% IMPORTANT: The following is UTF-8 encoded. This means that in the presence
% of non-ASCII characters, it will not work with BibTeX 0.99 or older.
% Instead, you should use an up-to-date BibTeX implementation like “bibtex8” or
% “biber”.
@PHDTHESIS{Otten:1018302,
author = {Otten, René},
title = {{S}pin qubit control and readout using cryogenic
electronics},
school = {RWTH Aachen},
type = {Dissertation},
publisher = {RWTH Aachen University},
reportid = {FZJ-2023-04687},
pages = {pages 1 Online-Ressource : Illustrationen, Diagramme},
year = {2023},
note = {Dissertation, RWTH Aachen, 2023},
abstract = {Recent progress in the field of quantum computation has
further substantiated the claim of a computational advantage
for quantum computers and recent developments in multiple
platforms such as superconducting qubits, ion traps, optical
systems, and neutral atoms create hope that large scale
quantum computers are within reach. In the meantime, spin
qubits in semiconductors have achieved error rates for all
aspects of their operation beyond the error correction
threshold, and now face the challenge of scaling to larger
system sizes, while maintaining their 'semiconductor'
scaling advantage. In this spirit multiple ideas for
architectures have emerged that integrate the quantum layer
with local control electronics creating a QPU with high
interconnect density between qubits and control while
maintaining a sparse amount of connection in and out of the
cryostat. In this thesis, I demonstrate multiple steps
towards this vision: dc-bias voltage generation and readout
amplification. I show that the operation of traditional CMOS
electronics at mK temperatures on the mixing chamber is
feasible and use a custom ultra-low power DAC to bias
electrodes of a GaAs qubit chip. We achieve an
integrated-circuit power consumption of 30 µW, of which 13
µW can be attributed to the DAC, while generating 8 output
voltages with a refresh rate of 4 kHz. We observe no leakage
while storing the voltage on sample-and-hold capacitors on
this timescale. Voltage drift due to leakage is quantified
to be 125 µV/s, well below room temperature values and
likely limited by the Schottky contacts on the qubit chip.
Nevertheless, this highlights that custom IC designs for
ultra-low temperatures are a valuable tool for achieving
ultra-low power, scalable control electronics. We manage to
maintain qubit and IC temperatures below 1 K with thermal
gradients at material interfaces presenting the main
obstacle for reaching lower temperatures. This allows us to
measure a single electron transistor only with bias voltages
generated by our CryoDAC. Extrapolating from our results, we
demonstrate that bias voltage generation with a power
consumption of ~4 nW/ch is feasible paving the way for
integrated generation on up to 107 voltages within the
cooling power budget available in a cryostat optimized for
integrated control electronics. As a second step towards
integrating readout and control, we investigate
heterojunction-bipolar transistors as cryogenic amplifiers
in direct proximity to a qubit chip. By biasing the
transistor away from its nominal operating point near the
saturation regime, we maintain gain and transconduction, 200
and gm = 0.1 mS @ Ib = 2 nA, away from the forward-active
regime, while lowering the power consumption significantly
by a factor of 8. I investigate biasing options for these
transistors and investigate their noise performance, showing
1/f noise all the way to the noise floor of our measurement
setup. I demonstrate functional amplifier behavior at power
consumption as low as 50 nW, paving the way for integrated,
close-to-qubit readout readout amplification for large qubit
numbers. To underline the usefulness for qubit operation, I
also show charge detection and fast recording of charge
stability diagrams using a charge sensor amplified by the
HBT. To complement these advancements, I summarize automated
tuning methods we developed across multiple experiments,
cryostats and cooldowns. I show that we are able to reliably
tune devices to the single electron regime automatically
with relatively simple methods across different readout
amplification techniques. We use the data gathered in these
experiments to train a convolutional neural network to
automatically classify charge stability diagrams by the
number of observed quantum dots with an accuracy of $94\%,$
enabling more sophisticated automated tuning algorithms in
the future. I complement the previous results by summarizing
the state-of-the-art in flip-chip bonding and 3D integration
technologies and putting the locally available processes
into this context. Overall, my thesis provides advances,
outlines ways forward, and provides guidance on many
complementing aspects of scaling quantum computing systems
to the qubit numbers needed for fully error-corrected
systems. This includes signal generation and control,
readout, and cryogenic amplification, automated-tuning and
calibration, as well as heterogeneous integration and
cryostat design. By doing so I substantiate the dream of a
useful quantum computing system in the (not so distant)
future.},
keywords = {Hochschulschrift (Other) / physics ; quantum computing ;
quantum information ; quantum mechanics ; qubits ;
semiconductors ; solid state physics ; spin qubits ;
cryoelectronics ; heterojunction-bipolar transistors ; qubit
control ; qubit readout ; automated tuning ; machine
learning (Other)},
cin = {PGI-11},
cid = {I:(DE-Juel1)PGI-11-20170113},
pnm = {5221 - Advanced Solid-State Qubits and Qubit Systems
(POF4-522)},
pid = {G:(DE-HGF)POF4-5221},
typ = {PUB:(DE-HGF)11},
doi = {10.18154/RWTH-2023-07133},
url = {https://juser.fz-juelich.de/record/1018302},
}