Enhanced photoresponse in curled graphene ribbons
Nanoscale View Article Online
PAPER
View Journal | View Issue
Published on 23 September 2013. Downloaded by Vanderbilt University on 27/02/2014 23:08:06.
Enhanced photoresponse in curled graphene ribbons† Cite this: Nanoscale, 2013, 5, 12206
Zeynab Jarrahi,‡a Yunhao Cao,‡b Tu Hong,‡b Yevgeniy S. Puzyrev,a Bin Wang,a Junhao Lin,a Alex H. Huffstutter,b Sokrates T. Pantelidesabc and Ya-Qiong Xu*ab Graphene has become one of the most promising materials for future optoelectronics due to its ultrahigh charge-carrier mobility, high light transmission, and universal absorbance in the near-infrared and visible spectral ranges. However, a zero band gap and ultrafast recombination of the photoexcited electron–hole pairs limit graphene's potential in photovoltaic generation. Recent studies have shown that hot carriers can enhance photovoltaic generation in graphene p–n junctions through the photothermoelectric effect (PTE). It is, therefore, desirable to synthesize graphene nanostructures with an intrinsic PTE-induced photocurrent response. Here we report a simple method to synthesize quasi-one dimensional (quasi-1D)
Received 31st July 2013 Accepted 14th September 2013
curled graphene ribbons (CGRs) that generate a photocurrent response with two orders of magnitude enhancement. Scanning photocurrent and photoluminescence measurements reveal that the photocurrent response is primarily attributed to the PTE and that the infrared emission may arise from
DOI: 10.1039/c3nr03988a
thermal radiation. These results offer a new way to fabricate graphene-based optoelectronics with an
www.rsc.org/nanoscale
enhanced photoresponse.
Graphene optoelectronics has become an attractive eld to both theoretical and experimental researchers. The extremely high charge-carrier mobility, optical transparency, ultrafast photoluminescence, broadband absorption, and enormous tensile strength make graphene an ideal candidate for the next generation optoelectronic devices.1–15 Enhancing the photon-to-electron conversion rate in graphene is the next step towards an efficient energy harvesting technology. Recent studies have shown that hot carriers can enhance the photocurrent response in graphene p–n junctions and graphene/FeCl3-intercalated fewlayer graphene heterostructures by the photothermoelectric effect (PTE).10–12,15 Moreover, this novel nonlocal hot-carrierassisted transport regime is expected to increase power conversion efficiency in graphene-based energy-harvesting devices.10 It is, therefore, desirable to synthesize graphene nanostructures with an intrinsic PTE-induced photocurrent response. A large number of theoretical and experimental studies have focused on modifying the mechanical structure of pristine graphene to alter its electrical and optical properties. Molecular dynamics (MD) simulations indicate that twisting a graphene nanoribbon leads to a tunable modication of the electrical structure of graphene.16–20 The ability to change the physical a
Department of Physics and Astronomy, Vanderbilt University, Nashville, TN 37235, USA. E-mail: [email protected]
b
Department of Electrical Engineering and Computer Science, Vanderbilt University, Nashville, TN 37235, USA
c
Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA
† Electronic supplementary 10.1039/c3nr03988a
information
(ESI)
‡ These authors contributed equally to this work.
12206 | Nanoscale, 2013, 5, 12206–12211
available.
See
DOI:
properties of graphene simply by varying its morphology is an attractive option that makes the move to graphene-based photovoltaic technology more viable. Various graphene structures, such as carbon nanoscrolls,21,22 crumpled graphene lms,23 and stacked graphene membranes24 have been fabricated and have displayed distinct properties from pristine graphene. However, none of the above has demonstrated an enhanced photocurrent response, a key component for future photovoltaics. In this paper, we report a practical method to synthesize quasi-one-dimensional (quasi-1D) curled graphene ribbons (CGRs) that can produce a strong photocurrent, about two orders of magnitude greater than the photocurrent generated at the contact areas in at graphene ribbon devices. We also investigated the nature of the photoresponse in free-standing CGRs via scanning photocurrent and photoluminescence microscopy (SPPM). Our experimental results show that the enhanced photocurrent response in CGRs mainly originates from the PTE, while signicant infrared emission may result from thermal radiation. Single-layer graphene (SLG) was grown on copper foil at 950 C in the presence of 35 sccm of methane and 8 sccm of hydrogen via chemical vapor deposition. Then a poly-methyl methacrylate (PMMA) layer was spin-coated on top of the graphene lm that was grown on the copper foil to hold the graphene lm. The copper foil was later removed through wet etching in iron chloride solution, and then the PMMA–graphene lm was transferred onto the pre-patterned chip, where 5 mm-wide and 5 mm-deep trenches were etched on 170 mm-thick transparent fused silica substrates via an Oxford 80 RIE and source and drain electrodes were deposited with 5 nm of Ti and
This journal is ª The Royal Society of Chemistry 2013
View Article Online
Published on 23 September 2013. Downloaded by Vanderbilt University on 27/02/2014 23:08:06.
