www.acsnano.org
Plasmonic Silicon Quantum Dots Enabled High-Sensitivity Ultrabroadband Photodetection of Graphene-Based Hybrid Phototransistors Zhenyi Ni,†,⊥ Lingling Ma,‡,⊥ Sichao Du,‡ Yang Xu,*,‡ Meng Yuan,† Hehai Fang,§ Zhen Wang,§ Mingsheng Xu,‡ Dongsheng Li,† Jianyi Yang,‡ Weida Hu,*,§ Xiaodong Pi,*,† and Deren Yang*,† †
State Key Laboratory of Silicon Materials and School of Materials Science and Engineering and ‡Key Laboratory of Micro-Nano Electronics and Smart System of Zhejiang Province, College of Information Science and Electronic Engineering, Zhejiang University, Hangzhou, Zhejiang 310027, China § State Key Laboratory of Infrared Physics, Shanghai Institute of Technical Physics, Chinese Academy of Sciences, Shanghai 200083, China S Supporting Information *
ABSTRACT: Highly sensitive photodetection even approaching the singlephoton level is critical to many important applications. Graphene-based hybrid phototransistors are particularly promising for high-sensitivity photodetection because they have high photoconductive gain due to the high mobility of graphene. Given their remarkable optoelectronic properties and solution-based processing, colloidal quantum dots (QDs) have been preferentially used to fabricate graphene-based hybrid phototransistors. However, the resulting QD/ graphene hybrid phototransistors face the challenge of extending the photodetection into the technologically important mid-infrared (MIR) region. Here, we demonstrate the highly sensitive MIR photodetection of QD/graphene hybrid phototransistors by using plasmonic silicon (Si) QDs doped with boron (B). The localized surface plasmon resonance (LSPR) of B-doped Si QDs enhances the MIR absorption of graphene. The electron-transition-based optical absorption of B-doped Si QDs in the ultraviolet (UV) to near-infrared (NIR) region additionally leads to photogating for graphene. The resulting UV-to-MIR ultrabroadband photodetection of our QD/ graphene hybrid phototransistors features ultrahigh responsivity (up to ∼109 A/W), gain (up to ∼1012), and specific detectivity (up to ∼1013 Jones). KEYWORDS: silicon quantum dots, boron doping, graphene, phototransistor, mid-infrared, localized surface plasmon resonance
G
high-performance photodetection in the wavelength range from the ultraviolet (UV) to near-infrared (NIR) by taking advantage of photogating effect.5,7,13 With the growing demand for the photodetection in the mid-infrared (MIR) region, it has been considered that narrow-bandgap CQDs such as colloidal HgTe QDs are promising to form hybrid phototransistors with graphene.22 However, the stability and toxicity of narrowbandgap CQDs seriously challenge their practical use on a large scale.23,24 As one type of the most important CQDs, Si QDs hold an advantageous position given the abundance and nontoxicity of
raphene is emerging as a viable alternative to conventional optoelectronic materials.1 Breakthroughs are highly expected in graphene-based optoelectronic devices that can be monolithically integrated into silicon (Si)based mass-production platforms.2−4 Among all kinds of graphene-based optoelectronic devices, phototransistors are attracting significant attention due to their highly sensitive photodetection and compatibility with Si-based read-out electronics.5−7 Since the intrinsic light absorption of graphene is rather weak,8−10 graphene is usually hybridized with other nanostructures to enhance the light absorption of graphenebased phototransistors.7,11−18 Among these nanostructures colloidal quantum dots (CQDs) are particularly interesting because they have excellent tunable optical absorption and suitability for low-cost solution-based processing.19−21 Up to now, CQD/graphene hybrid phototransistors have exhibited © 2017 American Chemical Society
Received: May 22, 2017 Accepted: September 18, 2017 Published: September 18, 2017 9854
DOI: 10.1021/acsnano.7b03569 ACS Nano 2017, 11, 9854−9862
Article
Cite This: ACS Nano 2017, 11, 9854-9862
Article
ACS Nano
Figure 1. Structure of a Si-QD/graphene phototransistor. (a) Schematic diagram of the structure of the hybrid phototransistor based on Bdoped Si QDs and graphene. (b) Two distinct optical phenomena of B-doped Si QDs may be exploited during the phototransistor operation. In the MIR region, the intense near field associated with the LSPR of B-doped Si QDs may enhance the direct excitation of graphene, enabling the device to efficiently respond to the MIR light. The UV-to-NIR absorption of Si QDs leads to photogenerated carriers (electrons and holes) in the QDs. Transfer of one type of carriers from the QDs to graphene and trapping of the other type of carriers in the QDs (i.e., photogating) then take place, rendering the photoresponse of the device.
