Room-Temperature Amplification of Terahertz Radiation by Grating-Gate Graphene Structures

We study terahertz (THz) radiation transmission through grating-gate graphene-based nanostructures. We report on room-temperature THz radiation amplification stimulated by current-driven plasmon excitation. Specifically, with an increase of the dc current under periodic charge density modulation, we observe a strong redshift of the resonant THz plasmon absorption, followed by a window of complete transparency to incoming radiation and subsequent amplification and blueshift of the resonant plasmon frequency. Our results are, to the best of our knowledge, the first experimental observation of energy transfer from dc current to plasmons leading to THz amplification. Additionally, we present a simple model offering a phenomenological description of the observed THz amplification. This model shows that in the presence of a dc current the radiation-induced correction to dissipation is sensitive to the phase shift between oscillations of carrier density and drift velocity. And, with an increasing current, the dissipation becomes negative, leading to amplification. The experimental results of this work, as all obtained at room-temperature, pave the way toward the new 2D plasmon-based, voltage-tunable THz radiation amplifiers.

Popular Summary

More than 40 years ago, a new direction in physics opened up with the arrival of plasma wave electronics. The possibility that plasma waves could propagate faster than electrons fascinated all, suggesting that so-called “plasmonic” devices could work at terahertz (THz) frequencies, too high for standard electronic devices. However, numerous experimental attempts to realize such amplifiers or emitters failed, and the creation of compact, tunable, room-temperature THz devices remains a challenge. To that end, we explore THz light-plasmon coupling, light absorption, and amplification in a graphene-based system, whose excellent room-temperature electrical and optical properties are ideal for solving this problem.

Using monolayer graphene grating-gate structures, we demonstrate tunable resonant plasmon absorption which, with an increase of the current, turns to THz radiation amplification. The observed gain of up to 9% is far beyond the well-known landmark level of 2.3% that is maximal available in monolayer graphene when photons directly interact with electrons. To interpret our results, we use a dissipative plasmonics crystal model, which captures the main trends and basic physics of the amplification phenomena. Specifically, the model predicts that increasing current drives the system into an amplification regime, wherein the plasma waves may transfer energy to the incoming electromagnetic waves.

All results were obtained at room temperature. Therefore, our experimental setup paves the way toward future THz plasmonic technology with a new generation of all-electronic, resonant, voltage-controlled THz amplifiers.

Figure 1

The microphoto of the A-DGG3.1 sample (a), schematic top and cross-sectional device images illustrating the $h$-BN/graphene/$h$-BN heterostructure including the asymmetric dual grating-gate metallization (b), and the gate-controlled charge density distribution (green area) together with potential distribution along the channel (black solid line) (c).

Figure 2

Electrical characterization of the samples. Ambipolar resistance versus back gate (a) and top gate for the $C1$ cavity (b) of the A-DGG3.1 structure. We also show resistance versus inverse swing voltage $[1/(U_g-U_{\mathrm{CNP}}]$ that allows extraction of the resistance of the contacts (a) and the resistance of total ungated part of samples for the $C1$ cavity (b). $C1$ cavity carrier density $n$ and mobility as a function of the gate voltage (c): The lower horizontal axis corresponds to the top-gate bias, and the upper horizontal axis is for the back-gate bias.

Figure 3

Measurement setup. (a) THz TDS. (b) Total reflection configuration. The incident THz radiation pulse is linearly polarized. The sample stage is rotated to obtain the radiation polarization perpendicular or parallel to the drain-source direction. (c) Typical time trace for samples with (orange) and without gate biasing (green) in the measurement system are shown in the upper panel. Spectra are obtained by FFT transformation of the same signal in the lower panel.

Figure 4

Typical time traces obtained for the A-DGG1 sample with (orange) and without gate biasing (green). Left panels (a)–(c) are experiments without drain-source biasing ($U_d=0$), and right panels (e)–(g) the results obtained with a strong drain biasing at $U_d=700\mathrm{mV}$. (b) and (f) are a magnification of the shadowed regions in (a) and (e), respectively. One can observe that the drain-source biasing inverses the order of magnitude of orange and green traces in the time traces and spectra row data. (d) and (h) show spectra obtained by Fourier transform for time traces with and without gate biasing, respectively.

Figure 5

Gate-length-dependent extinction spectra of the graphene structures for $U_d=0\mathrm{V}$ with incident light polarized parallel (a) and perpendicular (b) to the gate fingers. The measured data points and lines are the fits by the Drude model fit for parallel polarization and the damped oscillator model for perpendicular polarization. Black solid squares are a maximum of experimental resonant plasma frequencies as a function of the wave vector $q=\pi /L_1$. The black straight line through these dots is the best linear fit (right-hand scale).

Figure 6

Gate-voltage-dependent extinction spectra of the graphene structures for $U_d=0\mathrm{V}$ with incident light polarized perpendicular to the gate fingers. Panels (a), (b), (c), and (d) presents results for $C1$, $C2$, $C3$, and $C4$ cavities, correspondingly. The experimental data are marked as dots, and continuous lines are the results of fits using standard physical models of gated 2D plasmons in graphene.

Figure 7

Gate-voltage dependence of the extracted resonance frequency. Continuous lines are results of calculations according Eq. (7) with corrections to fringing and plasma leakage phenomena—see Supplemental Material [51] for more details. The dotted line is $¼$ power dependence as a function of the gate bias, typical for plasmons in graphene.

Figure 8

Drain-bias-dependent spectra for the $C2$ plasma cavity. Blue marks absorption, the green zone is the gap between absorption and emission, and red corresponds to amplification. The arrow indicates raising drain bias.

Figure 9

Drain-bias-dependent properties of the $C1$, $C2$, $C3$, and $C4$ plasma cavities. The left-hand panels show the spectra (a), (b), (c), and (d), correspondingly. The right-hand panels show the contour plots (e), (f), (g), (h), correspondingly. The dotted lines on contour plots are results of the fit with a phenomenological formula Eq. (10).

Figure 10

The experimental results showing the resonant frequency versus drain voltage bias in normalized scales. Dotted lines show results of calculations according to Eq. (10).

Figure 11

1D plasmonic crystal driven by dc current in the oscillating field of THz radiation.

Figure 12

Active regions of the plasmonic crystal with a large fundamental frequency are separated by passive regions with a small frequency.

Figure 13

Relaxation time as a function of the gate bias. ${\tau }_1$ is the momentum relaxation time in the active region shown as the black diamonds, ${\tau }_{\mathrm{eff}}$ is the plasmon relaxation rate calculated using Eq. (15) (solid lines), and color dots are experimental ${\tau }_{\mathrm{eff}}$.

Figure 14

Dissipation in the channel for different values of ${\mathcal{V}}_1/s_1$.

Figure 15

The amplitude of the dissipation as a function of $x={\mathcal{V}}_1/s_1$. Dissipation-amplification transition corresponds to $x=x_0=1/\sqrt{3}$.