JPH0622239B2 - Semiconductor device - Google Patents

Description

TECHNICAL FIELD The present invention relates to a PN junction forming a semiconductor device and a bipolar transistor using the same.

(Prior Art) In a PN junction used in a conventional semiconductor device,
Usually, the doping level of either or both of P-type and N-type impurities is low, and it is limited to the case where Boltzmann statistics in which carriers are not degenerate are observed. In this case, the forward current through the junction is exponential in voltage. The slope of logarithm of current vs. voltage is proportional to the reciprocal of temperature, and increasing this slope can reduce the logical oscillation. Therefore, the current voltage can be lowered, which is desirable in a transistor circuit. For this reason, low temperature operation is desired.

(Problems to be Solved by the Invention) However, at low temperature, at least one of P-type and N-type carriers freezes out in a low-doped PN junction, and parasitic resistance is generated in series in the junction. Therefore, good current / voltage characteristics cannot be obtained. On the other hand, if both the P region and the N region are heavily doped to prevent freeze-out at low temperatures, the junction becomes a tunnel junction,
A large current flows in reverse bias, and a negative resistance appears in the forward direction, so good rectification characteristics cannot be obtained.

(Means for Solving the Problems) In the present invention, both the P region and N region semiconductors are strongly degenerated, and the thickness is such that a tunnel current does not flow in the PN junction, and no impurities or low impurities are contained. It is a PN junction having a concentration region between the P region and the N region and in which carriers are injected over the potential barrier. Further, in the present invention, this PN junction is used as an emitter-base junction or a base-junction.
A semiconductor device used for either or both of collector junctions.

(Embodiment) FIG. 1 shows a structure of a first embodiment of the present invention, a band diagram in thermal equilibrium, and a band diagram when a bias V is applied. As shown in FIG. 1 (a), a donor acceptor is provided between the semiconductor 1 (N ⁺ layer) doped with a high concentration of P-type impurities and the semiconductor 2 (P ⁺ layer) doped with a high concentration of N-type impurities. I layer 3 not doped with is formed.

FIG. 1 (b) is a band diagram showing a thermal equilibrium state in which a bias is not applied to the PN junction of the structure shown in FIG. 1 (a). 4 is Fermil level in thermal equilibrium state, 5N, 5I, 5P are N, I, and
The conduction band in the P region, 6N, 6I, and 6P are valence bands in the N, I, and P regions, respectively.

At this time, in the band structure, the band gap of the P ⁺ layer is narrow and the band gap of the N ⁺ layer is wide. Further, the band gap of the I layer is continuously changed from the band gap of the N ⁺ layer to the band gap of the P ⁺ layer. In order to form such a band structure, for example, InAl _1-x Ga _x As is grown as an I layer on P ⁺ InGaAs while continuously changing the value of x from 1 and grown on this I layer. An N ⁺ type InAl _1-x Ga _x As layer with a fixed x value may be grown. Or Al _1-x Ga as an I layer on P ⁺ GaAs
_x As is formed by continuously changing the value of x from 1.
N ⁺ Ga _1-x Ga _x As with a fixed x value on this I layer
The layers can be grown.

Thus, crystal growth is currently performed by molecular beam epitaxy (MB
E) technique, metal-organic vapor phase epitaxy (MOSVD method), etc. Si is usually used as the III-V group donor impurity, and Be is used as the acceptor. However, the present invention is not limited by the selection of such semiconductor material, the growth method, and the type of impurities. Here, the electrons in the N ⁺ region and the holes in the P ⁺ region are strongly degenerate, and the Fermi level 4 is in the conduction band 5N and the valence band 6P, respectively. The condition of such strong degeneracy is realized if the doping is sufficiently high. That is, when the average distance between donors or acceptors becomes smaller than the respective Bohr radii due to heavy doping, the wavefunctions of these impurity levels are overlapped with each other to form an impurity band. This impurity band spreads as the overlap of the wave functions increases, and the gap with the conduction band or the valence band disappears. In such a case, the semiconductor exhibits metallic conduction and the Fermi level is in the conduction band or valence band.

