Metal–semiconductor transition in atomically thin Bi2Sr2Co2O8 nanosheets Yang Wang1, Rui Cheng1, Jianjin Dong1, Yuan Liu1, Hailong Zhou2, Woo Jong Yu2, Ichiro Terasaki3, Yu Huang1,4, †, Xiangfeng Duan2,4, a 1Department of Materials Science and Engineering, University of California, Los Angeles, California 90095, USA 2Department of Chemistry and Biochemistry, University of California, Los Angeles, California 90095, USA 3Department of Physics, Nagoya University, Nagoya 464-8602, Japan 4California a) Nanosystems Institute, University of California, Los Angeles, California 90095, USA Electronic mail: xduan@chem.ucla.edu Supporting Information 1. Low-temperature localization in BSCO bulk and N≥4 nanosheets. (FIG. S1) 2. Transport properties of BSCO nanosheets. (FIGS. S2-S5) 3. Raman spectroscopy results. (FIG. S6) 4. The effect of SiO2/Si substrate on BSCO. 1 1. Low-temperature localization in N≥4 nanosheets and bulk BSCO The N≥4 nanosheets and bulk BSCO have almost the identical transport characteristics. They typically exhibit a metallic behavior in resistance-temperature dependence at higher temperatures. With decreasing temperature, the localized spins induced by the interlayer magnetic correlation result in a localized transport behavior. As a result, the metallic conduction does not persist to the low temperatures. In contrast, for the few-layered nanosheets (N<4), with weak interlayer coupling and the absence of localized interlayer magnetic correlation, we should expect that the metallic conduction could persist at low temperature. However, this is not the case. Our results show that these few-layered samples (N<4) exhibit much stronger localization throughout the temperature of 2 to 300 K, which can be attributed to disorder potential and Coulomb charging effect. Therefore, the localized characteristics of N<4 nanosheets and N≥4 nanosheets (and bulk) have totally different origins. This supported by the temperature-dependent transport characteristics. In particular, the plot of lnG as a function of T-1/3 shows that the N=4 sample in the semiconducting-like range (below 140 K) does not obey the Mott-VRH mechanism (Fig. S1), in stark contrast to the cases of N=1, 2, and 3 samples. Furthermore, the G-T behavior obeys neither ES-VRH nor thermal-activation-type conduction. This supports that the localization below 140 K in the N=4 sample is not originated from disorder potential or Coulomb charging effect, which is different from the cases of N=1, 2, and 3 samples. FIG. S1. (a) Temperature dependence of conduction G of the N=4 nanosheet. (b) The corresponding logarithmic conductance lnG as a function of T-1/3. 2 2. Transport properties of BSCO nanosheets FIG. S2. Temperature dependence of transport behavior of BSCO nanosheets. (a-d) Ids-Vds curves of N=4, 3, 2, and 1 nanosheets under Vbg = 0 measured at different temperatures. FIG. S3. Temperature dependence of transport behavior of BSCO monolayer. (a) Temperature dependence of sheet resistance R (=1/G). (b) Temperature dependence of mobility μ, calculated by = dIds /dVbg [L/(WCi Vds )] with the capacitance between the channel and back-gate per max unit area Ci = 1.18×10-4 Fm-2. 3 FIG. S4. Temperature dependence of transport behavior of BSCO nanosheets under Vbg = 0. (a-d) Temperature dependence of sheet conductance G of BSCO nanosheets. The G-T behavior under Vbg = 0 is consistent with that under Vbg = -84 V. The N=4 nanosheets have similar electric transport behavior with the BSCO bulk, whereas the N=3, 2, and 1 nanosheets remain semiconducting behavior throughout the temperature range. (e-h) Solid lines: Ids-Vds curves of BSCO nanosheets at 10 K (at 2 K for the quadrilayer) under Vbg = 0 in a small range of ±20 mV. Dash line: the corresponding dIds/dVds curves (for the monolayer, the dIds/dVds is shown as black squares and the dash lines are guide for eyes). Similar with the case under Vbg = -84 V, the nonlinear conductance suppression is observed in monolayer, bilayer, and trilayer but it is absent in quadrilayer. Under zero gate, the scale of the drain source gap is slightly larger than that under -84 V gate. ∆bias is 6.1, 1.4, and 0.6 meV for the monolayer, bilayer, and trilayer, respectively, as denoted by the arrows. 