ar
X
iv
:0
90
6.
19
95
v1
[
co
nd
-m
at.
me
s-h
all
]
10
Ju
n 2
00
9 Device fabrication and transport measurements of
FinFETs built with 28Si SOI wafers towards donor
qubits in silicon
Cheuk Chi Lo1,3, Arun Persaud3, Scott Dhuey2, Deirdre
Olynick2, Ferenc Borondics2, Michael C. Martin3, Hans A.
Bechtel3, Jeffrey Bokor1,2 and Thomas Schenkel3
1 Department of Electrical Engineering and Computer Sciences, University of
California, Berkeley, CA 94720, USA
2 The Molecular Foundry, E.O. Lawrence Berkeley National Laboratory, Berkeley,
CA 94720, USA
3 Accelerator and Fusion Research Division, E.O. Lawrence Berkeley National
Laboratory, Berkeley, CA 94720, USA
4 Advanced Light Source, E.O. Lawrence Berkeley National Laboratory, Berkeley,
CA 94720, USA
E-mail: cclo@eecs.berkeley.edu
Abstract. We report fabrication of transistors in a FinFET geometry using
isotopically purified silicon-28 -on-insulator (28-SOI) substrates. Donor electron spin
coherence in natural silicon is limited by spectral diffusion due to the residual 29Si
nuclear spin bath, making isotopically enriched nuclear spin-free 28Si substrates a
promising candidate for forming spin quantum bit devices. The FinFET architecture
is fully compatible with single-ion implant detection for donor-based qubits, and the
donor spin-state readout through electrical detection of spin resonance. We describe
device processing steps and discuss results on electrical transport measurements at 0.3
K.
PACS numbers: 03.67.Lx 85.35.-p

21. Introduction
Silicon based quantum computation with donor qubits has attracted much attention
since its original proposal by Kane [1]. The prevalence and maturity of silicon processing
technologies offer a convenient platform for pursuing silicon-based quantum computation
schemes. More importantly, it has been long known that donor electron and donor
nuclei spins have extraordinary long spin-coherence times at cryogenic temperatures in
isotopically purified 28Si substrates [2, 3], with the long spin lifetime being attributed
to the weak spin-orbit interaction in the crystal lattice and the nuclear spin-free matrix.
However, many challenges have to be met in order to demonstrate donor spin qubits,
including the fabrication of a readout device doped with single donors [4, 5, 6, 7, 8] ,
and the implementation of a suitable single spin-state measurement.
Single-ion implantation is a technique that enables single atom placement for
donor qubit device formation. Early single-ion implants were achieved by sensitive
detection of secondary electrons from single-ion implantation events [9]. More recently,
electrical detection of single-ion impacts have been achieved with the use of p-i-n
photodetector-like structures [6, 10] as well as with metal-oxide-semiconductor (MOS)
field-effect transistor (FET) architectures [7, 8]. In our earlier work [7], conventional
planer FETs were used with an aperture formed in the middle of the gate, allowing
implanted ions to travel through. While the experiment serves as a proof-of-concept
single-ion detection scheme using FETs, the planer device structure is not ideal as the
gate has to be partially removed in order for the single-ion implantation event to take
place. The partial gate removal potentially damages the silicon channel and introduces
instability issues into the device. To circumvent this problem, we explore the possibility
of utilizing the non-planer device architecture of the FinFET. The FinFET architecture
allows both single-ion implantation detection and spin-state readout capabilities via
electrically detected magnetic resonance (EDMR) [11, 12]. Recently, we reported studies
of ion implant detection and ion impact mapping using FinFETs [13]. In this paper,
we describe our device design strategies, the device fabrication process and first low
temperature transport measurements of FinFETs formed in isotopically purified silicon-
28 -on-insulator (28-SOI).
2. Device design
While FinFETs were invented as an end-of-roadmap CMOS technology with extremely
scaled dimensions (< 50 nm gate length) [14], they also provide a convenient architecture
to achieve single-ion implantation and spin-state detection with the same device for
realizing donor qubits. Several factors with the objective of silicon quantum computation
in mind establishes the baseline for the modified device design: Shallow donors in silicon
have Bohr radii of approximately 2 nm, hence the donor electron wavefunction can
extend to >10 nm from the position of the donor nuclei. While donor electrons in bulk
silicon have extraordinarily long spin coherence lifetimes, the presence of the oxide-