JPH0311108B2 - - Google Patents

Description

[Detailed Description of the Invention] [Summary] An element having a single quantum well (SQW) as a channel structure in a field-effect transistor, which has a thin at least one channel structure in which carriers can tunnel. Doping is performed only in narrow regions of two or more wells separated by one barrier. As a result, subthreshold characteristics are improved, threshold fluctuations due to temperature are reduced, and carrier mobility is increased.

[Industrial application field]

The present invention relates to a field effect transistor, and in particular includes a single quantum well (SQW) as a channel structure, gives the channel formed in the well two-dimensionality by a heterojunction of the SQW, and scatters carriers. The present invention relates to an element that achieves high mobility by reducing the .

[Conventional technology]

Conventionally, various attempts have been made to improve the characteristics of field effect transistors, such as increasing mutual conductance (g _n ) and reducing short channel effects.

Figure 5 shows a conventional GaAs MESFET. In the figure, 51 is a semi-insulating GaAs substrate, 52 is an n
-GaAs layer, 53 and 54 are n ⁺ layers for source and drain contacts, 56 and 57 are source,
55 is a drain electrode and a gate electrode. By controlling the channel with the depletion layer 58 extending by applying a bias voltage to the gate electrode 55.
FET operation is performed, but when the channel length is shortened, the short channel effect becomes a problem. As shown in FIG. 7, when the channel length becomes about 1 μm or less, the threshold value Vth of the field effect transistor changes as shown in the figure. This variation becomes much smaller as the impurity concentration N of the active layer of the channel decreases. Therefore, the active layer has conventionally been highly doped in order to reduce the short channel effect. Also,
When the active layer is highly doped, the depletion layer 91 becomes thinner and the number of modulated carriers 92 (the amount of charge that can be induced for a unit gate bias change) increases, as shown in the energy hand diagram in FIG. This means that g _n can be improved.

Further, FIG. 6 shows a HEMT (high electron mobility transistor) as another conventional example. In the figure, a GaAs buffer layer 62 is provided on a GaAs substrate 61,
Undoped i-GaAs layer 63, carrier supply layer n-AlGaAs layer 64, cap layer GaAs layer 65
is formed. Then, source and drain alloy electrodes 66 and 67 and a gate electrode 68 are formed. In this HEMT, AlGaAs/GaAs
i-GaAs with larger heterojunction electron flux
A two-dimensional electron gas 69 is generated on the layer 63 side, and the i
- Since the GaAs layer is undoped, there is less electron scattering and high electron mobility is achieved.

[Problem that the invention seeks to solve]

However, in the conventional MESFET,
There are problems such as a decrease in device breakdown voltage or a decrease in mobility as the active layer becomes highly doped. On the other hand, HEMT
However, even if the amount of impurity doping in the carrier supply layer is increased, the sheet carrier concentration Ns in the two-dimensional electron gas layer becomes saturated, so there is a problem that Ns cannot be made sufficiently large. Furthermore, since the AlGaAs of the carrier supply layer has a deep donor level and a DX center exists, the amount of supplied electrons varies depending on the temperature, resulting in a drawback that the threshold value varies greatly depending on the temperature.

The present invention aims to solve these conventional problems and provide an element with excellent characteristics.

[Means for solving problems]

The present invention is an element having a single quantum well (SQW) as a channel structure in a field effect transistor, in which at least one thin barrier capable of carrier tunneling is formed within the single quantum well (SQW). , only the narrower region of the two or more well regions separated by the doping is doped.

The present invention will be described using the case shown in FIG. 2 in which a single barrier divides a quantum well into a narrow region and a wide region. FIG. 2 is an energy band diagram of a field effect transistor to which the present invention is applied, showing the energy band in the direction across the channel. In the figure, the gate electrode
Cap layer 9 of Al14, i-GaAs layer, i-
AlGaAs layer 8, single quantum well (SQW), i-
This is AlGaAs layer 2. A single quantum well (SQW) has a barrier 4 that is thin enough to allow electron tunneling at asymmetric positions. Then, impurities are doped only in the narrow width region of the well area delimited by this. The width of the single quantum well (SQW) is preferably about 100 mm in order to provide two-dimensionality.

[Effect]

According to the above-mentioned configuration of the invention, the following effects can be obtained.

As mentioned above, when a barrier thin enough to allow electrons to tunnel is provided at an asymmetric position within a single quantum well (SQW), the probability of electron existence is small in the narrow region of the Yoshiko well, and the width of the quantum well The probability of its existence in a wide area increases. This is because when considering each independent quantum well, the energy of electron motion in a narrow well region is large, and the quantum level of the ground state becomes high. That is, when independent wells are brought close to each other so that they are close enough to each other to allow tunneling, the effect is similar to that of electrons transferring from a region with a high quantum level to a region with a lower quantum level. Normally, the quantum level in a single quantum well (SQW) (see Figure 4) is as follows, assuming that the barrier is infinitely high.

