Control of Multi-level Voltage States in a Hysteretic SQUID Ring-Resonator System
Extracted source text ยท arXiv cond-mat/0405320
\documentclass[twocolumn,pre,a4paper]{revtex4}
\usepackage{amsfonts}
\usepackage{amsmath}
\usepackage{amssymb}
\usepackage{graphicx}
\begin{document}
\title{Control of Multi-level Voltage States in
a Hysteretic SQUID Ring-Resonator System}
\author{P. Stiffell}
\affiliation{School of Science and Technology,
University of Sussex,
Brighton,
Sussex,
BN1 9QT,
U.K.}
\author{M.J. Everitt}
\affiliation{School of Science and Technology,
University of Sussex,
Brighton,
Sussex,
BN1 9QT,
U.K.}
\author{T.D. Clark}
\email{t.d.clark@sussex.ac.uk}
\affiliation{School of Science and Technology,
University of Sussex,
Brighton,
Sussex,
BN1 9QT,
U.K.}
\author{J.F. Ralph}
\affiliation{Department of Electrical Engineering and Electronics,
University of Liverpool,
Brownlow Hill,
Liverpool,
L69 3GJ,
U.K.}
\keywords{SQUID, Non-linear, Multi-level States}
\begin{abstract}
In this paper we study numerical solutions to the quasi-classical
equations of motion for a SQUID ring-radio frequency (rf) resonator
system in the regime where the ring is highly hysteretic. In line
with experiment, we show that for a suitable choice of of ring
circuit parameters the solutions to these equations of motion
comprise sets of levels in the rf voltage-current dynamics of the
coupled system. We further demonstrate that transitions, both up and
down, between these levels can be controlled by voltage pulses
applied to the system, thus opening up the possibility of high order
(e.g. 10 state), multi-level logic and memory.
\end{abstract}
\maketitle
\section{Introduction}
In an earlier paper~\cite{PranceWCPSRAE99} we reported on a new phenomena generated by
the non-linear interaction of a SQUID ring (here, a single Josephson
weak link enclosed by a thick superconducting ring) with a parallel
$LC$ resonant (tank) circuit. The block diagram for this coupled
system is shown in
\begin{figure}[tb]
\begin{center}
\resizebox*{0.4\textwidth}{!}{\includegraphics{figure01.eps}}
\caption{ Block diagram for an inductively coupled SQUID ring-tank circuit
system together with excitation, static flux bias and readout
circuitry. }
\label{fig:tc+s}
\end{center}
\end{figure}
figure~\ref{fig:tc+s}. As has now become apparent, SQUID rings can
display behaviour ranging from fully quantum
mechanical~\cite{FriedmanPCTL00,vanderWaltWSHOLM00,Spiller00,ChiorescuNHM03}
through to
quasi-classical~\cite{SchieveBJ88,SoerensenBCB85,Likharev96,ChuaDK87}
depending on the circuit parameters of the ring and the temperature
($T$) of the environment. At temperatures of a few K, and for
relatively large weak link capacitances $\left( 10^{-12}-10^{-13}
\mathrm{F}\right) $, it is well
established~\cite{SchieveBJ88,Likharev96} that a SQUID ring (inductance
$\Lambda$, weak link critical current $I_{c}$) can be treated
quasi-classically, i.e. as a particle moving, with damping, in a
cosine modulated parabolic potential,
\begin{equation}
U\left( \Phi_{s},\Phi_{x}\right)
=\frac{\left( \Phi_{s}-\Phi_{x}\right)^{2}}{2\Lambda} -
\frac{I_{c}\Phi_{0}}{2\pi}\cos\left( \frac{2\pi\Phi_{s}}
{\Phi_{0}}\right)
\label{squidpot}
\end{equation}
where $\Phi_{x}$ is the external magnetic flux applied to the ring,
$\Phi_{s}$ is the total included flux in the ring and $\Phi_{0}\left(
\triangleq h/2e\right) $ is the superconducting flux quantum. In
this quasi-classical regime it is usual practice to describe the SQUID
ring, and its interactions, using the Resistively Shunted Junction
plus Capacitance (RSJ+C) model~\cite{Likharev96,BaroneP82}. Here, we denote the ring
weak link effective capacitance by $C_{s}$ and the parallel link
resistance by $R_{s}$. Furthermore, in making use of this model it is
convenient to introduce the $\beta$-parameter for the SQUID
ring $\left( \triangleq2\pi\Lambda I_{c}/\Phi_{0}\right)$. This allows
us to distinguish between the parameter space in which $I_{c}$ is
always a single-valued function of $\Phi_{x}$ (the inductive regime)
and the hysteretic regime where $I_{c}$ can be multi-valued in
$\Phi_{x}$~\cite{Likharev96,BaroneP82}.
The cosine in the potential (\ref{squidpot}) leads to a
superconducting screening current response in the SQUID ring which is
manifestly a non-linear, and $\Phi_{0}$-periodic, function of
$\Phi_{s}$. When at least part of this external flux is time
dependent, and applied through an inductively coupled resonant
circuit, the ring-resonator system displays non-linear dynamical
behaviour which has been the focus of much research over the last
three decades~\cite{SchieveBJ88,SoerensenBCB85,ClarkRPPDW98} It is precisely this behaviour that forms the
basis of the well know ac-biased SQUID magnetometer~\cite{Likharev96,BaroneP82,Lounasmaa74}.
