ELECTRONIC COMPONENT BASED ON P-DOPED GALIUM NITRIDE

Abstract

ELECTRONIC COMPONENT BASED ON P-DOPED GALIUM NITRIDE The subject of the invention is an electronic component (100) comprising: A substrate (107) An active stack (101i) formed above the substrate and including: a layer (110i) of p-doped Galium Nitride GaN, arranged above the substrate (107); a layer (111i) of a semiconductor material arranged on the layer (110i) of p-doped Galium Nitride GaN; the component comprising two lateral zones (105i) located on either side of the layer (110i) of p-doped GaN, the two lateral zones being implanted with oxygen. Figure to be published with the abstract: Figure 4

Description

Title of the invention: ELECTRONIC COMPONENT BASED ON P-DOPED GALIUM NITRIDE TECHNICAL FIELD OF THE INVENTION

[0001] The present invention relates to electronic components based on p-doped gallium nitride (GaN) and more particularly to GaN-based heterojunction power electronic components, such as high electron mobility transistors (or HEMTs, for "High Electron Mobility Transistor" in English) and Schottky diodes. TECHNOLOGICAL BACKGROUND OF THE INVENTION

[0002] Gallium nitride is today a highly prized material for the production of various electronic components. This is particularly the case for light-emitting diodes, Schottky diodes and so-called high electron mobility transistors (HEMTs). These components require the use of p- and n-doped semiconductors. Organometallic epitaxy is the most widespread growth technique for obtaining these components. However, hydrogen is one of the most significant contaminants in semiconductors produced from organometallic precursors. One of the disadvantages linked to the presence of hydrogen is that it will neutralize the acceptors when one seeks to obtain a p-doped semiconductor. This neutralization is achieved by the formation of electrically inactive acceptor-hydrogen complexes, making it more difficult to obtain effective p-doping.

[0003] The problem mentioned above concerning obtaining efficient p-doping hindered by the presence of hydrogen is found in particular during the production of GaN-based components such as HEMT-type heterojunction transistors both for RF applications (amplification, low noise amplification LNA Low Noise Amplifier, switches, oscillators, etc.) and for power applications (power transistor for energy conversion). A two-dimensional electron gas (or 2DEG, for "2-Dimensional Electron Gas" in English) is formed under the interface between a barrier layer (for example in AlGaN, InAIN, InAIN, AIN, ScAIN, AIN, InAlGaN, etc.) and the channel (for example in GaN). This 2DEG is connected at the source and drain by metallic ohmic contacts and is controlled by a gate (of the Schottky or MIS Metal-Insulator-Semiconductor type).

[0004] A heterojunction is formed by joining two semiconductor materials with different band gaps. For example, an AlGaN / GaN heterojunction comprises a layer of gallium nitride (GaN) topped by a layer of aluminum gallium nitride (AlGaN). The 2DEG forms under the interface between the AlGaN layer and the GaN layer, due to spontaneous polarization and piezoelectric polarization. This two-dimensional electron gas serves as a conduction channel within heterojunction electronic components, such as HEMT transistors and Schottky diodes.

[0005] An AlGaN / GaN heterojunction electronic component is generally manufactured from a semiconductor substrate (for example silicon, sapphire or SiC), by successively growing by epitaxy a nucleation layer, one or more transition layers, a thick buffer layer and the heterojunction layers (GaN channel, then AlGaN barrier). The buffer layer, several microns thick, makes it possible to limit lateral and vertical leakage currents in the component and to better confine the two-dimensional electron gas of the heterojunction. The semi-insulating buffer layer is for example formed of carbon-doped GaN.

[0006] This type of component supports high current densities in the on-state, due to the high density of charge carriers (electrons) and the high mobility of these carriers in the two-dimensional electron gas. However, it suffers from a transient phenomenon of current collapse in the on-state. This phenomenon is linked to charge trapping in the epitaxial structure, and more particularly in the carbon-doped GaN buffer layer. Traps become negatively charged when the component is in the off-state and then deplete the two-dimensional electron gas back to the on-state, by capacitive coupling effect. The depletion of the two-dimensional electron gas leads to an increase in the dynamic resistance and the drain saturation current in the on-state of the component (and therefore a decrease in the on-state current).

[0007] Patent application FR3047607 describes an example of a heterojunction transistor structure, which addresses this problem of current collapse in the on state.

