JPH02231733A - Semiconductor device - Google Patents

Abstract

PURPOSE:To improve mutual conductance interruption frequency by arranging the constitution such that the thicknesses of the layers, which have narrow forbidden band widths, on both sides of a layer, which has wide forbidden band width are formed different from each other. CONSTITUTION:In an alternate layer 11, the thicknesses of layers, which have narrow forbidden band widths and are provided on both sides of a wide forbidden band width, are different. For this reason, the energy levels of minibands, which are formed in narrow forbidden band widths on both sides, are put in different conditions. Accordingly, even if a gate electrode 19 is deeply and negatively biased, electrons cases to overflow from an electric conductive layer to this alternate layer 11 causing real space transition. Hereby, short channel effect is suppressed, and mobility becomes great, and mutual conductance improves and also interruption frequency improves.

Description

[Detailed description of the invention] Industrial application field> This invention relates to a field effect transistor, more specifically a high electron mobility transistor having a heterostructure (hereinafter referred to as HEM).
(referred to as T).

BACKGROUND ART One of the important performances of a field effect transistor (PET) is the cutoff frequency ft. As is widely known, when the gate length i2g is shortened, the mutual conductance gl increases in inverse proportion to this h, and at the same time, the input capacitance 11Cgs decreases, which improves f t (=gs/2 πCgs). For example, in the case of a HEMT used for microwave band amplification, when Qg=0.25u-, ft=on the order of 70 GHz.

However, when the gate length Qg is made shorter than this, to approximately 0.1μ, a phenomenon in which the g length actually decreases, a so-called short channel effect, occurs, which becomes an obstacle to improving the ft of PET.

Conventionally, HEMTs aiming to improve the cut-off frequency have a fourth
There are some as shown in Fig. 5 or Fig. 5.

The HEMT shown in FIG. 4 has 1!
A GaAs layer 37, a GaAs channel layer 32, and an n 12 GaAs layer 38 as an electron supply layer are sequentially laminated, and a source electrode 47 is formed on this n-A 17 GaAs layer 38. Drain 1! It has a structure in which a fiJ8 and a gate electrode 49 are provided. This HEMT consists of an AI2G aAs layer 37 having a larger energy gap than the GaAs channel layer 32.
By forming a potential barrier at the hetero interface 32a, Ga
By preventing the electrons 5 in the As channel layer 32 from overflowing to the Ai2GaAs layer 37, the above short channel effect is suppressed.

On the other hand, in the HEMT shown in FIG. 5, alternating layers 6l consisting of a superlattice of an AI2GaAs layer 57 and a GaAs layer 62 are provided on a GaAs substrate 51, and a GaAs channel layer 52
,n-AI2GaAs layer 58. Source electrode 67. It has a structure in which a drain electrode 68 and a gate electrode 69 are provided. The short channel effect is suppressed by the potential barrier of the hetero interface 52a.

Problems to be Solved by the Invention However, in the conventional HEMT shown in FIG. 4, l! When electrons lO enter the GaAs layer 37, as is widely known, A(
Since the crystallographic quality of 20aAs is inferior to that of GaAs,
The electron IO is placed in the defect level ll in the A12GaAs layer 37.
There are problems in that the threshold voltage fluctuates and the noise current increases. Furthermore, since the AQGaAs layer 37 is a single layer and is thicker than other layers, a significant lattice mismatch occurs between the AQGaAs layer 37 and the GaAs channel layer 32, which disturbs the flatness of the Peter interface 32a at the atomic layer level. . Therefore, the crystal quality of the GaAs channel layer deteriorates, the mobility of the electrons 35 decreases, the mutual conductance g decreases, and the cutoff frequency is not improved.

On the other hand, HEM T shown in FIG.
Unlike T, a GaAs channel layer 52 is formed on an alternating layer 61 consisting of a superlattice of an AQGaAs layer 57 and a GaAs layer 62.
, the lattice mismatch is relaxed, the hetero interface 52a becomes flat, and the Ai2G above the alternating layer 6l
Impurities in the aAs layer are confined to this hetero interface 52a. Therefore, if there are no other factors, the above GaAs
The mobility of electrons 60 within channel layer 52 should be improved. However, as shown in Figure 6, this HEM
T means that when the gate electrode is deeply negatively biased, electrons 60 pass through the mini-bands 63 formed in the GaAs layer 62 by the superlattice structure of the alternating layers 6l.
1 (real space transition effect), the mobility of the electrons 60 decreases, making it impossible to improve the cutoff frequency ft. Furthermore, when electrons overflow in this way, the threshold voltage fluctuates and the noise current also increases.