device is a phototransistor with the backgate of the heavily doped p-type Si substrate and the channel of Si-QD/graphene. Figure 1b illustrates that two distinct optical phenomena of Si QDs may be exploited during the phototransistor operation. The LSPR of Si QDs induces an intense electric field, which enhances the direct excitation of the underlying graphene in the MIR region. Hence, the device is able to efficiently respond to the MIR light. The UV-to-NIR absorption of Si QDs leads to photogenerated carriers (electrons and holes) in the QDs. Transfer of one type of carriers from the QDs to graphene and trapping of the other type of carriers in the QDs (i.e., photogating) then take place, rendering the photoresponse of the device. Si QDs heavily doped with B used in the phototransistors are synthesized by using SiH4-based nonthermal plasma.27,28,30 The nominal B concentration of Si QDs is 40%, which has been measured by using inductively coupled plasma atomic emssion spectroscopy. Colloidal Si QDs can be readily formed by dispersing Si QDs in ethanol, as shown in Figure 2a. The lowresolution transimission electron microscopy (TEM) image (Figure 2b) demonstrates that nearly spherical Si QDs are synthesized in this work. The crystallinity of Si QDs is evidenced by the selected-area electron diffraction (the inset of Figure 2b) and the high-resolution TEM image of an individual Si QD (Figure 2c). A statistical size analysis indicates that Si QDs have a mean size of ∼6 nm (Figure 2d). Figure 2e shows the MIR absorption of Si QDs. It is seen that a LSPR peak occurs at ∼3 μm, which can be well fitted by the Mie absorption theory including the Drude contribution (Supporting Information).26,28,34 The overtone at ∼4 μm is related to the vibration of boron hydride at the surface of Si QDs.27 From the fitting, we work out that the hole concentration of Si QDs is ∼4.5 × 1020 cm−3. This indicates that only ∼2.2% of doped B atoms are effectively activated to produce free holes, consistent with previously observed partial electrical activation of B in Si QDs.26,35 It is the collective oscillation of free holes in Si QDs that gives rise to the LSPR of Si QDs in the MIR region.26−28 The UV-to-NIR absorption spectrum of colloidal Si QDs is shown in Figure 2f. In addition to the conventional band-toband transistion based optical absorption in the UV−vis region (Figure S1a in the Supporting Information), the heavy B
Si, the stability of Si QDs, and the compatibility of Si QDs with Si-based technologies.23,25 It has been recently found that the doping of Si QDs with boron (B) and phosphorus (P) enables the localized surface plasmon resonance (LSPR) of Si QDs in the MIR region.26−29 Compared with P-doped Si QDs, Bdoped Si QDs are currently better positioned for device applications in terms of dopant activation efficiency, oxidation resistance, and dispersibility.26,29,30 This inspires the probable use of B-doped Si QDs as plasmonic antennas to enhance the optical absorption of graphene in the mid-infrared region, analogous to classical plasmonic metal nanostructures.31−33 Given the fact that heavy-B-doping-induced band-tail states have already extended the optical absorption of Si QDs from the traditional UV−vis region into the NIR region,28 hybrid phototransistors based on graphene and Si QDs should exhibit the capability of UV-to-MIR ultrabroadband photodetection, which is in stark contrast with the narrowband photodetection of those based on graphene and plasmonic metal nanostructures. In this work, we use colloidal Si QDs heavily doped with B to form hybrid phototransistors with graphene. The LSPR of Bdoped Si QDs significantly enhances the MIR absorption of graphene, enabling the MIR photodetection of QD/graphene hybrid phototransistors. In addition, the photogating effect following the band-to-band and sub-bandgap optical absorption of B-doped Si QDs causes QD/graphene hybrid phototransistors to effectively operate in the UV-to-NIR region. Therefore, ultrabroadband UV-to-MIR photodetection of QD/ graphene hybrid phototransistors with high sensitivity is realized. The current work advances the integration of graphene with Si technologies for high-performance optoelectronic devices.
