In the format provided by the authors and unedited. Photon-triggered nanowire transistors Jungkil Kim, Hoo-Cheol Lee, Kyoung-Ho Kim, Min-Soo Hwang, Jin-Sung Park, Jung Min Lee, Jae-Pil So, Jae-Hyuck Choi, Soon-Hong Kwon, Carl J. Barrelet & Hong-Gyu Park * *E-mail: hgpark@korea.ac.kr This PDF file includes: Supplementary Figures 1 7 Supplementary Text Supplementary References NATURE NANOTECHNOLOGY www.nature.com/naturenanotechnology 1
Supplementary Figures Supplementary Figure 1 TEM characterization of a Si NW with a single PSi segment. a, Scanning transmission electron microscopy (STEM) image of a 200 nm-diameter Si NW with a single PSi segment. Scale bar, 500 nm. b,c, HR-TEM images recorded at the red dotted circles in a, which correspond to PSi (b) and CSi (c). Scale bars, 5 nm. Inset, FFT patterns for PSi (b) and CSi (c). Single crystals of Si are clearly observed in CSi, whereas nanoscale crystal lattices are observed surrounding the pores in PSi. The typical FFT patterns of Si in the [110] zone axis indicate that the crystallographic orientation of the Si NW is [100]. d, TEM image of the PSi segment which effectively includes PSi and CSi. Scale bar, 100 nm. The strong contrast is observed between CSi and PSi. PSi region is wider at the centre of the segment (inset). We note that the localized structural changes in the PSi segment have no impact on the devices because the incident pump spot covers the whole region of the PSi segment in our experiments. e, HR- TEM image of the red dotted circle in d. The white dotted line marks the interface between PSi NATURE NANOTECHNOLOGY www.nature.com/naturenanotechnology 2
and CSi. Scale bar, 5 nm. f, Energy-dispersive X-ray spectroscopy (EDX) Si mapping image taken around the PSi segment. Scale bar, 100 nm. A weaker colour density is observed at the PSi segment. g, Normalized EDX-integrated signal profile of Si in the PSi segment, which is obtained along the line-scan of the white dotted line in f. The Si element is reduced and reaches a minimum value (~30%) at the centre of the PSi segment. This indicates that the porosity of Si is ~70% at the centre of the PSi segment. NATURE NANOTECHNOLOGY www.nature.com/naturenanotechnology 3
Supplementary Figure 2 Measurement setups. a, Experimental setup for the measurement of current voltage (I V) curves in PTNTs with varying pump laser power (Figs. 1, 3 and 4c). The PTNTs mounted on a precise XYZ translation stage are optically pumped at room temperature using a 658-nm laser diode (LD). A 50 objective lens is used to focus the LD with a spot size of ~1 m. A source measure unit (SMU) and customized probe station are used to control the applied bias voltage and measure the current. b, Experimental setup for the measurement of the transient current of PTNTs (Supplementary Fig. 7). A preamp is connected to the SMU and PTNTs in series to convert current to voltage. The transient response of the PTNTs is recorded by connecting the pulse generator and the preamp to CH1 and CH2 of the oscilloscope, respectively. The time constant of the preamp is <1 s in the measurement settings (sensitivity = 10-6 A/V and bandwidth = 200 khz). c, Experimental setup using two identical pump lasers, LD1 and LD2 (Figs. 2 and 4e g). The two lasers are controlled independently to simultaneously pump the PTNTs. NATURE NANOTECHNOLOGY www.nature.com/naturenanotechnology 4
Supplementary Figure 3 Structure analysis and modelling of the band structure of PSi. a, Left, SEM image and schematic (inset) of a bundle of Si NWs with three PSi segments. The NW diameter is 25 nm and the length of each PSi segment (white dotted lines) is 100 nm. Scale bar, 300 nm. Right, magnified image of the red dotted box in the left panel. Scale bar, 100 nm. b, HR-TEM image of a PSi segment. Nanoscale crystal lattices are observed surrounding the pores. Scale bar, 5 nm. Inset, FFT pattern of PSi in the [110] zone axis. c, Schematics of band diagrams of a Si NW with a single PSi segment in dark (left) and light (right) conditions. The electronic structure of crystalline silicon (CSi) is intrinsically different from that of PSi, which consists of nanoscale Si lattices surrounding the pores as shown in b. The electronic structure of CSi is similar to that of bulk Si with a well-defined band gap. However, the electronic structure of PSi is more complicated, showing a wider electronic gap due to the quantum confinement effect in nanocrystals S1. In the dark, injected carriers are trapped in the localized states of the PSi segment NATURE NANOTECHNOLOGY www.nature.com/naturenanotechnology 5
and the current is blocked. When light is incident to the PSi segment, the trapped electrons are excited into higher electronic states, triggering a current across the electrodes. d, Schematics of band diagrams of a small-diameter Si NW with a single PSi segment in dark (left) and light (right) conditions at a low bias voltage (Fig. 3b). The localized states of the PSi segment are partially occupied due to the small number of injected electrons (inset, Fig. 3c). Then, optically excited electrons are re-trapped in the localized states. NATURE NANOTECHNOLOGY www.nature.com/naturenanotechnology 6