Therefore, an object of the present invention is to suppress the short channel effect, increase the mobility, improve the mutual conductance, and improve the cutoff frequency by preventing the above-mentioned real space transition from occurring in the alternating layers. HEMT
Our goal is to provide the following.

Means for Solving the Problems> In order to achieve the above object, the HEMT of the present invention has an electrically conductive layer on the surface of a semiconductor substrate parallel to the surface, and a connection between this electrically conductive layer and the surface of the substrate. In a semiconductor device having an alternating semiconductor layer in which layers with a wide bandgap width and layers with a narrow bandgap width are alternately formed between the layers, narrow bandgap widths on both sides of the layer with a wide bandgap width are formed. It is characterized in that the layers have different thicknesses.

Further, it is preferable that one of the two layers constituting the alternating semiconductor layer is made of the same material as the electrically conductive layer.

In the alternating layers, since the layer thicknesses of the layers with narrow forbidden band widths provided on both sides of the layer with wide forbidden band widths are different from each other, the mini-bands formed within the narrow forbidden band widths on both sides are The energy levels become different from each other. Therefore,
Even if the gate electrode is deeply negatively biased, electrons will not overflow from the electrically conductive layer into this alternating layer due to real space transition. Therefore, short channel effects are suppressed, mobility is increased, mutual conductance is improved, and cut-off frequency is improved.

Note that since electrons no longer overflow, fluctuations in threshold voltage and noise current are reduced.

In addition, when one of the two layers constituting the alternating layers is made of the same material as the electrically conductive layer, the lattice mismatch between the electrically conductive layer and the alternating layers is reduced, and the mobility is further reduced. The larger the transconductance, the higher the cutoff frequency.

Embodiments> Hereinafter, the HEMT of the present invention will be explained in detail with reference to illustrated embodiments.

As shown in Figure 1, this HEMT consists of semi-insulating GaA
On the surface of the s-substrate l, layers l2 and l! made of GaAs are formed. Ga
2g+layers (in this case, 1=
5 and 10 layers), an undoped GaAs layer 2 as an electrically conductive layer, an undoped AQGaAs layer 14 as a spacer layer, a Si-doped n-Ai2GaAs layer 8 as an electron supply layer, and a contact resistance. Si-doped n-GaAslB is used as a reduction cap layer, a source electrode 17 and a drain electrode 18 are formed on the n-GaAs layer 16, and a gate is formed on the n-AllGaAS layer 8 within the central opening of the n-GaAs layer 16. electrode l
9 in sequence.

This HEMT is manufactured as follows.

■First, GaAs is grown on the surface of the GaAs substrate 1 which has been cleaned in advance by the molecular beam epitaxial growth method (MBE method).
Ji l 2 , AI2G. The aAs layers 7 are stacked alternately. At this time, the layer thickness of the GaAB layer l2 grown at the (2k-1)th (the value of k is 2, 3, ..., (a-1)) is set to the (2k-3)th and The layers are grown to have different thicknesses from the (2k+1)th GaAs layer 12. Note that the thickness of each GaAs layer 12 is preferably an integral multiple of the lattice constant of GaAs.

On the other hand, the layer thicknesses of the even-numbered Al2GaAs layers 7 are the same, 20 atomic layers. In addition, AQGaA
The composition of s is expressed as 12xGa+-xAs. A above
The Al2 mixed crystal ratio of the QGaAs layer 7 is set to 0.25.

In this way, each layer is grown and laminated up to the 211th layer to form alternating layers 2.

■Subsequently, by MBE method, undove GaAs was
An undoped GaAa layer 7 is grown by 0 atomic layer, an undoped AI2GaAs layer 14 is grown by 4 atomic layers of undoped Al2GaAs, and a carrier concentration 1X 1 0 ”
n-AQGaA doped with Si so that cR″″3
s (1! Mixed crystal ratio is 0.25) by growing 100 atomic layers to form an n-Al2GaAs layer 8, carrier concentration 2xl016
2 Si-doped n-GaAs to form G wing-3
The n-GaAs layer 16 is continuously formed in this order by growing only 0 atomic layer.

(2) Next, using the photoresist as an etching mask, mesa etching is performed to isolate the elements. After removing the above photoresist, the film thickness was reduced to 10% by lift-off method.
Source electrode l7 and drain electrode l made of Au-Ge film, Ni film, and Au film of 00, 100, and too many people.
form 8. And the above n-GaAslil
The temperature was set at 400°C in a nitrogen gas atmosphere so that the electrodes were in contact with the ohmic. ■Perform syntax in minutes.