Although not invariably the case, it has been common practice~\cite{Lounasmaa74}
to make use of a radio frequency (rf $\approx$ 20\textrm{MHz}),
parallel $LC$ (tank circuit) resonator in these coupled magnetometer
systems. In the work described here we adopt this frequency regime for
the resonant circuit. Considered on its own, of course, the tank
circuit circuit is strictly linear, i.e. on resonance the rf voltage
$\left( V_{out}\right) $ across it is linearly dependent on the level
of rf drive input current $\left( I_{in}\left( t\right) \right) $
applied. However, when a SQUID ring is coupled to such a tank circuit
the situation can alter radically. With the SQUID ring potential
above, and at finite critical current $I_{c}$, the screening
supercurrent $I_{s}=\Phi_{s}/\Lambda$ flowing in the ring to oppose
any externally applied flux $\Phi_{x}$ is a non-linear function of
this flux. Since both the SQUID ring and the tank circuit are
macroscopic in nature, the back reaction through the inductive
coupling leads to the system displaying non-linear, even chaotic,
dynamics. In general the stronger the non-linearity in $I_{s}\left(
\Phi_{x}\right) $, the more pronounced the non-linear behaviour can
be. This non-linear behaviour is most easily seen in plots of
$V_{out}$ versus $I_{in}\left( t\right) $, these also being dependent
on the level of static or quasi-static bias flux $\left(
\Phi_{x}\right) $ applied to the SQUID ring. In SQUID magnetometer
systems these characteristics display essentially constant rf voltage
steps in the (time averaged) $V_{out}$ versus $I_{in}\left( t\right) $
at intervals periodic in $I_{in}$. These SQUID steps are modulated in
a $\Phi_{0}$-periodic manner by the external bias flux~\cite{Likharev96,Lounasmaa74,ZimmermanTH70}.
The ring-tank circuit system is described dynamically by two (coupled)
equations of motion, one for the tank circuit and the other for the
ring. These are given by
\begin{widetext}
\begin{alignat}{2}
C_{tc}\frac{d^{2}\Phi_{tc}}{dt^{2}}+\frac{1}{R_{tc}}\frac{d\Phi_{tc}}
{dt}+\frac{\Phi_{tc}}{L_{tc}\left( 1-K^{2}\right) }
\ &=
\ I_{in}\left( t\right) + \frac{\mu\Phi_{s}}{\Lambda\left( 1-K^{2}\right) }
&\qquad\text{(Tank circuit)}
\label{tcequmotion}\\
C_{s}\frac{d^{2}\Phi_{s}}{dt^{2}}+\frac{1}{R_{s}}\frac{d\Phi_{s}}{dt}
+I_{c}\sin\left( \frac{2\pi\Phi_{s}}{\Phi_{0}}\right) +\frac{\Phi_{s}
}{\Lambda\left( 1-K^{2}\right) }
\ &=
\ \frac{\mu\Phi_{tc}}{\Lambda\left(1-K^{2}\right) }
&\qquad\text{(SQUID ring)}
\label{squidequmotion}
\end{alignat}
\end{widetext}
where the subscripts $tc$ and $s$ refer, respectively, to the tank
circuit and the SQUID ring and $\mu\left( =M/L_{tc}\right) $ is the
fraction of the flux coupled between the ring and the tank circuit.
Thus, $C_{tc}$ and $L_{tc}$ are, respectively, the tank circuit
capacitance and inductance, $\Phi_{tc}$ is the flux in the tank
circuit inductor, $R_{tc}$ is the resistance of the parallel tank
circuit on resonance and $K\left( =\sqrt{M^{2}/\Lambda L_{tc}}\right)
$ quantifies the strength of the inductive coupling between the ring
and tank circuit with a mutual inductance of $M$. In
(\ref{tcequmotion}) and (\ref{squidequmotion}) the last terms on the
right hand sides of these equations describe the back reaction between
the ring and the tank circuit.
In the original paper~\cite{PranceWCPSRAE99} we used (\ref{tcequmotion}) and
(\ref{squidequmotion}) to compute the $V_{out}$ versus $I_{in}$
dynamics of the ring-tank circuit system in the highly hysteretic,
strongly underdamped, regime to model the observed experimental
behaviour. What was recorded experimentally was a succession of
plateau regions in the time averaged $V_{out}$ versus $I_{in}$
characteristics of the ring-tank circuit system. Each plateau region
consisted of a set a parallel steps at regular separations along the
$I_{in}$ axis. The lengths of these steps along this axis were much
greater than those observed in standard hysteretic SQUID magnetometer
characteristics, where, typically, $\beta\approx$ a few~\cite{Likharev96,Lounasmaa74}. In
addition, for each individual plateau the rf voltage $\left(
V_{out}\right) $ at which these steps occurred varied
$\Phi_{0}$-periodically in $\Phi_{s}$. In experiment~\cite{PranceWCPSRAE99} the
ring-tank circuit system was observed to jump
stochastically~\cite{GammaitoniHJM98,BulsaraG96,LindnerMNBDB01} between the various steps associated with
each plateau. Since, in practice, it was easy experimentally to access
circuit parameters in which many steps per plateau region could be
seen (a maximum of 19 steps/plateau was recorded over the course of
these experiments), there seemed a possibility that the plateaux and
steps could be utilised to create multi-level logic as an alternative
to the standard binary logic. We found that with simulated low
temperature noise on the tank circuit drive current, introduced via a
noise distribution for a thermal bath, the solutions to these coupled
equations of motion modelled our experimental results very well,
including the stochastic jumping between steps on particular plateaux.
We argued in the paper~\cite{PranceWCPSRAE99} that when the SQUID ring is sufficiently
underdamped it no longer follows the potential (\ref{squidpot})
adiabatically as the rf flux coupled in from the tank circuit changes
with time. Essentially, it is this non-adiabatic response that
generates the large multi-step plateaux in the ring-tank circuit
$V_{out}$ versus $I_{in}$ characteristics, each step corresponding to
a different flux jump trajectory (local well to local well) in
(\ref{squidpot}). Nevertheless, even though we had demonstrated that
the underdamped RSJ+C description could model the experimentally
observed plateaux and steps, the stochastic jumping between steps
generated in the theoretical calculations required more explanation.
In the experiments we took the view, we believed quite validly, that
these jumping processes were caused by ambient noise. However, for
technical reasons we were not able to measure the full spectral
density function of this noise and with the computer power then
available to us we could not be certain that the stochastic jumping
processes seen in the simulations did not arise because of
computational inaccuracies. New calculations, at much higher
accuracy, are the basis of the work reported here. At this new level
we have been able to show that, from the viewpoint of the simulations
presented in our previous paper~\cite{PranceWCPSRAE99}, the jumping processes arose due
to the build up of computational error in the numerical integration
for the evolution of the system. Repeating the calculations, and
eliminating this potential problem by using more accurate integration
methods, only single steps are generated in each plateau region, even
with large variance current noise added. Nevertheless, it was clear
from experiment that the ring-tank circuit system could be perturbed
to generate jumps. We now show how this can be achieved in a
controlled way, pointing to possible device applications in
multi-level logic.
