1. Introduction

Graphene is one atomic layer of honeycomb carbon lattice with electron transport exhibiting massless Dirac fermions mimicking relativistic particles [1]. It was first reported as a single-layer carbon via reducing graphite oxide by H. P. Boehm et al. in 1962 [2]. Graphene has become one of the most promising nano-materials after K. S. Novoselov, A. K. Geim, et al. demonstrated that such a 2-dimensional one-atom-thick film is stable in ambient environments with many specific and charming physical and chemical properties [3,4]. For example, it possesses a unique band structure described by Dirac equation and holds high carrier mobility up to 10⁶ cm²/Vs and the light transmittance up to 97.7%. It has, thus, been broadly investigated as a next generation electronic and optoelectronic material for many applications such as replacing the commonly used semiconductor-based transparent electrical conductive coatings based on indium tin oxide [5]. Carbon-based electronics and photonics [6] have, therefore, emerged among the most promising and studied modern sciences and technologies.

Many applications of single-layer graphene are based on its outstanding electron transport properties, which are significantly environmental dependent. Being only one atom thin, minute impurities on the graphene surface and in the interface between it and the supporting substrate might alter the electronic structure and transport significantly. This is because charge carriers in graphene are within atomic distance from these impurities. Even without impurities, electronic properties of single-layer graphene could also be modified due to the ambient environments and the properties of the supporting substrates. Light illumination could also change the electronic characteristics of graphene through electromagnetic interactions. For example, C. Biswas et al. reported that the photoconductivity of graphene on SiO₂/Si substrate switched from negative to positive under prolonged illumination by ultra violet laser (wavelengths between 200nm and 300nm) in ambient atmosphere [7]. The variation of the photoconductivity is attributed to illumination induced surface and interfacial charges by photo-ionization and desorption of surface adsorbates as well as photo-induced electron-hole pairs in silicon and their diffusion to the interfacial silicon dioxide to neutralize trapped charges [8–10]. The unique optoelectronic characteristics of graphene offer opportunities for many possible applications, such as graphene-based opto-electrical switches. More efforts are desirable to understand how the photoconductivity of graphene can be effectively controlled by illumination of light, especially visible light. In this work, photoconductivity of single-layer graphene transferred onto pristine and Ag-nanoparticle-coated SiO₂/Si substrates has been investigated. Both positive and negative photoconductivities have been measured. Effects of plasmonic coupling and electron current induced annealing of graphene on the photoconductivity are discussed based on photo-induced surface and interfacial charges, desorption of surface adsorbates, and modification of charge trapping in silicon dioxide due to photo-generated electron-hole pairs in silicon.

2. Experimental

Single-layer graphene is synthesized by thermal chemical vapor deposition (CVD) on polycrystalline copper foils of 99.8% purity in a gas mixture of H₂, and CH₄ at 1000 °C in a high temperature furnace shown schematically in Fig. 1(a) . A turbomolecular pump backed by a rotary vane pump is used to evacuate the reaction chamber made of a 3-inch diameter quartz tubing. The gas pressure is controlled by a throttle valve between the reaction chamber and the pumping system. The flow rate of each feeding gas is controlled by an electronic mass flow controller. Furnace temperature is controlled in three zones independently by precise electronic controllers with thermocouple sensors for measuring the furnace temperature in each zone. An additional thermocouple sensor is inserted inside the quartz tubing to monitor the temperature near the copper foils. Water cooled flanges are used to seal the quartz tubing. A linear vacuum feed-through was employed to shift the samples from one furnace zone to the other without breaking vacuum.

 

Fig. 1 (a) Schematic diagram representing the thermal CVD system for growing graphene. (b) A Raman spectrum of a graphene film. (c) Apparatus for measuring photoconductivity of graphene. SMU stands for source-measurement-unit while CCD stands for charge coupled devices. (d) Schematic diagram of a test device with an exposed graphene film of 20μm x 400μm in size between two gold electrodes.

