Showing papers by "Guanxiong Liu published in 2012"

Graphene quilts for thermal management of high-power GaN transistors

Abstract: Self-heating is a severe problem for high-power gallium nitride (GaN) electronic and optoelectronic devices. Various thermal management solutions, for example, flip-chip bonding or composite substrates, have been attempted. However, temperature rise due to dissipated heat still limits applications of the nitride-based technology. Here we show that thermal management of GaN transistors can be substantially improved via introduction of alternative heat-escaping channels implemented with few-layer graphene-an excellent heat conductor. The graphene-graphite quilts were formed on top of AlGaN/GaN transistors on SiC substrates. Using micro-Raman spectroscopy for in situ monitoring we demonstrated that temperature of the hotspots can be lowered by ∼20 °C in transistors operating at ∼13 W mm(-1), which corresponds to an order-of-magnitude increase in the device lifetime. The simulations indicate that graphene quilts perform even better in GaN devices on sapphire substrates. The proposed local heat spreading with materials that preserve their thermal properties at nanometre scale represents a transformative change in thermal management.

Selective gas sensing with a single pristine graphene transistor.

Abstract: We show that vapors of different chemicals produce distinguishably different effects on the low-frequency noise spectra of graphene. It was found in a systematic study that some gases change the electrical resistance of graphene devices without changing their low-frequency noise spectra while other gases modify the noise spectra by inducing Lorentzian components with distinctive features. The characteristic frequency fc of the Lorentzian noise bulges in graphene devices is different for different chemicals and varies from fc = 10–20 Hz to fc = 1300–1600 Hz for tetrahydrofuran and chloroform vapors, respectively. The obtained results indicate that the low-frequency noise in combination with other sensing parameters can allow one to achieve the selective gas sensing with a single pristine graphene transistor. Our method of gas sensing with graphene does not require graphene surface functionalization or fabrication of an array of the devices with each tuned to a certain chemical.

Graphene-on-Diamond Devices with Enhanced Current-Carrying Capacity: Carbon sp2-on-sp3 Technology

Abstract: Graphene demonstrated potential for practical applications owing to its excellent electronic and thermal properties. Typical graphene field-effect transistors and interconnects built on conventional SiO2/Si substrates reveal the breakdown current density on the order of 1 uA/nm2 (i.e. 10^8 A/cm2) which is ~100\times larger than the fundamental limit for the metals but still smaller than the maximum achieved in carbon nanotubes. We show that by replacing SiO2 with synthetic diamond one can substantially increase the current-carrying capacity of graphene to as high as ~18 uA/nm2 even at ambient conditions. Our results indicate that graphene's current-induced breakdown is thermally activated. We also found that the current carrying capacity of graphene can be improved not only on the single-crystal diamond substrates but also on an inexpensive ultrananocrystalline diamond, which can be produced in a process compatible with a conventional Si technology. The latter was attributed to the decreased thermal resistance of the ultrananocrystalline diamond layer at elevated temperatures. The obtained results are important for graphene's applications in high-frequency transistors, interconnects, transparent electrodes and can lead to the new planar sp2-on-sp3 carbon-on-carbon technology.

Graphene-on-diamond devices with increased current-carrying capacity: carbon sp2-on-sp3 technology.

Abstract: Graphene demonstrated potential for practical applications owing to its excellent electronic and thermal properties. Typical graphene field-effect transistors and interconnects built on conventional SiO(2)/Si substrates reveal the breakdown current density on the order of 1 μA/nm(2) (i.e., 10(8) A/cm(2)), which is ~100× larger than the fundamental limit for the metals but still smaller than the maximum achieved in carbon nanotubes. We show that by replacing SiO(2) with synthetic diamond, one can substantially increase the current-carrying capacity of graphene to as high as ~18 μA/nm(2) even at ambient conditions. Our results indicate that graphene's current-induced breakdown is thermally activated. We also found that the current carrying capacity of graphene can be improved not only on the single-crystal diamond substrates but also on an inexpensive ultrananocrystalline diamond, which can be produced in a process compatible with a conventional Si technology. The latter was attributed to the decreased thermal resistance of the ultrananocrystalline diamond layer at elevated temperatures. The obtained results are important for graphene's applications in high-frequency transistors, interconnects, and transparent electrodes and can lead to the new planar sp(2)-on-sp(3) carbon-on-carbon technology.

Direct Probing of 1/f Noise Origin with Graphene Multilayers: Surface vs. Volume

Abstract: Low-frequency noise with the spectral density S(f)~1/f^g (f is the frequency and g~1) is a ubiquitous phenomenon, which hampers operation of many devices and circuits. A long-standing question of particular importance for electronics is whether 1/f noise is generated on the surface of electrical conductors or inside their volumes. Using high-quality graphene multilayers we were able to directly address this fundamental problem of the noise origin. Unlike the thickness of metal or semiconductor films, the thickness of graphene multilayers can be continuously and uniformly varied all the way down to a single atomic layer of graphene - the actual surface. We found that 1/f noise becomes dominated by the volume noise when the thickness exceeds ~7 atomic layers (~2.5 nm). The 1/f noise is the surface phenomenon below this thickness. The obtained results are important for continuous downscaling of conventional electronics and for the proposed graphene applications in sensors and communications.

Epitaxial graphene nanoribbon array fabrication using BCP-assisted nanolithography.

Abstract: A process for fabricating dense graphene nanoribbon arrays using self-assembled patterns of block copolymers on graphene grown epitaxially on SiC on the wafer scale has been developed. Etching masks comprising long and straight nanoribbon array structures with linewidths as narrow as 10 nm were fabricated, and the patterns were transferred to graphene. Our process combines both top-down and self-assembly steps to fabricate long graphene nanoribbon arrays with low defect counts. These are the narrowest nanoribbon arrays of epitaxial graphene on SiC fabricated to date.