[0008] With reference to [Fig.l], this first transistor structure comprises a substrate 11, at least one transition layer 12 arranged on the substrate 11, a buffer layer 13 (for example made of carbon-doped GaN) arranged on the transition layer 12, a p-doped GaN layer 14 arranged on the buffer layer 13, an n-doped GaN layer 15 arranged on the p-doped GaN layer 14, an unintentionally doped GaN layer 16 arranged on the n-doped GaN layer 15 and an AlGaN layer 17 arranged on the unintentionally doped GaN layer 16. The unintentionally doped GaN layer 16 and the AlGaN layer 17 form the heterojunction. A two-dimensional electron gas is intrinsically formed under the interface between the unintentionally doped GaN layer 16 and the AlGaN layer 17. This two-dimensional electron gas is illustrated in [Fig.l] as a layer 18 disposed between layers 16 and 17.

[0009] The p-doped GaN layer 14 and the n-doped GaN layer 15 together form a particularly high potential barrier (depleted PN junction) under the electron gas layer 18. This potential barrier, also called a confinement barrier (or back barrier BB according to English terminology) improves the confinement of the two-dimensional electron gas by reducing the number of electrons trapped in the buffer layer 13.

[0010] The transistor structure described in patent application FR3047607 makes it possible to limit the current collapse phenomenon but assumes the use of a p-doped GaN layer 14. This p-doping is for example obtained by a magnesium-based doping Mg used as an acceptor. To be efficient, the confinement barrier must have a maximum activation rate of Mg. Otherwise, it is necessary to use a higher concentration of Mg to reach the target acceptor concentration, which leads to a higher concentration of Mg in the channel up to the 2DEG gas and degrades the electronic transport properties in the channel. However, as we have seen above, the presence of hydrogen, particularly of organometallic origin, limits the activation of Mg.

[0011] A known solution to this problem of activation of the acceptors, here Mg atoms, consists of carrying out annealing under oxygen to avoid the creation of complexes between hydrogen and magnesium. This solution is however not ideal in the case of the HEMT transistor of [Fig.l] because the p-doped GaN layer 14 is located below a series of layers formed by the n-doped GaN layer 15, the unintentionally doped GaN layer 16 and the AlGaN layer 17. Thus, the fact that the p-GaN layer 14 is covered prevents the exo-diffusion of the hydrogen which remains present in the p-GaN layer 14. It is the configuration of the electric field in the upper layers which prevents the exo-diffusion of the hydrogen.

[0012] This difficulty is found in other electronic components using p-doped GaN such as light-emitting diodes or certain power components where the use of a buried p-layer makes it possible to obtain a positive threshold voltage. In this case too it is important to optimize the activation of the Mg atoms.

[0013] A known solution to the above problem has been proposed for applications in optoelectronics (case of light-emitting diodes) in the paper “Lateral Hydrogen Diffusion at p-GaN Layers in Nitride-Based Light Emitting Diodes with Tunnel Junctions”. Yuka Kuwano et al, Japanese Journal of Applied Physics 52 (2013) 08JK12 08JK12-1 # 2013 The Japan Society of Applied Physics. This solution is illustrated in Figures 2a and 2b. [Fig.2a] shows a conventional stack of a light-emitting diode with a buried p-GaN layer. According to [Fig.2b], the hydrogen responsible for the low activation rate of Mg is evacuated through the sides of the component. The IILN semiconductor stack is etched around the component and annealing in an atmosphere containing Oxygen allows the diffusion and evacuation of Hydrogen over a distance of the order of a few tens of microns and thus improves the activation of Mg.

[0014] This solution is however not satisfactory for all GaN-p based components. Thus, for a transistor, the production of etchings can lead to problems of integration (topology) and parasitic currents. In addition, a transistor often has dimensions greater than the hydrogen diffusion length and it is difficult to envisage etchings within the transistor. This approach is therefore not particularly suitable for the manufacture of transistors. More generally, the production of etchings to obtain mesa makes the manufacturing process of an electronic component more complex. Summary of the invention

[0015] There is therefore a need to provide an electronic component based on p-doped gallium nitride with satisfactory activation of the acceptors responsible for p-doping, without degrading the performance of the component and easy to manufacture, in particular when said electronic component is a HEMT type transistor.

[0016] To do this, the invention relates to an electronic component comprising: • a substrate; • a p-type doped Galium Nitride GaN layer, placed above the substrate; • a layer of a semiconductor material placed on the p-type doped Galium Nitride layer; the component comprising two lateral zones located on either side of the p-type doped GaN layer, the two lateral zones being implanted in oxygen.

[0017] A lateral oxygen-implanted zone is understood to mean a zone in which a large portion of the oxygen is available to form covalent bonds with hydrogen. In other words, this oxygen-implanted zone is not an oxygen-doped zone because in this case the majority of the oxygen present would be in a substitutional site in the GaN crystal lattice. Thus, the oxygen-implanted zone must be seen as a zone that has undergone an oxygen atom implantation step and thermal annealing but has not been subjected to a sufficient thermal budget for the oxygen to behave as a donor and form an n-doped GaN zone. The oxygen is here, on the contrary, randomly distributed within the crystal lattice and mainly in an interstitial position. After implantation there is also a large quantity of crystal defects in the matrix (each implanted oxygen atom displaces several hundred atoms from the matrix).