\section{SQUID ring-tank circuit dynamics in the highly hysteretic regime}
\begin{figure}[tb]
\begin{center}
\resizebox*{0.48\textwidth}{!}{\includegraphics{figure02.eps}}
\caption{Dynamical rf voltage $\left( V_{out}\right) $ versus rf current
$\left( I_{in}\right) $ characteristics for a highly hysteretic
$\left( \beta=137\right) $ SQUID ring-tank circuit system with
circuit parameters ring: $C_{sq}=1\times10^{-13}\mathrm{F}$,
$\Lambda_{sq}=6\times10^{-10}\mathrm{H}$, $R_{sq}=10\Omega$,
$I_{c}=75.2\mu \mathrm{A}$; tank circuit:
$C_{tc}=7.6\times10^{-10}\mathrm{F}$, $L_{tc}=63\mathrm{nH}$ and $\mu\left(
=M^{2}/L_{tc}\Lambda;\right) =0.0087$, a bare tank circuit
resonant frequency $23\mathrm{MHz}$ and $\Phi_{x}=0.0\Phi_{0}$ (modulo
$n\Phi_{0}$, $n$ integer).\vspace*{-12pt}}
\label{fig:VvI}
\end{center}
\end{figure}
\begin{figure}[b]
\begin{center}
\resizebox*{0.48\textwidth}{!}{\includegraphics{figure03.eps}}
\caption{ Time dependence of the normalised SQUID ring flux $\Phi_{sq}/\Phi
_{0}$ as a function of time, normalised to the tank circuit period
$2\pi \sqrt{L_{tc}C_{tc}}$. }
\label{fig:SFlx}
\end{center}
\end{figure}
With access to more computational power we considered it important to
establish the origin of the stochastic jumping in the computer
simulations as previously published~\cite{PranceWCPSRAE99}. As an example of the
possible origin of this jumping phenomenon we computed the $V_{out}$
versus $I_{in}$ characteristics for a SQUID ring-tank circuit in the
large $\beta$ regime. In this we chose suitable ring and tank circuit
parameters for this regime, specifically for the SQUID ring:
$\Lambda_{s}=6\times10^{-10}$\textrm{H}, $I_{c}=75.2\mu$\textrm{A}
$\left( \beta=137\right)$, $C_{s}=10^{-13}$\textrm{F},
$R_{s}=10\Omega$ and for the tank circuit: $C_{tc}=7.6\times
10^{-10}$\textrm{F}, $L_{tc}=63$\textrm{nH} and a quality factor $Q$
of $500$, the latter parameters yielding a bare (uncoupled to the
ring) tank circuit resonant frequency of $23\mathrm{MHz}$. We also set
$\mu\left( =M^{2}/L_{tc}\Lambda;\right) =0.0087$, a coupling quite
typical of low noise SQUID ring-tank circuit systems. These are the
circuit parameters used throughout this paper.
In figure~\ref{fig:VvI} we show two sets of computed $V_{out}$ versus
$I_{in}$ characteristics using these ring and tank circuit parameters.
With these values of $R_{s}$ and $C_{s}$ the SQUID is underdamped. The
originally reported characteristic~\cite{PranceWCPSRAE99}, calculated at relatively low
accuracy using a fourth order Runge-Kutta numerical integration
routine with an adaptive step size algorithm,is plotted in
figure~\ref{fig:VvI}(a) for the first plateau region. In generating
this characteristic 4.2 Kelvin current noise, with a thermal noise
distribution, has been introduced and the bias flux has been set at
$\Phi_{x}=0.0\Phi_{0}$ (modulo $n\Phi_{0}$). Here, the multi-step
solutions on the first plateau, and the stochastic jumping between
these, are perfectly clear. Since the initial calculations were
performed, we have had access to much greater computational power and
this has allowed us to apply significantly higher accuracy to our
numerical integration. Subsequently we have found that this greatly
improved accuracy leads to the suppression of the jumps between the
multiple levels in the $V_{out}$ versus $I_{in}$ characteristics. This
is very apparent in the other solution shown in figure~\ref{fig:VvI}(b)
for which we used exactly the same SQUID and tank circuit parameters
as in the characteristic of figure~\ref{fig:VvI}(a). This second solution
implies that the experimentally observed jumps may be driven by noise
processes but not through current noise as we previously thought.
Within the circuit model we have adopted, the alternative is that the
driving force arises from some form of voltage perturbation. In order
to investigate this possibility we present a number of simulations of
the response of this system to voltage pulses applied to the SQUID
ring.
\begin{figure*}[t]
\begin{center}
\resizebox*{0.95\textwidth}{!}{\includegraphics{figure04.eps}}
\caption{The SQUID ring potential for the system is shown in grey for
\textbf{(a)} a positive $\Phi_{s}$ going, outward, multi-$\Phi_{0}$
trajectory; \textbf{(b)} return trajectory for (a); \textbf{(c)} a
negative going, multi-$\Phi_{0}$ trajectory; \textbf{(d)} return
trajectory for (c).}
\label{fig:SFTraj}
\end{center}
\end{figure*}
In considering the dynamical basis for the plateaux (and the step
levels) it is helpful to examine the way in which the SQUID ring flux
$\left( \Phi _{s}\right) $ changes in time with the rf drive current
$\left( I_{in}\right) $. Such a response, is shown in
figure~\ref{fig:SFlx} for a maximum (sinusoidal) drive flux in the tank
circuit of $13.5\Phi_{0}$ peak to peak (equivalent to a drive flux in
the SQUID ring of $0.1175\Phi_{0}$) . In this figure the SQUID ring
flux is normalised to $\Phi_{0}$ and the time axis is plotted in units
of the reciprocal tank circuit period $\left(
2\pi\sqrt{L_{tc}C_{tc}}\right)$ and, for comparative timing
purposes, we show the form of $I_{in}$ in light grey. What this
computed plot demonstrates is that, for the most part, the oscillating
flux $\left( \Phi_{s}\right)$ in the underdamped SQUID ring
corresponds to the ring being confined to a single well. In
figure~\ref{fig:SFlx} this single well SQUID ring flux response is shown
oscillating over small amplitudes around $\Phi_{s}/\Phi_{0}=0$.