Paper 40 nm of Pt via an e-beam evaporator. Aer the device was dried in air, the sample was heated up to 420 C within 1 min and held for 30 min in a furnace in the presence of 200 sccm argon at one atmosphere, which led to a complete evaporation of PMMA.25,26 During the evaporation, the non-suspended graphene adhered to the substrate due to the van der Waals interaction with the substrate, while the suspended part became wrinkled and crumpled and then shrank and curled into a CGR (Fig. 1a). We then separated each device by physically scratching between the electrodes with sharp needles. We also studied how different annealing conditions affected the morphology of CGRs and noticed that graphene ribbons started transforming into CGRs at 320 C, which corresponds to the PMMA's degradation and desorption temperature.25,26 However, a considerable amount of PMMA residues were still found aer annealing at this temperature, which might potentially inuence graphene properties. The threshold temperature at which PMMA evaporated completely was 420 C25,26 at which a CGR tended to have a narrow width in the curled multi-layer region. These experimental results indicate that the PMMA evaporation process lead to the formation of CGRs. To explore this process, we performed MD simulations on a 1 mm-long and 0.1 mm-wide graphene ribbon containing 3 840 000 carbon atoms. Each end (10 nm) of the ribbon (shown in blue in
Fig. 1 Structures and Raman spectra of CGRs. (a) SEM image of a CGR device after annealing. The CGR was suspended across a 5 mm-wide and 5 mm-deep trench on a 170 mm-thick transparent fused silica substrate. The white arrows specify the spots where Raman spectroscopy was performed. The scale bar is 1 mm. (b) A MD simulation cell with a restoring elastic force F ¼ kx at each end and random momenta given to randomly selected regions (red dots) along the ribbon (left) and a CGR (right) resulted from the ribbon on the left side. (c) TEM image of a CGR. The scale bar is 200 nm. The inset shows a close-up image of the curled area with a scale bar of 50 nm. (d) Raman spectra of six different regions along the CGR in (a) at 532 nm. The 2D-to-G intensity ratios are greater than 1 in the regions R1, R2, R5 and R6, indicating the presence of a single layer graphene membrane. The broad 2D bands in the regions R3 and R4 may result from the interlayer interactions between different graphene layers within the CGR.
This journal is ª The Royal Society of Chemistry 2013
Nanoscale Fig. 1b le) was kept at with a restoring elastic force F ¼ kx, according to the experimental conditions. Random momenta were given to a randomly selected group of atoms (shown in red dots in Fig. 1b le and Fig. S2† top) to simulate the impulses given by PMMA thermal desorption, while the total momentum remained zero. The calculated morphology (Fig. 1b right) is in good agreement with the TEM image of a CGR (Fig. 1c), where we found that the CGR shows a single layer near the edge area and has a multi-layer structure in the central region (Fig. 1c inset). The MD simulation results indicate that random momenta produced during the simulated PMMA evaporation process induce the formation of CGRs. Raman spectroscopy was performed to inspect the resonance from the curled structure. We focused the 532 nm laser on six
Fig. 2 Optoelectronic response comparison between a CGR and a flat graphene ribbon. (a) Schematic diagram of the device geometry. A CGR is suspended across a 5 mm-deep and 5 mm-wide trench on a 170 mm-thick transparent fused silica substrate. Source and drain electrodes are used to apply a voltage across the CGR and the third electrode is used as an electrolyte gate. A diffraction-limited laser spot (1). The CGR device still showed a similar behavior at R2 and R5 except a subtle shi of peak positions, possibly induced by the wrinkled and crumpled structures. As we moved further into the regions R3 and R4, where the graphene curled structure was formed, the 2D-to-G intensity ratio decreased substantially. The 2D bands also became broadened and asymmetrical, indicating that more scattering cycles were involved during the secondorder double resonance, which may result from the interlayer interaction between different graphene layers within the CGR. To evaluate the photon-to-electron conversion efficiency of CGRs, we performed spatially resolved scanning photocurrent measurements on a suspended CGR via SPPM in comparison with a at graphene ribbon. Fig. 2a presents a schematic diagram of a CGR device used in this study. When a diffractionlimited continuous-wave laser spot (1.2 mW, 785 nm) scanned over a CGR transistor suspended on the top of a 170 mm-thick transparent fused silica substrate, the photocurrent signals were collected via a preamplier and the refection image was recorded through a photodetector. As shown in Fig. 2e, the photocurrent generated along a CGR is in the range of tens of nA, about two orders of magnitude greater than that generated at graphene–metal contacts in a suspended at graphene ribbon transistor (Fig. 2d) and in non-suspended at graphene ribbon transistors reported previously.7,10 In order to investigate the photocurrent generation mechanisms in CGRs, we performed gate-dependent scanning photocurrent measurements on CGR