WO1997005661A1

WO1997005661A1 - Modulation-doped field-effect transistor with a composition-modulated barrier structure

Info

Publication number: WO1997005661A1
Application number: PCT/DE1996/001427
Authority: WO
Inventors: Karl-Heinz Bachem; Lester F. Eastman
Original assignee: Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V.
Priority date: 1995-08-01
Filing date: 1996-08-01
Publication date: 1997-02-13
Also published as: DE19528238A1; DE19528238C2; EP0842542A1

Abstract

The invention concerns a modulation-doped field-effect transistor comprising a channel layer (3) to which is applied a barrier layer (4) adapted as concerns the lattice to the channel layer crystal structure. The barrier layer is composed of a mixed crystal structure and doping. In particular, the buffer layer (2) consists of GaAs, the channel layer (3) of GaInAs and the barrier layer (4) of GaxIn1-xP. The invention is characterized in that the ratio between the elements of the mixed crystal composition is site-dependent such that the ratio at at least one point within the barrier layer (4) has an extreme value. In particular, the Ga content x of the mixed crystal within the barrier layer (4) is site-dependent and comprises at least one region in which x has a maximum value.

Description

Modulation-doped field effect transistor with composition-modulated barrier structure

Description

Technical field

The invention relates to a modulation-doped field effect transistor with a channel layer, on which a barrier layer is applied, which is lattice-matched to the crystal structure of the channel layer, which is composed of a mixed crystal structure and has a doping.

State of the art

Modulation-doped field effect transistors, called MODFET components for short, are further developed field effect transistors whose conductive channel through which the source / drain current flows is undoped, whereas the barrier layer adjoining the channel layer has dopants in a characteristic manner. The doping is placed in the barrier layer, for example, in atomically sharp planes, so-called delta doping, so that the banding pattern in the barrier layer can be influenced in a characteristic manner, which, as will be explained in more detail later, also results in the banding pattern in¬ is "bent" within the channel area in such a way that energetic conditions result in this area, which preferably lead to the formation of 2-dimensional electron gas densities.

A special technical quality feature of MODFET Components is the very high achievable 2-dimensional electron gas density in the current-carrying channel area. The electron gas density that forms limits the maximum current in the component given the width of the component and thus the channel, which essentially determines the microwave power available at the output of the component. For this reason, an essential aspect of the development work on MODFET components is aimed at developing semiconductor structures which permit the formation of ever larger 2-dimensional electron gas densities in the current-carrying channel region.

In addition to the aforementioned aspect, a second, essential quality criterion of the components in question is the dielectric strength of the barrier structure lying between the control electrodes and the channel area. The dielectric strength of the barrier structure determines the operating voltage level of the component, which as a second factor significantly influences the microwave power available on the component.

The currently most widespread MODFET is the so-called AlGaAs / GalnAs / GaAs-MODFET. In the articles by J. Dickmann and H. Däembkes, IEEE Tr.o.El.Dev. , Vol. 42, No. 1, Jan. 1995, pp. 2-6 and by L. Jelloian, et al. , IEEE Electrical. Dev. Lett., Vol 15, No. 5, May 1994, pp. 172-174, such MODFETs are described. However, this known electronic component has mixed-crystal-bound aluminum in its barrier layer, which, as a highly corrosive element, makes the manufacture of such layers considerably more difficult. As is well known, aluminum-containing materials are only difficult to breed in good quality, but what is only possible on an industrial scale using complex molecular beam epitaxy systems. In addition to this manufacturing-related complexity, the presence of aluminum also leads to corrosion problems on the electrical contact areas of the component itself.

In addition to the mixed crystal AlGaAs, the compound GalnP has become known for use for the barrier structure. It has been shown that the advantageously modified MODFET has better noise behavior than its predecessor and, moreover, is easier to manufacture. By eliminating the aluminum component, the epitaxial deposition of the barrier layer is considerably simplified. In particular, aluminum-free structures with the gas-phase epitaxy which is particularly suitable for mass production with organometallic source materials are easier to produce in the industrial sector. In addition, the combination of phosphidic and arsenid layers in one structure offers advantages in terms of production technology, especially since certain etching selectivities can be advantageously used in the structuring of the components.