(2) Thereafter, a resist mask 2l having a window of 0.1 .mu.l and width of 200 .mu.m is formed in the center of the element by electron beam exposure. Then, the n-GaAs layer 16 is recess-etched using a chemical etching solution, and the n-GaAs layer 16 is recessed.
The central portion of the surface of the Al2GaAs layer 8 is exposed. Subsequently, as shown in FIG. 2, Al2 is deposited on the entire surface and a gate electrode 19 is formed by a lift-off method to complete the fabrication. In addition, corresponding to the above resist mask 2l,
The gate length of this HEMT is 0.1μ. Gate width is 20
It becomes 0μ1.

In this way, in the alternating layers 1l, a wide forbidden band width of Al
Narrow forbidden band width G provided on both sides of layer 7 made of 2GaAa
When the thicknesses of the layers 12 made of aAs are different from each other, the energy levels 13 of the mini-bands formed in the layers 12 made of GaAs on both sides are different from each other, as shown in FIG.

Therefore, even if the gate electrode 19 is operated with a deep negative bias, electrons will not overflow from the GaAs layer 2 to the alternating layers 11 due to real space transition. Therefore, the short channel effect can be suppressed, the mobility can be increased, the mutual conductance gm can be improved, and the cutoff frequency ft can be improved. Accordingly, fluctuations in threshold voltage and noise current can be reduced.

Further, in the above-mentioned HEMT, one layer l2 of the two layers constituting the alternating layer I1 is replaced with a GaAs layer 2 which is an electrically conductive layer.
Since the GaAs layer 2 and the alternating layer 11 are made of the same material, the lattice mismatch in the hetero interface 2a between the GaAs layer 2 and the alternating layer 11 is reduced, further increasing the mobility and improving gmJr.

In this embodiment, the alternating layers 11 are composed of 10 layers (115 in the number of layers of 2 m), but the structure is not limited to this, and other layers may be used as long as the lattice mismatch with the GaAs layer 2 does not become noticeable. It may also be the number of layers.

Effects of the Invention> As is clear from the above, the present invention has an electrically conductive layer on the surface of a semiconductor substrate that is parallel to the surface. both,
A semiconductor device having alternating semiconductor layers formed by alternately forming layers having a wide bandgap and layers having a narrow bandgap between the electrically conductive layer and the surface of the substrate, the semiconductor device having a wide bandgap. Since the layer thicknesses of the layers with narrow forbidden band widths on both sides of the layers are different from each other, it is possible to avoid real space transitions in the alternating layers, thus suppressing the short channel effect and increasing the mobility. It is possible to improve the mutual conductance and cut-off frequency.

In addition, when one of the two layers constituting the alternating layers is made of the same material as the electrically conductive layer, the lattice mismatch between the electrically conductive layer and the alternating layers can be reduced, and even more easily moveable. The degree of mutual conductance can be increased. The cutoff frequency can be improved.

[Brief explanation of the drawing]

FIG. 1 is a schematic sectional view showing an embodiment of the HEMT of the present invention, FIG. 2 is a schematic sectional view showing the state of the HEMT during the manufacturing process, and FIG. 3 is a diagram showing the energy band of electrons in the HEMT. , Fig. 4. Figure 5 shows the conventional HEM
A schematic sectional view showing T, FIG. 6 is a conventional }I shown in FIG.
It is a diagram showing energy bands of electrons in EMT. l... Semi-insulating GaAs substrate, 2... GaAs electrically conductive layer, 2a... Hetero interface,
? -A layer made of GaAs, n-12GaAs electron supply layer, 1l...alternating layer, 12...layer made of GaAs, 1 4--A (lGaAa spacer layer, 16...r
(-GaA8 contact resistance reduction cap layer, l7...source electrode, 18...drain electrode, 19
...Gate electrode. Figure 1

Claims (2)

[Claims] (1) Having an electrically conductive layer parallel to the surface on the surface of the semiconductor substrate, and a layer having a wide forbidden band width and a layer having a narrow forbidden band width between this electrically conductive layer and the surface of the substrate. A semiconductor device having alternating semiconductor layers formed alternately, characterized in that the layers having a narrow bandgap on both sides of the layer having a wide bandgap have different thicknesses. (2) The semiconductor device according to claim 1, wherein one of the two layers constituting the alternating semiconductor layer is made of the same material as the electrically conductive layer.