You are on page 1of 4

A Memristor SPICE Model for Designing

Memristor Circuits
Mohammad Mahvash Alice C. Parker
Ming Hsieh Department of Electrical Engineering Ming Hsieh Department of Electrical Engineering
University of Southern California University of Southern California
Los Angeles, CA USA Los Angeles, CA USA
Email: mahvashm@usc.edu Email: parker@usc.edu

Abstract—The memristor, a circuit element, was first presented


by Leon Chua in 1971. The physical implementation of the
memristor was created by scientists at HP Labs in 2008 and
the coupled variable resistor model was proposed. Here we show
a SPICE model for such a memristor using dependent voltage
sources. The model is validated by simulating simple circuits and
comparing with the expected results. The proposed model makes
it possible to design and simulate memristor circuits using SPICE.
We simulate two circuits, a low pass filter in which a memristor is
in series with a resistor and an integrator circuit with operational
amplifier. The results are compared with inductor circuits in
which a memristor is replaced by an inductor. The comparison
shows that a memristor is acting like an inductor under certain
conditions. Since the memristor has great performance in terms
of power dissipation and with its nanometer size there might be
a possible application of the memristor to be used as an inductor.

I. I NTRODUCTION Fig. 1. The coupled variable resistor model

In 1971 Leon Chua presented the missing circuit element


that he called a memristor (memory + resistor) [1]. He noted present another possible application of the memristor, using
among the six possible combinations of the four fundamental the memristor as an inductor under certain conditions. We also
circuit variables, i, v, q and φ, five have well-known relation- show the use of the memristor in an integrator circuit and in
ships. Two of them are the definition of current and Faraday’s a low-pass filter.
law and the rest are given by the three circuit elements,
dφ dq II. S PICE MODEL OF A MEMRISTOR
resistor(R = dv dt ), inductor(L = di ) and capacitor(C = dv ).
Based on the symmetry, he claimed there should be a forth In the SPICE model of a memristor presented here, there
fundamental element called a memristor which is a relation is a thin semiconductor film that has two regions, one with a
between charge and magnetic flux(M = dφ dq ). Although he
high concentration of dopant that behaves like a low resistance
presented memristor’s laboratory realization in the form of called RON and the other with a low dopant concentration
active circuits, before May 2008 the existence of a memristor with higher resistance called ROF F . The film is sandwiched
in the form of a physical device had not been discovered. between two metal contacts (Fig. 1). When we apply a voltage
In May 2008, scientists at HP labs, led by R. Stanley v(t) to the device, the dopants drift from low to high or high
Williams, announced an invention of a physical device for the to low concentration depending on the voltage polarity. For
memristor [2], [3]. They also presented a physical model of simplicity, we assume linear ionic drift in a uniform electric
a memristor called the coupled variable resistor model, which field with average ion mobility µV . In this case, the V-I
works like a perfect memristor under certain conditions. In this characteristic of the device is
paper, we present a simplified SPICE model for memristors
w(t) w(t)
based on the coupled variable resistor model. While a propri- v(t) = (RON + ROF F (1 − ))i(t) (1)
D D
etary SPICE model for the HP technology exists, no model has
been available for general use. In order to design and simulate w is the state variable which is the length of the doped
circuits with memristor elements, a SPICE model is required. region in the thin film (Fig. 1). dw(t)
dt is proportional to the
Applications of the memristor include implementing neu- current and therefore w is a function of charge.
romorphic circuits using memristive nanodevices [4], [5], RON
[6] and ultra dense nonvolatile memories. In this paper we w(t) = µV q(t) (2)
D

978-1-4244-7773-9/10/$26.00 ©2010 IEEE 989


Fig. 2. 100 ohm resistor subcircuit
Fig. 3. Memristor subcircuit

Based on Equations (1) and (2), in fact a memristor is acting


like a variable resistor; its resistivity M (q) is a function of
charge.

