The identity of Zener breakdown is interpreted as metal-insulator transition (MIT). For a negative-differential-resistance (NDR) Si-transistor as a sort of MIT transistor, a structure of “reverse-pn-junction (insulator role for tunneling) and MIT” is proposed. Its characteristics are investigated through the reverse active mode of a donor-acceptor-donor bipolar transistor, similar to the NDR-transistor structure. As evidence of the MIT at outlet layer, the Ohmic behavior in I-V measurements and the NDR in a 100 KHz power pulse are observed. It switches a much higher current than a bipolar transistor when the MIT occurs and can be used as a power device.

We are confronted with a problem to reduce heat energy induced by high resistance at the conducting channel in a power semiconductor transistor. This can be an alternative to develop a new switching transistor with lower resistance, which is the case with a metal-insulator-transition (MIT) transistor not using the semiconductor characteristic.1 The MIT transistor can be made by two methods of electrostatic field2–5 generating the critical carrier density, nc, for MIT and supplying a MIT critical current to an insulator or a semiconductor.6 The former has been attempted in non-Si materials to enhance gate effect.3–5 

The latter has been studied in the diode structure in many materials, including Si. As exemplified in Si, an avalanche diode and an impact-ionization-transit-time diode were presented,7 as interpreted by a type of MIT device. However, a triode MIT transistor has not yet been disclosed. In particular, a Si-MIT transistor is very important, since the transistor can be directly applied to the Si-manufacturing process. Moreover, as for the Si MIT, it was known that highly doped Si undergoes the MIT in very low temperatures,8,9 and the current jump in Si-devices with low doping is Zener breakdown. It is unclear whether the identity of current-jump is due to Zener breakdown or MIT.

Here, we discuss the identity of current jump with Zener breakdown and MIT and demonstrate a structure of a Si-based negative-differential-resistance (NDR) transistor, as a sort of the MIT transistor, in order to show its operation with the NDR phenomenon induced by a 100 KHz power pulse supplying the MIT critical current.

The structure of the NDR transistor, as depicted in Fig. 1(a), is composed of three layers: an inlet (I) layer that supplies current, an outlet (O) layer that drains current, and a control (C) layer that controls current. The inlet layer has electron doping of high concentration and low resistance. The outlet layer also has electron doping, but of very low concentration, since it is in an insulated state. The hole doping in the control layer is of a moderate concentration, near an order of nc, MIT ≈1 × 1018 cm−3,3,4 and is very thin. A reverse-biased P-N (acceptor-donor) junction is formed from inlet to control (symbol in Fig. 1(a)), which plays the role of an insulator. This blocks current leakage and forces electrons coming from the outlet to tunnel through the P-N junction barrier. The MIT occurs in the outlet layer, when holes from the control layer are injected into the outlet layer (hole-driven MIT10–12 or a hole-carrying critical current) (white arrow in Fig. 1(a)). Electron carriers generated by the MIT at the outlet layer tunnel through the control layer to the inlet (red arrow). Then, a much larger current than that at semiconductor transistor flows from the inlet to the outlet (black arrow). When hole carriers in the control layer do not flow to the outlet, the inlet current ceases to flow. Layer materials, extrinsic semiconductors4 or universal materials,13 undergo the MIT. Thus, its structure is suggested as “metal-(reverse-pn junction; insulator role for tunneling)-MIT.”

It is possible to find or fabricate a device with similar or identical properties as the above structure. For example, the reverse structure of a bipolar transistor can be similar to that of the NDR transistor, in that a donor(N)-acceptor(P)-donor(N) transistor, collector, base, and emitter have respective low electron, moderate hole, and high hole doping. In the reverse active mode, in which current flows from emitter to collector, current gain at the semiconductor regime is insignificant, different from a high gain in the forward active mode. Thus, the reverse active mode is not used.

This study carried out experiments for operating the NDR transistor, using commercial, general purpose, small-signal NPN Si-transistors of 2N3904 consisting of: electron doping of 1018 cm−3 at the emitter, hole doping of 1016 cm−3 (smaller than the expected doping in the NDR transistor) at the base, and electron doping of 1015 cm−3 at the collector, which is regarded as the insulating layer inducing the MIT. The chip dimension was 300 × 300 × 180 μm (Base 100 × 100 μm, Emitter 100 × 100 μm). Top and bottom metals are Al and Au, respectively. The base thickness is less than 0.4 μm. The maximum-current specification is approximately 200 mA. Base-collector resistance is about 40 MΩ. Current gain is about 3 at the semiconductor regime in the reverse mode. A photo of the sample chip is shown in Fig. 1(a). Moreover, I-V curves in Figs. 1 and 2 were measured by a parameter analyzer of Agilent B1500A. The system in Fig. 3(b) used a function generator of Agilent 33120A, a digital oscilloscope (HP Infinium 500 MHz) and a power amplifier (FLC electronics A600).