However, when the drive current reaches maximum amplitude it is found
that the SQUID ring can move outside the confines of a single well and
make more extended trajectories in the potential (\ref{squidpot}),
i.e. the flux in the ring can jump non-locally (and non-adiabatically)
several $\Phi_{0}$ in $\Phi_{s}$-space between initial and final
(target) wells~\cite{BenjacobA81}. We now provide some illustrative examples of
stable solutions of equations (\ref{tcequmotion}) and
(\ref{squidequmotion}) which give rise to the different voltage levels
seen in figure~\ref{fig:VvI}(a).
In the particular example of figure~\ref{fig:SFlx} these non-local jumps
take place over $\pm$ six wells in $\Phi_{s}$-space relative to the
lowest energy well in the potential. Transposed to the rf dynamics of
the coupled system, each multi-$\Phi_{0}$ traversal leads to one of
the set of plateau step levels in $V_{out}$ versus $I_{in}$, as
exemplified in figure~\ref{fig:VvI}. Examples of these multi-$\Phi_{0}$
trajectories (jumps) in the SQUID ring potential can be seen in
figure~\ref{fig:SFTraj}, again for the system of figure~\ref{fig:VvI}.
Taking the origin as the lowest well in this potential, the SQUID flux
trajectories shown in figures~\ref{fig:SFlx} and~\ref{fig:SFTraj}
correspond to the level denoted $l_{0}$ in figure~\ref{fig:VvI}(a). If
the origin is displaced to the next lowest well in the potential, the
$V_{out}$ versus $I_{in}$ characteristic of the coupled system moves
down by one level from the original solution. This level is denoted
$l_{1}$ in the solutions plotted in figure~\ref{fig:VvI}(a). Moving
the origin again to the next lowest well repeats the process, i.e. the
dynamics now correspond to the level denoted by $l_{2}$ in the
solutions plotted in figure~\ref{fig:VvI}(a). Again, moving the origin
to the next lowest well generates the level denoted $l_{3}$ in
figure~\ref{fig:VvI}(a). However, in our example of a $6\Phi_{0}$
excursion this level ($l_{3}$), is equivalent to having displaced our
origin to the middle well. We find that beyond this central well in
the excursion, attempting to localise our trajectory around one of the
remaining three wells simply reverses the shift between the levels on
the plateau in figure~\ref{fig:VvI}(a), i.e. starting at $l_{3}$ and
returning to $l_{0}$ through levels $l_{2}$ and $l_{1}$. To illuminate
this discussion, we can examine, for level $l_{0}$, the excursions in
the potential shown in the figure~\ref{fig:SFTraj}. Starting in the
absolute minimum well of the potential at $\Phi_{s}/\Phi _{0}=0$, with
$\Phi_{x}=0$ (modulo $n\Phi_{0}$), the ring can either execute local
motion in this well or occasionally move several $\Phi_{0}$ to a
target well. With the ring potential plotted in light grey in
figure~\ref{fig:SFTraj}, we show by computation in
figure~\ref{fig:SFTraj}(a) one such extended trajectory, starting in
the lowest well and followed by a positive going trajectory in
$\Phi_{s}$. In this example, having reached the target well the ring
completes almost half a tank circuit period in this well before
returning to its original starting point in $\Phi_{s}$, as shown in
figure~\ref{fig:SFTraj}(b). This mirrors the dynamical behaviour
shown in the first four jumps of figure~\ref{fig:SFlx}. It is also
important to note that these non-local excursions do not occur every
time the rf drive flux reaches its maximum amplitude, although as the
amplitude of the rf drive grows this process happens more frequently.
Further multi-$\Phi_{0}$ excursions in the ring potential, but now for
negative going $\Phi_{s}$, can be seen in figures~\ref{fig:SFTraj}(c)
and~(d) for the outward and return paths, respectively. These
correspond to the last two pulses in figure~\ref{fig:SFlx}. In general the
target well in the SQUID ring is not symmetric about its centre
(minimum) and this is seen to affect the lifetime of the ring in this
target well, as evidenced in plots presented in figure~\ref{fig:SFlx}. In essence
the consequence of this is that the ring can reach a new traversal
(target) point in drive flux before it has completed half a $\Phi_{0}$
period in $\Phi_{s}$. This phenomenon, arising from the localisation
in a non-symmetric well in the SQUID ring potential, means that the
excursions in $\Phi_{s}$ are only made in one direction.
With the details of figure~\ref{fig:SFTraj} in mind, and as a possible means
to navigate the solutions of (\ref{tcequmotion}) and
(\ref{squidequmotion} ), we now consider the effect of rapidly
changing voltages applied to the SQUID ring. In order make this
relatively simple, and physically transparent, we simulate the
application of appropriately shaped voltage pulses to the SQUID ring,
and follow the tank circuit voltage ($V_{out}$) response. In our
example we shall now consider positive amplitude trapezoidal voltage
pulses (inset in figure~\ref{fig:MLJ}) with an upper voltage state time
duration of 0.01 tank circuit periods which is activated during the
flux traversal shown in bold in figure~\ref{fig:SFTraj}(d). We found that it
was only within such highlighted activation regions that the system
could be made to respond to these voltage pulses, leading to a level
change in $V_{out}$. For reference we term pulses with positive
amplitude as A type pulse and, conversely, negative amplitude pulses
are denoted as B type.
