Hot Electron-Based Near-Infrared Photodetection Using Bilayer MoS2

Oct 1, 2015 - The experimentally measured photocurrent spectrum is shown in Figure 2b for source–drain biases of 0.6, −0.6, and 0 V (red, blue, an...
3 downloads 5 Views 969KB Size
Subscriber access provided by University of Otago Library

Communication 2

Hot Electron-based Near-infrared Photodetection Using Bilayer MoS Wenyi Wang, Andrey Klots, Dhiraj Prasai, Yuanmu Yang, Kirill Bolotin, and Jason Valentine Nano Lett., Just Accepted Manuscript • DOI: 10.1021/acs.nanolett.5b02866 • Publication Date (Web): 01 Oct 2015 Downloaded from http://pubs.acs.org on October 5, 2015

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Nano Letters is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 19

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

Hot Electron-based Near-infrared Photodetection Using Bilayer MoS2 Wenyi Wang1, Andrey Klots2, Dhiraj Prasai3, Yuanmu Yang3, Kirill I. Bolotin2, Jason Valentine4 1

Department of Electrical Engineering and Computer Science, Vanderbilt University, Nashville, Tennessee 37212, USA 2 Department of Physics and Astronomy, Vanderbilt University, Nashville, Tennessee 37235, USA 3 Interdisciplinary Graduate Program in Materials Science, Vanderbilt University, Nashville, Tennessee, 37235, USA 4 Department of Mechanical Engineering, Vanderbilt University, Nashville, Tennessee 37212, USA

ABSTRACT Recently, there has been much interest in the extraction of hot electrons generated from surface plasmon decay as this process can be used to achieve additional bandwidth for both photodetectors and photovoltaics. Hot electrons are typically injected into semiconductors over a Schottky barrier between the metal and semiconductor, enabling generation of photocurrent with below bandgap photon illumination. As a twodimensional semiconductor single and few layer molybdenum disulfide (MoS2) has been demonstrated to exhibit internal photogain and therefore becomes an attractive hot electron acceptor. Here, we investigate hot electron-based photodetection in a device consisting of bilayer MoS2 integrated with a plasmonic antenna array. We demonstrate sub-bandgap photocurrent originating from the injection of hot electrons into MoS2 as well as photoamplification that yields a photogain of 105. The large photogain results in a photoresponsivity of 5.2 A/W, which is far above similar silicon-based hot electron photodetectors in which no photoamplification is present. This technique is expected to

ACS Paragon Plus Environment

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 19

have potential use in future ultra-compact near-infrared photodetection and optical memory devices. KEYWORDS: Plasmon resonance, hot electron injection, bilayer MoS2, near infrared photodetection MANUSCRIPT Due to their ability to confine and trap light in deep subwavelength volumes, surface plasmons (SPs) provide manipulation of light at the nanoscale and have been harnessed

in

numerous

techniques

including

ultra-compact

modulators

and

photodetectors1,2, plasmonic enhanced photovoltaic cells3, and subwavelength imaging and lithography1,4. Surface plasmons can decay either radiatively into photons or nonradiatively into energetic hot electrons5. If not harnessed, these electrons will eventually dissipate their energy through thermalization. However, recent studies have demonstrated that hot electrons can be extracted before thermalization to participate in chemical reactions6–8 or be injected into semiconductors for generating photocurrent9,10. The injection of hot electrons into semiconductors is typically achieved using a Schottky interface between a metal and the semiconductor11, enabling the harvesting of photons with energy lower than the band gap, providing additional bandwidth and spectral tunability12. In addition to the demonstration of hot electron injection into bulk semiconductors such as silicon9,10,12 and TiO26, the injection of hot electrons into twodimensional (2D) materials such as graphene13 and MoS214,15 is also attractive due to their unique optoelectrical properties16,17. In particular, MoS2 is an attractive hot electron acceptor due to its semiconductor electronic band structure and internal photogain owing to the various traps at the metal/MoS2 and/or MoS2/substrate interfaces18,19. Using these