RESULTS AND DISCUSSION Si-QD/Graphene Phototransistor Structure. Figure 1a schematically shows the structure of a hybrid phototransistor based on graphene and Si QDs. Graphene chemically vapordeposited on Cu foil is transferred to the 10 μm long channel region between two Cr/Au electrodes on the top of a SiO2/Si substrate. Colloidal B-doped Si QDs are spin-coated onto graphene to form a complete device structure. The prepared 9855
DOI: 10.1021/acsnano.7b03569 ACS Nano 2017, 11, 9854−9862
Article
ACS Nano
Figure 2. Characterization of Si QDs and Si-QD/graphene. (a) A photograph of an ethanol solution of B-doped Si QDs with the concentration of 0.025 mg/mL. (b) Low-resolution TEM image of B-doped Si QDs. The selected area electron diffraction is shown in the inset. (c) High-resolution TEM image of a B-doped Si QD. (d) Size distribution with a log-normal fit for B-doped Si QDs. The mean size of Bdoped Si QDs is ∼6 nm. (e) FTIR spectrum (black line) of B-doped Si QDs. The LSPR-induced absorption peak is fitted with the Drude model (red line). (f) Optical absorption spectrum of B-doped Si QDs (red line) and absorption coefficient (α) of a B-doped Si-QD film. (g) Raman spectra of graphene and Si-QD/graphene.
doping extends the optical absorption of Si QDs into the NIR region because the heavy B doping introduces band-tail states to Si QDs, giving rise to sub-bandgap transistion.28 By fitting the UV-to-NIR absoprtion onset of Si QDs, we find that the width of the band-tail states is ∼0.74 eV (Figure S1b in the Supporting Information). When colloidal Si QDs are spincoated to form a film, the UV-to-NIR absorption of the film is also significantly strong. We have worked out the absorption coefficient of the Si-QD film, as shown in Figure 2f. It is significant that the absorption coefficient of the Si-QD film is much larger than that of bulk Si in nearly the whole UV-to-NIR region.36
Figure 2g shows the Raman spectra of graphene and Si-QD/ graphene. The ratio of the intensity of the 2D peak at ∼2692 cm−1 to that of the G peak at ∼1591 cm−1 for graphene is larger than 2, implying that monolayer graphene is used in the current work.37 The absence of the D peak at ∼1350 cm−1 for graphene means that negligible defects occur to our graphene.38 The Si− B related peak at ∼2434 cm−1 (Figure S2 in the Supporting Information) together with the aforementioned 2D and G peaks in the Raman spectrum indicates that the hybrid structure of Si QDs and graphene is indeed formed in the channel region of a phototransistor. The thickness of the Si-QD film is ∼105 nm (Figure S3a in the Supporting Information), which has 9856
DOI: 10.1021/acsnano.7b03569 ACS Nano 2017, 11, 9854−9862
Article
ACS Nano
Figure 3. Ultrabroadband UV-to-MIR photoresponse. (a) Photocurrent density (JPh) as a function of VDS at VG = 0 V from the UV to MIR. The irradiances of the MIR illumination and the UV-to-NIR illumination are 375 mW/cm2 and 0.2 μW/cm2, respectively. Please note that the JPh is plotted in a semilogarithmic scale. (b) Responsivity as a function of laser irradiance at different laser wavelengths at VG = 0 V and VDS = 1 V. (c) Gain as a function of illumination wavelength at VG = 0 V and VDS = 1 V. The illumination irradiances in the MIR and UV-to-NIR regions are 375 mW/cm2 and 0.2 μW/cm2, respectively. (d) Time-dependent photocurrent density of the Si-QD/graphene photodetector at different illumination wavelengths at VG= 0 V and VDS = 1 V. (e) Gain as a function of the inverse of carrier transit time (ttr−1). The wavelength and irradiance of the illumination are 532 nm and 0.2 μW/cm2, respectively. (f) Spectral dependence of the NEP and specific detectivity (D*) of the device. The UV-to-NIR and MIR measurements were conducted at room temperature and 77 K, respectively.