transistors. A freestanding CGR was sealed into a microuidic chamber lled with 1.5 mM phosphate buffered saline solution and a gold electrode was used to change the electrochemical potential of the system. The chamber was kept in a steady stream to ensure a homogeneous concentration of ions. Fig. 3a displays the SEM image of a CGR projected on its corresponding reection image; the photocurrent image of this CGR is shown in Fig. 3b, which was taken at a zero source–drain bias with a gate voltage Vg ¼ 1.9 V. By sweeping the gate voltage from a value smaller than Vdirac (Vdirac ¼ 1.62 V represents the Dirac point of this device) to larger than Vdirac while recording the photocurrent along the CGR, we obtained the gate-dependent scanning photocurrent map (Fig. 3c). Three regions (R1, R2, and R3) along the CGR were selected to study their photovoltage signal (Vpc ¼ IpcR) evolution as a function of the sweeping gate voltage. As shown in Fig. 3d, the photovoltage signals in the region R3 (R1) exhibit strong nonmonotonic gate voltage dependence and have a similar pattern to the calculated thermoelectric power (S, Fig. 3d, bottom), which may result from the PTE. However, the photovoltage response in the region R2 shows monotonic gate voltage dependence, indicating that the photovoltaic effect (PVE) that results from the built-in electric eld, plays an important role in
12208 | Nanoscale, 2013, 5, 12206–12211
Paper
Fig. 3 Photocurrent responses of a CGR device. (a) A SEM image of a CGR device projected on the corresponding reflection image. The scale bar is 1 mm. (b) The corresponding photocurrent image at Vg ¼ 1.9 V and a zero source–drain bias. The laser scanning position is indicated by the green dotted line. (c) The gatedependent scanning photocurrent image as Vg varying from 1.4 V to 2.0 V. (d) The horizontal cuts along the dotted lines for different regions (R1, R2, and R3) in the CGR as specified in the photocurrent images. The bottom curve shows the calculated Seebeck coefficient in the R3 region. (e) Conductance measurement of the CGR device as a function of Vg. The flowing directions for different major carriers are illustrated in the inset diagrams.
its photovoltage generation. It is therefore necessary to consider both PVE and PTE in the photoresponse generation in CGRs, which can be expressed as ð h VPC ¼ VPVE þ VPTE ¼ nx evV þ SðxÞvTe ðxÞ dx (1) sðnÞ where s(n) is the local conductivity at the curled area, Te is electron temperature, h and nx are the mobility and the density of the photoexcited carriers, respectively, and S is the Seebeck coefficient. According to the Mott relation,
This journal is ª The Royal Society of Chemistry 2013
View Article Online
Paper
Nanoscale
Published on 23 September 2013. Downloaded by Vanderbilt University on 27/02/2014 23:08:06.
S¼
p2 kb 2 Te dðln sÞ dVg | dVg dE E¼EF 3e
(2)
where kb is the Boltzmann constant, e is the electronic charge, and EF is the Fermi energy. d(ln s)/dVg derived from the conductance measurement plays a key role in the photovoltage generated from the PTE, whereas the contribution of PVE largely depends on the local potential gradient DV. As discussed previously, the relatively low 2D-to-G intensity ratio and the broad 2D bands in the curled regions (Fig. 1d) may result from the interlayer interactions between graphene planes, which may lead to an increase of the density of states (DOS) in the curled multi-layer region of a CGR. It is well-known that the interlayer interaction can induce a parabolic dispersion of the energy bands in multi-layer graphene as compared to a linear dispersion of the energy bands in SLG.27–29 As a result of the Fermi level alignment, the Dirac point of a CGR is higher than that of a SLG, leading to the formation of a built-in electric eld. As illustrated in Fig. 3e, for the n-doped graphene, the photoexcited electrons ow from the CGR to the SLG due to the built-in electric eld. However, according to the second law of thermodynamics, the hot carriers induced by PTE tend to diffuse to the regions with larger DOS to maximize the entropy, leading to the electron ow from the SLG to the CGR. In the region R1 (R3), electrons ow from the SLG to the CGR and produce a negative (positive) photocurrent, which mainly results from the PTE (Fig. 3d). In the highly curled region R2, the contribution of PVE increases, which overwhelms the PTEinduced electron ow and produces a negative current with the present experimental setup. This indicates that the photocurrent generation depends on the local morphology of a graphene structure. We, therefore, investigated how the local morphology of a CGR affects its photocurrent generation. In order to eliminate the interference from the electrodes, we picked CGRs located far
Fig. 4 Photocurrent response of CGRs. (a) and (d) SEM images of suspended CGRs. The red circles specify the “junction” areas. The scale bars are 1 mm. (b) and (e) the corresponding photocurrent images of CGRs. (c) and (f) Line-cuts from the photocurrent images along the CGR devices as marked by the red dotted lines. The solid arrows and the dashed arrows refer to the contributions from PVE and PTE, respectively. Blue color represents the negative current in the present experimental setup and red color corresponds to the positive current.