\begin{figure}[tb]
\begin{center}
\resizebox*{0.48\textwidth}{!}{\includegraphics{figure05.eps}}
\caption{$V_{out}$ versus normalised time characteristics for the SQUID
ring-tank circuit system of figure~\ref{fig:VvI} showing the controlled step to
step jumping which can be induced using suitably configured positive
(A-type) ramp pulses in the absence of extraneous voltage noise;
here we show the effect of increasing the ramp amplitude to induce
jumps over one, two or three step intervals.}
\label{fig:MLJ}
\end{center}
\end{figure}
In figure~\ref{fig:MLJ} we have plotted the computed tank circuit voltage
response as a function of normalised tank circuit time (i.e. in terms
of the tank circuit period) for the rf drive amplitude
($I_{in}$=222nA) set in the middle of the first plateau for the level
denoted $l_{0}$ in figure~\ref{fig:VvI}(a). In this calculation there was
no extraneous flux or voltage noise present and the system was allowed
to reach a steady state within this plateau (we note that the initial
sharp rise at the beginning of the voltage dynamics is due to
transient behaviour). However, after a set time, denoted $A_{i}$ in
figure~\ref{fig:MLJ}, a voltage pulse of the form shown inset in this
figure was applied. Here, the amplitudes of the pulses $A_{i},i=1,2,3$
used were 15, 75 and 120$\mu$V, respectively. As is demonstrated in
figure~\ref{fig:MLJ}, the application of a single voltage pulse can
change the voltage level of our system by one or more levels at a
time, as summarised in table~\ref{tbl:ranges}.
\begin{table}[th]
\begin{tabular}{|c|c|c|c|} \hline
Pulse Type & Lower Bound & Upper Bound & Value Used \\ \hline
$A_1$ & 15 & 55 & 15 \\
$A_2$ & 56 & 90 & 75 \\
$A_3$ & 91 & 126 & 120 \\ \hline
\end{tabular}
\caption{Summary of pulse amplitude ranges ($\mu$V) as described in the text.
\label{tbl:ranges}}
\end{table}
If the noiseless (or, in practice, a low enough voltage noise)
situation can be realised experimentally, the computational results of
figure~\ref{fig:MLJ} indicate that multi-level logic, based on SQUID
ring-resonator systems could be a feasible proposition. If so, it is
reasonable to assume that other forms of voltage pulse could be used
for this purposes. With regard to our example, we note that the
number of levels on the plateaux depends on the ring parameter values.
In the example of figure~\ref{fig:MLJ} this is four but could be
significantly larger. To further illustrate the control possible we
show in
\begin{figure}[tb]
\begin{center}
\resizebox*{0.48\textwidth}{!}{\includegraphics{figure06.eps}}
\caption{$V_{out}$ versus normalised time characteristics for the SQUID
ring-tank circuit system of figure~\ref{fig:VvI} showing the effect
of two sequential A-type voltage pulses in the absence of extraneous
voltage noise; as is apparent these step jumps can be induced in a
controlled manner.}
\label{fig:AAJ}
\end{center}
\end{figure}
figure~\ref{fig:AAJ} the result of applying two A-type pulses, in time
sequence, to the SQUID ring-tank circuit system.
Of course, the activation pulse in voltage can be negative (B-type) as
well as positive. The effect of using such a pulse after an A-type
pulse is shown in
\begin{figure}[tb]
\begin{center}
\resizebox*{0.48\textwidth}{!}{\includegraphics{figure07.eps}}
\caption{$V_{out}$ versus normalised time characteristics for the SQUID
ring-tank circuit system of figure~\ref{fig:VvI} in the absence of extraneous
voltage noise showing the effect of applying sequential A and B-type
voltage ramps.}
\label{fig:ABJ}
\end{center}
\end{figure}
figure~\ref{fig:ABJ} with the form of the B-type pulse shown explicitly
in the inset of this figure. This shows that in the absence of
extraneous noise, and by a suitable choice of voltage pulse form, we
can move at will between the steps on any particular plateau in
$V_{out}$ versus $I_{in}$. To demonstrate that we can induce step to
step jumping at essentially any time of our choosing, we show in
\begin{figure}[tb]
\begin{center}
\resizebox*{0.48\textwidth}{!}{\includegraphics{figure08.eps}}
\caption{$V_{out}$ versus normalised time characteristics for the SQUID
ring-tank circuit system of figure~\ref{fig:VvI} in the absence of
extraneous voltage noise showing the effect of applying sequentially
at differing time intervals as a set A-type voltage ramps.}
\label{fig:AAAAJ}
\end{center}
\end{figure}
figure~\ref{fig:AAAAJ} a sequence of jumps induced by a sequence of A-type
pulses spaced by different time intervals. We note that the
application of the last pulse causes the tank circuit voltage to step
up rather than down.
The computed solutions of figures~\ref{fig:MLJ} to~\ref{fig:AAAAJ}
show very well the level to level jumping induced by various pulse
sequences applied to a highly hysteretic (large $\beta$) SQUID
ring-tank circuit system. These solutions, which are the end result
of very non-linear interactions between the ring and the tank circuit,
may be appreciated more clearly by the following qualitative argument.
As regards the jumping process between levels on any particular
plateau in $V_{out}$ versus $I_{in}$, the system operates cyclically
from an initial (local) well in the SQUID potential (for example the
$w_{0}$ well in the potential of figure~\ref{fig:SFTraj}). This is shown
diagrammatically in
\begin{figure}[tb]
\begin{center}
\resizebox*{0.48\textwidth}{!}{\includegraphics{figure09.eps}}
\caption{Graphical representation in the manner in which the wells in the
SQUID potential correspond to voltage levels in the $V_{out}$
versus $I_{in} $ characteristic of a highly hysteretic SQUID
ring-tank circuit system. The vertices around the polygons indicate
which well (in a six well example - see figure~\ref{fig:SFTraj}) the SQUID
oscillation is localised in, whilst the letters on the projection
denote which one of the set of levels on a particular plateau in
$V_{out}$ versus $I_{in}$ the coupled system occupies. See text
for more details.}
\label{fig:GrphRep}
\end{center}
\end{figure}
figure~\ref{fig:GrphRep}. The way in which movement between wells in the
SQUID potential relates to a given voltage level within a particular
plateau can be determined by thinking of each well being mapped on to
one of the vertices of a polygon (where the order of the polygon is
dependent on the parameter values of the ring-tank circuit system
under investigation). This is illustrated in figure~\ref{fig:GrphRep}(a) for
the system considered throughout this paper with the vertices labelled
$v_{i}$ for $i=0,...,5$ and where the mapping is applied cyclically
from the lowest (reference) well of the potential on to each vertex
(for our example using the labels for the wells $w_{i}$ in
figure~\ref{fig:SFTraj}(a) we have the mapping $w_{i}\rightarrow
v_{i\operatorname{mod}6}$ for $i \in \mathbb{Z}$). Here, we can find
the particular tank circuit voltage level associated with each vertex
by projection on to the horizontal axis. For example, the vertices
$v_{1}$and $v_{5}$ correspond to having the SQUID ring-tank circuit
system in level $l_{1}$ in figure~\ref{fig:VvI}. Clearly, therefore, if
the system is moved through an appropriate number of wells it will
return to the same state of operation in which it began.