As a standard for the concentration at which the donor forms an impurity band, a concentration at which the distance between donor impurities is about twice the Bohr radius may be used. When the Coulomb integral of the overlap at this time is calculated, it becomes about 0.17 times the depth of the donor, which is sufficient to form a band. When the effective mass of electrons or holes is m ^{*, and} the specific derivative of the semiconductor is ε, the effective radius r of the donor or acceptor is r = 0.53 × ε / m ^* Å. For example, ε = 13 of semiconductor, effective mass of electron m ^*
= 0.1, r = 73Å. Therefore, diameter d = 14
6Å, donor concentration N _D = 1 / d ³ = 3 × 10 ¹⁷ cm
^-3 . When the effective mass m of holes is m ^* = 0.2, the effective radius r of the acceptor is r = 34Å, and the acceptor concentration at that time is 3 × 10 ¹⁸ cm ⁻³ . When m ^* is small, r is large and the impurity band is formed at a low concentration.

Now, under such conditions, the difference between the Fermi level 4 and the conduction band 5P in the P ⁺ region will be referred to as the electron threshold voltage V _T. This is because, as shown in FIG. 1 (c), when a bias V is applied and V> V _T , electrons are transferred from the conduction band 9N in the N ⁺ region to P ^+.
It is named because it is injected into the conduction band 9P of the region.

However, the bias V is the difference between the N-side pseudo-Fermi level 8N and the P-type pseudo-Fermi level 8P. V _T means the minimum bias voltage at which electrons are injected from the N ⁺ layer to the P ⁺ layer.
Similarly, in FIG. 1B, the difference from the Fermi level 4 to the valence band 6N in the N region will be referred to as the hole threshold voltage V _T2 . This is shown in FIG. 1 (c), when the bias becomes V> V _T2 , P ⁺
It means the minimum voltage at which holes are injected from the valence band 10P in the region to the valence band 10N in the N ⁺ region. Since V _T <V _{T2 in} this embodiment, only injection of electrons from N ⁺ to P ⁺ occurs when V _T <V <V _T2 . The electron injection efficiency at this time is 1.

The relationship between the current density J and the voltage V at this time is calculated. At low temperature, when the carriers are degenerate, the electron diffusion coefficient D
Can be written as

Here, μ is electron mobility, q is electron charge, h is Planck's constant, m is effective mass of conduction electron, and n is electron density. The electron density n of the injected electrons follows the following diffusion equation.

Here, τ is the lifetime in which the injected electrons recombine with holes in the P layer 1 and disappear. In this equation, D is n as in equation (1).
It is a nonlinear equation because it depends on. However, the origin of the x coordinate is set at the interface 7 between the I layer 3 and the P layer 1 shown in FIG.
Now, when the bias is V, the electron density n (0) at x = 0
Is given by the following equation.

This formula is accurate at absolute 0 degree, but it is finite temperature T
But if kT << q (V−V _T ), it is correct. However, k is the Boltzmann constant. x = 0 and n = 0 and dn / dx = 0,
When the equation (2) is solved with this and the equation (3) as boundary conditions, the equation of the current density J is based.

In this current formula, when V ≦ V _T , J = 0, but V> V _T
Therefore, J is a formula that rises sharply with voltage. The physical reason is that the electron degenerates at a low temperature, the distribution function of the electron sharply rises near the Fermi level, and the quantum mechanical density of states is large.

Here, a numerical example of Expression (4) is shown.

q = 1.6 × 10 ⁻¹⁹ C, μ = 1000 cm ² / v · se
c, τ = 10 ⁻⁹ sec, m = 0.1 × 9.1 × 10 ⁻²⁸ g, Substituting for, we obtain

J = 1.15 × 10 ⁷ (V−V _T ) ² A / cm ² Due to this, for example, J = 1 × 10 ³ to 1 × 10 ⁵ A / cm ²
In order to obtain the above, a very small bias of V−V _T = 9.3 mV is sufficient. That is, J is a function that rises very sharply as a function of V.

Thus, in the embodiment shown in FIG. 1, the current becomes a function of abruptly increasing the voltage, and this property can be utilized in various devices.

Next, let us consider a case where the band gap of the P ⁺ region 1 in FIG. 1 becomes smaller as it goes away from the interface 7.
For simplicity, consider the case where the band gap Eg (x) can be written as

Eg (x) = Eg (0 ) -Ex (5) (5) surface 7 Nitori in FIG. 1 the origin of the x in formula, P ⁺ region 1
The direction towards x is positive at x.