This journal is ª The Royal Society of Chemistry 2013
away from the electrodes (SLG hence acted as the interconnections between CGRs and electrodes). The SLG membranes used in the experiments were shown to be p-doped aer growth and transfer processes. Fig. 4b shows a CGR device that exhibits a pronounced negative (positive) photocurrent response in the region R1 (R2). Fig. 4c is the extracted cut of the photocurrent image that illustrates the current changes along the suspended CGR. The PTE-induced photocurrent might be responsible for the negative (positive) current in R1 (R2), resulting from the hole injection from the SLG side at the trench edge to the curled area in the middle of the CGR. The symmetric current change indicates that this CGR device might uniformly curl from the trench edge to the middle of the CGR (the junction between regions R1 and R2). Unlike the previous device, the CGR device in Fig. 4e shows a dominant negative photocurrent response which may result from the shi of the highly curled region from the middle to the right side of the CGR (the junction between regions R3 and R4). As illustrated in Fig. 4a and 4d, we found that the dividing point of the photocurrent was strongly correlated with the junction-like structures (in red circles), where large morphology variance could be seen. Depending on the curling degree, different strengths of interlayer interactions may take place in those junctions, which may be responsible for the photocurrent spreading from 40 nA to 100 nA in different CGRs. Photoexcited electron–hole pair relaxation occurs mainly in two pathways: photocurrent and photo-induced emission. We therefore investigated the simultaneous photocurrent response and emission of CGRs (Fig. 5a) and explored their relationships to the incident laser power. As shown in the insets of Fig. 5b and 5c CGR has not only an enhanced photocurrent response, but
Fig. 5 Photocurrent and infrared emission intensities of a CGR versus incident laser power. (a) Schematic diagram of scanning photocurrent and photoluminescence microscopy (SPPM). (b) Photocurrent intensities increase linearly with incident laser power. Inset: the photocurrent image of a p-doped CGR. (c) Infrared emission intensities show a non-linear relationship with incident laser power. Inset: the emission image of the corresponding CGR. White dashed lines correspond to the edges of the trench. The scale bars represent 2 mm.
Nanoscale, 2013, 5, 12206–12211 | 12209
View Article Online
Published on 23 September 2013. Downloaded by Vanderbilt University on 27/02/2014 23:08:06.
Nanoscale also strong infrared emission. When the laser power increases from 0.8 mW to 1.9 mW, the photocurrent intensities of a CGR show a linear relationship with the incident laser power (Fig. 5b). A higher incident laser power generates more hot carriers; the photocurrent intensity is thus proportional to the laser power until saturation. Fig. 5c shows that the infrared emission intensity increases nonlinearly when the incident power rises, which is likely due to thermal radiation. When hot carriers relax in CGRs, electronic energy is transformed into Joule heat. Since the CGRs are suspended above the trench, a small fraction of heat can dissipate into the substrate.30 Therefore, the majority of the heat either dissipates into the metal contacts31,32 or radiates into free space as a grey body.33,34 At an elevated laser power, Umklapp scattering reduces the thermal conductivity of CGRs and decreases the heat transfer to metal contacts.32,35 Thus, the thermal emission into free space plays an important role in heat dissipation of the CGRs. The total amount of infrared emission is modeled by the Stefan–Boltzmann law I f Tn+1, where T is the temperature and n is the dimension of the material. Fitting of the power dependence reveals that the emission intensity I is proportional to PL5.1, where PL is the incident laser power (Fig. 5c). On the other hand, T f PL2.1 for suspended graphene membranes in air as shown in the previous literature (Fig. S4†).36 Therefore, we nd that n is about 1.4, which indicates that CGRs are quasi-1D materials.
Conclusions We have developed a simple method to synthesize free-standing quasi-1D CGRs and achieved two orders of magnitude enhancement of the photocurrent response in CGRs. Simultaneous photocurrent and photoluminescence measurements indicate that the photocurrent signals mainly result from the PTE and infrared emission may arise from thermal radiation. The capability of changing electrical and optical properties of graphene by modifying its morphology may provide a new way to build future graphene-based photovoltaics.