4 FIG. S5. The fittings of conductance under Vbg = -84 V vs temperature. (a,b) Logarithmic conductance lnG as a function of T-1/3 of BSCO trilayer and bilayer, respectively. The solid lines are linear fittings. Similar with the case under zero gate, the linear relationship between lnG and T-1/3 is valid throughout the whole temperature range, indicating the Mott 2D-VRH transport mechanism. (c) Plot of lnG vs T-1/3 of BSCO monolayer, where the linear relationship (Mott 2D-VRH) is valid from 300 to ~55 K. (d) Plot of lnG vs T-1/2 of monolayer. The linear relationship below ~55 K reveals an ES-VRH mechanism in this temperature range. The dash lines in (c) and (d) are guide for eyes. It can be seen that the transport behaviors under Vbg = -84 V are all similar with the case under zero gate, but the characteristic temperatures (TM, TES, and Tcross) become smaller, indicating that the disorder is partially suppressed under a large negative gate voltage. 3. Raman spectroscopy spectra of BSCO. According to the Raman spectra (Fig. S5a), the two phonon peaks at ~440 cm-1 and ~615 cm-1, corresponding to the E1g and A1g modes, represent the in-plane and out-of-plane vibrations of oxygen atoms, respectively (1). The A1g peak can be extracted by subtracting the background of silicon (Fig. S5b). As the layer number decreases, the intensity of out-of-plane vibration mode A1g only slightly decreases until N=3, where it suddenly weakens (Fig. S5c), suggesting a discontinuous interlayer exchange coupling happens at N=4 to 3. From the Raman peaks, the strength of electron-phonon coupling λ can be extracted by 5 I I 0 [1 ( 0 ) / ]2 /[1 ( 0 ) 2 / 2 ] , where ω0 is the discrete phonon frequency, I and Γ are intensity and width of the phonon peak (2). By fitting the Raman spectra, the extracted λ values are presented in Fig. S5d. It can be seen that with reducing layer number, λ increases slowly until N=4, which means that the electron-phonon interaction in this system is less affected by layer number for the N≥4 flakes. However, λ suddenly increases with the change of N=4 to 3 as well as N=2 to 1. Since electron-phonon coupling is crucial for the formation of polarons and thus induces the localization of electrons in semiconductors, herein the noticeable enhancement of λ indicates a large enhancement of carrier localization at N=4 to 3 and N=2 to 1. FIG. S6. Raman spectra and electron-phonon coupling strength of the BSCO system. (a) Normalized Raman spectra of BSCO nanosheets and bulk. The measurement was performed at room temperature under ambient conditions by a Renishaw 1000 system. The two phonon peaks at ~440 cm-1 and ~615 cm-1 represent the in-plane and out-of-plane vibration modes E1g and A1g, respectively. For the nanosheets, the strong Raman scattering at 520 cm-1 is from silicon substrate. Raman signal is relatively weak in few-layered BSCO because less material is excited, so the Raman spectra are normalized by the E1g peak for comparison. (b) Fitting of the silicon Raman peak for the N=1 nanosheet. By subtracting the background of silicon, the A1g peak of BSCO nanosheets can be extracted. (c) Comparison of the A1g peak of BSCO nanosheets and 6 bulk. (d) Layer number N dependence of the electron-phonon coupling strength λ; the dash lines are guide for eyes. 4. The effect of SiO2/Si substrate on BSCO Disorder effect caused by underlying SiO2/Si substrate has been widely reported in 2D materials. This disorder can be screened better with increasing layer number, which could partly contribute to the layer-dependent transport behavior. Nonetheless, it is important to note that SiO2 disorder potential likely plays much a less important role in the BSCO system than in graphene or MoS2 for several reasons. First, the surface roughness of SiO2/Si substrate is typically around ±0.2 nm (3), whereas the thickness of BSCO single-layer is 3 nm. Therefore, different from the case of graphene or MoS2, of which the single-layer thickness is on the order of 10-1 nm, the effect of surface roughness of SiO2/Si on BSCO is much more limited. Furthermore, BSCO monolayer has a BiO-SrO-CoO2-SrO-BiO structure (see Fig. 1A, structure sketch), in which the inner CoO2 sublayer dominates the electric transport behavior and is not in direct contact with SiO2/Si substrate. Therefore, the disorder potential due to the underlying SiO2 surface is less important to the transport properties when compared to graphene or MoS2. References S1. S. K. Yuan, M. An, Y. Wu, Q. M. Zhang, X. G. Luo, and X. H. Chen, J. Appl. Phys. 101, 113527 (2007). S2. P. Zhou, K.-A. Wang, P. C. Eklund, G. Dresselhaus, and M. S. Dresselhaus, Phys. Rev. B 48, 8412 (1993). S3. C. H. Lui, L. Liu, K. F. Mak, G. W. Flynn, and T. F. Heinz, Nature 462, 339 (2009). 7