The extremely big advantage that can be achieved from the manufacturing side in the production of the GalnP / GalnAs / GaAs-MODFET has to be contrasted with the not to be despised disadvantage, namely that compared to the above-described AlGaAs / GalnAs / GaAs-MODFET considerably smaller electron gas densities in the channel area and also lower electrical breakdown voltage ratios. This night ile are material-specific and ultimately result from the smaller so-called conduction band discontinuities between the area of the channel and the barrier layer. The only known way to eliminate these disadvantages so far goes back to the inventor of this application and consists in the substitution of the barrier material GalnP by the quaternary mixed crystal AlGalnP. Although this manipulation achieves the desired high electron density in the channel, the advantages of an aluminum-free structure listed above are also lost again.

Presentation of the invention

The invention is based on the object of further developing a modulation-doped field effect transistor with a channel layer on which a barrier layer which is matched to the crystal structure of the channel layer and which is composed of a mixed crystal structure and has a doping is applied, so that on the one hand the electron density determining the performance increase of the component is increased and on the other hand the breakdown voltage between the channel and the control electrode connections, which is also important for the quality of the component, is optimized. These optimizations should, if possible, be carried out in such a way that the element aluminum, which has the disadvantages mentioned, can be dispensed with. In particular, it is important to optimize the known GalnP / GalnAs / GaAs MODFET in the manner described. The production of such an optimized component should also be suitable for production on an industrial scale without great technical effort, so that the pro- production costs can be kept as low as possible.

The solution on which this task is based is to be described on the basis of an example which does not restrict the general inventive concept in more detail. Starting from a general structure of a MODFET, as can be seen in FIG. 1, an undoped buffer layer is on a semi-insulating substrate layer 1

2 applied. In the example described, layers 1 and 2 consist of GaAs. Of course, alternative substrate materials can also be used.

The channel layer applied over the buffer layer 2

3 is characteristically undoped in the MODFETs in question and in this example consists of GalnAs which is pseudomorphically deposited on the GaAs buffer layer 2. A barrier structure 4, which consists of GalnP, is applied to the channel layer 3 in a lattice-adapted manner. A doped barrier region 5 is provided within the barrier structure 4 and is, for example, offset with suitable doping atoms within an atomically sharp plane in accordance with a delta doping. Contact layers 6 made of highly doped GaAs are also provided on the barrier structure 4, on which the ohmic contacts 7 are provided for the source and drain connections. In the center of the ohmic contacts 7, a control electrode 8 made of metal is placed directly on the barrier structure 4, via which the gate voltage can be applied.

FIG. 2 shows the associated band diagram for the known MODFET structure shown in FIG. 1. In the energy / location diagram, the energetic see tape runs through the layers 2 to 6 described above. The axis of abscissa X represents the cross-sectional line through the layer structure, the energy values of the associated band profiles are plotted on the ordinate.

The continuous graph EC corresponds to the course of the lower conduction band edge energy as a function of the position coordinate for the control electrode voltage VG = 0 V. In comparison, the course of the Fermi energy in the individual layer sections is shown with EFH. It is assumed here that the position of the Fermini levels within the buffer layer 2 is largely independent of the applied control electrode potential, since its influence only affects channel 3. In the area of the control electrodes 6 it can be seen how the Fermi energy changes as a function of different control electrode potentials (VG = 0 V or VG V-. 0 V).

For the formation of a 2-dimensional electron gas, it is necessary for the conduction band edge energy to drop below the Fermi energy, since this represents the energy of the uppermost occupied electron state. It is only in this band constellation that it is possible for the electrons to pass into the conduction band and thus allow current to flow. This characteristic band course is shown in FIG. 2 in the channel layer 3. The aim of the current efforts is now to lower the conduction band edge energy within the channel layer as far as possible below the Fermi energy so that as many electrons as possible can get into the conduction band in order to achieve a high to contribute dimensional electron gas density.

A lowering of the conduction band edge energy within the channel layer is to be carried out in a known manner by a targeted increase in the doping within the barrier structure. The reference symbol ED represents the local and energetic position of the donor state introduced in isolation in the barrier layer structure 4. The characteristic course of the conduction band edge energy between the channel layer 3 and the barrier structure 4 is commonly referred to as conduction band edge discontinuity, the size and nature of which can be derived from the Maxwellian relationships.