Methods
Paper laser beam (l ¼ 785 nm) was expanded and focused onto a diffraction-limited laser spot (
PAPER
View Journal | View Issue
Published on 23 September 2013. Downloaded by Vanderbilt University on 27/02/2014 23:08:06.
Enhanced photoresponse in curled graphene ribbons† Cite this: Nanoscale, 2013, 5, 12206
Zeynab Jarrahi,‡a Yunhao Cao,‡b Tu Hong,‡b Yevgeniy S. Puzyrev,a Bin Wang,a Junhao Lin,a Alex H. Huffstutter,b Sokrates T. Pantelidesabc and Ya-Qiong Xu*ab Graphene has become one of the most promising materials for future optoelectronics due to its ultrahigh charge-carrier mobility, high light transmission, and universal absorbance in the near-infrared and visible spectral ranges. However, a zero band gap and ultrafast recombination of the photoexcited electron–hole pairs limit graphene's potential in photovoltaic generation. Recent studies have shown that hot carriers can enhance photovoltaic generation in graphene p–n junctions through the photothermoelectric effect (PTE). It is, therefore, desirable to synthesize graphene nanostructures with an intrinsic PTE-induced photocurrent response. Here we report a simple method to synthesize quasi-one dimensional (quasi-1D)
Received 31st July 2013 Accepted 14th September 2013
curled graphene ribbons (CGRs) that generate a photocurrent response with two orders of magnitude enhancement. Scanning photocurrent and photoluminescence measurements reveal that the photocurrent response is primarily attributed to the PTE and that the infrared emission may arise from
DOI: 10.1039/c3nr03988a
thermal radiation. These results offer a new way to fabricate graphene-based optoelectronics with an
www.rsc.org/nanoscale
enhanced photoresponse.
Graphene optoelectronics has become an attractive eld to both theoretical and experimental researchers. The extremely high charge-carrier mobility, optical transparency, ultrafast photoluminescence, broadband absorption, and enormous tensile strength make graphene an ideal candidate for the next generation optoelectronic devices.1–15 Enhancing the photon-to-electron conversion rate in graphene is the next step towards an efficient energy harvesting technology. Recent studies have shown that hot carriers can enhance the photocurrent response in graphene p–n junctions and graphene/FeCl3-intercalated fewlayer graphene heterostructures by the photothermoelectric effect (PTE).10–12,15 Moreover, this novel nonlocal hot-carrierassisted transport regime is expected to increase power conversion efficiency in graphene-based energy-harvesting devices.10 It is, therefore, desirable to synthesize graphene nanostructures with an intrinsic PTE-induced photocurrent response. A large number of theoretical and experimental studies have focused on modifying the mechanical structure of pristine graphene to alter its electrical and optical properties. Molecular dynamics (MD) simulations indicate that twisting a graphene nanoribbon leads to a tunable modication of the electrical structure of graphene.16–20 The ability to change the physical a
Department of Physics and Astronomy, Vanderbilt University, Nashville, TN 37235, USA. E-mail: [email protected]
b
Department of Electrical Engineering and Computer Science, Vanderbilt University, Nashville, TN 37235, USA
c
Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA
† Electronic supplementary 10.1039/c3nr03988a
information
(ESI)
‡ These authors contributed equally to this work.
12206 | Nanoscale, 2013, 5, 12206–12211
available.
See
DOI:
properties of graphene simply by varying its morphology is an attractive option that makes the move to graphene-based photovoltaic technology more viable. Various graphene structures, such as carbon nanoscrolls,21,22 crumpled graphene lms,23 and stacked graphene membranes24 have been fabricated and have displayed distinct properties from pristine graphene. However, none of the above has demonstrated an enhanced photocurrent response, a key component for future photovoltaics. In this paper, we report a practical method to synthesize quasi-one-dimensional (quasi-1D) curled graphene ribbons (CGRs) that can produce a strong photocurrent, about two orders of magnitude greater than the photocurrent generated at the contact areas in at graphene ribbon devices. We also investigated the nature of the photoresponse in free-standing CGRs via scanning photocurrent and photoluminescence microscopy (SPPM). Our experimental results show that the enhanced photocurrent response in CGRs mainly originates from the PTE, while signicant infrared emission may result from thermal radiation. Single-layer graphene (SLG) was grown on copper foil at 950 C in the presence of 35 sccm of methane and 8 sccm of hydrogen via chemical vapor deposition. Then a poly-methyl methacrylate (PMMA) layer was spin-coated on top of the graphene lm that was grown on the copper foil to hold the graphene lm. The copper foil was later removed through wet etching in iron chloride solution, and then the PMMA–graphene lm was transferred onto the pre-patterned chip, where 5 mm-wide and 5 mm-deep trenches were etched on 170 mm-thick transparent fused silica substrates via an Oxford 80 RIE and source and drain electrodes were deposited with 5 nm of Ti and
This journal is ª The Royal Society of Chemistry 2013
View Article Online
Published on 23 September 2013. Downloaded by Vanderbilt University on 27/02/2014 23:08:06.