Using this graphical map as the guide, in this paper we have
investigated a system with 4 levels, corresponding to
figure~\ref{fig:GrphRep}(a). We note that in this example the system
must traverse 6 wells in the potential before it returns to its
original state of operation. If we choose different SQUID parameters
the system can access a variable number of wells within its cycle of
operation. However, the system will always display a symmetry about
the horizontal axis. For comparison, we show in
figure~\ref{fig:GrphRep}(b) a diagram illustrating the situation
where the ring-tank circuit system traverses an odd number of wells in
the SQUID potential, in this case five.
By the application of suitable voltage pulses the system can be moved
around the vertices of the polygon, so translating the SQUID ring from
one potential well to another. Thus, in figure~\ref{fig:GrphRep}(a)
application of positive (A-type) pulses causes the ring to move around
the polygon in a clockwise direction, whilst negative (B-type) pulses
lead to anti-clockwise movement. This movement around the polygon can
be performed utilising nearest neighbour, single well, translations, or
by multi-$\Phi_{0}$ traversals between non-local wells in the
potential, depending on the magnitude of the applied pulse The actual
level in $V_{out}$ versus $I_{in}$ into which the system settles is
determined by three factors: (i) the well in which the SQUID ring is
currently localised (ii) the size of voltage pulse used to change
location in the potential and (iii) the direction of traversal of the
polygon (whether A or B-type pulses are utilised).
\section{Conclusions}
In this work we have demonstrated that, in principle, it is possible
to use highly hysteretic (large $\beta$) SQUID ring-tank circuit
systems as the basis for multi-level logic or memory
devices~\cite{BraytonHS90,DevadasMNS88,BartlettCDH86,LinD95,LiRGWWDW98}. Following earlier experiments, where giant SQUID
magnetometer plateaux were observed containing sets of constant
voltage $\left( V_{out}\right)$ steps, we have shown that the jumping
processes between steps on a particular plateau can be generated by
voltage pulses applied to the system. In the absence of noise (in our
computational modelling achieved by the use of high accuracy numerical
integration techniques), we find that a high $\beta$ SQUID ring-tank
circuit system can remain stably on one of the steps on any particular
plateau until a suitably shaped voltage pulse is applied to the ring.
Following such a pulse, the ring can be made to jump in a controlled
manner to other steps at different levels of $V_{out}$. This new level
can be either above or below the original voltage level, depending on
the present state of the system and the type of pulse used. Provided
it is possible to reduce the ambient voltage noise on the SQUID ring
sufficiently, for example by the use of cryogenically cooled
GaAsFET~\cite{PranceWCPSRAE99} or HEMT-based pre-amplifier electronics~\cite{BoutezCDCFB97}, it
should prove possible to develop devices with a controlled voltage
response which can be selected, and modified, at will. This could
prove useful in such areas as multi-level logic or finite state
machinery~\cite{BraytonHS90,DevadasMNS88,BartlettCDH86,LinD95,LiRGWWDW98}. As part of any such developments we would expect
to see interesting new fields open up in the non-linear dynamics of
these highly hysteretic systems.
\section{Acknowledgements}
We would like to thank the Engineering and Physical Sciences Research
Council for its generous support of this work. We would also to thank
Dr. R.J. Prance and Professor A.R. Bulsara for interesting and
informative discussions.
\bibliographystyle{apsrev}
\bibliography{ref}
\end{document}
\begin{thebibliography}{23}
\expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
\expandafter\ifx\csname bibnamefont\endcsname\relax
\def\bibnamefont#1{#1}\fi
\expandafter\ifx\csname bibfnamefont\endcsname\relax
\def\bibfnamefont#1{#1}\fi
\expandafter\ifx\csname citenamefont\endcsname\relax
\def\citenamefont#1{#1}\fi
\expandafter\ifx\csname url\endcsname\relax
\def\url#1{\texttt{#1}}\fi
\expandafter\ifx\csname urlprefix\endcsname\relax\def\urlprefix{URL }\fi
\providecommand{\bibinfo}[2]{#2}
\providecommand{\eprint}[2][]{\url{#2}}
\bibitem[{\citenamefont{Prance et~al.}(1999)\citenamefont{Prance, Whiteman,
Clark, Prance, Schollmann, Ralph, Al-Khawaja, and Everitt}}]{PranceWCPSRAE99}
\bibinfo{author}{\bibfnamefont{R.~J.} \bibnamefont{Prance}},
\bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Whiteman}},
\bibinfo{author}{\bibfnamefont{T.~D.} \bibnamefont{Clark}},
\bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{Prance}},
\bibinfo{author}{\bibfnamefont{V.}~\bibnamefont{Schollmann}},
\bibinfo{author}{\bibfnamefont{J.~F.} \bibnamefont{Ralph}},
\bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Al-Khawaja}},
\bibnamefont{and} \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Everitt}},
\bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{82}},
\bibinfo{pages}{5401} (\bibinfo{year}{1999}).
\bibitem[{\citenamefont{Friedman et~al.}(2000)\citenamefont{Friedman, Patel,
Chen, Tolpygo, and Lukens}}]{FriedmanPCTL00}
\bibinfo{author}{\bibfnamefont{J.~R.} \bibnamefont{Friedman}},
\bibinfo{author}{\bibfnamefont{V.}~\bibnamefont{Patel}},
\bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{Chen}},
\bibinfo{author}{\bibfnamefont{S.~K.} \bibnamefont{Tolpygo}},
\bibnamefont{and} \bibinfo{author}{\bibfnamefont{J.~E.}
\bibnamefont{Lukens}}, \bibinfo{journal}{Nature}
\textbf{\bibinfo{volume}{406}}, \bibinfo{pages}{43} (\bibinfo{year}{2000}).