that JPh is in a logarithmic scale in Figure 3a). When the phototransistor is illuminated at the irradiance of 375 mW/cm2 in the MIR region (2.5−3.9 μm), JPh increases with the increase of the incident light wavelength (λ) from 2.5 to 3.0 μm and then decreases as λ further increases from 3.0 to 3.9 μm. For the illumination at the irradiance of 0.2 μW/cm2 in the UV to NIR region (λ = 375−1870 nm), the maximum JPh is obtained as λ is ∼532 nm. The responsivity (R) of the phototransistor as a function of the illumination irradiance at VDS = 1 and VG = 0
been found to render optimum overall device performance. (Figure S3b in the Supporting Information). Ultrabroadband Photoresponse. Figure 3a shows the drain-source photocurrent density (JPh) as a function of the drain-source bias (VDS) for a Si-QD/graphene phototransistor under the UV-to-MIR illumination at the gate voltage (VG) of 0 V. The UV-to-NIR and MIR measurements were conducted at room temperature and 77 K, respectively. It is found that JPh linearly increases as VDS increases from 0 to 1 V (please note 9857
DOI: 10.1021/acsnano.7b03569 ACS Nano 2017, 11, 9854−9862
Article
ACS Nano
Figure 4. LSPR, photogating, and carrier transfer. (a) Cross-section distribution of the square of electric field (|E|2) at B-doped Si QDs and graphene under the illumination at the wavelength of 2.7, 3.0, and 3.6 μm. (b) Drain-source current density (JDS) as a function of VG at 3 μm and 532 nm with irradiances of 375 mW/cm2 and 0.2 μW/cm2, respectively. The MIR measurements were conducted at 77 K. Those for SiQD/graphene and graphene-only devices in the dark are also shown. (c) Schematic diagram of the band structures of graphene and Si QDs and the tendency of charge transfer between them. (d) Schematic diagram of the band structure of Si-QD/graphene and charge transfer after the UV-to-NIR illumination.
nm with the irradiance of 0.2 μW/cm2. As ttr decreases from 256 to 4 ps, the gain increases from ∼3 × 1010 to 2 × 1012. The linear dependence of the gain on the inverse of ttr is consistent with the photoconductive structure of our devices.42 We should be able to further increase the gain by reducing the Si-QD/ graphene channel length because ttr decreases with the decrease of L. As shown in Figure S6 in the Supporting Information, we have also calculated the external quantum efficiency (EQE) and internal quantum efficiency (IQE) of the device by using EQE = Ip/q × ℏν/P and IQE = Ip/q × ℏν/αP, respectively, where Ip is the photocurrent of the device, q is the elemental charge, ℏ is the Planck constant, ν is the frequency of light, P is the power of light, and α is the absorbance of the Si-QD/graphene structure).36 The EQE of the device varies from ∼0.07 to 18 in the MIR region, while it changes from ∼108 to 5 × 109 in the UV-to-NIR region. The IQE of the device ranges from ∼7 to 539 and from ∼109 to 6 × 1010 in the MIR and UV-to-NIR regions, respectively. The figure of merit used to evaluate the capability of weak light detection for a photodetector is the noise equivalent power (NEP), which can be evaluated by considering the 1/f noise, shot noise, and thermal noise of the device.43,44 The calculations of the spectral density of these noises are detailed in the Supporting Information. At the modulation frequency of 1 Hz, the values of spectral density of 1/f noise (SI (1/f)), shot noise (SI (shot)), and thermal noise (SI (thermal) are ∼1.70 × 10−19, 8.91 × 10−22 and 4.41 × 10−23 A2/Hz, respectively (Figure S7 in the Supporting Information). With the measured responsivity (R) of the device at the modulation frequency of 1 Hz at VG = 0 V and VDS = 1 V, the NEP can be calculated by using NEP = √SI/R(SI = SI (1/f) + SI (shot) + SI (thermal)).39 The obtained values of NEP are as low as ∼10−10 and 10−18 W/ Hz0.5 in the MIR region and the UV-to-NIR region, respectively