[0018] The invention is based on the creation of oxygen pockets (i.e. lateral zones) in areas located on either side of the GaN area where effective activation of the acceptors is sought. These oxygen pockets near the doped area and therefore atoms likely to act as acceptors (for example magnesium Mg in GaN) allow hydrogen, under the effect of post-implantation annealing, to diffuse towards these pockets and react with the implanted oxygen so as to reduce the formation of hydrogen-acceptor complexes. Unlike the solutions of the prior art, the component according to the invention does not require etching around the p-type layer. The lateral areas can be easily produced starting from the p-doped GaN layer in which oxygen is implanted on its lateral areas; in other words, the lateral areas are in the same plane as the p-GaN layer and form the end parts of this layer.

[0019] The component according to the invention may also have one or more of the characteristics below, considered individually or in all technically possible combinations: • The acceptors of the p-type doped Galium Nitride GaN layer are Magnesium atoms. • The width of the active stack and the width of each of the oxygen-implanted zones are less than or equal to 20 pm and greater than or equal to 1 pm, the width being measured parallel to the plane of the layers in alignment with the stack and the oxygen-implanted zones. • The active stack includes: • a layer of n-type doped GaN deposited on the layer of p-type doped GaN; • an unintentionally doped GaN layer deposited on the n-type doped GaN layer; • a semiconductor layer deposited on the unintentionally doped GaN layer to form a two-dimensional electron gas; the component comprising a source region, a drain region and a control gate region formed on or in the semiconductor layer to form a two-dimensional electron gas, the two oxygen-implanted lateral regions also extending on either side of the n-type doped GaN layer, the unintentionally doped GaN layer and the semiconductor layer to form a two-dimensional electron gas. • The electronic component includes: • a plurality of active stacks, each active stack comprising: • a layer of n-type doped GaN deposited on the GaN layer p-type doped; • an unintentionally doped GaN layer deposited on the n-type doped GaN layer; • a semiconductor layer deposited on the unintentionally doped GaN layer to form a two-dimensional electron gas; • two lateral oxygen-implanted zones extending on either side of the p-type doped GaN layer, the n-type doped GaN layer, the unintentionally doped GaN layer and the semiconductor layer to form a two-dimensional electron gas the source, drain and control gate areas being common to the plurality of stacks, each stack being electrically isolated from the adjacent stack by one of the oxygen-implanted lateral areas. • all of said stacks are surrounded by a zone, called an isolation zone, said isolation zone being an oxygen-filled zone. • According to one embodiment, each of the two lateral zones implanted with oxygen extends into or under the grid zone and into or under the drain zone and the source zone; • According to another embodiment, each of the two oxygen-implanted lateral zones extends into or under the gate zone and does not extend into or under the drain zone and the source zone.

[0020] The invention also relates to a method of manufacturing a component according to the invention comprising the following steps: • Production on a substrate of a stack formed above the substrate and including: • a p-type doped Galium Nitride GaN layer, placed above the substrate; • a layer of a semiconductor material arranged on the layer of semiconductor material; • Creation of a mask above the stack, said mask masking the part of the stack forming the active stack; • Ion implantation to introduce oxygen atoms into the two zones of the stack located on either side of the masked zone so as to obtain two lateral zones in the p-type doped Galium Nitride GaN layer implanted with oxygen; • Annealing following ion implantation.

[0021] The method according to the invention may also have one or more of the characteristics characteristics below, considered individually or in all technically possible combinations: • the annealing temperature is between 700 and 850°C. • the implantation is a multi-energy implantation. • the energy of the ionized oxygen atoms used for implantation is less than or equal to 200 keV. • the dose of implanted oxygen atoms is: • greater than 2.1014cm2x (Wa / Wo) where Wa denotes the width of the active stack and Wo denotes the width of the implanted lateral zones and; • less than 6.1015cm2 x (Wa / Wo). • the ion implantation step to introduce oxygen atoms is accompanied by an ion implantation step to introduce argon or nitrogen atoms.

[0022] The invention and its various applications will be better understood upon reading the following description and examining the accompanying figures. BRIEF DESCRIPTION OF THE FIGURES

[0023] The figures are presented for information purposes only and in no way limit the invention.

[0024] [Fig. 1] represents an example of a heterojunction transistor according to the prior art.

[0025] [Fig.2a] and [Fig.2b] represent a solution for activating magnesium in an optoelectronic component according to the prior art.

[0026] [Fig.3] represents a top view of an electronic component according to a first embodiment of the invention.

[0027] [Fig.4] represents a sectional view of the component of [Fig.3] along a plane perpendicular to the plane of the layers and along the grid finger.