Paper 40 nm of Pt via an e-beam evaporator. Aer the device was dried in air, the sample was heated up to 420 C within 1 min and held for 30 min in a furnace in the presence of 200 sccm argon at one atmosphere, which led to a complete evaporation of PMMA.25,26 During the evaporation, the non-suspended graphene adhered to the substrate due to the van der Waals interaction with the substrate, while the suspended part became wrinkled and crumpled and then shrank and curled into a CGR (Fig. 1a). We then separated each device by physically scratching between the electrodes with sharp needles. We also studied how different annealing conditions affected the morphology of CGRs and noticed that graphene ribbons started transforming into CGRs at 320 C, which corresponds to the PMMA's degradation and desorption temperature.25,26 However, a considerable amount of PMMA residues were still found aer annealing at this temperature, which might potentially inuence graphene properties. The threshold temperature at which PMMA evaporated completely was 420 C25,26 at which a CGR tended to have a narrow width in the curled multi-layer region. These experimental results indicate that the PMMA evaporation process lead to the formation of CGRs. To explore this process, we performed MD simulations on a 1 mm-long and 0.1 mm-wide graphene ribbon containing 3 840 000 carbon atoms. Each end (10 nm) of the ribbon (shown in blue in
Fig. 1 Structures and Raman spectra of CGRs. (a) SEM image of a CGR device after annealing. The CGR was suspended across a 5 mm-wide and 5 mm-deep trench on a 170 mm-thick transparent fused silica substrate. The white arrows specify the spots where Raman spectroscopy was performed. The scale bar is 1 mm. (b) A MD simulation cell with a restoring elastic force F ¼ kx at each end and random momenta given to randomly selected regions (red dots) along the ribbon (left) and a CGR (right) resulted from the ribbon on the left side. (c) TEM image of a CGR. The scale bar is 200 nm. The inset shows a close-up image of the curled area with a scale bar of 50 nm. (d) Raman spectra of six different regions along the CGR in (a) at 532 nm. The 2D-to-G intensity ratios are greater than 1 in the regions R1, R2, R5 and R6, indicating the presence of a single layer graphene membrane. The broad 2D bands in the regions R3 and R4 may result from the interlayer interactions between different graphene layers within the CGR.
This journal is ª The Royal Society of Chemistry 2013
Nanoscale Fig. 1b le) was kept at with a restoring elastic force F ¼ kx, according to the experimental conditions. Random momenta were given to a randomly selected group of atoms (shown in red dots in Fig. 1b le and Fig. S2† top) to simulate the impulses given by PMMA thermal desorption, while the total momentum remained zero. The calculated morphology (Fig. 1b right) is in good agreement with the TEM image of a CGR (Fig. 1c), where we found that the CGR shows a single layer near the edge area and has a multi-layer structure in the central region (Fig. 1c inset). The MD simulation results indicate that random momenta produced during the simulated PMMA evaporation process induce the formation of CGRs. Raman spectroscopy was performed to inspect the resonance from the curled structure. We focused the 532 nm laser on six
Fig. 2 Optoelectronic response comparison between a CGR and a flat graphene ribbon. (a) Schematic diagram of the device geometry. A CGR is suspended across a 5 mm-deep and 5 mm-wide trench on a 170 mm-thick transparent fused silica substrate. Source and drain electrodes are used to apply a voltage across the CGR and the third electrode is used as an electrolyte gate. A diffraction-limited laser spot (1). The CGR device still showed a similar behavior at R2 and R5 except a subtle shi of peak positions, possibly induced by the wrinkled and crumpled structures. As we moved further into the regions R3 and R4, where the graphene curled structure was formed, the 2D-to-G intensity ratio decreased substantially. The 2D bands also became broadened and asymmetrical, indicating that more scattering cycles were involved during the secondorder double resonance, which may result from the interlayer interaction between different graphene layers within the CGR. To evaluate the photon-to-electron conversion efficiency of CGRs, we performed spatially resolved scanning photocurrent measurements on a suspended CGR via SPPM in comparison with a at graphene ribbon. Fig. 2a presents a schematic diagram of a CGR device used in this study. When a diffractionlimited continuous-wave laser spot (1.2 mW, 785 nm) scanned over a CGR transistor suspended on the top of a 170 mm-thick transparent fused silica substrate, the photocurrent signals were collected via a preamplier and the refection image was recorded through a photodetector. As shown in Fig. 2e, the photocurrent generated along a CGR is in the range of tens of nA, about two orders of magnitude greater than that generated at graphene–metal contacts in a suspended at graphene ribbon transistor (Fig. 2d) and in non-suspended at graphene ribbon transistors reported previously.7,10 In order to investigate the photocurrent generation mechanisms in CGRs, we performed gate-dependent scanning photocurrent measurements on CGR