Coexistence of fcc- and bcc-like crystal structures in ultrathin Fe films grown on $\mathrm{Cu}\left(111\right)$

We report on bcc-like phases in ultrathin Fe films grown by thermal deposition on $\mathrm{Cu}\left(111\right)$ previously thought to consist exclusively of fcc phases distinguished only by their magnetic order. Our scanning tunneling microscopy and spectroscopy data together with published x-ray photoelectron diffraction results [M. T. Kief and W. F. Egelhoff, Jr., Phys. Rev. B 47, 10785 (1993)] provide us with sufficient detail to deduce the film structure. Two growth regimes are considered: (1) films with 1–2 monolayer average thickness grown near 200 K, which nucleate as bcc-like bilayer islands: Larger islands show bcc-like fringes coexisiting with an fcc center domain; i.e., the bcc-like phase is stable only within a certain distance to a step edge. The presence of a bcc-like bilayer phase provides a straightforward explanation for the ferromagnetism previously observed in these films. In addition we find that the bcc-like phase can be promoted by H adsorption at $80\mathrm{K}$. The bcc domains form “displacement vortex” structures to simultaneously minimize film stress and interface energy. (2) In films grown at room temperature, between pseudomorphic fcc areas, we observe a more ideal but still strained bcc phase in regions with a local thickness of at least 4 monolayers. Also in this growth regime, the fcc-bcc transformation is facilitated by step edges, which are abundant due to the imperfect layer-by-layer growth.

Figure 1

fcc-bcc transformation of pseudomorphic fcc Fe films on an fcc (111) substrate. (a) Top view with consecutive out-of-plane and in-plane shear deformations. The dotted circles indicate atom locations before the respective shear strain is applied. In the Kurdjumov-Sachs (KS) orientation one of the two close-packed rows of the bcc (110) surface is aligned with a close-packed row of the fcc substrate. In the Nishiyama-Wassermann (NW) orientation both close-packed rows of the bcc (110) film are symmetrically misaligned with respect to the close-packed rows of the fcc substrate. (b) Side view of the fcc-bcc transformation (substrate not shown) with in-plane and out-of-plane shear simultaneously applied, demonstrating the dominant out-of-plane shear deformation of $\approx 19°$. The body-centered tetragonal cell that becomes cubic in the bcc product is highlighted. The arrows indicate the $⟨110⟩$ and $⟨100⟩$ forward-scattering directions used in Refs. 16, 17, 18 to identify the bcc phase by XPD.

Figure 2

STM images of 1-ML Fe films grown on $\mathrm{Cu}\left(111\right)$ and representative line profiles along the lines in the images. (a) The growth mode at $90\mathrm{K}$ is fully three dimensional, producing narrow pyramidal islands. The arrow indicates a small island part where the first monolayer is exposed (better visible in line profile). (b) Films grown at $300\mathrm{K}$ consist predominantly of bilayer islands. Tunneling conditions are $-1\mathrm{V}$ sample voltage and $1\mathrm{nA}$ tunneling current. The image header of this and most other figures indicates the average film thickness, growth, and imaging temperatures.

Figure 3

STM image of a 1-ML film deposited at $200\mathrm{K}$ substrate temperature. (a) Original image: two different classes of islands can be distinguished: bilayer islands with a larger footprint and narrow, pyramidal islands, showing additional 2–3 atomic levels. (b) Same image with the gray scale adjusted to enhance the contrast for the second monolayer. (c) Different islands of increasing size (circles in overview image) are interpreted as stages of the island evolution: very small bilayer islands always appear “bright” and either evolve towards the pyramidal form (3D growth) or remain flat but develop a “dark” phase in their center region. Tunneling conditions: $-0.1\mathrm{V}$, $0.3\mathrm{nA}$.

Figure 4

(a) STM image of a bilayer island also shown in Figs. 3b, 3c (imaged at $-0.1\mathrm{V}$ and $0.3\mathrm{nA}$). (b) Contrast inverted and high-pass-filtered atomically resolved image of a large part of the island surface imaged at $-0.5\mathrm{mV}$ and $0.3\mathrm{nA}$. The “dark” phase in the center shows a perfect hexagonal symmetry indicative of the fcc (111) Fe phase. The “bright” phase shows also a nearly hexagonal surface structure but is misaligned by about $3°$ relative to the ideal fcc phase. (c) Line profiles of the raw STM images between points indicated by greek letters in the STM images above. The contrast between the “dark” fcc and “bright” phases strongly depends on the tunneling voltage (note the different height scales).