[0028] [Fig.5], [Fig.6], [Fig.7] and [Fig.8] schematically represent the steps of the manufacturing process of a component according to the invention.

[0029] [Fig.9] represents a top view of an electronic component according to a second embodiment of the invention.

[0030] [Fig. 10a] and [Fig. 10b] represent two sectional views of the component of [Fig.3].

[0031] [Fig.11a] and [Fig. 11b] represent two sectional views of the component of [Fig.9].

[0032] [Fig. 12] represents an example of an oxygen implantation profile used for the production of the components according to the invention. DETAILED DESCRIPTION

[0033] [Fig. 3] represents a top view of an electronic component 100 according to the invention in an Oxy plane. According to this embodiment, the electronic component 100 is here a high electron mobility transistor (or HEMT, for “High Electron Mobility Transistor" in English). Transistor 100 comprises: • An active zone 101 including a plurality of elementary active zones 101 (i here ranging from 1 to 4); • A drain zone 102; • A source zone 103; • A grid area 104; • An area implanted with oxygen 105 including: • a peripheral isolation zone 106; • a plurality of elementary zones implanted in oxygen 105j (here j ranging from 1 to 3) arranged between each elementary active zone lOli.

[0034] It should be noted that the numbers i=4 of elementary active zones and j=3 of elementary zones implanted with oxygen are given here for purely illustrative purposes and that these numbers may vary depending on the intended applications.

[0035] As we will see later, the assembly formed by the active elementary zones 1011 and the oxygen-implanted zones 105j is produced from a single stack of epitaxial layers, certain zones of which have undergone oxygen implantation and thermal annealing, the thermal annealing being applied to the entire wafer.

[0036] The active elementary zones 1011 and 105j implanted in oxygen are arranged successively in the form of parallel strips along the direction Ox so that an active elementary zone 1011 is surrounded by two elementary zones implanted in oxygen 105j and 105j+i (case of the active elementary zones 1012 and 1013) or by an elementary zone and the peripheral insulation zone 106 (case of the active elementary zones 1011 and 1014).

[0037] The gate zone (also called gate finger) 104 is arranged between the source zone 103 and the drain zone 102 along the axis Oy perpendicular to the axis Ox along which the elementary active zones 101 and the oxygen-implanted elementary zones 105j extend. According to the invention, the source, drain and gate zones are common to all the active elementary zones 101 (i.e. there is a single source zone, a single drain zone and a single gate zone for the plurality of elementary active zones 101). In other words, the single gate zone controls all the elementary active zones.

[0038] According to this embodiment, the source 103 and drain 102 zones are both in contact (surface contact or buried in the zones) with the active zones 101 but also with the elementary zones implanted in oxygen 105j.

[0039] According to the invention, the gate finger 104 is both in contact (surface contact or via a dielectric layer, or buried in the AlGaN layer) with the active zones of Olim but also with the elementary zones implanted in oxygen 105j.

[0040] In the remainder of the description, Wa will be the width of each active zone lOli measured along the Oy direction and Wo the width of each elementary zone implanted in oxygen 105j also along the Oy direction. Lg will also be the gate length measured along the Ox axis. By convention, the “length” (L) corresponds to the transverse dimension (Lg,...) and the “width” (W) corresponds to the development of the transistor in the direction perpendicular to the plane of the transverse sections.

[0041] [Fig.4] shows a sectional view of the component 100, the section being made along a plane PI parallel to the plane Oyz and passing through the grid zone 104. The plane of the different layers is parallel to the plane Oxy of [Fig.3].

[0042] The component 100 comprises a substrate 107. The substrate 107 is for example made of silicon (intrinsic or doped), silicon carbide (SiC) or sapphire on which is deposited a transition layer 108 acting as a nucleation layer and adaptation of the lattice parameters between the substrate 107 and the subsequent GaN buffer layer 109. The transition layer 108 makes it possible to manage the mechanical constraints between the substrate 107 and the layers formed by the epitaxy. The layer 108 may include the superposition of a nucleation layer (typically AIN) and several adaptation layers (for example several layers of AIGaN with a decreasing molar fraction of AIN, or a superlattice comprising several AlxGa(lx)N / GaN bilayers).Such an intermediate layer 108 proves to be particularly advantageous in the event of a strong mismatch of lattice parameters between the GaN layer deposited above and the substrate 107, which could result in a prohibitive level of mechanical dislocations in these layers. In order not to unnecessarily burden the figures, this transition layer 108 will not be shown in the other figures.