It has also been shown that the lowering of the conduction band edge energy within the channel layer 3 can be increased by the

Dopant concentration is increased within the doping introduced locally in the barrier layer 4. This relationship can be seen in particular in FIG. 3, top representation, in which a diagram representation comparable to FIG. 2 is shown. In this diagram only the conduction band edge energies of the buffer, channel and barrier structure layers are shown. The different band energy profiles a, b and c correspond to different levels of doping within the barrier layer 4, where b represents the highest doping. The curve a corresponds to a vanishingly small doping in the barrier 4 (see also the very thin line for the donor state). In this case, channel 3 is almost free of charge carriers. For the curves b and c there are higher dopings accepted. The conduction band edge energy in the channel now touches the Fermi level, see curve b, or is below the Fermi level, see curve c within the channel layer. Curve c shows the case of the highest possible electron concentration within the channel layer. This case arises when the minimum of the conduction band edge energy within the barrier layer 4 is only a few lOmeV above the Fermi level. However, a further increase in the electron concentration within the channel layer by increasing the doping is not possible, especially since a further increase in the doping would not allow the conduction band edge energy within the barrier structure to be set at the Fermi level. Rather, the conduction band drops below the Fermini level, as a result of which an electron gas would likewise form in the barrier layer in the vicinity of the minimum. The formation of a further electron gas in addition to that within the channel layer would increase the capacitance of the gate electrode very much, especially since in this case not only the channel charge but also the parasitic charge formed within the barrier layer would have to be modulated when the MODFET component is operating . As a result, however, the MODFET would suffer a very great loss in performance.

The above considerations are now based on the usual assumption that the ratio of gallium and indium within the barrier structure layer is 50% in each case. This relationship can also be seen from the lower representation in FIG. 3, from which the composition factor x of GaxInl-xP is plotted in relation to the barrier thickness. Experiments have shown that the lattice-matched GalnP barrier layer 4, in which the Ga content is equal to the In content, can be replaced by a so-called pseudomorphic GaAnP barrier layer with an increased Ga content. Here, the Ga content has been increased uniformly in the entire barrier layer 3 by more than 50% content. By this manipulation of the band gap is increased, which means that the whole band energy Leitungs¬ is shifted away from the Fermi level, ^'whereby the conduction band discontinuity is also enlarges ver¬ at the same time. However, the formation of this effect is very weak.

However, the use of this measure leads to the problem that an increase in the Ga content changes the lattice structure of the GalnP barrier material in such a way that lattice-adapted growth of the barrier layer material increased in the Ga content is not possible without lattice dislocations over large layer thicknesses. For example, the GaAs buffer layer has a cubic lattice cell structure, whereas the modified GalnP lattice has a tetragonal lattice cell structure. A growth of larger layer thicknesses on the buffer layer material is therefore not possible without the formation of internal mechanical crystal stresses, which relax when certain mechanical limit stresses are exceeded, i.e. the crystal lattice assumes its original lattice constant, which ultimately results in an unusable component.

The maximum growth thickness up to which no relaxation processes occur is also referred to as the critical layer thickness. With usual barrier layer Thicknesses of approximately 20 nm may therefore only be raised slightly above 50%, so that internal stresses are avoided as far as possible in order to adapt the critical layer thickness to the desired layer thickness. Due to the only slight increase in the Ga content, the above-described effect of increasing the bandgap is only very slight.

According to the invention, it has now been recognized that the effect of increasing the Ga content within the barrier layer while simultaneously increasing the critical layer thickness can be achieved by not increasing the Ga content uniformly in the entire barrier layer 4, but only at the points where maximum efficiency should be achieved.

According to the invention, therefore, a modulation-doped field effect transistor with a channel layer, on which a barrier layer, which is lattice-matched to the crystal structure of the channel layer, is applied, which is composed of a mixed crystal structure and has a doping, is further developed in that the ratio of the element composition of the mixed crystal is such It is location-dependent that the ratio has an extreme value at least at one point within the barrier layer. In the example of a GalnP / GalnAs / GaAs MODFET which does not limit the inventive concept, the Ga content of the mixed crystal within the barrier layer is location-dependent and is greatest at least in one area.

According to the invention, it has been recognized that by specifically introducing gallium within the barrier layer preferably at the areas where the conduction band edge energy has a minimum value in the case of a uniform distribution in the mixed crystal between Ga and In, the benefit being greatest. This is usually the case at the points within the barrier layer where the highest dopant concentrations are.