V is illustrated in Figure 3b. The decrease of R with the increase of the irradiance is typical for photodetectors.7,14,17 R is in the range of 0.22−44.9 A/W for the MIR illumination at the lowest irradiance of 375 mW/cm2, while R changes between 1.2 × 108 and 2.2 × 109 A/W for the UV-to-NIR illumination at the lowest irradiance of 0.2 μW/cm2. It should be pointed out that a phototransistor with only graphene in the channel hardly responds in the MIR region. We find that Si QDs may enhance R by the factor of ∼1011 in the UV-to-NIR region by comparing the Si-QD/graphene phototransistor with the graphene-only phototransistor (Figure S4 in the Supporting Information). Figure 3c shows the dependence of the photoconductive gain of the Si-QD/graphene phototransistor on λ at VDS = 1 and VG = 0 V. Ultrahigh gains in the range from ∼7 × 1011 to 2.6 × 1012 have been obtained for the photodetection from the UV to MIR. The gain is calculated by using τ/ttr (τ is the time for the carrier trapping, and ttr is the time for the transfer of carriers through the device channel).39 Since we know ttr = L2/(μVDS) (L is the length of the device channel, and μ is the carrier mobility), we can work out that ttr is ∼4 ps. τ may be approximated by the fall time of the transient Jph during the ON/OFF illumination, as shown in Figure 3d. It is found that for the UV-to-MIR illumination, τ is in the range of ∼3.4−9.0 s, which is comparable to those for the phototransistors fabricated by Roy et al.6 and Lopez-Sanchez et al.40 Please note that τ for the Si-QD/graphene photodetector is similar to that for the photodetector only with Si QDs in the channel (Figure S5 in the Supporting Information). This implies that the carrier trapping is mainly associated with Si QDs, which may have trap states induced by defects (e.g., dangling bonds) at the QD surface.41 It is the large difference between τ and ttr that leads to the ultrahigh gain in this work. Figure 3e shows the change of the gain with the inverse of ttr under the illumination at λ = 532 9858
DOI: 10.1021/acsnano.7b03569 ACS Nano 2017, 11, 9854−9862
Article
ACS Nano
Figure 5. Comparison of the current results with those reported in literature. (a) Responsivity and (b) specific detectivity obtained in the current work are compared with those of related devices reported in literature.
Supporting Information). Hence, for the Si-QD/graphene phototransistor with the lower B doping level, the wavelength at which the maximum photoresponse appears increases, while the photoresponse of the device is attenuated (Figures S4 and S9 in the Supporting Information). It should be noted that the LSPR of Si QDs is also dependent on the QD size.28,34 The distribution density of Si QDs determines the spacing between Si QDs and thus the coupling of the LSPR of neighboring Si QDs.26 Our FDTD simulations have verified that both the size and distribution density of Si QDs indeed affect the electromagnetic field in graphene via the LSPR (Figures S10 and S11 in the Supporting Information). Therefore, it can be envisioned that the performance of the Si-QD/graphene phototransistor may be further improved by enhancing the LSPR-induced absorption of graphene through the systematic optimization of the doping level, size, and distribution density of Si QDs. Figure 4b shows that the device with only graphene in the channel has a Dirac point (VD) at ∼39 V. After Si QDs are placed atop graphene, VD is shifted to ∼9 V. This implies the electron transfer from Si QDs to graphene, as schematically shown in Figure 4c. The Si QDs induced redshift of the 2D and G peaks of graphene in the Raman spectra (Figure 2g) actually also indicates this electron transfer.47 Moreover, our ultraviolet photoelectron spectroscopy (UPS) study shows that the work functions of Si QDs and graphene are −4.62 and −4.78 eV, respectively, further validating the electron transfer from Si QDs to graphene. After the Fermi level of Si QDs becomes the same as that of graphene, the valence band maximum (VBM) and conduction band minimum (CBM) of Si QDs bend upward at the Si-QD/graphene interface (Figure 4d). When band-to-band or sub-bandgap electron transition occurs in Si QDs under the UV-to-NIR illumination, the resulting photogenerated holes in Si QDs readily transfer to graphene because of the upward bending of the energy bands at the Si-QD/ graphene interface (Figure 4d). Therefore, the transfer of photogenerated holes from Si QDs to graphene makes VD shift to a more positive value, as representatively shown in Figure 4b for the illumination at λ = 532 nm. It is seen that the highest device responsivity is obtained at λ = 532 nm in the UV to NIR region (Figure 3a,b). This indicates that the absorption depth of the Si-QD film at λ = 532 nm is rather comparable with the current Si-QD film thickness (∼105 nm). As λ decreases toward 375 nm, the optical absorption of the Si-QD film mainly occurs in the near-surface region of the Si-QD film. The photogenerated holes may not efficiently transfer to the
