A CNTFET Oscillator at 461 MHz A. Taghavi, C. Carta, T. Meister, F. Ellinger, M. Claus, and M. Schroter Abstract—This letter presents design, implementation, and characterization of the first reported carbon nanotube field-effect transistor (CNTFET) RF oscillator. The circuit is implemented with discrete CNTFETs mounted on standard FR-4 substrate with off-the-shelf surface-mount-device components. The oscilla- tor topology is similar to the phase shifter oscillator and uses two cascaded common source stages to provide enough gain for self-sustained and self-startup oscillation. The oscillator tank is merged with the matching network between two stages. The circuit oscillates at the frequency of 461 MHz with a phase noise of -115 dBc/Hz at 1 MHz and power consumption of 60 mW. While limited in output power by the driving capabilities of prototype CNTFETs, which still have a large density of residual metallic tubes, both power consumption and phase noise compare well with established and mature technologies. Moreover, the presented phase noise measurements provide a useful benchmark for the physical noise models being currently developed for this category of devices. Index Terms— Carbon nanotube transistors, carbon nanotube field-effect transistor (CNTFET), oscillator, phase noise. I. I NTRODUCTION I T HAS been recently shown that the carbon nanotube field-effect transistors (CNTFETs) are emerging devices with the potential of bringing improvements into the RF frontend of highly integrated wireless communication systems. Design, implementation, and characterization of fundamental buildings blocks for such systems are necessary steps toward the experimental validation of such a promises and device- level considerations in design phase. As openly acknowledged within the device engineering community, the current state of CNTFET technologies is still far from its best possible performance, which will enable the expected benefits in RF applications [12]. Nevertheless discrete experimental devices have been fabricated for several years and offer performance sufficient for operation in the ultrahigh-frequency band of the RF spectrum. Indeed, the first amplifiers have been already demonstrated [1], [2] even up to 2 GHz. Within this frame, this letter presents the characterization of the first reported RF oscillator based on CNTFETs as active devices [13]. While the available discrete devices are early prototypes, which still suffer from limitations such as high density of metallic tubes in the channel, hysteresis, and gain degradation, direct measurements of a complete and functional circuit provide useful information toward the validation of the large-signal and noise models of the devices in the context of their intended circuit uses, as well as for essential technological optimizations. In particular, the phase noise of an oscillator Manuscript received January 30, 2017; accepted February 17, 2017. Date of publication May 23, 2017; date of current version June 5, 2017. This work was supported in part by the Center for Advancing Electronics Dresden, path B (Carbon), and in part by the German Research Foundation under Project DFG SCHR695/6. (Corresponding author: A. Taghavi.) The authors are with Technische Universität Dresden, 01069 Dresden, Germany (e-mail: amirali.taghavi@tu-dresden.de). Digital Object Identifier 10.1109/LMWC.2017.2701312 Fig. 1. Schematic of common source topology with matching network (OMN: Output matching network and IMN: Input matching network). is a strong function of the noise characteristics of the active device employed in the circuit, but CNTFETs—being a new technology—still lack the low-frequency noise models neces- sary for prediction of the oscillator phase noise. Also there are only limited publications regarding noise measurements of multitube devices [3] and none for single-tube CNTFETs. For this reason, the measurement results presented in this letter cannot be compared with circuit simulations, but rather provide an experimental data point for the validation of the required device noise model, and available device noise performance. II. DESIGN The schematic of the circuit is shown in Fig. 1, it consists of cascade of two common-source stages ( T 1 and T 2 ), connected with a narrow-band matching network, which serves at the same time as power-matching network and resonator tank. The output of the second stage is fed back to the input through two additional matching networks and a λ-long coaxial cable. This installment will satisfy the Barkhausen criterion regarding the unity closed-loop gain and zero phase at the specific frequency. Each stages consist of a common source amplifier which will contribute to the half of 360° phase shift and the coaxial cable serves to close the loop and does not affect the phase shift (mod 360°) The use of two gain stages is beneficial in providing additional gain for starting and sustaining the oscillation, as well as providing more degrees of freedom for compensating process tolerances and possible device degradation over time. The printed circuit board (PCB) has been designed to minimize interferences from the measurement environment. We used a solid ground plane in a FR-4 PCB with dielectric thickness of 1.5 mm, multiple parallel capacitors—scaled in value—to block power supply noise, and a double braided RG142 coaxial cable to close the circuit loop. In its open-loop configuration, the circuit is a narrow- band two-stage amplifier with input and output matching networks to a 50- impedance. The circuit is implemented using discrete CNTFETs having W = 50 μm, L = 0.8 μm, and t ox = 20 nm with a channel consists of 2500 paral- lel tubes having average diameter of 1.7 nm with a ratio © 2017 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. DOI: 10.1109/LMWC.2017.2701312.