JP7458904B2 - Nitride Semiconductor Substrate - Google Patents

Description

The present invention particularly relates to the structure of a nitride semiconductor substrate doped with iron (Fe).

In HEMTs using nitride semiconductors suitable for high-output high-frequency amplifiers, in order to suppress off-leakage current flowing between the source electrode and drain electrode through the buffer layer during pinch-off, or to impurate donor impurities in the buffer layer, A technique is known in which a buffer layer is doped with Fe and, if necessary, with carbon (C) for the purpose of activation and shortening the recovery time.

Patent Document 1 discloses a substrate, a buffer layer formed by adding Fe and C to GaN on the substrate, and a channel layer through which electrons travel, formed of GaN on the buffer layer. , an electron supply layer formed on the channel layer for forming a two-dimensional electron gas in the channel layer, a gate electrode formed on the electron supply layer, and an electron supply layer formed on the electron supply layer. a drain electrode formed and a source electrode formed on the electron supply layer, the channel layer has a layer thickness of 0.5 μm or more, and the buffer layer has a layer thickness of 1 to 1.5 μm. A semiconductor device is described in which the total amount of Fe and C added is 1×10 ¹⁶ cm ⁻³ or more and 1×10 ¹⁸ cm ^{−3 or} less.

Patent Document 2 describes an impurity element that provides high resistance to a semiconductor material formed on a substrate, as a field effect transistor that suppresses Fe from entering the electron transport layer and suppresses the occurrence of cracks in the semiconductor layer. A high resistance layer doped with , a multilayer intermediate layer formed on the high resistance layer, an electron transport layer formed of a semiconductor material on the multilayer intermediate layer, and a semiconductor material formed on the electron transport layer. The multilayer intermediate layer is formed of a multilayer film in which GaN layers and AlN layers are alternately laminated, and is formed between the high resistance layer and the electron transport layer. A large amount of Fe is incorporated into the region where the intermediate layer made of AlN is formed, and the Fe concentration has a peak, so the amount of Fe that has entered the electron transit layer is smaller than that of the semiconductor stack without AlN. There is an invention disclosed that the amount of Fe entering into the electron transport layer can be reduced compared to that of a film.

Patent No. 6419418 JP2013-74211A

Both Patent Documents 1 and 2 state that it is preferable to move the peak of Fe concentration away from the interface between the electron transit layer and the buffer layer, that is, to move the region with high Fe concentration away. However, it is difficult to say that such a technical idea alone has been able to provide a nitride semiconductor substrate suitable for HEMT that has excellent other characteristics.

In view of the above, the present invention aims to provide an Fe-doped nitride semiconductor substrate that can improve multiple characteristics.

A nitride semiconductor substrate according to the present invention includes a substrate, a buffer layer made of a nitride semiconductor on the substrate, and an active layer made of a nitride semiconductor layer on the buffer layer, and the buffer layer includes Fe. The concentration of Fe gradually increases uniformly from the interface between the substrate and the buffer layer in the thickness direction of the buffer layer, and is 2×10 ¹⁷ atoms/cm ³ or more and 1.1×10 ²⁰ atoms/cm 3 or more. After taking a maximum value in a range of cm ³ or less, a concentration profile is formed that gradually decreases uniformly to the interface between the buffer layer and the active layer, and the position where the maximum value is taken is determined by the thickness of the buffer layer. It is characterized in that it is within a range of ±50 nm from the midpoint in the direction, and is 1875 nm or more away from the interface between the buffer layer and the active layer.

By having such a configuration, it is possible to obtain a nitride semiconductor substrate suitable for a HEMT that has a plurality of excellent characteristics.

The present invention makes it possible to effectively improve, with a simple structure, a number of characteristics that were previously considered difficult to achieve in a nitride semiconductor substrate using Fe as a dopant.

A schematic cross-sectional diagram showing a layer structure of a nitride semiconductor according to one embodiment of the present invention FIG. 2 is a schematic diagram showing an Fe concentration profile of a nitride semiconductor according to one embodiment of the present invention.

The present invention will be described in detail below with reference to the drawings. The present invention is a nitride semiconductor substrate comprising a substrate, a buffer layer made of a nitride semiconductor on the substrate, and an operating layer made of a nitride semiconductor layer on the buffer layer, the buffer layer containing Fe, the Fe concentration gradually increases uniformly from an interface between the substrate and the buffer layer in a thickness direction of the buffer layer, reaches a maximum value in a range of 2×10 ¹⁷ atoms/cm ³ to 1.1×10 ²⁰ atoms/cm ³ , and then forms a concentration profile gradually decreasing uniformly to the interface between the buffer layer and the operating layer, the position of the maximum value being within a range of ±50 nm from a midpoint in the thickness direction of the buffer layer and being 500 nm or more away from the interface between the buffer layer and the operating layer.