Download Full Size | PDF

Graphene is analyzed by Raman spectroscopy. Figure 1(b) shows a Raman spectrum for a graphene used for the investigation. The symmetric 2D-peak with a low full width at half maximum of 34 cm⁻¹ indicates that the graphene is a single-layer graphene. The small ratio of the D peak intensity to the G peak intensity, I(D)/I(G), and the I(2D)/I(G) intensity ratio of 2.2 indicate that the graphene is of high quality [11]. The synthesized graphene is transferred from a copper foil onto pristine or Ag-nanoparticle coated silicon dioxide/silicon substrates [12]. Silicon dioxide with a thickness of 300 nm is grown thermally on silicon. The silver coating is carried out by thermal evaporation followed by thermal annealing. More details of the graphene/substrate structure, the silver coating process and the SEM images of silver nanoparticle coated SiO₂/Si substrates [13], and optical characteristics of silver nanoparticles [14] have previously been reported.

The setup for measuring photoconductivity is schematically plotted in Fig. 1(c). Lift-off process is applied to pattern two gold contact pads on the sample surface resulting in an exposed graphene surface of 20μm x 400μm in size between these two gold contact pads as shown in Fig. 1(d). Two tungsten probes connecting to a Keithley 236 source-measure unit (SMU) are used to contact the gold contact pads in the ambient atmosphere. The applied voltage by the SMU is pre-set to be either 1 V or 10 mV. Electrical current flowing through the graphene film between two gold contact pads is also measured by the SMU. A 405-nm diode laser is used to illuminate the sample surface including the contact pads. A charge coupled device (CCD) is used to confirm the illumination areas and probe positions. Several light-off and light-on cycles are preformed to investigate the corresponding electrical current responses under a constant applied voltage. More detailed information related to photoconductivity measurement is described in [14].

3. Results and discussion

Electrical current flowing through a transferred graphene film is measured at a constant applied voltage between two gold contact pads while the illuminating laser light is switched on and off. A variety of off- and on-durations ranging from 0.5 hr to 12.7 hr are listed in Table 1 to reveal the long-term evolution of the persistent photoconductivity of the graphene films.

Table 1. Duration when light illumination is off or on corresponding to data presented in Figs. 2 and 3.

View Table

Figure 2(a) shows the measured electrical current flowing through a graphene film transferred onto a pristine silicon dioxide covered silicon substrate under an applied voltage of 1 V between two gold contact pads. In the initial phase, the illumination is off, during which the electrical current gradually increases from 1.15 to 1.2 mA in a period of 1.4 hr. It is apparent that the current quickly and dramatically reduces to 1.15 mA after graphene is illuminated by the laser light. When measured electrical conductivity is lower than the dark electrical conductivity which corresponds to that before light illumination begins, it will be described as negative photoconductivity in the following discussion. When the electrical conductivity changes to a level higher than the dark electrical conductivity during the light illumination is described as positive photoconductivity. Therefore, Fig. 2(a) shows that negative photoconductivity of the graphene film has been measured upon the illumination by the blue laser. Although the electrical current gradually increases and, therefore, the electrical conductivity is gradually restored towards the original dark level, it remains below the dark level even after 2.5-hr laser illumination. Once the laser illumination is turned off, the electrical current rises quickly, further indicating the negative photoconductivity of the graphene. The dark current increases with time and eventually exceeds the dark current measured before the laser illumination begins. This rising dark current is attributed to the high electrical current, induced by the applied voltage of 1V, which anneals of graphene by Joule heating [15]. Annealing by flowing electrical current assists in desorption of surface adsorbents, the reduction of electron scattering by charged molecules on the surface of the graphene, and, thus, the increase in the electrical conductivity of graphene [7]. Charged molecules on the surface of graphene can be generated by photoionization or charge transfer processes between the surface molecules and the graphene. Some surface molecules act as dopants and contribute to the charge carrier density in the graphene. The others act as compensating dopants which reduces the charge carrier density in the graphene. Desorption of compensating dopants increases the charge carrier density in the graphene and enhances the electrical conductivity.