(Figure 3f). The rather small values of NEP indicate that the SiQD/graphene photodetectors may be well used for weak light detection. By using D* = √S/NEP (S is the area of the device), we have calculated the specific detectivity of D*, which is also shown in Figure 3f. The values of D* are on the orders of 105 and 1013 Jones for the MIR photodetection and the UV-to-NIR photodetection, respectively. This signifies the excellent photodetection sensitivity of the Si-QD/graphene phototransistor. LSPR and Photogating. To elucidate the mechanism for the photoresponse of Si-QD/graphene phototransistors in the MIR region, we have carried out finite-difference time-domain (FDTD) simulation for the distribution of square electric field at Si QDs and the underlying graphene. Figure 4a representatively shows the results for the illumination at λ = 2.7, 3.0, and 3.6 μm. It is clear that the LSPR of Si QDs introduces intense electromagnetic fields in their proximity. The electromagnetic field beneath a Si QD is the most significantly enhanced when the illumination is at λ = 3.0 μm. The enhanced electromagnetic fields would significantly increase the optical absorption of graphene, resulting in a prominent photoresponse.45,46 The highest responsivity obtained at λ = 3.0 μm (Figure 3b) is consistent with the most intense LSPR of Si QDs at λ = 3.0 μm (Figure 2e). We have measured the dependence of the drain-source current density (JDS) on VG before and after the MIR illumination. It is seen that the Dirac point (VD) of graphene hardly changes with the MIR illumination, as representatively shown in Figure 4b for the illumination at λ = 3.0 μm. This means that the photoresponse in the MIR region is merely induced by the LSPR enhanced optical absorption of graphene. There is no transfer of photogenerated carriers from Si QDs to graphene. We have found that a phototransistor with the structure of undoped Si-QD/graphene in the channel has negligible photoresponse in the MIR region, similar to the grapheneonly phototransistor (Figure S4 in the Supporting Information). This is consistent with the fact that the LSPR does not occur to undoped Si QDs.26 Since it is well-known that the LSPR of Si QDs depends on their doping level,26 one may expect that the photoresponse of the Si-QD/graphene phototransistor in the MIR region should be readily tuned by the doping level of Si QDs. We have demonstrated such tuning by using Si QDs with a lower B doping level (the hole concentration of ∼3.3 × 1020 cm−3). With the decrease of the B doping level, the peak and intensity of the LSPR of Si QDs redshifts and decreases, respectively (Figure S8 in the 9859
DOI: 10.1021/acsnano.7b03569 ACS Nano 2017, 11, 9854−9862
Article
ACS Nano underlying graphene. As λ increases toward 1870 nm, however, the optical absorption in the Si-QD film is reduced because of the increased absorption depth. From the dependence of Jph on VG for the illumination at λ = 532 nm (Figure 4b), we can infer that R in the UV-to-NIR region may be also modulated by VG. This modulation is based on the VG-induced change of the Fermi level of graphene. For example, R can increase to ∼3 × 109 A/W as VG becomes −1 V (VG − VD = −16 V) for the illumination at λ = 532 nm (Figure S12 in the Supporting Information). In fact, the optical absorption of Si QDs in the UV-to-NIR region also depends on the doping level.28 For a lower B doping level (the hole concentration of ∼3.3 × 1020 cm−3), we see the attenuated photoresponse in the UV-to-NIR region (Figure