Supplementary Figure 4 Wavelength-dependent responsivity and device characteristics with and without PSi segment. a, Log-scale plot of the responsivity, the current normalized by the pump power, as a function of the incident wavelength on the PSi segment at a bias voltage of 5 V. We use a PTNT device with a NW diameter of 200 nm and a PSi length of ~400 nm. A supercontinuum laser with a wide wavelength range from 500 to 1600 nm (SuperK EXTREME EXB-4, NKT Photonics) is used as a pumping light source. We observe that the longer the incident wavelength the smaller the resulting responsivity. This wavelength-dependent current generation can be rationalized by a broad range of energy levels for the trap states (Supplementary Fig. 3c). As an example, for shorter pumping wavelength, a larger current is measured because the trapped electrons from a broad range of energy levels are excited into higher electronic states. But for longer pumping wavelength, only the trapped electrons from a narrow range of energy levels are excited and therefore a smaller current is measured. Notably, a current is measured even when the wavelength is between 1100 and 1600 nm, which is below the band gap of Si. We estimate the activation energy of the trap states to be smaller than 0.78 ev. b, Log-scale plot of the current laser power in the 180 nm-diameter NW devices with (blue) and without (red) PSi segment at a bias voltage of 5 V. The PSi length is ~450 nm in the PTNT. The wavelength of the CW pump laser is 658 nm. Under illumination, the current in the device with a PSi segment exceeds the current in the device without a PSi segment. A significant current increase occurs in the PSi segment because the trapped electrons from a broad range of energy levels are simultaneously excited into higher electronic states by optical pumping (Supplementary Fig. 3c). Considering the resistivity of the wafer used to synthesize the NWs as well as the NW diameter of 180 nm and its length of 8 m, the current of the NW without a PSi NATURE NANOTECHNOLOGY www.nature.com/naturenanotechnology 7
segment is estimated to be ~300 na for a bias voltage of 5 V. The measured current (red) is significantly lower than this calculated value, because of non-ideal Ohmic contacts S2, surface depletion effects S3, and enhanced carrier scattering by NW surface roughness S4. Note that the champion device in a indicates the PTNT in b (device 1 in Fig. 1e). NATURE NANOTECHNOLOGY www.nature.com/naturenanotechnology 8
Supplementary Figure 5 Measured I V curves by 1550-nm laser pumping. A PTNT device with a NW diameter of 200 nm is optically pumped at room temperature using a CW laser with a wavelength of 1550 nm. The laser is focused to a spot size of ~4.3 m on the NW device using a 50 objective lens with a numerical aperture (NA) of 0.42 (Supplementary Fig. 2a). The pump power varies from 0 to 2.3 W. We note that a current is measured even when the photon energy is less than the band gap of Si. This strongly suggests that the current is generated by photon triggering of the trapped electrons in the PSi segment (Supplementary Figs. 3c and 4a), not by the conventional interband charge separation S5,S6. NATURE NANOTECHNOLOGY www.nature.com/naturenanotechnology 9
Supplementary Figure 6 Measurement of currents in a long Si NW with a PSi segment. a, Schematic illustration of the measurement process. A 658-nm CW pump laser with a spot size of ~1 m is injected onto either the PSi segment or CSi region of a long PTNT. When the CSi region is pumped, the pump laser is far enough from the PSi segment (>7 m) so that the effect of the pump laser on the PSi segment is minimized. b, Measured currents in the PSi segment (0 m) and the CSi region (7.3 m) at a bias voltage of 5 V. The NW diameter is 200 nm and the power of the pump laser is 0.85 mw. Lower inset, magnified graph for the measured current in the CSi region, which is 7.3 m from the PSi segment. The dotted line indicates the measured current when the pump laser is turned off. The ratio of the two measured currents in the PSi segment and the CSi region is ~2 10 5. Upper insets, optical microscope images of the device when the pump laser is focused on the PSi segment (I) or the CSi region (II). Scale bars, 5 m. NATURE NANOTECHNOLOGY www.nature.com/naturenanotechnology 10
Supplementary Figure 7 Measurement of transient current in PTNTs. Normalized transient current of device 1 (Fig. 1e), measured when a pulsed pump laser is injected onto the PSi segment. The laser pulse is generated using a pulse generator to modulate the laser diode. The pulse width, interval, and peak power of the pump laser are 250 s, 500 s, and ~0.74 mw, respectively. The bias voltage is 5 V. The rise/fall times were measured to be ~28 s. Similar photoresponse times were measured when the spacing of the metal contacts of the PTNT was changed from 8 to 1 m. NATURE NANOTECHNOLOGY www.nature.com/naturenanotechnology 11