transistors. A freestanding CGR was sealed into a microuidic chamber lled with 1.5 mM phosphate buffered saline solution and a gold electrode was used to change the electrochemical potential of the system. The chamber was kept in a steady stream to ensure a homogeneous concentration of ions. Fig. 3a displays the SEM image of a CGR projected on its corresponding reection image; the photocurrent image of this CGR is shown in Fig. 3b, which was taken at a zero source–drain bias with a gate voltage Vg ¼ 1.9 V. By sweeping the gate voltage from a value smaller than Vdirac (Vdirac ¼ 1.62 V represents the Dirac point of this device) to larger than Vdirac while recording the photocurrent along the CGR, we obtained the gate-dependent scanning photocurrent map (Fig. 3c). Three regions (R1, R2, and R3) along the CGR were selected to study their photovoltage signal (Vpc ¼ IpcR) evolution as a function of the sweeping gate voltage. As shown in Fig. 3d, the photovoltage signals in the region R3 (R1) exhibit strong nonmonotonic gate voltage dependence and have a similar pattern to the calculated thermoelectric power (S, Fig. 3d, bottom), which may result from the PTE. However, the photovoltage response in the region R2 shows monotonic gate voltage dependence, indicating that the photovoltaic effect (PVE) that results from the built-in electric eld, plays an important role in
12208 | Nanoscale, 2013, 5, 12206–12211
Paper
Fig. 3 Photocurrent responses of a CGR device. (a) A SEM image of a CGR device projected on the corresponding reflection image. The scale bar is 1 mm. (b) The corresponding photocurrent image at Vg ¼ 1.9 V and a zero source–drain bias. The laser scanning position is indicated by the green dotted line. (c) The gatedependent scanning photocurrent image as Vg varying from 1.4 V to 2.0 V. (d) The horizontal cuts along the dotted lines for different regions (R1, R2, and R3) in the CGR as specified in the photocurrent images. The bottom curve shows the calculated Seebeck coefficient in the R3 region. (e) Conductance measurement of the CGR device as a function of Vg. The flowing directions for different major carriers are illustrated in the inset diagrams.
its photovoltage generation. It is therefore necessary to consider both PVE and PTE in the photoresponse generation in CGRs, which can be expressed as ð h VPC ¼ VPVE þ VPTE ¼ nx evV þ SðxÞvTe ðxÞ dx (1) sðnÞ where s(n) is the local conductivity at the curled area, Te is electron temperature, h and nx are the mobility and the density of the photoexcited carriers, respectively, and S is the Seebeck coefficient. According to the Mott relation,
This journal is ª The Royal Society of Chemistry 2013
View Article Online
Paper
Nanoscale
Published on 23 September 2013. Downloaded by Vanderbilt University on 27/02/2014 23:08:06.
S¼
p2 kb 2 Te dðln sÞ dVg | dVg dE E¼EF 3e
(2)
where kb is the Boltzmann constant, e is the electronic charge, and EF is the Fermi energy. d(ln s)/dVg derived from the conductance measurement plays a key role in the photovoltage generated from the PTE, whereas the contribution of PVE largely depends on the local potential gradient DV. As discussed previously, the relatively low 2D-to-G intensity ratio and the broad 2D bands in the curled regions (Fig. 1d) may result from the interlayer interactions between graphene planes, which may lead to an increase of the density of states (DOS) in the curled multi-layer region of a CGR. It is well-known that the interlayer interaction can induce a parabolic dispersion of the energy bands in multi-layer graphene as compared to a linear dispersion of the energy bands in SLG.27–29 As a result of the Fermi level alignment, the Dirac point of a CGR is higher than that of a SLG, leading to the formation of a built-in electric eld. As illustrated in Fig. 3e, for the n-doped graphene, the photoexcited electrons ow from the CGR to the SLG due to the built-in electric eld. However, according to the second law of thermodynamics, the hot carriers induced by PTE tend to diffuse to the regions with larger DOS to maximize the entropy, leading to the electron ow from the SLG to the CGR. In the region R1 (R3), electrons ow from the SLG to the CGR and produce a negative (positive) photocurrent, which mainly results from the PTE (Fig. 3d). In the highly curled region R2, the contribution of PVE increases, which overwhelms the PTEinduced electron ow and produces a negative current with the present experimental setup. This indicates that the photocurrent generation depends on the local morphology of a graphene structure. We, therefore, investigated how the local morphology of a CGR affects its photocurrent generation. In order to eliminate the interference from the electrodes, we picked CGRs located far
Fig. 4 Photocurrent response of CGRs. (a) and (d) SEM images of suspended CGRs. The red circles specify the “junction” areas. The scale bars are 1 mm. (b) and (e) the corresponding photocurrent images of CGRs. (c) and (f) Line-cuts from the photocurrent images along the CGR devices as marked by the red dotted lines. The solid arrows and the dashed arrows refer to the contributions from PVE and PTE, respectively. Blue color represents the negative current in the present experimental setup and red color corresponds to the positive current.