FIG. 1.

(a) Structure and symbol of an NDR transistor. A photo of the surface of an NPN transistor chip of 2N3904 with a similar structure to that of the NDR transistor. (b) Icontrol-out–Vcontrol-out curves measured at the diode structure in the V-mode when the inlet terminal is open. Inset shows Ohmic behaviors. (c) The resistance dependence of Iout−Vin-out curves observed inV-mode. The inset shows Ohmic behaviors above the jump voltage in Fig. 1(c). (d) Density of states (conductance) d(ln Iout)/d(ln V) of the data measured at Icontrol = 1 μA and R = 100 Ω in Fig. 1(c). The inset is the measurement circuit.

FIG. 1.

(a) Structure and symbol of an NDR transistor. A photo of the surface of an NPN transistor chip of 2N3904 with a similar structure to that of the NDR transistor. (b) Icontrol-out–Vcontrol-out curves measured at the diode structure in the V-mode when the inlet terminal is open. Inset shows Ohmic behaviors. (c) The resistance dependence of Iout−Vin-out curves observed inV-mode. The inset shows Ohmic behaviors above the jump voltage in Fig. 1(c). (d) Density of states (conductance) d(ln Iout)/d(ln V) of the data measured at Icontrol = 1 μA and R = 100 Ω in Fig. 1(c). The inset is the measurement circuit.

Close modal
FIG. 3.

(a) Schematic diagram of the measurement system. The input-power-source voltage (about 4 W) of the control signal was measured by on of SW1 and off of SW2. Pulse experiments were carried out by off of SW1 and on of SW2. (b) The dependence of the inlet-power voltage measured ata square-wave of 100 KHz with the schematic diagram exhibited in Fig. 3(a). The upper figure is a pulse peak of NDR, measured at the control terminal when Vinput control = 10 V. Voltage-drop near the peak corresponds to NDR. The bottom figure shows NDR pulse peaks, measured at resistor of 1 Ω at the outlet terminal.

FIG. 3.

(a) Schematic diagram of the measurement system. The input-power-source voltage (about 4 W) of the control signal was measured by on of SW1 and off of SW2. Pulse experiments were carried out by off of SW1 and on of SW2. (b) The dependence of the inlet-power voltage measured ata square-wave of 100 KHz with the schematic diagram exhibited in Fig. 3(a). The upper figure is a pulse peak of NDR, measured at the control terminal when Vinput control = 10 V. Voltage-drop near the peak corresponds to NDR. The bottom figure shows NDR pulse peaks, measured at resistor of 1 Ω at the outlet terminal.

Close modal
FIG. 2.

The dependence of the control current with respect to Iout vs. Vin-out measured at I-mode in which observes voltage at constant current. Inset shows hysteresis at NDR.

FIG. 2.

The dependence of the control current with respect to Iout vs. Vin-out measured at I-mode in which observes voltage at constant current. Inset shows hysteresis at NDR.

Close modal

Figure 1(b) shows Icontrol-out–Vcontrol-out curves measured at the diode structure in the V-mode, which measures current at a constant voltage, when the inlet terminal is open and current flows from control to outlet. Lines 1 and 2 in figure are demonstrated in linear explanations in Fig. 1(d). Line 3 is the Ohmic behavior appearing without current jump at more than 46 mA at 0.88 V (1.4 Ω from the linear slope in inset) in 0 Ω data (inset). Fig. 1(c) exhibits the resistance dependence of Iout−Vin-out curves observed in the V-mode. The measurement circuit is exhibited in Fig. 1(d), and current jumps near 8 V are evident. The position and the magnitude of the jump decrease as the control current increases. Ohmic behaviors of lines above the jump are depicted in the inset. We interpret that the Ohmic behavior appears when the outlet layer becomes metal (see Fig. 1(b)) by flowing a current from control terminal to outlet both at the increased Vin-out voltage and by Icontrol. Thus, it can be concluded that NDR transistor's structure, “metal-(reverse-pn junction; insulator role for tunneling)-MIT,” as mentioned in a previous section, is similar to the Zener's interband-tunneling structure14 or metal-insulator-superconductor tunneling one. Note that the Ohmic behavior appears, when superconductor becomes metal by breakdown of the superconducting gap (gap-nogap transition).