If x> Eg (0) / E≡x ₁ , then Eg (x) <0, but in the following story, this does not matter. In reality, x ₂
When x> x ₂ is satisfied with respect to x ₂ which satisfies <x ₁ , it is considered that the bandgap is uniform or bonded to the metal electrode. Now, let us consider a case where this bandgap gradient is steep enough and the electric field acceleration due to this bandgap gradient is dominant over the diffusion current. In this case, the electron density transport equation can be written as

Solving this first-order differential equation with the boundary condition of Eq. (3) To get However, L _B = qμEτ. The formula for the current is become that way. In this equation as well, V <V _T J = 0 and V> V _T is a formula in which J rises sharply as a function of voltage. The reason is the same as described in equation (4).

An example of the numerical value of Expression (8) is shown.

q = 1.6 × 10 ⁻¹⁹ C, μ = 1000 cm ² / v · se
When c, E = 0.1V / (100 × 10 ^-8 cm), the following formula is obtained.

^{J = 2.3 × 10 8 (V} -V T) A / cm 2 V-V T to obtain ^{J = 10 3 ~10 5 A /} cm 2 in this formula
A very small bias of 0.26 to 5.58 mV may be applied. Equation (8) with a bandgap gradient generally causes a larger current to flow than equation (4) of the diffusion current.

Next, the case of E _F <V−V _T <V _T2 will be qualitatively described by the formulas (4) and (8). However, E _F is the N ⁺ layer 2 as shown in FIG. 1 (b).
To the Fermi level 4 from the bottom 5N of the conduction band of the electron. This does not change even if a bias is applied, and is equal to the depth from the bottom 9N of the conduction band of the N layer to the pseudo-Fermi level 8N of the electron in FIG. 1 (c). At this time, Fig. 1 (c)
And pseudo-Fermi level 8 in N ⁺ region 2 and I region 3
N, 8I and the bottom 9N of the conduction band have a downward sloping slope. At this time, it means that a voltage drop has occurred in the N ⁺ region, which means that a series resistance component is added to the junction. At this time, the relationship between the external voltage V and the current density J has a relationship in which the voltage drop is corrected by the equations (4) and (8).

This time is represented by an expression obtained by replacing the _{V-V T -V R (J} ) instead of the V-V _T. However, V _R (J) is a voltage drop with respect to the current density J. It is roughly given by the following formula.

V _R (J) = (thickness of N ⁺ layer) × J / (specific conductivity of N ⁺ layer) Further, when V> V _T2 , injection of holes from P ⁺ to N ⁺ is further started, The current flowing through the junction is the sum of both injected currents. Therefore, in this region, the JV characteristic starts to have a slightly gentle slope.

Next, consider a case where the P ⁺ semiconductor 1, the N ⁺ semiconductor 2, and the I layer 3 all have the same band gap and are homojunctions. At this time,
From P ⁺ to N ^+, or minority carrier injection from the N ⁺ to P ⁺ must consider the case that occur at the same time. However, in each case, the threshold voltages V _T and V _T2 can be determined by making the doping concentrations of P and N different, and the same argument holds if V _T2 > V _T , for example. If the doping concentrations of P and N are the same and V _T = V _T2 , the current-voltage characteristic having the same threshold value is exhibited, and the sharp rising current-voltage characteristic is obtained.

FIG. 2 shows a second embodiment of the present invention. The structure is shown in FIG. 2 (a). High concentration N ⁺ layer 2-1 High concentration P ⁺ 2-2
Sandwich the low-concentration impurity layer I ₂ 2-3. FIG. 2 (b) shows a band diagram in the thermal equilibrium state. In this case, unlike the first embodiment, there is an offset in the valence band. N,
The conduction bands in the I and P regions are 2-5N, 2-5I, 2 respectively.
-5P. The valence band of each is 2-6N,
2-6I and 2-6P. Fermi level is 2-4
Is. FIG. 1 shows this band structure in the first embodiment.
The difference from (b) is that band offsets ΔE _{v1 and} ΔE _v2 in valence electrons exist at the interface between the N ⁺ I ₂ layer and the I ₂ -P ⁺ layer. At this time, the threshold voltage V _{T of the} electrons is Fermi level 2-
4 and the conduction band 2-5P of the P ⁺ layer. On the other hand, the threshold voltage V _'T2 of the hole Fermi level and I
It is defined as the lowest value of the energy of the valence band of the _two layers, that is, the difference between the valence band energy of the N ⁺ layer and the I ₂ interface.