 

Fig. 2 Electrical current evolution of graphene on SiO₂/Si under a series of light-off and light-on cycles. The white and blue areas represent the durations of light off and on, respectively. The applied voltage in (a) is 1 V while that in (b) is 10 mV.

Download Full Size | PDF

During the second light-on cycle, the illumination is kept on for 12.5 hr, during which, the current barely recovers to the dark level. The sample still exhibits negative photoconductivity and the electrical current rises even faster after turning off the laser. Therefore, the current increment during light-on can be attributed to neutralization of trapped charges [7] in the silicon dioxide by photo-generated electron-hole pairs in the silicon and the electrical current induced annealing of graphene. The electrical current induced annealing of graphene continues to different extents according to electrical current no matter the light illumination is on or off.

In order to study the effects of electrical current induced annealing of graphene, the applied voltage between two gold contact pads is reduced by 100 times from 1V to 10mV. Figure 2(b) shows the electrical current flowing through a graphene on a pristine silicon dioxide grown on silicon substrate. Although the applied voltage is decreased by 100 times, the dark current still increases gradually following the same trend as that in Fig. 2(a) and negative photoconductivity is measured. The electrical current and the conductivity of the graphene continue to decrease with increasing illumination time in the entire first light-on cycle. The rate of decrease in photoconductivity is lower near the end of the light-on cycle implying that the graphene is approaching to a balance between the effects of net decrease in the photoconductivity due to adsorption and desorption of molecules from the ambient environments and the increasing photoconductivity due to neutralization of trapped charges in the underlying silicon dioxide by photo-generated electron-hole pairs. The period for the second light-off cycle is intentionally made short to allow the conductivity of graphene to recover only partially to its original dark level. When the light is turned on again, it takes only 15 min for the conductivity to reach the minimum in the second light-on cycle. Near the end of the second light-on cycle, the conductivity starts to show an up-trend. The time it takes for the conductivity to reach its minimum during a light-on cycle for the applied voltage of 1V in Fig. 2(a) is between 9 and 14 min. This is shorter than the time period shown in Fig. 2(b) corresponding to a lower applied voltage of 10mV. This implies that a higher applied voltage, and thus a higher electrical current density, accelerates the ending of the decreasing trend in photoconductivity evolution with illumination time during the light-on cycle. This might be attributed to accelerated desorption of photoionized surface molecules by electron current induced Joule heating and electron wind.

When a two-dimensional array of silver nanoparticles is illuminated with light of a proper wavelength, plasmonic coupling induces very strong electromagnetic fields on some surface areas of silver nanoparticles as well as the space between silver nanoparticles. The strong electromagnetic fields are expected to elevate the photo-induced effects on the variation of electrical conductivity for single-layer graphene as shown in Figs. 2(a) and (b). In order to reveal the effects of plasmonic coupling, measurements similar to those shown for Figs. 2(a) and (b) are carried out for a graphene film which is transferred to cover a two-dimensional array of silver nanoparticles coated on a silicon dioxide grown on silicon wafer. As a comparison, other features of the test structure and experimental conditions remain the same.

Figure 3(a) displays measured electrical current evolution for an Ag-nanoparticle-coated sample under an applied voltage of 1 V. Similar to the pristine case, the dark current in the initial phase gradually increases from 8.1 to 8.8 mA in a period of 3.3 hr. The current quicklydecreases to 7.6 mA after the laser illumination is switched on. The electrical current then gradually increases after having reached the minimum level. However, the rate of increase in electrical current from the minimum is much faster than that for graphene without underlying silver nanoparticles. It takes only 24 minutes for the current to fully recover to its dark current level measured before the illumination begins. The faster rising electrical current accelerates the switching from negative photoconductivity to positive photoconductivity when the light illumination time period is long enough. The photoconductivity of graphene under the effects of plasmonic coupling does switch to positive photoconductivity from the initial negative photoconductivity. This is further confirmed by the decrease in electrical current, and the lower electrical conductivity of the graphene, when the light illumination is terminated as shown in Fig. 3(a). After the conditioning of the graphene under the effects of plasmonic coupling by blue laser illumination, the second and the third light-on cycles only show positive photoconductivity. It is apparent that the pristine and Ag-nanoparticle-coated samples exhibit qualitatively different light-induced modification of electrical conductivity of graphene. Silver nanoparticles play important roles on the differences since their existence is the main discrepancy between the two sets of measurements. It has been reported that porous anodic aluminum oxide films encapsulated with an Ag-nanoparticle array exhibit positive photoconductivity because of light-induced plasmonic coupling [16–18]. It has also been reported that silver nanoparticles encapsulated by hydrogenated graphene exhibit a higher surface enhanced Raman scattering, than those encapsulated by graphene, of molecules adsorbed on the surface of graphene. Hydrogenation of graphene reduces the electrical conductivity of graphene by many orders of magnitude [13].

