September 1969 Electronics World
Table of Contents
Wax nostalgic about and learn from the history of early electronics. See articles
from
Electronics World, published May 1959
 December 1971. All copyrights hereby acknowledged.

If you do a search on solid state inductors, you will not find much with a date later than
the time when this article appeared in Electronics World in 1969. It appears a
patent was issued in 1965, but the concept seems to never have caught on. The theory and
construction is simple enough according to the information here. Fundamentally, it
involves exploiting the properties of a Hall device when loaded with capacitive or
inductive loads to effect inductive or capacitive properties, respectively. The ability to
integrate capacitive elements into solid state substrates means the current and voltage
phase relationship of an inductor can be obtained in an IC without a physical inductive
element. Evidently practical implementation was difficult.
The SolidState Inductor
By David L. Heiserman
Fig. 1  Current flows lengthwise through a thin slice of semiconductor material. A magnetic
field cuts the flow of current carriers at right angles and creates a Hall voltage along the
edges of the semiconductor material as shown.
Fig. 2  Equivalent circuit of unloaded Hall device. In the case of a Halleffect
device with very large impedance (or open circuit) across output terminals, the longitudinal
voltage across the Hall material, V1, is simply the input current times the intrinsic d. c.
resistance of the material, R1. Since the Hall voltage, V_{h}, delivers virtually
no current to a very high impedance load, then voltage V2 equals V_{h}.
Fig. 3  Equivalent circuit of loaded Hall device. When loaded with a fairly low impedance,
Z_{e} a Hall current I2 flows through the intrinsic transverse resistance, R2, and
the external circuit. Since this current is subjected to the same Hall phenomenon as I1, the
Hall current creates a secondary Hall voltage, V'_{h}, that tends to oppose input
current.
Fig. 4  Phase relations between voltages and currents in a.c.operated
Hall device. In (A) the device is resistively loaded and all waveforms are in phase. In (B)
a pure capacitance has been connected across the output terminals. The Hall current, I2, leads
its source voltage, V_{h}, by 90 degrees. This voltage lag is reflected in the secondary
Hall voltage which is mainly responsible for the V1 potential. As shown, the voltage across
the input of the device leads the applied current by 90 degrees  just like an inductor. In
(C) the device has been loaded with a pure inductor. The device now behaves as capacitor to
input.
By combining the Hall effect with integrated circuit technology, a solidstate device with
inductive reactance has made its appearance.
Conventional thinking about inductors seems to rule out any possibility of developing solidstate
versions of inductive devices. Dr. S. Kataoka and his coworkers at the Tanashi Electrotechnical
Laboratory in Tokyo, however, departed from conventional thinking recently and produced what
may be the first practical approach to truly solidstate inductors.
Semiconductor Halleffect devices have been playing an increasingly important role in electronics
technology for over a decade. Dr. Kataoka and his staff took advantage of some littleused
Halleffect phenomena to develop their new device.
The Hall Effect
Fig. 1 shows the basic experimental arrangement for demonstrating the Hall effect. A current,
I_{i}, from a constantcurrent source passes lengthwise through a thin slice of semiconductor
material such as indium antimonide (InSb). A magnetic field, B, at right angles to the input
current bends the current carriers toward one edge of the Hall device. With an excess of charge
carriers thus gathering on one edge, a potential difference, V_{h}, develops between
the edges. This potential is the Hall voltage, and is given by the equation:
V_{h} = KI_{i}B sin θ;
where K is the Hall constant, and θ
is the angle between the input current and the magnetic field. The Hall constant depends upon
the density and mobility of the current carriers, and the thickness of the Hall device. If
the magnetic field is fixed and at rightangles to the input current, the Hall output voltage
is proportional to the input current. Likewise, holding the input current constant and changing
the strength of the magnetic field varies the Hall output voltage accordingly. The latter
effect is employed in the design of Halleffect gaussmeters.
Fig. 2 shows the equivalent circuit for a Hall device with very large impedance connected
across the output terminals. The input impedance of a Hall device with a large output load
can be approximated by the simple relationship: Z_{i} + R1. Used in this way, the
input impedance of the Hall device is simply the inherent longitudinal resistance of the Hall
material. Most presentday Hall devices are used in this mode.
Shortcircuiting the output terminals or making the load impedance very low, however, produces
quite a different input impedance (Fig. 3). Making Z_{e} very small makes the
input
impedance approximate: Z_{i} = R1 + K^{2}B^{2}/R2. With the Hall output
terminals shortcircuited, then, the input impedance rises and responds to the square of the
magnetic field strength. Such a device is used as a magnetoresistance  an effect just now
becoming popular in research and development laboratories.
Hall Devices with A. C. Inputs
Fig. 4A shows the phase relationships between I1, V_{h}, 12, and V1 for a Hall
device with a purely resistive load. Taking the constantcurrent input as a reference phase,
the figure shows that V_{h} is in phase with I1. The Hall current developed by V_{h}
is also in phase with the input as is the resulting voltage across the input terminals of
the Hall device. Since the voltage across the device, V1, is in phase with the input current,
the resistively loaded Hall device appears to be a purely resistive load to its associated
current source.
If the external load is purely capacitive, on the other hand, current through the external
load will lead the Hall voltage by 90°. The result of this phase shift, as shown in Fig.
4B, is that the voltage across the input terminals leads the input current by 90°. Thus,
with a capacitor across the output terminals, a Hall device appears inductive to the current
source.
On the other hand, connecting an inductor across the output terminals, as shown in Fig.
4C, makes the Hall device appear capacitive to the input current source.
The Kataoka SSI
Once it was established that a Hall device could simulate an inductor by connecting a capacitor
across its output terminals, Kataoka carried the circuit idea one step further and integrated
the capacitor into the Halleffect device. The Kataoka SSI (solidstate inductor) consists
of two thin slabs of InSb  one of the ptype and the other of the ntype  separated by a
thin layer of a metal oxide dielectric. With a permanent magnet fixed to the device, the two
slabs of InSb act as plates of a capacitor, and the voltage across the input terminals leads
the input current, as illustrated earlier in Fig. 4B.
Of course the metaloxide capacitor is not perfect, and the internal resistance and Hall
constant tend to limit the "Q" of the SSI. One of Dr. Kataoka's early inductors has a nominal
inductance of 570 μH, and a "Q" factor of 0.37 at 1 kHz. By conventional choke standards
this is, indeed, a poor inductor. Dr. Kataoka is quick to point out, however, that future
improvements in design will soon increase the SSI performance to the point that it may become
a major breakthrough in integratedcircuit technology.
Editor's Note: For readers who are interested in further information on the Hall effect
and on instruments that utilize this effect for magnetic measurements, refer to the article
"The Hall Effect" by John R. Collins which appeared in our April, 1963 issue.
Posted August 17, 2017