[0043] The component 100 here comprises a buffer layer 109 deposited on the transition layer 108. The buffer layer 109 may have a thickness depending on the voltage targeted for the transistor 100. A relatively large thickness of the buffer layer 109 makes it possible to limit the lateral and vertical leakage currents in the transistor 100 and also to better confine the layer of electron gas that we will describe later. The buffer layer 109 may for example be made of carbon-doped GaN-SI (semi-insulator) or by superimposing a layer of GaN-SI / AlxGa(lx)N with low x, for example between 4 and 8%.

[0044] The active zones 101 are for example formed by a stack comprising successively: • a layer of p-doped GaN 1 lOi placed on the buffer layer 109, • a layer of n 11 li doped GaN placed on the layer of p 1 lOi doped GaN, • a layer of unintentionally doped GaN (channel) 112i arranged on the layer of n-doped GaN 11 li and, • a layer of AlGaN (barrier) 113i placed on the layer of GaN not intentionally doped 112i. It should be noted that there may also be a “spacer” between the channel and the barrier, in the form of a very thin layer of AIN (0.5 to 2nm) which improves the density and transport properties in the 2DEG.

[0045] For each elementary stack lOli, the unintentionally doped GaN layer 112i and the AlGaN layer 113i form a heterojunction so that a two-dimensional electron gas 114i is intrinsically formed under the interface between the unintentionally doped GaN layer and the AlGaN layer. The AlGaN layers 113i may be covered with a passivation layer not shown in [Fig.4], for example made of silicon dioxide (SiO2), silicon nitride (Si3N4), aluminum nitride (AIN), gallium nitride (GaN) or alumina (A12O3). This passivation layer serves to properly control the surface of the AlGaN layer.

[0046] The p-doped GaN layer 110i and the n-doped GaN layer 11li together form a particularly high potential barrier (depleted PN junction) under the electron gas layer 114i.

[0047] Advantageously, the p doping of the layer 11 li is carried out with Magnesium and the n doping of the layer 112i is carried out with Silicon.

[0048] An oxygen-implanted zone is located on either side of each active stack 101. This may be either a part of the peripheral zone 106 or an elementary zone 105j. Thus, according to the invention, two lateral oxygen-implanted zones are arranged on either side of each p-type doped GaN layer 101. In the embodiment illustrated in FIGS. 3 and 4, two lateral zones, 106 or 105j, implanted with oxygen are arranged on either side of the entire stack 101. The interest of these lateral oxygen-implanted zones will be detailed later.

[0049] According to the invention, the active zone 101 and the oxygen-implanted zone 105 are produced from a single stack of layers. A method 200 for producing a component comprising the active zone 101 and the oxygen-implanted zone 105 including the peripheral zone and the plurality of elementary implanted zones is illustrated in FIGS. 6 to 8 as well as in [Fig. 5] representing the different steps of this method 200.

[0050] The first step 201 consists of depositing on a substrate 307 the desired stack, namely in the example of [Fig.6], a stack including: • a layer of p-doped GaN 310, • a layer of n-doped GaN 311 placed on the layer of p-doped GaN, • a layer of GaN not intentionally doped 312 placed on the layer n-doped GaN, • a layer of AlGaN 313 placed on the layer of GaN not intentionally doped, • a 314 passivation layer.

[0051] This stack is obtained by epitaxy techniques known to those skilled in the art, for example by organometallic vapor phase epitaxy MOCVD (Metalorganic Vapor Phase Epitaxy according to the English terminology). It will be noted that the transition and buffer layers have been deliberately omitted in Figures 6 to 8.

[0052] The method 200 then comprises a step 202 of masking the areas intended to form the elementary active areas of the component. A mask 315 made of photosensitive resin is thus shown in [Fig.7]. This masking step is carried out by photolithography techniques well known to those skilled in the art. The mask 315 is used here, for example, to delimit one of the elementary active areas 10li.

[0053] The method 200 then comprises a step 203 of ion implantation of oxygen atoms in the zones not protected by the resin, said implantation being intended to form the elementary zones implanted with oxygen and the peripheral isolation zone implanted with oxygen. The implantation step is followed by a step of removing the resin, for example by pickling or stripping according to the English terminology. The step of removing the resin is followed by a step of thermal annealing at a temperature of approximately 800°C (between 700°C and 850°C) for 1 to 30 minutes. The time required for the lateral diffusion of hydrogen depends on the temperature and the size W of the active zones.

[0054] The method 200 according to the invention thus makes it possible to produce the entire oxygen-implanted zone 105 shown in FIGS. 3 and 4 including the peripheral insulation zone 106 and the plurality of elementary oxygen-implanted zones 105j arranged between each elementary active zone 101i.