Due to the location-dependent composition of the mixed crystal composition within the barrier layer, the term composition-modulated MODFET has been created for the abbreviated description of the component.

Brief description of the invention

The invention is described below by way of example with reference to the drawings, without restricting the general inventive concept. Show it:

1 shows a schematic cross-sectional representation through a GalnP / GalnAs / GaAs MODFETs;

2 associated band diagram of the conduction band energies and Fermie energies in the layers involved;

3 shows a band diagram of a GalnP / GalnAs / GaAs MODFET with a non-location-dependent chemical composition in the GalnP barrier layer;

4 shows a band diagram of a MODFET modified according to the invention with a location-dependent chemical composition in the GalnP barrier layer; 5 alternative embodiment of the composition-modulated MODFET according to the invention;

6 representation of band diagrams, the same doping and variable control electrode potential with constant chemical composition in the barrier material;

7 shows band diagrams with constant doping and variable control electrode potential with composition-modulated barrier layer;

8 shows band diagrams to illustrate the differences between the tunnel barriers for electrons with a conventional barrier layer and the composition-modulated barrier layer;

9 shows band diagrams with an enlarged tunnel barrier within the barrier layer;

10, 11 and 12 advantageous embodiments of the composition-modulated barrier layer according to the invention.

Representation of exemplary embodiments

1 and 2 show the layer structure and the conduction band energy curves of a typical modulation-doped field effect transistor. As already explained in more detail above with reference to FIGS. 3 and 4, an enlargement of the conduction band discontinuity and thus the increase in the distance between the conduction band energy can are carried out according to the invention within the barrier layer and the Fermi energy, by changing the gallium content within the barrier structure depending on the location.

According to FIG. 4, lower representation, the gallium content assumes a proportion factor of 75% only at a single point, in a local minimum, the gallium content being lower in the other areas. By forming a local maximum of the gallium content x, which is also referred to as the composition factor x, it can be achieved that the average gallium content within the barrier structure 4 is increased only slightly above 50%, so that in this way a relatively high critical layer thickness can be achieved.

4 shows the conduction band energy curves with different doping within the barrier structure 4. The courses correspond to the band energies of the upper illustration in FIG. 3. The band curve a corresponds to the lowest doping within the barrier layer 4, the band curve d corresponds to that with the highest doping. In addition to the increase in the distance between the minimum of the conduction band energy within the barrier layer 4 and the fermi energy (reference line), it should be pointed out in particular that the reduction in the conduction band energy within the channel layer 3 below the fermi energy can be increased considerably with the measure according to the invention. In this way, the 2-dimensional electron gas density within the channel layer 3 can be increased and, on the other hand, the breakdown voltage behavior of the barrier layer can be improved.

5, lower representation, is an improved one Composition of the gallium content within the barrier layer 4 is shown. The formation of the local minimum with a composition factor of x equal to 75% is even sharper than in the illustration according to FIG. 4, so that the average gallium content within the barrier layer 4 is lower, as a result of which the internal mechanical stresses within the lattice structure can be further reduced.

The measure according to the invention, i.e. A targeted, location-dependent introduction of gallium within the barrier structure 4 thus considerably increases the band gap and thus also the distance of the conduction band edge from the Fermi level in the barrier exactly where it brings the greatest benefit in terms of component physics. The local increase in the band gap in the bypass of the doped barrier region also allows the electron concentration in channel 3 to be increased by increasing the doping in the barrier.

6 and 7 show band diagrams with constant doping in each case with variable control electrode potential VI to V4. These curves show how the channel density changes depending on the control voltage. In Fig. 6 the band course is assumed within a barrier 4 with a constant chemical composition. It can be seen that when the control electrode potential is lowered, the conduction band edges are shifted in the direction of the Fermini level. FIG. 7 shows how the conditions change when a composition-modulated barrier according to the invention with a gallium distribution within the barrier layer 4 according to FIG. 4, bottom illustration, is used. The arrow shown in FIG. 7 illustrates the amount by which the tax electrode potential and thus the channel density is to be increased.

The advantage of the composition-modulated barrier according to the invention can be seen, in particular, in the fact that with only a moderate increase in the average Ga content in the barrier, the conduction band edge energy of the barrier in the critical, doped region can be increased substantially more than with a general increase in the Gallium content over the entire barrier thickness is possible. With the composition-modulated barrier according to the invention, the "critical layer thickness" can thus also be used more effectively.