 

Fig. 3 Time evolution of electrical current in a graphene film on Ag-nanoparticle-coated SiO₂/Si under a series of light-off and light-on cycles. The white and blue areas represent the durations of light off and on, respectively. The applied voltage in (a) is 1 V while that in (b) is 10 mV.

Download Full Size | PDF

Due to its electrical conductivity, the graphene film tends to form an equi-potential surface on top of silver nanoparticles, where plasmonic coupling induces strong local electromagnetic fields when light is turned on. Therefore, when light is turned on, the graphene film is subjected to combined high electric fields due to the applied DC voltage of 1V across the graphene film of 20μm in length between two gold contact pads and the plasmonic coupling induced strong local electromagnetic fields across the underlying silver nanoparticles. The strong combined electromagnetic fields in the graphene and near the graphene surface and graphene/Si interface might enhance (i) the photo-induced desorption of surface adsorbents, (ii) the annealing of the graphene due to increased Joule heating, which results in further acceleration of the thermal induced desorption of surface adsorbents on graphene surface, and (iii) the generation of electron-hole pairs in silicon. Although the strong electromagnetic fields will also enhance the photo-induced ionization of surface molecules, which causes the increased electron scattering and the negative photoconductivity, the rapid decrease in the number density of surface molecules due to desorption might have made the electron scattering by charged molecules less effective.

For surface molecules which serve as dopants to contribute to the charge carrier density in the graphene for conducting electrical current, the enhanced desorption due to Joule heating and strong electromagnetic fields, decreases the electrical conductivity of the graphene. On the other hand, surface molecules which serve as compensating dopants to those contributing to the majority charge carriers in the graphene, the desorption reduces the compensation and increases the electrical conductivity. The opposite doping effects of different surface dopants and the electron scattering due to photo-ionization generated surface charges make the time evolution of the graphene depend very much on the actual ambient environments, which determines the concentration and characteristics of adsorbed molecules on the graphene surface.

Among effects on the electrical conductivity of graphene, the induction of strong electromagnetic fields by plasmonic coupling is a very fast process, the electron-hole pair generation and photo-ionization are fast processes while the diffusion of electron-hole-pairs to neutralize the trapped charges and the desorption of surface molecules due to light illumination and Joule heating are slow processes. Some surface molecules on graphene have lower ionization potentials than others. When light is turned on, those with low ionization potentials are ionized effectively and rapidly. The rapid photo-ionization of these surface molecules produces high concentration of electron scattering centers and is attributed to the rapid decrease in the electrical conductivity of the graphene when light is turned on during the first light-on period as shown in Fig. 3(a).

Slow processes which are in favor of increasing the photoconductivity take their turns after a different extent of delay to convert the negative photoconductivity to positive photoconductivity. Desorption of surface molecules with low ionization potentials is among slow processes which are in favor of positive photoconductivity. The Joule heating and the high electron concentration and speed in the graphene due to the applied high electrical current density with a constant DC voltage of 1 V across the gold contact pads and the plasmonic coupling induced strong electromagnetic fields further accelerate the desorption of surface molecules to result in the rapid turn-around from the decreasing electrical conductivity to increasing electrical conductivity of the graphene as observed during the initial phase of the first light-on period shown in Fig. 3(a).

