![Statistics of the switching performance
a–c, One-hundred measured switching transients for 1,000-nm-gap devices with 100-nm-thick gold (a), 100-nm-thick tungsten (b) and 500-nm-thick tungsten (c). d–f, Measured switching voltage at t = 20 ps (with standard deviation σa\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\sigma }_{{\rm{a}}}$$\end{document}) and measured noise level at t = −20 ps (with standard deviation σb\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\sigma }_{{\rm{b}}}$$\end{document}) corresponding to the waveforms shown in a–c, respectively. The normalized effective standard deviation σeff=σa2−σb2/VSW\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\sigma }_{{\rm{eff}}}=\sqrt{{\sigma }_{{\rm{a}}}^{2}-{\sigma }_{{\rm{b}}}^{2}}/{V}_{{\rm{SW}}}$$\end{document}, where VSW ≈ 120 V is the switching voltage, is 4.9%, 4.6% and 3.2% for the measured waveforms shown in a–c, respectively. g–i, Measured maximum dv/dt corresponding to the measured waveforms shown in a–c, respectively. Characterization of dv/dt is more subject to measurement errors because of the limited sampling time (5 ps per sample). It should be noted that the limited sampling time generally leads to an underestimation of dv/dt, as the sampling does not necessarily pick the maximum of dv/dt. All the results are presented without de-embedding.](https://www.researchgate.net/publication/340169509/figure/fig4/AS:873208068730904@1585200435594/Statistics-of-the-switching-performance-a-c-One-hundred-measured-switching-transients_Q320.jpg)
![Lifetime evaluation under harsh switching condition
a, Dissipated power inside a 700-nm-gap nanoplasma switch with tungsten pads under a short circuit test resulting in the highest possible current density for lifetime measurements (high current density is the main driver for electromigration). Measurements showed energy and peak power dissipation of 3 μJ and 0.4 kW at each short circuit pulse. Such a high power and energy dissipation are orders of magnitude higher than in practical applications. b, Degradation with the definition of (VTH[n] – VTH[0])/VTH[0], where VTH[n] is the threshold voltage at nth short circuit. The error bars show ±2σ, where σ is the standard deviation from ten measurements. The obtained results for the proposed devices with sputtered tungsten pads show their capability of withstanding repetitive short circuits, without any specific optimization. The devices with a thicker pads (thus lower current density) provide a more stable performance even under very harsh conditions, thus one could expect a very long lifetime in normal operations. In addition, electromigration has a solid background in silicon electronics with several demonstrated solutions, including the use of specific alloys, or single crystalline metals that result in nearly infinite lifetime even for submicrometre interconnections¹²³. Thus, even though the 100-nm-thick devices showed a larger degradation in such extreme conditions, they could also be useful in practical applications.](https://www.researchgate.net/publication/340169509/figure/fig5/AS:873208068730921@1585200435716/Lifetime-evaluation-under-harsh-switching-condition-a-Dissipated-power-inside-a_Q320.jpg)
A preview of this full-text is provided by Springer Nature.
Content available from Nature
This content is subject to copyright. Terms and conditions apply.
534 | Nature | Vol 579 | 26 March 2020
Article
Nanoplasma-enabled picosecond switches
for ultrafast electronics
Mohammad Samizadeh Nikoo1, Armin Jafari1, Nirmana Perera1, Minghua Zhu1,
Giovanni Santoruvo1 & Elison Matioli1 ✉
The broad applications of ultrawide-band signals and terahertz waves in quantum
measurements1,2, imaging and sensing techniques3,4, advanced biological treatments5,
and very-high-data-rate communications6 have drawn extensive attention to ultrafast
electronics. In such applications, high-speed operation of electronic switches is
challenging, especially when high-amplitude output signals are required7. For
instance, although eld-eect and bipolar junction devices have good controllability
and robust performance, their relatively large output capacitance with respect to
their ON-state current substantially limits their switching speed8. Here we
demonstrate a novel on-chip, all-electronic device based on a nanoscale plasma
(nanoplasma) that enables picosecond switching of electric signals with a wide range
of power levels. The very high electric eld in the small volume of the nanoplasma
leads to ultrafast electron transfer, resulting in extremely short time responses. We
achieved an ultrafast switching speed, higher than 10 volts per picosecond, which is
about two orders of magnitude larger than that of eld-eect transistors and more
than ten times faster than that of conventional electronic switches. We measured
extremely short rise times down to ve picoseconds, which were limited by the
employed measurement set-up. By integrating these devices with dipole antennas,
high-power terahertz signals with a power–frequency trade-o of 600 milliwatts
terahertz squared were emitted, much greater than that achieved by the state of the
art in compact solid-state electronics. The ease of integration and the compactness of
the nanoplasma switches could enable their implementation in several elds, such as
imaging, sensing, communications and biomedical applications.
Nanometre-scale transistors based on III–V compound semiconduc-
tors, such as gallium arsenide (GaAs), indium arsenide (InAs) and
indium phosphide (InP), are at the heart of many high-speed and high-
frequency electronic systems9. Owing to their high electron mobilities,
these devices exhibit very high small-signal cut-off frequencies, in
the terahertz range
10
. However, the high-frequency, large-signal per-
formance of transistors is still a challenge, as it is severely limited by
the output capacitance Cout, electron saturation velocity and critical
electric field11. The maximum switching speed of a transistor (Fig.1a)
with saturation current Imax is limited to
v
t
I
C
d
d=2(1
)
max
max
out
Equation(1) leads to a power (P)–frequency (f) trade-off, called the
Pf2 limit (Fig.1b)
Pf
Z
=
8π
(2
)
v
t
2max
d
dmax
2
2
where Z is the load resistance, v is the drain-source voltage and t is time.
Equation(1) is a self-normalized term, independent of device size and
gate length, that results in almost similar values of less than 1Vps−1, for
either power or radiofrequency devices (Fig.1c). For example, in lateral
devices, normalized saturation currents and output capacitances are
in range of 1mAμm
−1
and 1fFμm
−1
, respectively, resulting in switch
-
ing speeds of 0.5Vps
−1
. Johnson’s figure-of-merit (JFOM)
11
takes into
account the breakdown voltage and small-signal cut-off frequency of
devices, although these parameters are obtained from two different
measurements at two separate operating points
12
. As shown in Fig.1d,
JFOM indicates switching speeds of even less than 1Vps
−1
for GaAs and
InP, as two semiconductors with very high electron mobilities and the
main candidates for ultrafast electronics. As described in equation(2),
the maximum switching speed of 1Vps
−1
corresponds to an output
power of 2.5mW at 1THz for Z = 50Ω. Such a limitation can be seen in
the performance of not only transistor-based power amplifiers but
also other solid-state-based approaches, including frequency mul-
tipliers and negative resistance oscillators (Extended Data Fig.1). In
practice, several phenomena happening at high frequency and small
scales—such as radiofrequency transconductance (gm) collapse13,
https://doi.org/10.1038/s41586-020-2118-y
Received: 9 September 2019
Accepted: 15 January 2020
Published online: 25 March 2020
Check for updates
1Power and Wide-band-gap Electronics Research Laboratory (POWERlab), Institute of Electrical Engineering, École Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland.
✉e-mail: elison.matioli@epl.ch
Content courtesy of Springer Nature, terms of use apply. Rights reserved