Now, let V _T <V ′ _T2 and consider the relationship between the current and the voltage when the bias voltage V is applied. The band diagram at this time is shown in FIG. 2 (c). The bias voltage is the difference between the pseudo Fermi level 2-8P in the P region and the pseudo Fermi level 2-8N in the N region. As can be seen from the figure, when V _T <V <V ′ _T2 ,
Only the injection of electrons from N ⁺ to P ⁺ occurs. This means that the electrons, but can be more than a mountain of potential of the conduction band of the I ₂ layer, the hole is because it is impossible to exceed the mountain of the valence band of the I ₂ layer.

Now, the formula of the current of the electrons injected from N ⁺ to P ⁺ is the same as that of the first embodiment. When the P ⁺ layer has a uniform band cap, the formula (4) is obtained. Also, when there is a slope like the band cap formula (5) of the P ⁺ layer, the current formula is
It becomes like (8). Both of these rise sharply as a function of voltage.

The same effect is obtained when the energy of the valence band at the P ⁺ I ₂ interface acts as a barrier in the valence band of hole injection. In this case,
This can occur when ΔE _V2 is larger than ΔE _v1 . FIG. 3 is a diagram showing a third embodiment of the present invention. Fig. 3 (a)
The structure is shown in. High-concentration N ⁺ layer 3-1, high-concentration P ₊ 3-
The low concentration impurity layer I ₂ 3-3 is sandwiched between the _two . FIG. 3 (b) shows a band diagram in the thermal equilibrium state. In this case, the second embodiment is further generalized. The conduction bands of the N, I, P, regions are 3-5N, 3-5I, 3-5P, respectively.
Moreover, each valence band is 3-6N, 3-6I, 3-6
P. It is Fermi level 3-4. The difference between this band structure and FIG. 2 (b) in the second embodiment is that the band offsets are added to the conduction band and valence band at the interface between the N ⁺ I ₂ layer, the I ₂ layer and the I ₂ -P ⁺ layer. ΔE _c1, ΔE _c2, ΔE _v1,
This is the point where ΔE _v2 exists. The electron threshold voltage V ′ _T in this case is defined as the difference between the Fermi level 3-4 and the top of the conduction band of the I ₂ layer at the I ₂ -P ⁺ interface. The hole threshold voltage V ′ _T2 is the same as that in the second embodiment.

Well, as V _'T <V' _T2, consider the relationship between the current and the voltage at the time of applying a bias voltage V. The band diagram at this time is shown in FIG. 3 (c). The bias voltage is the difference between the pseudo Fermi level 3-8P in the P region and the pseudo Fermi level 3-8N in the N region. As can be seen from the _{Figure, V 'T <V T <} V' T2
When, only injection of electrons from N ⁺ to P ⁺ occurs. This means that the electrons, but can be more than a mountain of potential of the conduction band of the I ₂ layer, the hole is because it is impossible to exceed the mountain of the valence band of the I ₂ layer.

Now, in the third embodiment, when only electron injection occurs, various cases may occur depending on the values of the band offsets ΔE _{c1 and} ΔE _c2 of the conduction band. Here, in FIG. 3 (b), the relationship between the depth E _F from the electron conduction band 3-5N of the N layer to the Fermi level 3-4 and these offsets becomes important. Now, we are considering the case of E _F <ΔE _c1 . When ΔE _c2 = 0,
The current-voltage characteristics are the same when the band gap of the P ⁺ layer is uniform and when the band gap has a gradient. However, use of V _'T as the threshold voltage. Basically, the situation does not change when ΔE _c2 << E _F. Current rises sharply at V = V _'T. When ΔE _c2 becomes large, solving the transport equation of the injected electrons and solving the current equation is a complicated problem, but it can be qualitatively considered as follows. Third
As in FIG. (C), 'when the _T> 0, the density of electrons injected V instead of V _T in equation (3)' V-V become what put the _T. The difference from the case of ΔE _c2 = 0 is that when the injected electrons enter the conduction band 3-9P in the P ⁺ region, they have an extra kinetic energy of ΔE _c2 . Therefore, the electrons become hot electrons that are out of thermal equilibrium. 0 <V-
Given the V _'T E _F «ΔE _c2, the kinetic energy of the electrons at this time it is greater than the electron Fermi energy. Therefore, the electron is in a non-degenerate state. When P ⁺ has a uniform bandgap, there is no electric field acting on the electrons, so a diffusion current flows. The diffusion coefficient at this time is considered to be proportional to the electron temperature due to Einstein's relation, and does not depend on the electron density as in Eq. (1). Therefore, the diffusion equation (2) is a linear equation for the electron density n.