Graphene thickness-graded transistors with reduced electronic noise

Abstract: The authors demonstrate graphene thickness-graded transistors with high electron mobility and low 1/f noise (f is a frequency). The device channel is implemented with few-layer graphene with the thickness varied from a single layer in the middle to few-layers at the source and drain contacts. It was found that such devices have electron mobility comparable to the reference single-layer graphene devices while producing lower noise levels. The metal doping of graphene and difference in the electron density of states between the single-layer and few-layer graphene cause the observed noise reduction. The results shed light on the noise origin in graphene.

Selective Gas Sensing with a Single Pristine Graphene Transistor

Abstract: We show that vapors of different chemicals produce distinguishably different effects on the low-frequency noise spectra of graphene. It was found in a systematic study that some gases change the electrical resistance of graphene devices without changing their low-frequency noise spectra while other gases modify the noise spectra by inducing Lorentzian components with distinctive features. The characteristic frequency fc of the Lorentzian noise bulges in graphene devices is different for different chemicals and varies from fc=10 - 20 Hz to fc=1300 - 1600 Hz for tetrahydrofuran and chloroform vapors, respectively. The obtained results indicate that the low-frequency noise in combination with other sensing parameters can allow one to achieve the selective gas sensing with a single pristine graphene transistor. Our method of gas sensing with graphene does not require graphene surface functionalization or fabrication of an array of the devices with each tuned to a certain chemical.

Selective gas sensing by graphene

Abstract: Graphene might become an ultimate medium for sensing applications, enabling gas and vapor detection with high sensitivity. However, changes in graphene resistivity under the gas exposure might be quite similar for different gases, making the sensing selectivity to be one of the key barriers to overcome. In this paper, we report on using low frequency noise to define the new characteristic parameters, which, in combination with the resistance change, form a unique signature of a gas. The noise measurements can be also used in combination with evaluating “memory step” effect in graphene under gas exposure. (The “memory step” is an abrupt change of the current near zero gate bias at elevated temperatures T > 500 K in graphene transistors.) The “memory step” changes in graphene under gas exposure can be used for high-temperature gas sensors.

Graphene-diamond-silicon devices with increased current-carrying capacity: sp 2 -Carbon-sp 3 -Carbon-on-Silicon technology

Abstract: Graphene demonstrated potential for practical applications owing to its excellent electronic and thermal properties. Typical graphene field-effect transistors (FETs) and interconnects built on conventional SiO 2 /Si substrates reveal the breakdown current density on the order of 108 A/cm2, which is ∼100× larger than the fundamental limit for the metals but still smaller than the maximum achieved in carbon nanotubes. It was discovered by some of us that graphene has excellent thermal conduction properties with the thermal conductivity K exceeding 2000 W/mK at room temperature [1]. Few-layer graphene largely preserves the heat conduction properties [2]. However, the thermally resistive SiO 2 , with the thermal conductivity in the range from 0.5 to 1.4 W/mK, creates a bottleneck for heat removal. The latter does not allow graphene to demonstrate its true current-carrying potential. We show that by replacing SiO 2 with synthetic diamond one can substantially increase the current-carrying capacity of graphene to as high as ∼ 20×108 A/cm2 under ambient conditions. The two-terminal and three-terminal top-gated graphene devices (see Figure 1) were fabricated on synthetic single-crystal diamond (SCD) and ultrananocrystalline diamond (UNCD). To ensure Si integration, the UNCD layers were grown at low temperatures compatible with Si CMOS technology [3]. Our results indicate that graphene's current-induced breakdown is thermally activated. It was found that the current carrying capacity of graphene can be improved not only on SCD but also on an inexpensive UNCD. The latter was attributed to the decreased thermal resistance of UNCD at elevated temperatures (see Figure 2). The obtained results are important for graphene's hetero-integration on Si substrates. The enhanced current-carrying capacity is beneficial for the proposed applications of graphene in interconnects and high-frequency transistors.

Selective gas sensing with a single graphene-on-silicon transistor

Abstract: The low-frequency 1/f noise in graphene transistors has been studied extensively owing to the proposed graphene applications in analog devices and communication systems [1–5]. The studies were motivated by the fact that the low-frequency noise can be up-converted by device nonlinearity and contribute to the phase noise of the system. Similarly, the sensor sensitivity is often limited by the electronic low-frequency noise. Therefore, noise is usually considered as one of the main limiting factors for the device or overall system operation. However, the electronic noise spectrum itself can be used as a sensing parameter increasing the sensor sensitivity and selectivity. Here, we show that vapors of different chemicals produce distinguishably different effects on the low-frequency noise spectra of the graphene-on-Si transistor. Our study showed that some gases change the electrical resistance of pristine graphene devices without changing their low-frequency noise spectra while other gases modify the noise spectra by inducing Lorentzian components with distinctive features. The characteristic corner frequency f C of the Lorentzian noise bulges in graphene devices is different for different chemicals and varies from f C =10 − 20 Hz for tetrahydrofuran to f C =1300 − 1600 Hz for chloroform. We tested the selected set of chemicals vapors on different graphene device samples and alternated different vapors for the same samples. The obtained results indicate that 1/f noise in combination with other sensing parameters can allow one to achieve the selective gas sensing with a single pristine graphene transistor. Our method of gas sensing with graphene does not require graphene surface functionalization or fabrication of an array of the devices with each tuned to a certain chemical. The observation of the Lorentzian components in the vapor-exposed graphene can also help in developing an accurate theoretical description of the noise mechanism in graphene.