[0055] The oxygen-implanted areas must therefore be seen as areas that have undergone oxygen implantation and thermal annealing, thus creating pockets implanted both around the component to isolate it but also within the component to improve the activation of the acceptor (here Mg) of each layer 1 lOi of the confinement barrier. Indeed, there are two lateral areas 105i which are found both on either side of each layer of p-doped GaN 1 lOi but also on either side of the 2DEG gas 114i. The implantation of oxygen leads to the formation of deep levels (associated with implantation defects) in the gap of the GaN semiconductor which completely neutralize the 2DEG at the location of the implantation. In this area, the transistor is therefore not active, which ensures isolation. Post-implantation annealing also allows Hydrogen, which passivates the Mg dopant, to diffuse and react with the implanted Oxygen. The diffusion distance Since the Hydrogen supply is limited, it is necessary to split the Oxygen pockets in the active zone ([Fig.3] and 4). Each elementary Oxygen zone 105i of width Wo will interact with the active parts lOli of width Wa which are close to it.

[0056] Advantageously, the width of the elementary active zones Wa and the elementary implanted Oxygen zones Wo are at most 20 pm to allow an effective interaction between the Oxygen atoms and the Hydrogen atoms. They are at least I pm in order to avoid the complete neutralization of the 2DEG by the lateral dispersion of the Oxygen implantation in the active zones and to respect the masking constraints of the implantation step. In this window between I pm and 20 pm, the choice of Wo and the active zone ratio Wa / (Wa+Wo) is free and is dictated by the performance objective of the transistor in terms of frequency, thermal resistance and access resistances.

[0057] The oxygen implantation step must be carried out so that the oxygen is present in the depth of the stack and in particular at the level of the 2DEG gas and the p-doped GaN layer. To do this, several implantation energies will advantageously be used which make it possible both to neutralize the 2DEG (isolation) and to interact with the buried Mg doped layer. It is also advantageous to use a multi-energy implantation to distribute the Oxygen dose in depth. The thickness of the Oxygen implanted zone is noted t. It must be greater than or equal to the depth of the p-doped GaN (Mg) doped layer. The latter being of the order of 150nm, a maximum energy of 150-200keV is sufficient and achievable with an implantation frame conventionally used in a CMOS type clean room.It is known to the person skilled in the art to choose a sequence of energies and doses that allows obtaining an approximately flat Oxygen profile at the target oxygen concentration. An example of such a profile is given in [Fig.12]. This Oxygen implantation profile is obtained by triple implantation (40keV with an implantation dose of 5.1014cm-2, 80keV with an implantation dose of 1015cm-2 and 170 keV with an implantation dose of 3.1015cm-2). The Oxygen concentration is high (i.e. greater than 1020cm-3) around the 1 lOi layer of Mg-doped p GaN. The defects generated by the three implantations are sufficient to neutralize the 2DEG gas at the 114i interface between the 112i layer of AlGaN and the unintentionally doped 113i layer of GaN. Regarding the Oxygen implantation dose, . <t>Oxygenwhich can be considered as being approximately equal to [O] xt, where [O] is the oxygen concentration and t is the thickness of the zone implanted in Oxygen, it is: • sufficient to capture the Hydrogen in the p-doped GaN layer; typically, we will seek to have a dose 0Oxygen> 2.1014cm2x (Wa / Wo). This therefore corresponds for example to a concentration [O]> 1.1019cm3 if t = 200nm and Wa=Wo; • limited, in order to avoid substrate deformations and to facilitate the integration of this step into the transistor manufacturing steps; typically, we will seek to have a dose 3>Oxygen< 6e15cm2x (Wa / Wo). This corresponds to a concentration [O]< 3.1020cm3 if t=200nm and Wa=Wo. Such a dose range is also sufficient to isolate the transistor at the 2DEG gas level.

[0058] As mentioned previously, the oxygen-implanted zone is not an oxygen-doped zone because in this case the majority of the oxygen present would be in a substitutional site in the crystal lattice of the GaN and the oxygen would then not be available to capture the hydrogen preventing p-doping in the GaN by the creation of Mg-H complexes. Thus, the oxygen-implanted zone must be seen as a zone having undergone an oxygen atom implantation step followed by annealing but not having been subjected to a sufficient thermal budget for the oxygen to behave as a donor and form an n-doped GaN zone. However, in order to further prevent oxygen from playing a role as a dopant (i.e. donor) in the implanted GaN, the oxygen implantation step 203 may be a co-implantation step during which both oxygen and Argon Ar or nitrogen N are implanted.This ensures that no residual n-type doping appears and degrades the high resistivity of the implanted area.

[0059] In addition to the role of activator of the p-doping in the GaN layer of the confinement barrier and insulator of the 2DEG gas of the transistor, the use of the lateral zones 105j implanted in oxygen makes it possible to split (see top view in [Fig.3]) the active power zones 10li. The active zones of a power transistor are in fact localized heat sources whose extension is Wg x L, where Wg is the development of the gate and L is the length of the heat source (which can be greater than Lg) measured along the Ox axis, L being for example equal to 0.5 pm. Wg corresponds to the gate width, measured along the Oy axis opposite the active zone. By splitting the active zone of the transistor into unit heat sources Wa, spaced by Wo, the thermal coupling between these sources is limited. The thermal resistance of the transistor is thus reduced.