However, the diffusion coefficient may change depending on the location. x = 0
Since the boundary condition in is the equation (3), the expression of the current can be written as J∝n (0) ∝ (V−V ′ _T ) ^2/3 . This equation is still a sharp rise function V _'T as a function of V.

Next, when E _F <ΔE _c1 , the following is performed. At this time,
A highly doped N ⁺⁺ layer is sandwiched between the N ⁺ and I ₂ layers to locally deepen E _F , and at the interface between the N ⁺⁺ layer and I ₂ the electron pseudo-Fermi level is equal to that of the I ₂ layer. Try to be above the conduction band.

Incidentally, in Examples 1 and 3, the N ⁺ band gap is drawn wider than the P ⁺ band gap, but it is not always necessary and these may be equal or opposite.

FIG. 4 shows a fourth embodiment of the present invention. The structure is shown in FIG. 4 (a). A low impurity P − layer 4-2 and an N ⁻ layer 4-3 are sandwiched between a high impurity concentration P ⁺ layer 4-1 and an N ⁺ layer 4-4. Figure 4 (b) shows the band structure at thermal equilibrium. 4-
5 is the Fermi level. The conduction bands of the P ⁺ , P ⁻ , N ⁻ and P ⁺ layers are respectively 4-6P ⁺ , 4-6P ⁻ , 4-6N ⁻ and 4-6N ⁺ . Similarly husband Ataidenshi band people, 4-7P ^+, 4-7P ^-, 4
It is shown by −7N ⁻ and 4-7N ⁺ . Now, due to the existence of the N ⁻ and P ⁻ layers, no tunnel current flows between N ⁺ and P ⁺ .
The N ^+, P ⁺ doping level and the N ⁺ layer of the band gap width of, N ^-, P ^- if a tunnel current flows even without, N ^-, P ^- layer may be omitted. Even in such a structure, as described above, the current sharply rises as a function of the voltage with the threshold voltage as a boundary. However, the threshold voltage V _T
As Fermi level 4-5 and conduction band 4- of the P ⁺ layer
It is defined as the difference from 6P ⁺ . However, in some cases, the conduction band 4-8 at the interface between the N ⁻ layer and the P ⁻ layer may act as a barrier for electrons when the bias voltage V is V = V _T. However, in this case as well, the story is the same if V _T is appropriately corrected. The N ⁻ and P ⁻ layers may be I layers (non-doped layers).

A junction transistor can be formed using all the PN junctions described above. This will be explained next.

FIG. 5 shows an embodiment of the second invention of the present application. FIG. 5 (a) shows a 5000 Å-thick InGaA layer doped with 5 × 10 ¹⁸ cm ⁻³ donor (Si) on a semi-insulating InP substrate 5-1.
s-layer collector layer 5-1 ', 5 × 10 ¹⁶ cm ⁻³ donor (Si) -doped InGaAs collector layer 5-1 ″ with a thickness of 5000 × 1 × 10 ¹⁹ cm ⁻³
(Be) -doped base layer 5-2 having a thickness of 500 to 2000 Å, emitter layers 5-3 'and 5-3 on which a band gap is continuously changed, and 2 × 10 ¹⁹ cm ^-3 thereon. InAlAs emitter doped with donor (Si) of
4 is a cross-sectional view of the heterojunction bipolar transistor formed with FIG. The layer on the InP substrate 5-1 can be formed by a vapor phase epitaxial growth method or a molecular beam epitaxial growth method. A band diagram is shown in FIG. 5 (b). In order for the lattice constant to match that of the InP substrate, the proportion of In in the group III element may be kept at a constant value (x = 0.52). III
When the ratio of Ga and Al in the ratio of the elements of the group elements is changed, the band cap with a constant lattice constant of InP is changed from 1.45 eV of InAlAs to 0.7 of InGaAs.
It can be continuously changed up to 5 eV. In this embodiment, the base layer 5-2 is made of In _x (Ga _0.85 A
l _0.15 ) _1-x As, collector side is InGaAs,
The composition ratio of Ga and Al is continuously changed and the band cap is changed from 0.85 eV to 0.75 eV. In the emitter graded gap layer 5-3, 5-3 ', the composition ratio of Ga and Al up to the interface with the base is In _x (Ga).
_0.85 Al _0.15 ) _1-x As to InAlAs were changed, and the band gap was continuously changed from 0.85 eV to 1.45 eV. In this emitter graded gap layer, a layer 5 doped with a donor of 2 × 10 ⁻¹⁹ cm ⁻³ is used.
3 and the base 5-2 between the non-doped thin layer 5-3 '
Sandwiched between. This thickness may be such that no tunnel current flows when the emitter base is forward biased with a low bias voltage, and 100 Å is sufficient.
Another condition to consider is thicker to reduce the capacitance between the emitter and the base, but the forward resistance should not increase. Therefore, about 500 to 50Å
Between is suitable. Although this layer is not doped in this embodiment, it may be n-type or p-type having a low impurity concentration of 10 ¹⁶ cm ^-3 or less.