As already explained, the composition-modulated barrier according to the invention is not only suitable for increasing the maximum electron concentration in the channel, but can also be used with a corresponding refinement of the composition profile to increase the dielectric strength of the barrier structure.

8 shows band diagrams for barriers with a homogeneous distribution of gallium within the barrier layer 4 (see dashed line representation) and for barriers according to the invention composition-modulated (see solid line courses).

The upper dashed line in FIG. 8 corresponds to a conduction band course for a strongly negative control electrode potential, whereas the lower dashed line corresponds to the conduction band energy with a strongly positive control electrode potential. In contrast, the band diagrams represent composition-modulated barriers according to a gallium distribution according to FIG. 4, lower representation. Here, too, the upper continuous curve shape corresponds to the energetic conditions with a strong negative control electrode potential, the lower band shape with strongly positive control electrode potential. It can be seen that the tunnel barriers for electrons are always smaller with a uniformly distributed barrier material composition than with composition-modulated barriers.

This effect can be increased by increasing the band gap in the barrier below the control electrodes again. This is shown in Fig. 9. The associated composition profile of the gallium content within the barrier layer 4, which is shown in FIG. 10, has two maxima, the first maximum in the doped region of the barrier and the second maximum below the control electrode. The first maximum contributes to increasing the electron density in the channel, whereas the second maximum increases the tunnel barrier at a favorable point. Of course, the gallium concentrations of the maxima must be coordinated with one another if the mean gallium concentration is chosen to be constant. The highest possible electron density and dielectric strength cannot be set simultaneously when using GalnP as a barrier material.

To increase the dielectric strength, it is advantageous to add aluminum to the second maximum, which is evident from the composition profile according to FIG. 11. Here, aluminum has been added in the undoped barrier area 4 below the control electrode, especially since aluminum is the least disruptive in this area (no DX centers). 11 shows the composition factors x for gallium and y for aluminum.

FIG. 12 shows a further variant of the composition modulation within the barrier layer. Here, the band gap is only increased below the control electrode. This barrier profile would be chosen if a high maximum electron density in the channel is not required, for example in the case of low-noise small signal amplifiers, but the leakage currents of the control electrode are to be reduced.

Claims

Patent claims

1. Modulation-doped field effect transistor with a channel layer on which a barrier layer applied to the crystal structure of the channel layer is applied, which is composed of a mixed crystal structure and has a doping, characterized in that the ratio of the element composition of the mixed crystal is so location-dependent that the ratio has an extreme value at least at one point within the barrier layer.

2. Modulation-doped field effect transistor, which has a buffer layer made of GaAs, a channel layer made of GalnAs and a barrier layer made of GaxInl-xP, to which contact electrodes are applied for electrical control, characterized in that the Ga content x of the mixed crystal within the barrier layer depends on the location and has at least one area where x is greatest.

3. Modulation-doped field effect transistor according to claim 2, characterized in that the Ga content x has a local maximum within the barrier layer.

4. modulation-doped field effect transistor according to one of claims 1 to 3, characterized in that the channel layer is undoped,

5. Modulation-doped field effect transistor according to one of claims 1 to 4, characterized in that the doping in the barrier layer is delta doping or pulse doping.

6. Modulation-doped field effect transistor according to one of claims 3 to 4, characterized in that at the local maximum 'of the Ga content x is equal to 0.75.

7. Modulation-doped field effect transistor according to one of claims 1 to 6, characterized in that the barrier layer has a thickness of about 20 nm.

8. Modulation-doped field effect transistor according to one of claims 2 to 7, characterized in that the Ga content is set to a maximum at the locations in the barrier layer at which the conduction band edge energy in the case of a uniform distribution in the mixed crystal between Ga and In with x = 0.5 has a minimum value.

9. Modulation-doped field effect transistor according to one of claims 2 to 8, characterized in that the element ratio or the Ga content is extreme at the points of highest doping in the barrier layer.

10. Modulation-doped field effect transistor according to one of claims 1 to 8, characterized in that below one Contact electrode area within the barrier layer, the element ratio or the Ga content has a further extreme value.

11. Modulation-doped field effect transistor according to claim 1, wherein aluminum atoms are added to the mixed crystal below the contact electrode area within the barrier layer.