In an ambient environment where the concentration of molecules with low photo-ionization potential is low, the adsorption rate of these molecules during the light-off period is low. There are not many such molecules on the graphene surface after the first light-on andlight-off cycle. Therefore, the initial rapid decrease in the electrical conductivity of the graphene during the subsequent light-on period does not occur. More detailed research with graphene in a controlled environmental chamber with known gas species is necessary to confirm the proposed mechanism for interpreting the observed strong effect of plasmonic coupling on the persistent photoconductivity of graphene in the ambient environments.

To reveal the importance of the combined effects of plasmonic coupling and high DC electrical current and electric field in the graphene, the applied voltage between two gold contact pads is reduced to 10mV with other experimental conditions being kept the same. Figure 3(b) shows tine evolution of electrical current for a graphene with underlying silver nanoparticles. The photoconductivity is negative throughout the entire light-on cycle with the electrical current under illumination being lower than the dark current. The electrical current under light illumination remains almost constant at the low current level near 41μA. It implies that the increase in dark current is balanced by the photo-induced decrease of electrical current. Plasmonic coupling promotes the rate of decrease in current during the light-on cycle compared to the previous case without plasmonic coupling. The higher rate of decrease in electrical current with time, after deducting the rising dark current with time, results in the almost constant electrical current for the plasmonic coupling assisted photoconductivity compared to the measured rising electrical current during the light-on cycle in Fig. 2(b) for the same graphene without plasmonic coupling.

Switching of photoconductivity of graphene on SiO₂/Si from negative to positive has been reported by Biswas et al. [7]. The surface of a graphene in the ambient atmosphere is easy to adsorb many kinds of atoms and molecules. Photo-induced desorption of surface adsorbents can play both roles of increasing and decreasing electrical conductivity. Desorption of molecules and atoms which serve as dopants for the graphene reduces the charge carrier density in the graphene and thus decreases its conductivity. Desorption of non-dopant molecules and atoms, especially those with low ionization potentials, does not reduce the free charge carrier density in the graphene, but, instead, reduces the ionized scattering centers for free charge carriers in the graphene. In this case, photo-induced desorption promotes the transport of free charge carriers in the graphene and causes the conductivity to increase. Molecules and atoms can also be trapped in the interface between the graphene and substrate during the transfer process. These molecules and atoms will take more time to get out from under the graphene. Light illumination causes molecules and atoms with low ionization potential to be ionized and become additional scattering centers for free charge carriers in the graphene. Photoionization of these molecules and atoms thus contributes to the negative photoconductivity. Unlike suspending graphene, graphene studied in this work is supported by SiO₂/Si. On the other hand, charges trapped in SiO₂ are scattering centers for free charge carriers in the graphene. However, photo-induced electron-hole pairs help neutralize those trapped charges and reduce the scattering frequency of free charge carriers in the graphene causing the conductivity of graphene to increase. This will contribute to the increase in conductivity of the graphene. All these effects take relatively long time compared to the photo-induced generation of electron-hole pairs, which usually dominates the photoconductivity of semiconductors. Therefore, the time evolution of photoconductivity measured in this work is slow resulting in both positive and negative persistent photoconductivity.

Although the wavelength of 200 to 300 nm used in the work of Biswas et al. is much shorter than that of 405 nm used for this work, the mechanisms responsible for the photoconductivity switching are related. The penetration depth of UV light into the test structure is smaller than that of blue light. The photoionization yield of UV light is expected to be higher than the blue light for molecules with higher ionization potentials. UV induced desorption of some surface adsorbents has been known to be more effective than blue light. Both UV light and blue light are energetic enough to generate electron-hole pairs in silicon effectively. Due to the small penetration depth of UV light, electron-hole pairs are generated closer to the interface between silicon dioxide and silicon. This might have reduced the time for electrons to diffuse to the silicon dioxide for more quickly neutralizing trapped charges which cause undesirable charge scattering in the graphene film. . Whether this is significant is not known yet and needs more studies. Therefore, although the extent to which each of the above mentioned effects on the photoconductivity of single-layer graphene in the ambient air might be different, similar mechanisms apply to both wavelengths.