\bibitem[{\citenamefont{van~der Wal et~al.}(2000)\citenamefont{van~der Wal, ter
Haar, Wilhelm, Schouten, Harmans, Orlando, Lloyd, and
Mooij}}]{vanderWaltWSHOLM00}
\bibinfo{author}{\bibfnamefont{C.~H.} \bibnamefont{van~der Wal}},
\bibinfo{author}{\bibfnamefont{A.~C.~J.} \bibnamefont{ter Haar}},
\bibinfo{author}{\bibfnamefont{F.~K.} \bibnamefont{Wilhelm}},
\bibinfo{author}{\bibfnamefont{R.~N.} \bibnamefont{Schouten}},
\bibinfo{author}{\bibfnamefont{C.~J. P.~M.} \bibnamefont{Harmans}},
\bibinfo{author}{\bibfnamefont{T.~P.} \bibnamefont{Orlando}},
\bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Lloyd}}, \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{J.~E.} \bibnamefont{Mooij}},
\bibinfo{journal}{Science} \textbf{\bibinfo{volume}{290}},
\bibinfo{pages}{773} (\bibinfo{year}{2000}).
\bibitem[{\citenamefont{Spiller}(2000)}]{Spiller00}
\bibinfo{author}{\bibfnamefont{T.~P.} \bibnamefont{Spiller}},
\bibinfo{journal}{Fortschritte Phys.-Prog. Phys.}
\textbf{\bibinfo{volume}{48}}, \bibinfo{pages}{1075} (\bibinfo{year}{2000}).
\bibitem[{\citenamefont{Chiorescu et~al.}(2003)\citenamefont{Chiorescu,
Nakamura, Harmans, and Mooij}}]{ChiorescuNHM03}
\bibinfo{author}{\bibfnamefont{I.}~\bibnamefont{Chiorescu}},
\bibinfo{author}{\bibfnamefont{Y.}~\bibnamefont{Nakamura}},
\bibinfo{author}{\bibfnamefont{C.~J. P.~M.} \bibnamefont{Harmans}},
\bibnamefont{and} \bibinfo{author}{\bibfnamefont{J.~E.} \bibnamefont{Mooij}},
\bibinfo{journal}{Science} \textbf{\bibinfo{volume}{299}},
\bibinfo{pages}{1869} (\bibinfo{year}{2003}).
\bibitem[{\citenamefont{Schieve et~al.}(1988)\citenamefont{Schieve, Bulsara,
and Jacobs}}]{SchieveBJ88}
\bibinfo{author}{\bibfnamefont{W.~C.} \bibnamefont{Schieve}},
\bibinfo{author}{\bibfnamefont{A.~R.} \bibnamefont{Bulsara}},
\bibnamefont{and} \bibinfo{author}{\bibfnamefont{E.~W.}
\bibnamefont{Jacobs}}, \bibinfo{journal}{Phys. Rev. A}
\textbf{\bibinfo{volume}{37}}, \bibinfo{pages}{3541} (\bibinfo{year}{1988}).
\bibitem[{\citenamefont{Soerensen et~al.}(1985)\citenamefont{Soerensen,
Bartuccelli, Christiansen, and Bishop}}]{SoerensenBCB85}
\bibinfo{author}{\bibfnamefont{M.~P.} \bibnamefont{Soerensen}},
\bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Bartuccelli}},
\bibinfo{author}{\bibfnamefont{P.~L.} \bibnamefont{Christiansen}},
\bibnamefont{and} \bibinfo{author}{\bibfnamefont{A.~R.}
\bibnamefont{Bishop}}, \bibinfo{journal}{Phys. Lett. A}
\textbf{\bibinfo{volume}{109}}, \bibinfo{pages}{347} (\bibinfo{year}{1985}).
\bibitem[{\citenamefont{Likharev}(1986)}]{Likharev96}
\bibinfo{author}{\bibfnamefont{K.}~\bibnamefont{Likharev}},
\emph{\bibinfo{title}{Dynamics of Josephson Junctions and Circuits}}
(\bibinfo{publisher}{Taylor \& Francis}, \bibinfo{year}{1986}).
\bibitem[{\citenamefont{Chua et~al.}(1987)\citenamefont{Chua, Desoer, and
Kuh}}]{ChuaDK87}
\bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{Chua}},
\bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Desoer}}, \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{E.}~\bibnamefont{Kuh}},
\emph{\bibinfo{title}{Linear and Non-Linear Circuits}}
(\bibinfo{publisher}{McGraw-Hill}, \bibinfo{year}{1987}).
\bibitem[{\citenamefont{Barone and Paterno}(1982)}]{BaroneP82}
\bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Barone}} \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Paterno}},
\emph{\bibinfo{title}{Physics and Applications of the Josephson Effect}}
(\bibinfo{publisher}{John Wiley \& Sons Inc.}, \bibinfo{year}{1982}).
\bibitem[{\citenamefont{Clark et~al.}(1998)\citenamefont{Clark, Ralph, Prance,
Prance, Diggins, and Whiteman}}]{ClarkRPPDW98}
\bibinfo{author}{\bibfnamefont{T.~D.} \bibnamefont{Clark}},
\bibinfo{author}{\bibfnamefont{J.~F.} \bibnamefont{Ralph}},
\bibinfo{author}{\bibfnamefont{R.~J.} \bibnamefont{Prance}},
\bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{Prance}},
\bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Diggins}}, \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Whiteman}},
\bibinfo{journal}{Phys. Rev. E} \textbf{\bibinfo{volume}{57}},
\bibinfo{pages}{4035} (\bibinfo{year}{1998}).
\bibitem[{\citenamefont{Lounasmaa}(1974)}]{Lounasmaa74}
\bibinfo{author}{\bibfnamefont{O.}~\bibnamefont{Lounasmaa}},
\emph{\bibinfo{title}{Experimental Principles and Methods below 1K}}
(\bibinfo{publisher}{Academic Press}, \bibinfo{year}{1974}).