Figure 5

Scanning tunneling spectroscopy of 2-ML films. Two different but equally prepared films are shown on the left- and right-hand sides. (a) Simultaneous STM images ($50\mathrm{nm}$ wide, imaging at $-1\mathrm{V}$ and $1\mathrm{nA}$). The arrows indicate two single-phase bilayer islands. The white rectangle marks a trilayer island with coexisting fcc- and bcc-like phases. The inset is the same trilayer island with strongly enhanced contrast on the third atomic level. The black lines indicate regions identified as fcc by their spectroscopic “fingerprint” (cf. Sec. 2). (b) Surface areas colored with respect to their spectroscopic fingerprint. (c) Spectra averaged over the areas shown in (b).

Figure 6

Same spectroscopic data set as shown on the right-hand side of Fig. 5. Shown are tunneling spectra of islands of different sizes and morphologies near locations indicated in the simultaneously acquired STM image. The spectra are shown for each individual atomic level and in the case of the trilayer island also for “dark” fcc- and “bright” bcc-like regions within the same atomic level. The numbers inside the diagrams and the STM image indicate the local thickness in ML. The solid and dashed vertical lines indicate the location of the bcc-like and fcc-related peaks at $-0.2\mathrm{V}$ and $-0.4\mathrm{V}$, respectively. With higher local thickness, the peaks both in fcc- and bcc-like regions very slightly shift to higher energies and become more intense and narrower.

Figure 7

Schematic model of the bilayer island shown in Fig. 4. (a) A bcc-like stacking of the two Fe layers and a “floating” fcc-to-hcp stacking at the $\mathrm{Fe}\text{−}\mathrm{Cu}$ interface is proposed for the bcc-like “bright” domains, which are arranged around the triangular fcc kernel of the island. The rectangles indicate the bcc (110)-like unit cells of the differently oriented bcc-like domains. (b) Same model with surface Fe layer removed. The arrows indicate, strongly exaggerated, the displacements of the interface Fe atoms relative to the substrate fcc positions, which cause the $3°$ misalignment between the atom rows of the fcc and the bcc-like domains. (c) The relatively weak misalignment of $3°$ between the fcc substrate and bcc-like film avoids on-top coordinations at the $\mathrm{Fe}\text{−}\mathrm{Cu}$ interface provided that the bcc-like regions are not wider than $\approx 20\mathrm{atom}$ rows.

Figure 8

In-plane atom arrangement in the observed bcc-like bilayer phase compared to the ideal bcc phase in Kurdjumov-Sachs orientation and the fcc phase. The characteristic misalignment of the observed bcc-like phase in the bilayer islands is only $3°$ instead of $10.5°$ for the ideal bcc phase, and the structure therefore labeled “$\mathrm{KS}\text{−}3°$.”

Figure 9

${\mathrm{H}}_2$ adsorption on a triangular Fe bilayer island at $80\mathrm{K}$. The image sequence shows four distinct steps of the structural modification (images $20\mathrm{nm}$ wide, tunneling at $-0.22\mathrm{V}$, $0.41\mathrm{nA}$). (a) Steps 1–3. Step 1: the bcc-like phase is quickly covered with H and becomes dark (reduction of apparent height). Step 2: the bcc-like phase grows and the remaining fcc phase shrinks until nearly the entire island is bcc-like. Step 3: H adsorption shows an effect also on the (here tiny) fcc center triangle, which turns dark rather abruptly. (b) Schematic model of the bilayer film after adsorption step 3. The arrows indicate the displacements of the interface Fe atoms (exaggerated) relative to the substrate fcc positions. (c) Fourth and final step of H adsorption: The well-defined domain structure disappears, leaving diffuse darker and brighter areas.

Figure 10

$\mathrm{Fe}∕\mathrm{Cu}\left(111\right)$ film grown at $300\mathrm{K}$ after adsorption of $\approx 2\mathrm{L}{\mathrm{H}}_2$ from the residual gas. (a) Overview STM image ($70\mathrm{nm}$ wide, $-1\mathrm{V}$, $1\mathrm{nA}$) showing mainly bilayer islands. (b) Contrast-enhanced image of the multidomain bilayer island outlined in the overview image ($-20\mathrm{mV}$, $1.8\mathrm{nA}$). The island surface shows a near-hexagonal atomic pattern indicative of a $(2\times 2)$-H superstructure decorating the three bcc-like rotational domains. The domain orientation becomes apparent as a weak stripe pattern with twice the periodicity of the Fe lattice, as indicated by the striped boxes. These rotational domains are arranged around the large triangle-shaped fcc Fe domain forming a displacement vortex like that shown in Fig. 9. The angular misalignment of $\approx 2°$ between domains I and II is indicated (cf. Figs. 4, 7). A smaller second fcc domain with its own vortexlike environment is formed on the left-handside of the island with one domain (II) in common with the larger fcc domain. (c) Model of the film structure. Outside the two fcc center domains, the Fe layers are bcc like—i.e., bridge-site stacked. (d) Model of the film-substrate interface. The arrows indicate the displacement of the Fe atoms relative to the substrate’s fcc positions (dark triangles). The star symbol close to the lower island edge marks a small region where the double vortex structure cannot avoid on-top stacking causing a large apparent height and therefore brighter appearance in the STM image in (b). The narrow edge domains marked (NW) may be relaxed bcc in Nishiyama-Wassermann orientation (see text).