Now, in this transistor structure, the forward current J between the emitter and the base and the current gain h _{FE of the} grounded emitter are Given in. However, ΔE _g is a bandgap difference obtained by subtracting the bandgap between the base clecters from the bandgap at the emitter-base junction, and W _B is the base width.
m = 0.045 × 9.1 × 10 ⁻²⁹ g, μ = 1000c
When m ² / V · sec, W _b = 500Å, and ΔE _g = 0.1 eV, J = 1.4 × 10 ⁷ (V−V _T ) ^3/2 A / cm ² is obtained. In this formula, to obtain J = 10 ³ to 10 ⁵ A / cm ² , V
It is sufficient to take −V _T = 1 to 21.5 mV. In this embodiment, the operating temperature is an extremely low temperature of 4.2 k or less, but if the doping concentration of the emitter and the base is made higher, the operating temperature is about 77 k. As a second embodiment of the second invention of the present application, consider a hetero-bipolar transistor in which the base region 5-2 of FIG. 5 is InGaAs and has a uniform band gap of 0.75 eV. At this time, the composition is changed from InGaAs to I in the emitter graded gap layers 5-3 'and 5-3.
By continuously changing the composition ratio of Ga and Al up to nAlAs,
Change the bandgap from 0.75 eV to 1.45 eV. In addition, the emitter is, for example, In _x (Ga _0.5 Al _0.5 ) _1-x
As may be set to 1 eV, and in this case, the bad gap of the graded layer may be 0.75 to 1 eV. At this time, the base current is dominated by the diffusion current, and the forward-direction current density J of the base emitter and the current gain h _FE can be written as the following equation.

Using the same numerical values as in the first embodiment, the following equation is obtained.

J = 3.2 × 10 ⁸ (V−V _T ) In order to be J = 10 ³ to 10 ⁵ A / cm ² in this formula, V−
It is sufficient if V _T = 8 to 171 mV. As can be seen from the above, it is sufficient to apply a small bias to obtain a high current density level, so that a transistor having a logic amplitude extremely smaller than that of the conventional device can be realized. In this embodiment, the operating temperature is an extremely low temperature of 4.2 k or less, but if the doping concentration of the emitter base is made higher, the operating temperature is about 77 k.

Next, as a third embodiment of the second invention of the present application, a case where the base width is sandwiched in the first embodiment will be considered. It is assumed that the base width is thinner than the mean free path e of injected electrons. Electrons mobility and μ at the base of the semiconductor, the electric field F = ΔE _g / _g W _B of the band gap gradient, the electrons relaxation time and tau, the mean free path e is Here, v is the electron drift velocity, and τ is the collision relaxation time. Therefore, if μ is large and F is large, the free process can be increased. If the base width is smaller than this e, the electrons injected from the emitter reach the collector without being scattered. This condition is e = W _B
Then, solving for W _B yields the following equation.

In this formula, μ = 1000 cm ² / V · sec m = 0.1 ×
9.1 × 10 ⁻²⁸ g, q = 1.6 × 10 ⁻¹⁹ C ΔE _g =
When it is set to 0.1 eV, W _B = 238Å is obtained.

