Engineering the Dipole Orientation and Symmetry Breaking with Mixed‐Dimensional Heterostructures

Abstract Engineering of the dipole and the symmetry of materials plays an important role in fundamental research and technical applications. Here, a novel morphological manipulation strategy to engineer the dipole orientation and symmetry of 2D layered materials by integrating them with 1D nanowires (NWs) is reported. This 2D InSe –1D AlGaAs NW heterostructure example shows that the in‐plane dipole moments in InSe can be engineered in the mixed‐dimensional heterostructure to significantly enhance linear and nonlinear optical responses (e.g., photoluminescence, Raman, and second harmonic generation) with an enhancement factor of up to ≈12. Further, the 1D NW can break the threefold rotational symmetry of 2D InSe, leading to a strong optical anisotropy of up to ≈65%. These results of engineering dipole orientation and symmetry breaking with the mixed‐dimensional heterostructures open a new path for photonic and optoelectronic applications.


S1. The vibrational properties of multi-layer InSe in mixed-dimensional heterostructures.
A 532 nm laser excitation is used to investigate the possible strain effect in the mixeddimensional heterostructures. In the bent InSe region, compared with Raman response from flat InSe, the 176 cm -1 and 227 cm -1 modes show a shift of -0.78±0.1 cm -1 and -0.77±0.1 cm -1 , respectively. The previous study [1] on 30-40 layers of InSe under uniaxial tensile strain (produced by the substrate deflection method) reports a shift of -1.5±0.1 cm -1 /% for the 176 cm -1 Raman mode, and -2.3±0.1 cm -1 /% for the 227 cm -1 Raman mode. Therefore, we estimate ~ 0.3-0.5% strain in the sample. Nevertheless, with such a small strain, the Raman intensity enhancement is negligible. Therefore, we exclude the strain-induced Raman enhancement in our mixed-dimensional heterostructures. Figure S1: Comparison of InSe Raman spectra -with and -without NW.

S2. PL shift of multi-layer InSe in mixed-dimensional heterostructures.
InSe PL spectra, as shown below, are acquired under 532 nm laser excitation. Solid black and red lines are obtained after fitting the InSe PL data points for the flat and bent regions, respectively. From the fitted curves, we observe a slight redshift (<7 nm, or <9 meV) of PL peak position in the bent InSe region. An earlier study [2] on thick InSe (>30 nm) under tensile strain reports a persistent redshift of ~81 meV/% strain. Therefore, we can estimate a possible strain effect of ~ 0.1% in our mixed-dimensional heterostructures, which is negligible. Figure S2: Comparison of InSe PL peak position -with and -without NW.

S3. Assessment of optical interference, substrate, and energy transfer effects in mixeddimensional heterostructures.
We specially prepare a mixed-dimensional heterostructure sample with a large topographic change to study the optical interference effect. The sample is fabricated by sandwiching a thick hBN flake between InSe and NW. Figure S3(a) presents an optical image of the sample.
As shown in Figure S3(b), AFM measurements indicate that thickness of hBN and InSe flakes in the sample is ~40 nm and ~30 nm, respectively. Due to the thick hBN layer underneath, in the InSe/hBN/NW region, the height of the heterostructure area is large (comparable to the PL excitation and emission wavelengths). If the PL enhancement is due to the optical interference effect, we expect to see that the PL enhancement is strongly modulated with the height. We perform PL measurements on the sample and correlate the results with corresponding AFM topography. The white box in Figure S3 Figure S3(c)), we do not observe any optical interference effect (e.g., enhancement factor modulation as a function of the height in such a large topologic change). Therefore, we can rule out the optical interference effect as the reason behind enhanced optical properties in our proposed mixeddimensional heterostructures.
In the specially-prepared heterostructure sample, the lateral length of the bending region (i.e., suspension region) is also very large (~4 μm, see the AFM results in the bottom right figure of Figure S3(c)), much larger than the diameter of the NW of ~270 nm. If the PL enhancement is due to the sample suspension, we would expect to see that the PL enhancement factor should be quite uniform in the suspension region. However, in the bending suspension region, the enhancement factor increases significantly (e.g., at the lateral distance of ~3 μm) and starts to saturate (e.g., at the lateral distance of ~3.5 μm). Therefore, we can rule out the substrate and suspension effects as the reason behind enhanced optical properties in our proposed mixed-dimensional heterostructures.
Note that this specially prepared device also confirms that the PL enhancement is not because of the energy transfer effect, as we start to see the PL enhancement in the bending area (e.g., at the lateral distance of ~2.5 μm) that is far away from the NW position (at the lateral distance of ~5 μm), as shown in the bottom right figure of Figure S3(c).

S4. Angle-resolved SHG response of mixed-dimensional heterostructures.
The six-fold symmetry of SHG in InSe (black curve in Figure S4) is similar to the previous studies [3][4][5] . Interestingly, the enhanced SHG emission from the heterostructure region (red curve) no longer follows the six-fold symmetry pattern, indicating symmetry breaking in the InSe/NW heterostructure compared to the bare InSe sample. Figure S4: Comparison of polar plots of SHG intensity of InSe as a function of the rotation angle in a "parallel" configuration. During the measurements, first, the initial polarization of the incident light and the polarization of the analyzer are set along the NW axis. Afterwards, both the polarizer and the analyzer were rotated together from 0° to 360° degrees in a step of 4°.

S5. Efficacy of mixed-dimensional heterostructure approach to tune optical properties of
InSe.