Figure 11

(a) STM image of the same bilayer island as that shown in Fig. 10. During repeated scanning, the fcc center has become slightly smaller, although the overall domain arrangement remains unchanged $(-1\mathrm{mV}∕10\mathrm{nA})$. (b) Atomically resolved area marked by the white rectangle in (a) $(-1\mathrm{mV}$, $9\mathrm{nA}$). The atom displacements between the adjacent $\mathrm{KS}\text{−}3°$-I and -III domains are demonstrated by the grid of equidistant lines and the individual line following an atom row across the boundary. (c) Schematic model with exaggerated displacements (arrows) of the interface Fe atoms relative to the fcc positions of the Cu substrate. Note that the Fe atom density of the surface layer is significantly reduced along the domain boundary.

Figure 12

(a) Same atomically resolved image as in Fig. 11. The solid white lines now mark rows of brighter atoms producing the “stripe” pattern visible in images with lower resolution [Fig. 10b]. The dotted line marks the approximate location of the domain boundary center. (b) Possible arrangement of the H atoms in analogy to the H superstructure on the bulk $\mathrm{Fe}\left(110\right)$ surface (Ref. 42), explaining the different apparent heights (brightnesses) of the metal atoms by their different (average) H coordination. The $c(2\times 2)$-H superstructure unit cell and the strained bcc (110) unit cell are marked by the large and small rectangle, respectively.

Figure 13

STM images of 3.0 and 3.7 ML films grown and imaged at $300\mathrm{K}$. (a) Overview image of a 3.0-ML film with substrate step edges decorated by elongated pyramidal (“ridgelike”) islands (arrows). (b) Fourth ML of a 3.7-ML film with bcc-like regions appearing as bright stripes connecting deep holes in the film. (c) Individual bcc-like domain. The dark lines (arrow) are presumably boundaries between film regions, which are fcc twins with respect to each other. (d) Atomically resolved image of two grain boundaries between fcc- and bcc-like domains and between bcc-like reflection twins (upper left corner). The in-plane shear angle of this Kurdjumov-Sachs-like oriented bcc-like phase, termed “$\mathrm{KS}\text{−}7°$,” is typically $7°$ for a local thickness of 4 ML. (e) Schematic cross section illustrating the misalignment at the vertical interfaces between the bcc-like and fcc Fe phases. The line profile (location marked by dashed line in c) shows a distinct buckling at the right-hand side of the bcc-like grain, indicating lattice mismatch. Tunneling conditions: (a)–(c) $-1\mathrm{V}$, $1\mathrm{nA}$; (d) $-0.8\mathrm{mV}$, $11\mathrm{nA}$.

Figure 14

Illustration of the four bcc (110)-like phases observed in the systems $\mathrm{Fe}∕\mathrm{Cu}\left(100\right)$ and $\mathrm{Fe}∕\mathrm{Cu}\left(111\right)$ in relation to the fcc (100) and fcc (111) phases. The acronym NM stands both for the nanomartensite phase (Ref. 27) and the bcc-like needle-crystals (Ref. 26) in $\mathrm{Fe}∕\mathrm{Cu}\left(100\right)$ films, while $\mathrm{KS}\text{−}3°$ and $\mathrm{KS}\text{−}7°$ are the two Kurdjumov-Sachs-like phases in $\mathrm{Fe}∕\mathrm{Cu}\left(111\right)$ films. The directions and orientations of the different components of the strain-tensor ${\epsilon }_{ij}$ are shown both in the top view of the surface and the cross section parallel to the $yz$ plane. (a) Top view: the in-plane bond angles in nearest-neighbor atom triples are indicated. The stacking order of the phases is indicated by a representative atom in the adjacent atomic layer below. The observed interrow distances relative to that of the bcc (110) phase $\left(d_0\right)$ is shown below. (b) Side view: the interlayer distance relative to that of the bcc (110) phase $\left(h_0\right)$ is indicated. The value for the NM phase is taken from LEED experiments (e.g., Ref. 6), while those of the $\mathrm{KS}\text{−}3°$ and $\mathrm{KS}\text{−}7°$ structures are our estimates based on third-order elastic theory. The thick (near) vertical lines illustrate the perfect lattice matching between fcc (100) and NM phases and the imperfect lattice matching between fcc (111) and the KS phases. Note that while all other phases exist in films locally at least up to 8 monolayers thick, the $\mathrm{KS}\text{−}3°$ phase was observed only in bilayers although shown here 5 ML thick for the purpose of comparison.