ACS Paragon Plus Environment

Page 3 of 19

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

mechanisms, ultrasensitive photodetectors based on direct absorption of photons in monolayer MoS2 have been realized with responsivity reaching 880 A/W19 and a photogain exceeding 104. In this letter, we perform the first extensive investigation on the injection of hot electron into a bilayer MoS2 film that has been integrated with plasmonic resonators. The surface plasmon resonance in the Au nanostructures is excited by near infrared (NIR) light and will generate hot electrons via non-radiative decay. Provided the energy of the hot electrons are greater than the Schottky barrier between Au and MoS2 they can be injected into the bilayer MoS2 generating photocurrent. With an asymmetric plasmonic structure, we demonstrate sub-bandgap photoresponse and a photocurrent spectrum that varies with the bias polarity, features that are explained by hot electron injection and trap induced photoamplification. We verify the presence of hot electron injection by examining control devices in which an injection barrier layer is used to distinguish the relative contribution of hot electron injection and photothermoelectric20,21 effects. The hot electron injection efficiency is estimated to be comparable with Si-based hot electron photodetectors9 while an photocurrent amplification factor of 105 results in a photoresponsivity of 5.2 A/W under 1070 nm illumination. To begin, we examined the photocurrent from an interdigitated asymmetric plasmonic structure consisting of resonant wires and non-resonant wires, as is shown in Figure 1a. The non-resonant wires (NRWs, color coded with green for clarity) are narrow (w2 =80 nm) such that the resonance modes are outside the wavelength range of the measurement, while the resonant wires (RWs, color coded with yellow) have antennas with a dipolar resonance at 1250nm (a =280 nm, b =125 nm, p =340 nm, w1 =90 nm)

ACS Paragon Plus Environment

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

when illuminated with the electric field oriented along the y-direction (Ey, see Figure 1). The illumination is at a much lower energy than the indirect bandgap of bilayer MoS2 which is at 750 nm (1.65 eV)22. As such, direct photon absorption in MoS2 does not occur. Compared with symmetric structures, the asymmetric structure allows us to generate photocurrent in the unbiased case despite the fact that the device has symmetric Schottky barriers on the source and drain electrodes. In this case, current is generated due to the fact that the resonant wires generate far more electrons than the non-resonant wires. During device fabrication Au structures are first defined on a piranha (3:1 H2SO4:H2O2) cleaned quartz substrate using electron beam lithography (EBL) and deposition of 2 nm Ti and 15 nm Au films. Thick Au contacts were then fabricated using a second lithography step. The device was cleaned for a second time with O2 plasma (30s) and piranha (1 min) to ensure a pristine Au surface. Bilayer MoS2 that was mechanically exfoliated on a PDMS template was then transferred on top of the Au structures using a stamping method23,24, more details regarding the transfer process can be found in Section I of the Supporting Information. During fabrication, we found that the device yield when utilizing bilayer MoS2 was higher than that for monolayer devices, which is likely due to the higher stiffness of the bilayer films25. Because of that, only bilayer MoS2 devices are examined here, although we believe that injection of hot electrons and photocurrent amplification should also be observable in monolayer MoS2 due to their similar electronic band structures. After annealing in a H2/Ar environment at 160 ºC for 30 min to further remove any polymer residue the device was wire bonded and loaded into a vacuum chamber for measurements.

ACS Paragon Plus Environment

Page 4 of 19

Page 5 of 19

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

The experimental and full wave simulations of the absorption spectra of the entire structure are shown in Figure 2a and the simulated absorption in the individual components (RWs and NRWs) are shown with dotted lines. It can be seen that the RW component exhibits a strong resonance under Ey polarized excitation as is evident in the electrical field profile (|E|) shown in the inset, while there is no such resonance in the NRW component. The experimentally measured photocurrent spectrum is shown in Figure 2b for source-drain biases of 0.6V, -0.6V, and 0V (red, blue and green dots). These measurements were obtained with a chopper operating at 387 Hz and a lock-in amplifier. As can be observed, the photoresponsivity peaks at 1120 nm for 0.6V bias but disappears when the bias is -0.6V. This observation can be understood by the fact that for Ey polarized excitation the RW components dominate the hot electron generation due to their higher optical absorption. As illustrated with energy band diagrams in Figure 2c to 2e, at 0.6V bias only the hot electrons in the RWs are injected into MoS2 (Figure 2c), resulting in the responsivity peaking at the RW resonance. For the case of -0.6V bias only hot electrons generated in the NRWs are injected (Figure 2d) and thus the photoresponse spectrum shows no resonant feature. When there is no bias (0V), hot electrons in both RWs and NRWs are injected in opposite directions, therefore the net photocurrent will be proportional to the absorption difference between the RW and NRW components (Figure 2e). It should be noted that the 0V and ±0.6V bias conditions also exhibit different gain levels, which will be discussed later, leading to much lower photocurrent in the un-biased condition. The spectrum of the responsivity, R i ( ω ) , with the superscript i denoting the three cases of Vsd (0.6V, -0.6V and 0V), can be understood with the following expression,