S4 in the Supporting Information) consistent with the weakened Si-QD absorbance (Figures S1a in the Supporting Information). Ultrahigh Responsivity and Sensitivity. QDs and graphene are key players among materials used to fabricate high-performance photodetectors.5,39,48−54 We have compiled R and D* for QD- and graphene-based photoconductive detectors in Figure 5. It is clear that the values of R on the order of ∼10 A/W for the MIR region and those on the order of ∼109 A/W for the UV-to-NIR region obtained in the current work are all higher than previous ones. The present values of D* on the order of ∼1013 Jones for the UV-to-NIR region are at the top level compared with those reported before. Although the values of D* seriously decrease to be on the order of ∼105 Jones for the MIR region, they actually meet the photodetection requirement.36 We would like to point out that the Si-QD/graphene phototransistors now have relatively slow photoresponse dynamics, which should work well in a series of technologically important areas such as bioimaging,55 medical sensing,56 and low-light photography.40 It can be envisioned that both backgate pulses40,57 and topgate bias11 may be employed to sweep out the trapped carriers of Si QDs within an appropriate time frame in the future. This would decrease the device response time down to the order of microsecond, broadening the use of the Si-QD/graphene phototransistors toward high-speed photodetection.
of SiH4/Ar, B2H6/Ar, and Ar were set at 12, 157, and 105 sccm (standard cubic centimeter per minute), respectively. The pressure of the chamber was maintained at ∼2 Torr during the synthesis of the QDs. Plasma was generated with a 13.56 MHz power source and a matching network. The power coupled into the plasma was ∼105 W. B-doped Si QD colloid was prepared by adding as-synthesized Bdoped Si QDs into anhydrous ethanol with ultrasonication. As a homogeneous distribution of the QDs and a precise control of the film thickness are curial to the performance of photoconductive detectors, a spin coating method was used to control the thickness of Si-QD film on graphene. Multicoating and solvent evaporation were employed to obtain the appropriate thickness of Si-QD film. Si-QD Characterization. Inductively coupled plasma-atomic emission spectrometry (ICP-AES) measurements were performed by using Thermal Scientific iCAP6300. B-doped Si QDs were dissolved in a 2 mol L−1 NaOH solution and then diluted with ultrapure water to make the QDs concentration ∼10 μg/mL (10 ppm). HCl was used to neutralize the solution before the measurement. TEM measurements were performed by using JEM 2100F at an acceleration voltage of 200 kV. Samples were prepared by drop-casting diluted solutions of Bdoped Si QDs on copper grids coated with ultrathin carbon films. A UV-to-NIR spectrometer (HITACHI U4100) was employed to measure both diluted Si-QD colloid contained in a quartz cuvette with a path length of 10 mm and Si-QD films spun-coated on a quartz glass substrate at the resolution of 4 nm. FTIR measurements were carried out by using BRUKER VERTEX 80v with a resolution of 4 cm−1. Si QDs were drop-casted onto a potassium bromide (KBr) substrate for the FTIR measurements in the absorption mode. Each FTIR spectrum was collected after the solvent totally evaporated. UPS measurements were performed by using Thermo Scientific ESCALAB 250Xi with a 21.2 eV He-Ia source and a −9.8 eV bias. Samples were prepared by spin-coating Si QDs on ITO-covered glass substrates. SEM images were obtained by using a Hitachi S4800 field emission microscope at an acceleration voltage of 5 kV. Raman spectroscopy was measured by using a confocal Raman microscope (Senterra, BRUKER) and a Renishaw Invia system. The excitation was carried out with a 532 nm laser. Photodetector Fabrication. Monolayer graphene was grown by CVD using our previous standard process. Before transfer, we etched the graphene that was grown on the backside of Cu foil with O2 plasma to avoid the residue of Cu on graphene. After etching, graphene/Cu was spun-coated (with 500 rpm for 5 s and 4000 rpm for 60 s) with a PMMA (
