Vilfan, I. Step Patterns on Vicinal Reconstructed Surfaces. Surf. Sci. 1996, 350, 329−335. (25) Swamy, K.; Bertel, E.; Vilfan, I. Step Interaction and Relaxation at Steps: Pt(110). Surf. Sci. 1999, 425, L369−L375. (26) Carlon, E.; van Beijeren, H. Networks of Steps on Crystal Surfaces. Phys. Rev. Lett. 1996, 76, 4191−4194. (27) Momma, K.; Izumi, F. VESTA 3 for Three-Dimensional Visualization of Crystal, Volumetric and Morphology Data. J. Appl. Crystallogr. 2011, 44, 1272−1276. (28) Amann, P.; Cordin, M.; Gotsch, T.; Menzel, A.; Bertel, E.; Redinger, J.; Franchini, C. Halogen Phases on Pd(110): Compression Structures, Domain Walls, and Corrosion. J. Phys. Chem. C 2015, 119, 3613−3623. (29) Peng, Q.; Ji, W.; De, S. Mechanical Properties of the Hexagonal Boron Nitride Monolayer: Ab Initio Study. Comput. Mater. Sci. 2012, 56, 11−17. (30) Zhu, T.; Sun, S.-G.; van Santen, R. A.; Hensen, E. J. M. Reconstruction of Clean and Oxygen-Covered Pt(110) Surfaces. J. Phys. Chem. C 2013, 117, 11251−11257. (31) Bak, P. Commensurate Phases, Incommensurate Phases and the Devil’s Staircase. Rep. Prog. Phys. 1982, 45, 587−629. (32) Tolstova, Y.; Omelchenko, S. T.; Shing, A. M.; Atwater, H. A. Heteroepitaxial Growth of Pt and Au Thin Films on MgO Single Crystals by Bias-Assisted Sputtering. Sci. Rep. 2016, 6, 23232. (33) Lairson, B. M.; Visokay, M. R.; Sinclair, R.; Hagstrom, S.; Clemens, B. M. Epitaxial Pt(001), Pt(110), and Pt(111) Films on MgO(001), MgO(110), MgO(111), and Al2O3(0001). Appl. Phys. Lett. 1992, 61, 1390−1392. (34) Harp, G. R.; Farrow, R. F. C.; Marks, R. F.; Vazquez, J. E. Epitaxial Growth and Homoepitaxy of Pt(110) and Cu(110) on SrTiO3(110). J. Cryst. Growth 1993, 127, 627−633. (35) Nečas, D.; Klapetek, P. Gwyddion: An Open-Source Software for SPM Data Analysis. Open Phys. 2012, 10, 181−188. (36) Kresse, G.; Furthmüller, J. Efficiency of Ab-Initio Total Energy Calculations for Metals and Semiconductors Using a Plane-Wave Basis Set. Comput. Mater. Sci. 1996, 6, 15−50. (37) Kresse, G.; Hafner, J. Ab Initio Molecular Dynamics for Liquid Metals. Phys. Rev. B: Condens. Matter Mater. Phys. 1993, 47, 558−561. (38) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient Approximation Made Simple. Phys. Rev. Lett. 1996, 77, 3865−3868.