\bibitem[{\citenamefont{Zimmerman et~al.}(1970)\citenamefont{Zimmerman, Thiene,
and Harding}}]{ZimmermanTH70}
\bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Zimmerman}},
\bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Thiene}}, \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Harding}},
\bibinfo{journal}{J. Appl. Phys.}
\textbf{\bibinfo{volume}{41}}(\bibinfo{number}{4}), \bibinfo{pages}{1572}
(\bibinfo{year}{1970}).
\bibitem[{\citenamefont{Gammaitoni et~al.}(1998)\citenamefont{Gammaitoni,
Hanggi, Jung, and Marchesoni}}]{GammaitoniHJM98}
\bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{Gammaitoni}},
\bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Hanggi}},
\bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Jung}}, \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{Marchesoni}},
\bibinfo{journal}{Rev. Mod. Phys.} \textbf{\bibinfo{volume}{70}},
\bibinfo{pages}{223} (\bibinfo{year}{1998}).
\bibitem[{\citenamefont{Bulsara and Gammaitoni}(1996)}]{BulsaraG96}
\bibinfo{author}{\bibfnamefont{A.~R.} \bibnamefont{Bulsara}} \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{Gammaitoni}},
\bibinfo{journal}{Phys. Today} \textbf{\bibinfo{volume}{49}},
\bibinfo{pages}{39} (\bibinfo{year}{1996}).
\bibitem[{\citenamefont{Lindner et~al.}(2001)\citenamefont{Lindner, Mason,
Neff, Breen, Ditto, and Bulsara}}]{LindnerMNBDB01}
\bibinfo{author}{\bibfnamefont{J.~F.} \bibnamefont{Lindner}},
\bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Mason}},
\bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Neff}},
\bibinfo{author}{\bibfnamefont{B.~J.} \bibnamefont{Breen}},
\bibinfo{author}{\bibfnamefont{W.~L.} \bibnamefont{Ditto}}, \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{A.~R.} \bibnamefont{Bulsara}},
\bibinfo{journal}{Phys. Rev. E}
\textbf{\bibinfo{volume}{63}}(\bibinfo{number}{4}), \bibinfo{pages}{041107}
(\bibinfo{year}{2001}).
\bibitem[{\citenamefont{Benjacob and Abraham}(1981)}]{BenjacobA81}
\bibinfo{author}{\bibfnamefont{E.}~\bibnamefont{Benjacob}} \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Abraham}},
\bibinfo{journal}{Appl. Phys. Lett.} \textbf{\bibinfo{volume}{39}},
\bibinfo{pages}{835} (\bibinfo{year}{1981}).
\bibitem[{\citenamefont{Brayton et~al.}(1990)\citenamefont{Brayton, Hachtel,
and Sangiovannivincentelli}}]{BraytonHS90}
\bibinfo{author}{\bibfnamefont{R.~K.} \bibnamefont{Brayton}},
\bibinfo{author}{\bibfnamefont{G.~D.} \bibnamefont{Hachtel}},
\bibnamefont{and} \bibinfo{author}{\bibfnamefont{A.~L.}
\bibnamefont{Sangiovannivincentelli}}, \bibinfo{journal}{Proc. IEEE}
\textbf{\bibinfo{volume}{78}}, \bibinfo{pages}{264} (\bibinfo{year}{1990}).
\bibitem[{\citenamefont{Devadas et~al.}(1988)\citenamefont{Devadas, Ma, Newton,
and Sangiovannivincentelli}}]{DevadasMNS88}
\bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Devadas}},
\bibinfo{author}{\bibfnamefont{H.~K.} \bibnamefont{Ma}},
\bibinfo{author}{\bibfnamefont{A.~R.} \bibnamefont{Newton}},
\bibnamefont{and}
\bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Sangiovannivincentelli}},
\bibinfo{journal}{IEEE Trans. Comput-Aided Des. Integr. Circuits Syst.}
\textbf{\bibinfo{volume}{7}}, \bibinfo{pages}{1290} (\bibinfo{year}{1988}).
\bibitem[{\citenamefont{Bartlett et~al.}(1986)\citenamefont{Bartlett, Cohen,
Degeus, and Hachtel}}]{BartlettCDH86}
\bibinfo{author}{\bibfnamefont{K.}~\bibnamefont{Bartlett}},
\bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{Cohen}},
\bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Degeus}}, \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Hachtel}},
\bibinfo{journal}{IEEE Trans. Comput-Aided Des. Integr. Circuits Syst.}
\textbf{\bibinfo{volume}{5}}, \bibinfo{pages}{582} (\bibinfo{year}{1986}).
\bibitem[{\citenamefont{Lin and Devadas}(1995)}]{LinD95}
\bibinfo{author}{\bibfnamefont{B.}~\bibnamefont{Lin}} \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Devadas}},
\bibinfo{journal}{IEEE Trans. Comput-Aided Des. Integr. Circuits Syst.}
\textbf{\bibinfo{volume}{14}}, \bibinfo{pages}{974} (\bibinfo{year}{1995}).
\bibitem[{\citenamefont{Li et~al.}(1998)\citenamefont{Li, Wang, Wang, Dong, and
D.X.}}]{LiRGWWDW98}
\bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Li}},
\bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Wang}},
\bibinfo{author}{\bibfnamefont{Y.}~\bibnamefont{Wang}},
\bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{Dong}}, \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{D.X.}},
\bibinfo{journal}{1998 5th International Conference On Solid-State And
Integrated Circuit Technology Proceedings} pp. \bibinfo{pages}{588--589}
(\bibinfo{year}{1998}).
\bibitem[{\citenamefont{Boutez et~al.}(1997)\citenamefont{Boutez, Crozat,
Danelon, Chaubet, Febvre, and Beaudin}}]{BoutezCDCFB97}
\bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Boutez}},
\bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Crozat}},
\bibinfo{author}{\bibfnamefont{V.}~\bibnamefont{Danelon}},
\bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Chaubet}},
\bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Febvre}}, \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Beaudin}},
\bibinfo{journal}{Int. J. Infrared Millimeter Waves}
\textbf{\bibinfo{volume}{18}}, \bibinfo{pages}{85} (\bibinfo{year}{1997}).
\end{thebibliography}