Therefore, in the structure of the transistor as in this embodiment, when the base width is 238 Å or less, the injected electrons reach the collector without being scattered in the base region.

Under these conditions, the current between the emitter and the base can be calculated by calculating the current when the degenerated electron gas in the emitter is released into the electron-free base. The relation between the thus calculated current density J and the bias voltage V between the emitter and the base is given by the following equation.

However, V _T is the threshold voltage defined in the first embodiment.
Now, m = 0.1 × 9.1 × 10 ⁻²⁸ g, h = 1.05 × 1
Substituting 0 ⁻²⁷ erg · sec, q = 1.6 × 10 ⁻¹⁹ C, the following equation is obtained.

J = 7.9 × 10 ⁸ (V−V _T ) ² A / cm ² J is also a rapidly rising function of V. For example, J =
V-V _T = 1.1 mV to obtain 10 ³ to 10 ⁵ A / cm ^2.
It may be about 11 mV. That is, this transistor operates with a very small logic amplitude. Moreover, since the electrons are not scattered by the base, the cutoff frequency is very high.

The first and third examples show the case where electrons are scattered many times in the base and the case where the electrons are not scattered at all depending on the base width. However, it may be scattered several times in the base. In either case, J is still a rapidly rising function of V.

This is due to the structure of the emitter proposed in the present invention,
As described above, the present invention is effective even in the intermediate case between the first and third embodiments.

Further, in the first and second inventions of the second invention, regarding the structure of the junction between the emitter and the base, the second, third and fourth inventions of the first invention are provided.
The present invention also includes the case of using a PN junction as in the above embodiment.

(Effect of the Invention) By using the junction of the present invention, the junction can be turned on and off with a very small voltage amplitude. Further, if these junctions are used for a transistor, the logic amplitude can be made extremely small, and a high-speed and highly integrated bipolar LSI can be realized.

[Brief description of drawings]

1 (a), (b), and (c) are a cross-sectional view of a degenerated P ⁺ N ⁺ junction sandwiching a thin grading layer, a band diagram at thermal equilibrium, and a band diagram when a bias is applied. FIGS. 2 (a), (b) and (c) are similar views when the thin layer sandwiched by the P ⁺ N ⁺ junction has a wide gap and the conduction band is smoothly lowered. 3 (a) (b) (c) are similar to FIG. 2 except that the conduction band also has a band offset. FIGS. 4 (a) and 4 (b) are a sectional view and a band diagram when the layer sandwiched between N ⁺ and P ⁺ is N ⁻ , P ⁻ or one layer. 5 (a) and 5 (b) are a sectional view and a band diagram, respectively, showing an embodiment of the second invention of the present application.

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁵ Identification code Office reference number FI technical display location H01L 29/91

Claims (6)

[Claims] 1. In a PN junction, both the P region and the semiconductor in the N region are strongly degenerated, and a region which does not contain an impurity having a thickness such that a tunnel current does not flow in the PN junction or has a low impurity concentration is formed in the P region. A semiconductor device in which carriers in a strongly degenerate region between the N and N regions are injected over a potential barrier. 2. The semiconductor device according to claim 1, wherein one band gap between the P-type region and the N-type region is wider than the other. 3. A condition in which a semiconductor having a bandgap wider than that of a P-type or N-type semiconductor is sandwiched between PN junctions, and either the conduction band or the valence band has a band edge smoothly connected or close thereto. Claim 1 which is
Item 2. The semiconductor device according to Item 2. 4. A P-region and an N-region which have a thickness such that both the P-region and the N-region semiconductor degenerate strongly and a tunnel current does not flow in a PN junction and which does not contain impurities or has a low impurity concentration. A semiconductor device characterized by using a PN junction, in which a carrier in a strongly degenerate region between the two is injected over a potential, for either or both of an emitter-base junction and a base-collector junction. 5. The semiconductor device according to claim 4, wherein one band gap between the P-type region and the N-type region is wider than the other. 6. A condition in which a semiconductor having a bandgap wider than that of a P-type or N-type semiconductor is sandwiched between PN junctions, and either the conduction band or the valence band is smoothly connected or close to the band edge. Claim 4 which is
Item 5. The semiconductor device according to Item 5.