Supplementary Text 1. Field-effect transistor (FET) model for photon-triggered nanowire transistors (PTNTs). We analyse the electrical characteristics of PTNTs using a conventional FET model by making an analogy between PTNTs and FETs. The electric charge per unit length of the PTNT, Q, is given by e Q Nph NTH (S1) l where e is the elementary electric charge, l is the length of the PTNT, and is the carrier lifetime. The number of optically excited electrons per unit time from the localized states to the higher P 0 R 1 2 r e electronic states, N ph, is given by S7 N ph (S2) h where P 0 is the power of the pump laser on the PSi segment, R is the light reflectance from the surface of the NW, is the absorption coefficient, r is the radius of the NW, h is Planck constant, and is the frequency of the pump laser. We assume that the internal quantum efficiency is 1. We introduce the number of threshold electrons per unit time, N TH, which is the minimum number of excited electrons required to form a current channel through the NW S8. The charge Q in equation (S1) can be expressed in terms of effective capacitance, C eff. C eff is assumed to be a constant for a given NW S9 : C eff = e N ph /V g l = e N TH /V TH l. Then, eff g TH Q C V V (S3) where V g and V TH are the effective gate voltage and the threshold voltage, respectively. The laser power and its minimum value for the transistor operation correspond to V g and V TH, respectively. With the analogy of the conventional FET model S8,S9, the current I is represented by Ceff V I Vg VTH V (S4) l 2 where is the effective electron mobility in the PTNT and V is the applied bias voltage. As the bias voltage is greater than V g V TH, I is saturated to be I sat, Ceff I 2 sat Vg VTH. (S5) 2l Using equations (S2), (S4), and (S5), we can analyse the experimentally obtained I V characteristics of PTNTs. For example, Supplementary Text Fig. 1 presents the theoretical analysis of device 2 (Fig. 1f). Our measurements show that the l and of device 2 are ~8 m and NATURE NANOTECHNOLOGY www.nature.com/naturenanotechnology 12
~28 s, respectively (Fig. 1f and Supplementary Fig. 7). R and are set to 0.1 and 5000 cm -1, respectively S10. We made the simplifying assumption to ignore the in-plane heterostructure, i.e. porous and non-porous segments. Thus, the length between the metal contacts l is considered in the calculation of the effective electron mobility. To obtain, P TH, and C eff, we use the measured current light power (I L) curve for device 2, where P TH is the incident pump power corresponding to N TH., P TH, and C eff are determined to be ~0.99 10-3 cm 2 /V s, ~45 W, and ~8.79 10-8 C/V cm, respectively, by comparing the measured I L curve with equations (S4) and (S5). In Supplementary Text Fig. 1, the theoretically obtained I V curves with different pump powers closely agree with the experimentally obtained I V curves. Supplementary Text Figure 1 FET model for PTNTs. a, Measured I V curves for device 2 in Fig. 1f. b, Theoretically obtained I V curves with different laser powers using our model. Supplementary Text Figure 2 Carrier concentration versus laser power. Carrier concentration estimated as a function of laser power at a bias voltage of 5 V in devices 1 and 2. In Supplementary Text Fig. 2, we estimate the total carrier concentration as a function of laser power at a bias voltage of 5 V, in devices 1 and 2 (Figs. 1e and 1f). The carrier concentration n is NATURE NANOTECHNOLOGY www.nature.com/naturenanotechnology 13
obtained from the equation n = Il / Ae V (ref. S8), where I is the measured current, l and A are the length and cross-sectional area of the NW, e is the elementary electric charge, is the effective electron mobility, and V is the applied bias voltage. We use the values of : = 2.49 10-3 and 0.99 10-3 cm 2 /V s in devices 1 and 2, respectively. The graphs clearly show the features of FETs S8. NATURE NANOTECHNOLOGY www.nature.com/naturenanotechnology 14
Supplementary References S1. Wolkin, M. V., Jorne, J., Fauchet, P. M., Allan, G. & Delerue, C. Electronic states and luminescence in porous silicon quantum dots: the role of oxygen. Phys. Rev. Lett. 82, 197-200 (1999). S2. Jaeger, R. C. Introduction to Microelectronic Fabrication, 2nd ed (Prentice Hall, 2002). S3. Simpkins, B. S., Mastro, M. A., Eddy, Jr., C. R. & Pehrsson, P. E. Surface depletion effects in semiconducting nanowires. J. Appl. Phys. 103, 104313 (2008). S4. Ford, A. C. et al. Diameter-dependent electron mobility of InAs nanowires. Nano Lett. 9, 360-365 (2009). S5. Cao, L. et al. Engineering light absorption in semiconductor nanowire devices. Nature Mater. 8, 643-647 (2009). S6. Kim, S.-K. et al. Doubling absorption in nanowire solar cells with dielectric shell optical antennas. Nano Lett. 15, 753-758 (2015). S7. Soci, C. et al. ZnO nanowire UV photodetectors with high internal gain. Nano Lett. 7, 1003-1009 (2007). S8. Sze, S. M. & Ng, K. K. Physics of Semiconductor Devices (John Wiley & Sons, 2006). S9. Jaeger, R. C. & Blalock, T. N. Microelectronic Circuit Design, 4th ed (McGraw Hill, 2011). S10. Bisi, O., Ossicini, S. & Pavesi, L. Porous silicon: a quantum sponge structure for silicon based optoelectronics. Surf. Sci. Rep. 38, 1-126 (2000). NATURE NANOTECHNOLOGY www.nature.com/naturenanotechnology 15