This journal is ª The Royal Society of Chemistry 2013
away from the electrodes (SLG hence acted as the interconnections between CGRs and electrodes). The SLG membranes used in the experiments were shown to be p-doped aer growth and transfer processes. Fig. 4b shows a CGR device that exhibits a pronounced negative (positive) photocurrent response in the region R1 (R2). Fig. 4c is the extracted cut of the photocurrent image that illustrates the current changes along the suspended CGR. The PTE-induced photocurrent might be responsible for the negative (positive) current in R1 (R2), resulting from the hole injection from the SLG side at the trench edge to the curled area in the middle of the CGR. The symmetric current change indicates that this CGR device might uniformly curl from the trench edge to the middle of the CGR (the junction between regions R1 and R2). Unlike the previous device, the CGR device in Fig. 4e shows a dominant negative photocurrent response which may result from the shi of the highly curled region from the middle to the right side of the CGR (the junction between regions R3 and R4). As illustrated in Fig. 4a and 4d, we found that the dividing point of the photocurrent was strongly correlated with the junction-like structures (in red circles), where large morphology variance could be seen. Depending on the curling degree, different strengths of interlayer interactions may take place in those junctions, which may be responsible for the photocurrent spreading from 40 nA to 100 nA in different CGRs. Photoexcited electron–hole pair relaxation occurs mainly in two pathways: photocurrent and photo-induced emission. We therefore investigated the simultaneous photocurrent response and emission of CGRs (Fig. 5a) and explored their relationships to the incident laser power. As shown in the insets of Fig. 5b and 5c CGR has not only an enhanced photocurrent response, but
Fig. 5 Photocurrent and infrared emission intensities of a CGR versus incident laser power. (a) Schematic diagram of scanning photocurrent and photoluminescence microscopy (SPPM). (b) Photocurrent intensities increase linearly with incident laser power. Inset: the photocurrent image of a p-doped CGR. (c) Infrared emission intensities show a non-linear relationship with incident laser power. Inset: the emission image of the corresponding CGR. White dashed lines correspond to the edges of the trench. The scale bars represent 2 mm.
Nanoscale, 2013, 5, 12206–12211 | 12209
View Article Online
Published on 23 September 2013. Downloaded by Vanderbilt University on 27/02/2014 23:08:06.
Nanoscale also strong infrared emission. When the laser power increases from 0.8 mW to 1.9 mW, the photocurrent intensities of a CGR show a linear relationship with the incident laser power (Fig. 5b). A higher incident laser power generates more hot carriers; the photocurrent intensity is thus proportional to the laser power until saturation. Fig. 5c shows that the infrared emission intensity increases nonlinearly when the incident power rises, which is likely due to thermal radiation. When hot carriers relax in CGRs, electronic energy is transformed into Joule heat. Since the CGRs are suspended above the trench, a small fraction of heat can dissipate into the substrate.30 Therefore, the majority of the heat either dissipates into the metal contacts31,32 or radiates into free space as a grey body.33,34 At an elevated laser power, Umklapp scattering reduces the thermal conductivity of CGRs and decreases the heat transfer to metal contacts.32,35 Thus, the thermal emission into free space plays an important role in heat dissipation of the CGRs. The total amount of infrared emission is modeled by the Stefan–Boltzmann law I f Tn+1, where T is the temperature and n is the dimension of the material. Fitting of the power dependence reveals that the emission intensity I is proportional to PL5.1, where PL is the incident laser power (Fig. 5c). On the other hand, T f PL2.1 for suspended graphene membranes in air as shown in the previous literature (Fig. S4†).36 Therefore, we nd that n is about 1.4, which indicates that CGRs are quasi-1D materials.
Conclusions We have developed a simple method to synthesize free-standing quasi-1D CGRs and achieved two orders of magnitude enhancement of the photocurrent response in CGRs. Simultaneous photocurrent and photoluminescence measurements indicate that the photocurrent signals mainly result from the PTE and infrared emission may arise from thermal radiation. The capability of changing electrical and optical properties of graphene by modifying its morphology may provide a new way to build future graphene-based photovoltaics.
Methods
Paper laser beam (l ¼ 785 nm) was expanded and focused onto a diffraction-limited laser spot (