µV RON
M (q) = ROF F − (ROF F − RON ) q(t) (3)
D2
In order to model a variable resistor in SPICE, we first
model a regular resistor with a dependent voltage source
without using a resistor. Fig. 2 shows how to model a 100Ω
ohm resistor in SPICE. To sense current in the circuit, we
use an independent voltage source Vsense which is 0.0 volts
and therefore it has no effect on the output voltage. The other
source, Vr is a dependent voltage source that generates the
Fig. 4. Voltage and current of the memristor
voltage across the resistor based on the sensed current times
the desired resistance (Vr = I ∗ 100Ω), where I is measured
at Vsense .
Now to create a model for a memristor that is a variable
resistor, all we need to do is change the value of the voltage
source Vr to a function of q based on Equation (3). As shown
in Fig. 3, a capacitor Csense is added to sense the charge in the
circuit. Note that to cancel the effect of the capacitor voltage
on the output voltage, we subtract it from the Vr voltage value.
(Vr = I ∗ M − Vc ), where Vc is the voltage across Csense .
We validate the model by simulating for different input
voltages such as DC, sinusoidal and square wave signals and
show a curve here comparing with the expected results. Fig.
4 shows the current of the memristor when a sinusoidal input
voltage is applied. The V-I characteristic of the memristor
is shown in Fig. 5. (RON = 1Ω, ROF F = 160Ω, D =
10nm, µV = 10−14 m2 /sec.volts) Note that this model is valid Fig. 5. i-v curve
as long as the system remains in the memristor regime which,
in this case, is where the state variable w is bounded between
zero and D. Based on Equation 2, the condition for the circuit is non-linear, in addition to the main frequency, other
memristor regime is as follows: frequency components are generated. However the main fre-
quency has the highest magnitude. Fig. 7 shows the frequency
D2
0 ≤ q(t) ≤ (4) response when 0.3sin(ωt) is applied. The result is similar to
µV RON a simple RL filter (inductor in series with resistor) however
when the input signal amplitude changes, the response would
III. L OW PASS FILTER be different, which means the memristor is not acting like a
A low-pass filter using a memristor in series with a resistor fixed inductor but depends on the level of the voltage applied
is proposed (Fig. 6). In order to obtain the gain-magnitude to the memristor; the equivalent inductor is different. As shown
frequency response we apply a sine wave and measure the in Fig. 7, at a high frequency, the memristor is acting like a
magnitude of the main frequency in the output. Since the resistance and the circuit is changed to a voltage divider.

990
Fig. 8. Integrator with memristor

Fig. 6. Low pass filter with memristor

Fig. 9. Input and output waveforms in ramp shape regime

Fig. 7. Frequency response of the low-pass filter


(We retain only the negative sign in “±” to satisfy the
memristor regime condition in Equation 4). Using i = dqdt
IV. I NTEGRATOR we have
Fig. 8 shows an integrator using a memristor and an v0
i(t) = p 2 (8)
operational amplifier. When a square wave is applied to the ROF F − 2v0 rt
input, the output would be a ramp signal, which is in fact
the integral of the input voltage (Fig. 9). As compared to an v0 v02 r 12v03 r2 2
i(t) = + 3 t + 5 t + ... (9)
integrator with inductor, the memristor acts like an inductor ROF F ROF F ROF F
under certain conditions. If the period of the input signal or We can approximate the current using the first two terms
the amplitude of the input square wave increases, the output of the Taylor series, if the third term is much smaller than
is not a ramp shape any more (Fig. 10). the second term. This is the condition in which memristor is
Now we calculate the conditions in which the memristor equivalent to an inductor :
could be replaced by an inductor and also an equivalent
2
inductor value. Based on Equation 3 and the definition of the ROF F
v0 T  (10)
memristor (dφ = M (q)dq), we obtain the flux as a function 6r
of the charge: As compared to the output of an integrator with an inductor,
µV RON 2 the equivalent inductor is calculated.
φ(t) = ROF F q(t) − (ROF F − RON ) q (t) (5)
2D2 3
ROF F D2 1
Leqv = (11)
We identify r = (ROF F − RON ) µVDR2ON
. We assume the RON (ROF F − RON ) µV v0
input is v0 at time[0,T/2] where T is the period of the signal.
Using Faraday’s law (v(t) = dφ dt ) we find the flux as a function
V. C ONCLUSION
of time and then replace it in Equation 5 to to have the charge A SPICE model for the memristor based on the coupled
as a second-order equation of time as follows: variable resistor model was proposed and was validated by
r 2 simulating simple circuits and comparing with the expected
q (t) − ROF F q(t) + v0 t = 0 (6) results. A low pass filter and an integrator circuit using a
2
memristor were simulated and the results were analyzed. The
memristor is acting like an inductor under certain conditions.
p
ROF F ± ROF 2
F − 2v0 rt
q(t) = (7) The conditions were obtained and the equivalent inductor was
r

991
Fig. 10. Input and output waveforms in non linear regime

calculated. Comparing characteristics of the memristor such


as power and size with conventional inductors, we conclude a
possible application of the memristor is as an inductor.
R EFERENCES
[1] L. Chua, “Memristor-the missing circuit element,” Circuits Theory, IEEE
Transactions on [legacy, pre - 1988], vol. 18, no. 5, pp. 507-519, 1971.
[2] J. M. Tour and T. He, “The forth element,” Nature, vol. 453,pp. 42-43,
2008.
[3] D. B. Strukov, G. S. Snider, D. R. Stewart, and S. R. Williams, “The
missing memristor found,” Nature, vol. 453, no. 7191, pp. 80-83, 2008.
[4] L. O. Chua and S. M. Kang, “Memristive devices and systems.” Proc
IEEE, vol. 64, no. 2, pp. 209-223, 1976.
[5] L. Chua, “Nonlinear circuit foundations for nanodevices, I - The four
element torus,” Proc IEEE, vol. 91, no. 11, pp. 1830-1859, 2003.
[6] G. S. Sinder, “Spike-timing-dependent learning in memristive nanode-
vices,” ,2008

992