The width Δx of the single quantum well (SQW) is Δx=λ/2・n...(1) The energy En of the quantum level is En=Px ² /2m ^* =1/2m ^* (2πh/λx) ² ...(2) Substituting equation (1) into (2), En=1/2m ^* (πh/Δx・n) ² ...(3) (where, P _x : electron momentum, m ^* : effective mass,
h: Planck's constant, n=1, 2, ..., λ _x : wavelength of electron) That is, the energy En of the quantum level is proportional to the reciprocal of Δx ² and is sensitive to the width of the well Δx.
In practice, when tunneling distances are reached, narrow regions of single quantum wells (SQWs) (5, 6, 7)
Rather than independent levels existing in each wide area 3, a single level is formed, and as a result, as shown enlarged in Figure 3, there are narrow areas and wide areas. In this case, the electron distribution (indicated by diagonal lines) will change significantly.

In the present invention, by utilizing this physical phenomenon,
An electronic layer with high average mobility can be obtained by a method different from conventional methods.

That is, if a high-density n-layer 6 is formed in the narrow well region by a method such as planar doping, there is a high probability that the electrons in the system will exist in the wide well region, which is the undoped layer. Therefore, a high mobility electron layer can be formed.

The electron layer formed here has two-dimensional behavior because it is confined in the quantum well.

Although the thickness and height of the barrier within the quantum well can be set arbitrarily, it is necessary that the barrier allows electrons to tunnel therethrough.

Furthermore, the narrower the doping region of the quantum well, the greater the asymmetry and the higher the performance of high electron mobility.

For example, AlGaAs/
There is GaAs/AlGaAs. This structure is similar to the conventional
Compared to the HEMT structure, the same level of electron mobility is obtained, and since the GaAs layer is doped, there is little threshold variation with respect to temperature. That is, the doped layer is made of GaAs, which has a shallower donor level than i-AlGaAs or the like, and channel electrons are supplied from this layer, so that the amount of electrons supplied varies little with temperature, and the threshold value varies less with respect to temperature.

In contrast, in conventional HEMTs (high electron mobility transistors), the donor level is deep.
Since AlGaAs is used as the electron supply layer, the amount of electron supply changes easily depending on the temperature, and the threshold value changes greatly depending on the temperature.

Furthermore, since the electronic system is confined by a heterojunction, subthreshold characteristics are good and short channel effects are small. In other words, the doping concentration of the channel can be increased, and since electrons are confined in the SQW heterojunction, the channel becomes narrower, and the short channel effect can be prevented more than the conventional improved MESFET, and the short channel effect can be significantly reduced. become.

Furthermore, for the same reason, the linearity of the device characteristics becomes good and equal g _n can be achieved.

On the other hand, the conventional MESFET is a homojunction and has low impediments, and as shown in Fig. 8, the characteristic diagram of gate voltage Vgs and drain current Id does not become as shown in b of the example of the present invention, but as shown in a. As such, the closing is poor and the subthreshold characteristics are poor.

〔Example〕

FIG. 1 shows the main parts of a device according to an embodiment of the present invention. In the figure, semi-insulating (SI) GaAs substrate 1
On top, undoped i-AlGaAs layer 2, single quantum well (SQW), i-AlGaAs8, i-
Each layer of GaAs9 is formed. The molar ratio x of Al in the i-AlGaAs layers 2 and 8 is 0.2 to 1.0, and in this example is set to 0.2 to 0.3.

The single quantum well (SQW) layer is i-GaAs3, i-
It consists of an AlGaAs barrier layer 4, i-GaAs layers 5 and 7, and a doping layer 6. The doped layer is planar doped or highly doped.

Examples of each of the above layers are shown below.

2, 8: i-AlGaAs layer, undoped, several hundred Å thick
(Thickness at which carriers cannot tunnel) 3: i-GaAs layer, undoped, thickness 80 Å 4: i-AlGaAs layer, undoped, thickness 20 Å 5, 6, 7: Total thickness of GaAs layer 20 Å, 5, 7
is undoped, and 6 is planar doped (atomic planar dope: a monoatomic to several atomic layers of Si or Se is interposed between the i-GaAs layers).
Doping concentration due to impurity Si or Se
10 ¹⁹ cm ^-3 or more. Note that when the doped layer is highly doped, the impurity concentration is ¹⁰ cm
It should be around ^-3 .

The reason why the undoped i-GaAs layers 5 and 7 are provided on both sides of the doped n-GaAs layer 4 is to prevent dopants from diffusing into the i-GaAlAs layers 8 and 4 due to diffusion. Furthermore, 3,
The thickness of the SQW consisting of layers 4, 5, 6, and 7 is 100 Å or less to ensure two-dimensionality.