This leaves the major differences between this work and that of Biswas et al. being the effects of plasmonic coupling and the electrical current density, which is varied by changing the applied voltage between two gold contact pads from 1V to 10mV. From data presented in Figs. 2 and 3, it is apparent that a higher current density, when a higher voltage is applied, and the strong local electromagnetic fields induced by plasmonic coupling in the array of silver nanoparticles both promote the measured time evolution of the photoconductivity of a single graphene film towards positive photoconductivity. Annealing of graphene by electron transport in the graphene film is promoted by a higher electrical current. The annealing process accelerates the modification of the contents of the surface adsorbents and the trapped molecules in the interface between the graphene and the silicon dioxide. The detrapping of positive charges in the silicon dioxide might also be promoted depending on the extent of local temperature rise in the silicon dioxide. Higher light absorption due to plasmonic resonance increases the heating of the graphene by laser illumination besides the strong local electromagnetic fields. Plasmonic coupling induced promotion of desorption of surface adsorbates, especially when the electrical current in the graphene film is high, is believed to have contributed to the observed phenomena in time evolution of photoconductivity of graphene.

Biswas et al. reported that the period for such a switching increases with illumination wavelength [7]. The wavelength of blue laser is much longer than the UV they used. Therefore, for graphene on SiO₂/Si and subjected to low electrical current with an applied voltage of 10mV, the negative photoconductivity did not show the switching to positive photoconductivity like Biswas et al. reported even after tens of hours. It is unknown whether such a switching in photoconductivity will occur in an even longer period of time of illumination by blue light or not. Nevertheless, with the applications of plasmonic coupling and high conducting electrical current, both effects accelerate the switching from negative photoconductivity to positive photoconductivity as shown during the first light-on cycle in Fig. 3(a). The strong effects of plasmonic coupling and electrical current induced annealing of graphene not only accelerate the switching from negative photoconductivity to positive photoconductivity but also eliminate the negative photoconductivity in the subsequent light-on cycles. The graphene acts like typical semiconductor with positive photoconductivity, thereafter. More investigation in persistent photoconductivity due to illumination by visible light with longer wavelengths for a longer period of time is being undertaken and will be reported elsewhere.

4. Conclusions

Electrical conductivity of graphene exhibits a strong response to 405-nm laser illumination. The conductivity of graphene on SiO₂/Si decreases rapidly from its dark value upon illumination followed by slow increase which lasts for tens of hours to eventually exceeding its original dark value. On the contrast, the conductivity of graphene on Ag-nanoparticles-coated SiO₂/Si exhibits negative photoconductivity initially during the first light-on period. The negative photoconductivity reverses its trend with increasing conductivity and becomes positive in the later stage of a long light-on cycle. After the first light-on cycle, the graphene exhibits positive photoconductivity when the same blue laser is turned on in subsequent cycles of measurements. Plasmonic-coupling-induced strong electromagnetic fields near the graphene are attributed to the observed difference in photo-response. In addition, photoconductivity of graphene is significantly affected by the electrical current flowing in the graphene. At a low current, it exhibits persistent negative photoconductivity for graphene with and without underlying silver nanoparticles. At a high current density, the photoconductivity is still negative. However, when plasmonic coupling from Ag-nanoparticles is present, photoconductivity could switch from negative to positive. A positive persistent photoconductivity is hence demonstrated for single-layer graphene. Current level, plasmonic coupling, and illumination wavelength are found to play important roles in determining the time evolution of photoconductivity for a single-layer graphene film supported by SiO₂/Si. By properly applying both plasmonic coupling and a proper electrical current, the slow drift in the photoconductivity of a graphene film which is exposed to and interacting with the ambient atmosphere is minimized. This might help improve the reliability and reproducibility of graphene films for photoconductivity based sensor applications.