D.S. carried out the sample preparation and STM and LEED measurements. F.M. carried out the DFT calculations. E.B. conceived the experiment and wrote the manuscript together with F.M. All authors contributed to the interpretation. Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS We thank Stefano Agnoli (University of Padua) for communicating results prior to publication. D.S. thanks the Universität Innsbruck for financial support. F.M. gratefully acknowledges the Austrian Science Fund (FWF, project F45) for financial support and the Vienna Scientific Cluster for providing computing time. REFERENCES (1) Corso, M.; Auwärter, W.; Muntwiler, M.; Tamai, A.; Greber, T.; Osterwalder, J. Boron Nitride Nanomesh. Science 2004, 303, 217− 220. (2) Lyalin, A.; Gao, M.; Taketsugu, T. When Inert Becomes Active: A Fascinating Route for Catalyst Design. Chem. Rec. 2016, 16, 2324− 2337. (3) Elumalai, G.; Noguchi, H.; Uosaki, K. Electrocatalytic Activity of Various Types of h-BN for the Oxygen Reduction Reaction. Phys. Chem. Chem. Phys. 2014, 16, 13755−13761. (4) Cun, H.; Hemmi, A.; Miniussi, E.; Bernard, C.; Probst, B.; Liu, K.; Alexander, D. T. L.; Kleibert, A.; Mette, G.; Weinl, M.; Schreck, M.; Osterwalder, J.; Radenovic, A.; Greber, T. Centimeter-Sized Single-Orientation Monolayer Hexagonal Boron Nitride With or Without Nanovoids. Nano Lett. 2018, 18, 1205−1212. (5) Lee, J. S.; Choi, S. H.; Yun, S. J.; Kim, Y. I.; Boandoh, S.; Park, J.H.; Shin, B. G.; Ko, H.; Lee, S. H.; Kim, Y.-M.; Lee, Y. H.; Kim, K. K.; Kim, S. M. Wafer-Scale Single-Crystal Hexagonal Boron Nitride Film via Self-Collimated Grain Formation. Science 2018, 362, 817−821. (6) Pakdel, A.; Bando, Y.; Golberg, D. Nano Boron Nitride Flatland. Chem. Soc. Rev. 2014, 43, 934−959. (7) Aharonovich, I.; Englund, D.; Toth, M. Solid-State SinglePhoton Emitters. Nat. Photonics 2016, 10, 631−641. (8) Nagashima, A.; Tejima, N.; Gamou, Y.; Kawai, T.; Oshima, C. Electronic Dispersion Relations of Monolayer Hexagonal Boron Nitride Formed on the Ni(111) Surface. Phys. Rev. B: Condens. Matter Mater. Phys. 1995, 51, 4606−4613. (9) Auwärter, W.; Kreutz, T. J.; Greber, T.; Osterwalder, J. XPD and STM Investigation of Hexagonal Boron Nitride on Ni(111). Surf. Sci. 1999, 429, 229−236. (10) Nagashima, A.; Tejima, N.; Gamou, Y.; Kawai, T.; Oshima, C. Electronic Structure of Monolayer Hexagonal Boron Nitride Physisorbed on Metal Surfaces. Phys. Rev. Lett. 1995, 75, 3918−3921. (11) Morscher, M.; Corso, M.; Greber, T.; Osterwalder, J. Formation of Single Layer h-BN on Pd(111). Surf. Sci. 2006, 600, 3280−3284. (12) Ć avar, E.; Westerström, R.; Mikkelsen, A.; Lundgren, E.; Vinogradov, A. S.; Ng, M. L.; Preobrajenski, A. B.; Zakharov, A. A.; Mårtensson, N. A Single h-BN Layer on Pt(111). Surf. Sci. 2008, 602, 1722−1726. (13) Gómez Díaz, J.; Ding, Y.; Koitz, R.; Seitsonen, A. P.; Iannuzzi, M.; Hutter, J. Hexagonal Boron Nitride on Transition Metal Surfaces. Theor. Chem. Acc. 2013, 132, 1350. (14) Greber, T.; Brandenberger, L.; Corso, M.; Tamai, A.; Osterwalder, J. Single Layer Hexagonal Boron Nitride Films on Ni(110). e-J. Surf. Sci. Nanotechnol. 2006, 4, 410−413. (15) Corso, M.; Greber, T.; Osterwalder, J. h-BN on Pd(1 1 0): A Tunable System for Self-Assembled Nanostructures? Surf. Sci. 2005, 577, L78−L84. 7089
DOI: 10.1021/acsnano.9b02377 ACS Nano 2019, 13, 7083−7090
Article
ACS Nano (39) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. A Consistent and Accurate Ab Initio Parametrization of Density Functional Dispersion Correction (DFT-D) for the 94 Elements H-Pu. J. Chem. Phys. 2010, 132, 154104.
7090
DOI: 10.1021/acsnano.9b02377 ACS Nano 2019, 13, 7083−7090