ACS Paragon Plus Environment

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 19

 ( hω − qφB ) 2  i R i (ω ) = C f   k1α RW , Ld (ω ) + k2α NRW , Ld (ω )  g ω h  

(1)

Here, the first bracket is Fowler’s formula, which describes the internal quantum is the photon energy, qφB =0.67 eV is

efficiency of the photoemission process26,27,

the Schottky barrier height between Au and bilayer MoS228 and C f is a device specific constant that is treated as a universal fitting constant for the three source-drain voltages here. The second bracket represents the probability of a hot electron being generated and transported to the interface, where α RW , L and α NRW , L are the absorption in RW and d

d

NRW, respectively. Due to the limited hot electron diffusion length ( Ld ~20 nm29 at the electron energy considered here), hot electrons generated beyond 20 nm from the structure’s edge will not arrive at the Au/MoS2 interface, preventing injection into the channel. Therefore, only the absorption in Au that occurs within 20 nm to the edge is used in the calculation of α RW , L and α NRW , L (refer to Figure S1 for the spectrum of d

d

α RW , L and α NRW , L ). The values for ( k1 ,k2 ) are chosen to be (1, 0) for Vsd = 0.6V, (0, -1) d

d

for Vsd = -0.6V and (1, -1) for Vsd = 0V bias. The last fitting parameter g i represents a photogain that depends on the source-drain voltage and will be discussed in later sections. In Figure 2b we present the fit of equation 1 to the photocurrent spectra (lines), demonstrating good agreement with the line shape of the experimental measurements. The slight discrepancy between the experiments and theory can largely be attributed to fabrication imperfections, the uncertainty in the Schottky barrier height, and the error in estimating the electron diffusion length, Ld . Given the uncertainty in the electron diffusion length we calculated α RW , L and α NRW , L for Ld values ranging from 10 to 40 d

d

ACS Paragon Plus Environment

Page 7 of 19

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

nm (Figure S1). However, the spectral shape of α RW , L and α NRW , L vary only slightly d

d

with changing Ld values indicating the parameter is not a critical factor in achieving agreement with the experimental data. Another factor that could influence the fitting is the constant C f , which is generally a function of the wavelength and the profile of the plasmonic resonant mode. As such, it likely varies at short wavelengths where a grating mode is excited in addition to the dipolar mode in the antennas. The absorption spectrum and the photoresponse for illumination polarized along the x-direction (Ex polarization) can be found in Figure S2 of the Supporting Information. Neither RW nor NRW component responds to Ex polarization and therefore the absorption is minimal due to the lack of resonances. However, similar variations in the photocurrent magnitude under different bias polarities can still be observed although no resonance features are observed. As mentioned previously, ultrathin MoS2 can exhibit photoamlification due to traps at the interfaces18,19. To explore this effect we recorded the change of photoresponsivity as a function of source-drain voltage Vsd under illumination with a 150 nW Ey polarized 1070 nm laser (Figure 3a). The measurements show increasing responsivity with increasing Vsd and a peak responsivity of 4.5A/W at 3V bias. The responsivity was obtained by comparing the difference in source-drain current Isd under illumination and in a dark environment, as is shown in the inset of Figure 3a. The sweeping speed of Vsd was kept at 0.02V/s, and the hysteresis in the Isd under illumination indicates the existence of carrier traps in the system, which is the cause of the photocurrent amplification. To characterize the lifetime of the traps states and gain insight into the photogain mechanism and hot electron injection efficiency, we compared the response of the

ACS Paragon Plus Environment

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 19

structure at visible and infrared frequencies. Specifically, we recorded the time response of ∆Isd under a source drain bias of 0.8V and 1070 nm illumination (red curve in Figure 3b) and compared it with the well understood19 photovoltaic effect dominated photoresponse under 532 nm illumination (green curve). During the experiment, the laser was kept on for 500s and then turned off so that the entire rise and decay processes of ∆Isd were recorded. For both 1070 nm and 532 nm excitation, the photocurrent as a

function of time can be expressed as,

t ( 0 < t < 500 )  ∫ ( dn / dt)dt , ∂G ∂G  0 ∆n( t ) = Vsd I PC ( t ) = ∆I sd ( t ) = Vsd  t −500 ∂n ∂n  I t =500 − ∫ ( dn / dt)dt , ( t > 500 )  500 

(2)

where G is the conductance, Vsd ( ∂G / ∂n ) is the previously mentioned photogain ( g ),

n is the number of electrons in MoS2, and dn / dt at t