9: i-GaAs layer Undoped, film thickness several hundred angstroms In addition, in Fig. 1, 10 and 11 are n ⁺ regions (10 ¹⁷ to 10 ¹⁸ cm ^-3 ) formed by Si ⁺ ion implantation.
12 and 13 are source and drain electrodes (AuGe/Au), and 14 is a gate electrode (Al).

Although one embodiment has been described above, the present invention can be modified in various ways. For example, a plurality of tunnelable barriers may be provided, and impurities may be added only to narrow width regions among the regions separated by the plurality of barriers. You can dope it. FIG. 10 shows this embodiment in the form of an energy band diagram. In this embodiment, tunnelable barrier layers (i-AlGaAs) are formed in two layers 4, 4'. At this time, the molar ratio x of Al in the i-AlGaAs of the tunnelable barrier layers 4 and 4' may be the same, but may be different as shown in FIG. In this example, x=0.5 and x=0.2. In this embodiment, narrow well regions are formed on both sides of the quantum well, and doped layers 6 and 6' are provided, respectively. The central area is wider. Therefore, for the reasons mentioned above, there is a high probability that electrons will exist in the wide area in the center, but in this case, electrons will be supplied from the narrow areas on both sides, so the From the example, it is possible to increase the electron sheet carrier concentration Ns.

Next, FIG. 11 shows still another embodiment of the present invention. In the figure, the molar ratio of Al in i-AlGaAs x = 1, that is, the high barrier of AlAs (barrier layer 4,
In this way, by providing high barriers (large x value) on both sides of the doped layer 6 in the narrower region, the probability that carriers will gather in the wider region is higher. This is shown earlier.
In Equation (3), it is assumed that the barrier is infinitely high, but when the height of the barrier is actually taken into account, the carrier width is smaller when the barrier is higher, although the dependence is smaller than that on the width Δx of the quantum well. This is because the amount transferred to the wider area increases. As a result, if the width of each region of the quantum well is the same, it is better to provide a high barrier (larger x value) on both sides of the doped layer 6 in the narrower region, so that the carriers in the undoped layer 6 in the wider region are The amount of carriers that gather in the area increases, resulting in high mobility.

〔Effect of the invention〕

As is clear from the above description, the effects of the present invention can be summarized as follows.

While a high-density n-layer is formed in the narrow region of the quantum well using a method such as planar tope, the probability that the electrons in the system exist in the undoped layer, which is the wide well region, is This substantially allows the undoped layer to function as a channel, preventing scattering of electrons and forming a high mobility electron layer.

According to the present invention, an electron mobility comparable to that of a conventional HEMT structure can be obtained.

There is little threshold variation with respect to temperature. That is, GaAs doping layer has a relatively shallow donor level.
etc., and since the channel electrons are supplied from here, there is little variation in the amount of electrons supplied due to temperature,
The fluctuation of the threshold value with respect to temperature is reduced.

In principle, HEMT cannot generate a large amount of electron current even if the doping concentration is increased, but in the structure of the present invention, a sufficiently large current can be obtained by increasing the doping concentration.

Since electrons are confined in the SQP heterojunction,
It becomes possible to reduce short channel effects.
Furthermore, the subthreshold characteristics are very good even near pinch-off. For the same reason, the linearity of the device characteristics becomes good and it is possible to achieve equal g _n .

[Brief explanation of the drawing]

Fig. 1 is a cross-sectional view of the main part of an embodiment of the present invention, Fig. 2 is an energy band diagram of an embodiment of the present invention, Fig. 3 is a partially enlarged view of Fig. 2, and Fig. 4 is a diagram of the quantum level. Explanatory diagram, Figure 5 is a sectional view showing the outline of a conventional MESFET, Figure 6 is a sectional view of the main part of a conventional HEMT, and Figure 7 is a sectional view showing the outline of a conventional MESFET.
The figure is an illustration of the short channel effect, Figure 8 is a diagram showing the subthreshold, and Figure 9 is a diagram of the conventional MESFET.
FIG. 10 is an energy band diagram of another embodiment of the present invention, and FIG. 11 is an energy band diagram of still another embodiment of the present invention. Main code: 1...Semi-insulating (SI) GaAs substrate,
2...i-AlGaAs layer, 3...i-GaAs layer, 4
... i-AlGaAs layer: tunnelable barrier layer, 5
...i-GaAs layer, 6...doping layer: n-
GaAs layer, 7...i-GaAs layer, 8...i-
AlGaAs layer, 9...i-GaAs layer, 12, 13...
...source, drain electrode, 14...gate electrode.

Claims (1)

[Claims] 1. A field-effect transistor has a single quantum well (SQW) having at least one tunnelable barrier as a channel structure, and two of the single quantum wells are separated by the barrier. A field effect transistor characterized in that a doping layer is provided in at least one of the narrower regions among the above regions, and the wider region is undoped.