Density of states (quantity of change in the number of carriers), d(ln Iout)/d(ln Vin-out),4 and 3 lines distinguishing a change of the density of states are exhibited in Fig. 1(d). The constant slope of line 1 indicates the exponential increase of the carrier and is a typical characteristic of an extrinsic semiconductor; I/Is = (exp(eV/kBT) − 1), where eV = 2Δ is an exciton gap of an extrinsic semiconductor. Line 1 goes into the regime of the semiconductor. Lines 2 and 3 largely increase with voltage and correspond to the disappearance process of the main energy gap of Si, indicating the MIT. This indicates very rapid carrier generation—termed the “avalanche phenomenon” and is induced by impact ionization.4,15 Lines 2 and 3 are positioned in the MIT regime. Thus, the MIT (gap-nogap transition) occurs at lines 2 and 3 by excitation of critical carrier density like exciton, nc, in the semiconductor regime12 (line 1). The jump is attributed to the difference between semiconductor and metal conductivities as a result of the MIT (Fig. 1(b)).

The observed current jump shows the same behavior as Zener breakdown explained by a large tunneling probability in strong electric filed.14 The probability was calculated in a condition that bound charges in the valance band tunnel through the interband to the conduction band, appearing as band tilting in a strong electric field. However, we interpret the appearance of the conduction band as the emerging of a metal. Thus, the tilting and de-tilting of the band is regarded as the MIT; this is the identity of breakdown.

Figure 2 shows the dependence of the control current with respect to Iout vs. Vin-out, measured in I-mode, which observes voltage at a constant current. The NDR behavior, in which voltage abruptly decreases at a maximum voltage due to a decrease of resistance at the starting point of current jump, is exhibited by Icontrol = 80 μA. The NDR happens in the regime of the MIT.6,16–18 Above Icontrol = 80 μA, the MIT voltage decreases as Icontrol increases, as indicated by the arrow in the figure. Below the voltage of the arrow-crossing point, Iin-out behavior is the same as that at the semiconductor transistor, although current gain is much less, approximately 3. Above the arrow-crossing point voltage, Iin-out behavior is linear, with increasing voltage indicating the Ohmic characteristic of the metal. This is evidence that the outlet layer with high resistance in the transistor becomes metal. Moreover, the hysteresis at NDR is observed between 6.2 and 7.0 V (Inset of Fig. 2). This indicates first-order MIT. The small-scale of the hysteresis reveals that very little heat is generated.

Figure 3(b) shows the dependence of the inlet power voltage, measured at a square-wave power pulse (about 4 W) of 100 KHz. This experiment was carried out per the schematic diagram in Fig. 3(a). The upper figure shows pulse peaks of NDR measured at the control terminal, when the source voltage for the control input, Vinput control source = 10 V (Fig. 3(a)), is applied to the control terminal. The NDR phenomenon occurs at the peak power of the pulse and the voltage at the peak drops, due to a decreased resistance at constant current; the given current flows even though the voltage drops. The measured control voltage was reduced by 1.3–1.7 V from 10 V, with increasing VInlet. This low voltage is caused by the resistance-drop generated by the MIT, which is induced by the input-source control-voltage. The bottom figures show the pulse peaks measured at a resistance of 1 Ω at the outlet terminal (Fig. 3(a)).

When the inlet voltage is zero (i.e., the transistor becomes a diode, in which current flows from control terminal to outlet), NDR peaks are Vcontrol ≈ 1.3 V (red peak in the upper figure) and Voutlet ≈ 0.38 V (red peak in the bottom figure), corresponding to 0.38 A, due to the 1 Ω resistance. Then, the resistance between the control terminal and the outlet one is Rcontrol-outlet ≈ (1.3 V/0.38 A) − 1 Ω = 2.42 Ω. This suggests that MIT occurs at the conducting channel in the control layer of insulator, as mentioned in the structure section. In this condition, when inlet voltage above 1.3 V is applied (3-terminal transistor), Voutlet increases by about 0.8 V, as the peak of NDR, corresponding to 0.8 A (bottom figure), which is 4 times of the maximum specification current of 2N3904 TR. The temperature of the transistor then reached about 49 °C. The dependence on inlet voltage is small. This pulse experiment suggests that the NDR transistor is operated in the NDR mode by an input-control pulse. Moreover, it is noteworthy that the transistor was well operated without device breakdown, despite the fact that a current of Ioutlet ≈ 1.4 A flowed through it for two weeks, thereby increasing the control voltage.