[0060] [Fig.9] represents a variant of the transistor according to the invention represented in [Fig.3]. As in the case of [Fig.3], [Fig.9] represents a top view of a HEMT transistor 400 according to the invention in an Oxy plane. Like transistor 100, transistor 400 comprises: • An active zone 401 including a plurality of elementary active zones 401i; • A drain zone 402; • A 403 source zone; • A 404 grid area; • An area implanted with oxygen 405 including: • a peripheral isolation zone 406; • a plurality of elementary zones implanted in oxygen 405j arranged between each elementary active zone 401i.

[0061] The active, drain, source and gate areas as well as the peripheral isolation area are identical to the corresponding areas of [Fig.3].

[0062] On the other hand, this variant of the invention consists of implanting the oxygen only in the gate area but not up to the source and drain areas. In other words, the elementary oxygen implanted areas 405j are arranged to be under the gate area but not to be present at the drain and source areas. The manufacturing method of the oxygen implanted areas is identical to that of [Fig. 5], the difference residing in the shape of the resin pattern intended to cover the non-implanted areas. According to the variant of [Fig. 9], the width Wc of the elementary source and drain access areas is greater than in the case of the transistor of [Fig. 3] where the width of the source and drain access areas were equal to the width of the active area Wa. In addition, the access resistances (source and drain) are lower due to the absence of the highly resistive implanted areas on the drain and source side.Such an embodiment therefore increases the performance of the transistor. The shape of the Oxygen implanted zone must be sufficiently large around the gate (Wo) and gradually reduce towards the source and drain accesses. Of course, the surface of each elementary zone implanted in oxygen must be large enough to ensure a sufficient quantity of oxygen and allow the activation of the p-type dopants of the GaN layer at the gate level: in fact, the confinement barrier must be especially effective at the gate level (hence the importance of observing the phenomenon of hydrogen neutralization near the gate, at the location where the 2DEG electrons have a maximum energy).

[0063] In a known manner, the source and drain zones are either in direct contact with the AlGaN layer or buried ("recess" according to the English terminology) in the AlGaN layer and partially in the unintentionally doped GaN layer. Similarly, the gate zone may be a Schottky type gate (metal-semiconductor junction), a gate slightly buried in the stack of semiconductor layers or a MIS type gate (Metal Insulating Semiconductor). Thus, the elementary lateral zones implanted in oxygen may be under the gate zone (case of a Schottky or MIS contact) or be directly in contact with the gate zone (case of a gate buried in the elementary lateral zones implanted in oxygen). Similarly, according to the embodiment of [Fig. 3], the elementary lateral zones implanted in oxygen can be under the source and drain areas (case of direct contact with the AlGaN layer) or be directly in contact with the source and drain areas (case of a drain or source area buried in the AlGaN layer and partially in the unintentionally doped GaN layer).

[0064] The presence of the oxygen-implanted zones depending on whether the component according to the invention is that of [Fig.3] or that of [Fig.9] is respectively illustrated by Figures 10a and 10b (embodiment of [Fig.3]) and by Figures 11a and 11b (embodiment of [Fig.9]).

[0065] With reference to [Fig. 3], [Fig. 10a] shows a section of the device 100 along a plane P2 parallel to the plane Oxz and passing through an active zone lOli. [Fig. 10b] shows a section of the device 100 along a plane P3 parallel to the plane Oxz and passing through an elementary zone implanted in oxygen 105j. It will be noted that the source and drain zones 103 and 102 are here buried while the gate zone is in contact with the AlGaN layer. It can be seen here that the elementary zones implanted in oxygen 105j (see [Fig. 10b]) are at the level of the gate zone 104 and that they extend to the level of the source 103 and drain 102 zones. Figures 10a and 10b show the peripheral zone 106 implanted in oxygen isolating the component on its contour.

[0066] With reference to [Fig. 9], [Fig. 11a] shows a section of the device 400 along a plane P2 parallel to the plane Oxz and passing through an active zone 401i. [Fig. 10b] shows a section of the device 400 along a plane P3 parallel to the plane Oxz and passing through an elementary zone implanted in oxygen 405j. It can be seen here that the elementary zones implanted in oxygen 405j (see [Fig. 11b]) are only at the level of the gate zone 404 and that they do not extend to the level of the source 403 and drain 402 zones. Figures 11a and 11b show the peripheral zone 406 implanted in oxygen isolating the component on its contour.