The NDR transistor obtains a higher current gain than a semiconductor power transistor for the same Joule heat (=I2R) values, because it has lower resistance than the semiconductor transistor. Since resistance in metal increases as temperature increases, thermal runaway does not occur, although large amounts of heat are generated by high current. Furthermore, this nano-level transistor may be used for microelectronics.

In conclusion, operating and investigating the MIT characteristics demonstrated the capabilities of the NDR transistor. It can be used as a switching device managing power and a barrister lightning arrest. Electronics using the NDR phenomenon for the MIT are referred to as MIT Electronics (MITronics).

We acknowledge Sun-Kyu Jung for I-V measurements. This work was supported by Creative Research project in ETRI and Ukraine-ETRI collaboration project of MSIP.

1.
Z.
Yang
,
C.
Ko
, and
S.
Ramanathan
,
Annu. Rev. Mater. Res.
41
,
337
(
2011
).
2.
D. M.
Newns
,
J. A.
Misewich
,
C. C.
Tsuei
,
A.
Gupta
,
B. A.
Scott
, and
A.
Schrott
,
Appl. Phys. Lett.
73
,
780
(
1998
).
3.
G.
Stefanovich
,
A.
Pergament
, and
D.
Stefanovich
,
J. Phys.: Condens. Matter
12
,
8837
8845
(
2000
).
4.
H. T.
Kim
,
B. G.
Chae
,
D. H.
Youn
,
S. L.
Maeng
,
G.
Kim
,
K. Y.
Kang
, and
Y. S.
Lim
,
New J. Phys.
6
,
52
(
2004
).
5.
D.
Ruzmetov
,
G.
Gopalakrishnan
,
C.
Ko
,
V.
Narayanamurti
, and
S.
Ramanathan
,
J. Appl. Phys.
107
,
114516
(
2010
).
6.
H. T.
Kim
,
B. J.
Kim
,
S.
Choi
,
B. G.
Chae
,
Y. W.
Lee
,
T.
Driscoll
,
M. M.
Qazilbash
, and
D. N.
Basov
,
J. Appl. Phys.
107
,
023702
(
2010
).
7.
S. M.
Sze
,
Physics of Semiconductor Devices
, 2nd ed. (
John Wiley & Sons
,
1981
), pp.
566
636
.
8.
P. P.
Edwards
and
M. J.
Sienko
,
Phys. Rev. B
17
,
2575
(
1978
).
9.
T. F.
Rosenbaum
,
K.
Andres
,
G. A.
Thomas
, and
R. N.
Bhatt
,
Phys. Rev. Lett.
45
,
1723
(
1980
).
10.
11.
H. T.
Kim
, e-print arXiv:cond-mat/0110112.
12.
H. T.
Kim
,
B. J.
Kim
,
Y. W.
Lee
,
B. G.
Chae
,
S. J.
Yun
, and
K. Y.
Kang
,
Physica C
460–462
,
1076
(
2007
).
13.
P.
Stoliar
,
L.
Cario
,
E.
Janod
,
B.
Corraze
,
C. G.
Deudon
,
S. S.
Bourmand
,
V.
Guiot
,
J.
Tranchant
, and
M.
Rozenberg
,
Adv. Mater.
25
,
3222
(
2013
).
14.
C.
Zener
,
Proc. R. Soc. London, Ser. A
524
,
523
(
1934
).
15.
16.
F.
Sawano
,
I.
Terasaki
,
H.
Mori
,
T.
Mori
,
M.
Watanabe
,
N.
Ikeda
,
Y.
Nogami
, and
Y.
Noda
,
Nature
437
,
522
(
2005
).
17.
Y. W.
Lee
,
B. J.
Kim
,
J. W.
Lim
,
S. J.
Yun
,
S.
Choi
,
B. G.
Chae
,
G.
Kim
, and
H. T.
Kim
,
Appl. Phys. Lett.
92
,
162903
(
2008
).
18.
T.
Driscoll
,
J.
Quinn
,
M.
Di Ventra
,
D. N.
Basov
,
G.
Seo
,
Y. W.
Lee
,
H. T.
Kim
, and
D. R.
Smith
,
Phys. Rev. B
86
,
094203
(
2012
).