[0067] Of course, the invention is not limited to the embodiments which have just been described. Thus, even if the embodiments described relate to HEMT transistors, the invention also applies to other electronic components, in particular optoelectronic components, provided that the latter comprise at least one p-doped GaN layer covered with another semiconductor layer preventing the exo-diffusion of hydrogen.< / t>

Claims

Claims

1. Electronic component (100, 400) comprising: - A substrate (107) - A plurality of active stacks (101i), each active stack (101i) being formed above the substrate and including: • a layer (111i) of p-type doped GaN Galium Nitride, arranged above the substrate (107); • a layer (111i) of a semiconductor material arranged on the layer (111i) of p-type doped GaN Galium Nitride; characterized in that the component comprises two lateral zones (105i) located on either side of the respective p-type doped GaN layer (111i) of each active stack (101i), the two lateral zones being implanted with oxygen.

2. Electronic component (100, 400) according to the preceding claim characterized in that the acceptors of the p-type doped Galium Nitride GaN layer are Magnesium atoms.

3. Electronic component (100, 400) according to the preceding claim, characterized in that the width (Wa) of each active stack and the width (Wo) of each of the oxygen-implanted zones are less than or equal to 20 pm and greater than or equal to 1 pm, the width being measured parallel to the plane of the layers in alignment with each active stack and the oxygen-implanted zones.

4. Electronic component (100, 400) according to one of the preceding claims, characterized in that each active stack (101i) comprises: - a layer of n-type doped GaN (111i) arranged on the layer of p-type doped GaN; - a layer of unintentionally doped GaN (112i) arranged on the layer of n-type doped GaN; - a semiconductor layer (113i) arranged on the layer of unintentionally doped GaN to form a two-dimensional electron gas; the component further comprising a source zone (103), ... drain (102) and a control gate region (104) formed on or in the semiconductor layer to form a two-dimensional electron gas; the two oxygen-implanted side regions also extending across the n-type doped GaN layer, the unintentionally doped GaN layer, and the semiconductor layer to form a two-dimensional electron gas.

5. Electronic component according to claim 4 characterized in that - each active stack comprises: • a layer of n-type doped GaN arranged on the layer of p-type doped GaN; • a layer of unintentionally doped GaN arranged on the layer of n-type doped GaN; • a semiconductor layer arranged on the layer of unintentionally doped GaN to form a two-dimensional electron gas; • two lateral zones implanted with oxygen extending on either side of the layer of p-type doped GaN, the layer of n-type doped GaN, the layer of unintentionally doped GaN and the semiconductor layer to form a two-dimensional electron gas - the source, drain and control gate zones being common to the plurality of active stacks; each active stack being electrically isolated from the adjacent active stack by one of the lateral zones implanted with oxygen.

6. Electronic component according to the preceding claim, characterized in that the plurality of said stacks is surrounded by a zone, called an isolation zone, said isolation zone being an oxygen-implanted zone.

7. Electronic component according to one of claims 4 to 6 characterized in that each of the two lateral zones implanted with oxygen extends in or under the gate zone and in or under the drain zone and the source zone.

8. Electronic component according to one of claims 4 to 6 characterized in that each of the two lateral zones implanted with oxygen extends into or below the grid area and does not extend into or below the drain area and source area.

9. Method (200) for manufacturing a component according to one of the preceding claims, characterized in that it comprises the following steps: - Production (201) on a substrate (307) of a plurality of active stacks formed above the substrate, each active stack including: • a layer (310) of p-type doped Galium Nitride GaN, arranged above the substrate (307); • a layer (311) of a semiconductor material arranged on the p-type doped Galium Nitride GaN layer; - Production (202) of a mask (315) above the plurality of stacks, said mask masking each active stack; - Ion implantation (203) to introduce oxygen atoms into the two zones of each active stack located on either side of the masked zone so as to obtain two lateral zones in the p-type doped Galium Nitride GaN layer implanted with oxygen;- Annealing (202) following ion implantation.;

10. Method according to the preceding claim, characterized in that the annealing temperature is between 700 and 850°C.

11. Method according to one of claims 9 or 10 characterized in that the implantation is a multi-energy implantation.

12. Method according to one of claims 9 to 11 characterized in that the energy of the ionized oxygen atoms used for the implantation is less than or equal to 200 keV.

13. Method according to one of claims 9 to 12 characterized in that the dose of implanted oxygen atoms is: - greater than 2.1014cm2x (Wa / Wo) where Wa denotes the width of the active stack and Wo denotes the width of the implanted lateral zones and; - less than 6.1015cm2 x (Wa / Wo), the width being measured parallel to the plane of the layers in alignment with each active stack and the oxygen implanted areas.

14. Method according to one of claims 9 to 13 characterized in that the ion implantation step for introducing oxygen atoms is accompanied by an ion implantation step for introducing argon or nitrogen atoms.

Images