FIG. 1 is a schematic cross-sectional view showing a layer structure of a nitride semiconductor according to one embodiment of the present invention. That is, the nitride semiconductor substrate W has a structure in which a buffer layer B and an active layer G are sequentially stacked on one main surface of a substrate S serving as a base.

The schematic diagrams shown in this invention are schematic, simplified and emphasized shapes for the purpose of explanation, and the detailed shapes, dimensions and ratios differ from the actual ones. In addition, the same components are not designated by reference numerals, and other components that are not necessary for the explanation are not described.

In addition to silicon (Si), the substrate S may be made of silicon carbide (SiC), sapphire (Al ₂ O ₃ ), aluminum nitride (AlN), GaN, or the like. Furthermore, it may be made of a single material or made of different materials, and the structure of the surface orientation, dopant concentration, off angle, etc. can be set arbitrarily.

A nitride semiconductor is made of a combination of at least one group 13 element such as Ga, Al, and indium (In) and nitrogen (N). If necessary, various elements such as O, Si, and magnesium (Mg) may be doped.

The buffer layer B has a structure in which a plurality of nitride semiconductor layers are laminated, and depending on the use and purpose, a known method can be applied to the structure, but an appropriate initial layer is first formed on the substrate S. Therefore, a configuration in which nitride semiconductor layers having different compositions and impurity concentrations are stacked is more suitable.

In FIG. 1, the operating layer G includes a first layer 1 and a second layer 2 having a larger band gap than the first layer 1, but is not limited to this, and may include a second layer as necessary. A layer having a band gap larger than 2 or a layer containing a p-type conductive impurity at a high concentration may be added as appropriate.Furthermore, the layer thickness and impurity concentration of each layer described above may be appropriately designed depending on the purpose.

In a specific embodiment of the present invention, the substrate S is a Si single crystal substrate, and the buffer layer B is a structure in which a plurality of nitride layers represented by Al _x Ga _1-x N (0≦x≦1) are laminated. In the active layer G, the first layer 1 is made of GaN and the second layer 2 is made of Al _y Ga _1-y N (0<y<1).

The nitride semiconductor substrate W is not limited to the HEMT shown in FIG. 1, as long as the buffer layer B and the active layer G are formed on the substrate S, and can be used for other power devices that can achieve high frequency and high breakdown voltage. It is also suitably used for devices.

Further, the buffer layer B contains Fe. In the present invention, the primary purpose of doping with Fe is similar to the known technique described above, and is to increase the resistance of the buffer layer made of nitride semiconductor in order to reduce leakage current. I can say that.

The nitride semiconductor layer is preferably formed by heteroepitaxial growth on a heterogeneous substrate in the furnace of a vapor phase growth apparatus. Dislocations that occur during this process, or impurities that unintentionally mix with the nitride semiconductor from the components of the furnace, act as n-type dopants, causing so-called leakage current.

To suppress this leakage current, a method using the compensation effect of a p-type dopant has long been known. However, if the p-type dopant has a shallow level, it acts as a trap level and reduces the operating current. For this reason, a deeper dopant level is preferable, and Fe has been used as a suitable p-type dopant with such a deep level.

However, since nitride semiconductors doped with Fe tend to grow three-dimensionally, they cause unevenness at the film-forming interface, which may cause current leakage. As an example of a method for reducing this leakage current, a buffer layer without Fe is formed between the resistance layer to which Fe is added and AlN as the initial buffer layer, and the flatness of the interface is maintained by thermal diffusion during film formation. There is also a method of trying to increase the resistance of the buffer layer while keeping the buffer layer high.

However, in the above-mentioned method, while the electron transit layer and electron supply layer as the active layer G are subsequently formed on the Fe-doped buffer layer, Fe in the buffer layer is removed by thermal diffusion. Since it moves to the electron supply layer, it causes a decrease in electron mobility. Furthermore, unevenness occurs at the interface with the electron transport layer, which also causes a decrease in electron mobility.

In other words, the present invention lowers the Fe concentration near both the interface between the buffer layer B and the substrate S, and the interface between the operating layer G (electron transport layer as an example here) and the buffer layer B. , is characterized in that the peak position of the Fe concentration that achieves high resistance is kept as far away from both interfaces as possible.

In this way, the flatness of each interface can be maintained due to the low Fe concentration, and thermal diffusion to the electron transport layer can be suppressed, resulting in low leakage and high-speed operation. It can be a physical semiconductor.

In addition, because there is a peak in Fe concentration that realizes high resistance, the leakage current reduction effect is also appropriately obtained. In other words, it was discovered that it is possible to simultaneously achieve two improvements in characteristics: a reduction in leakage current and suppression of mobility degradation.

As a specific aspect of obtaining the above effect, the concentration of Fe contained in the buffer layer B is uniformly gradually increased from the interface between the substrate S and the buffer layer B toward the thickness direction of the buffer layer B, After taking a maximum value in the range of 2×10 ¹⁷ atoms/cm ³ to 1.1×10 ²⁰ atoms/cm ³ , a concentration profile is formed that gradually decreases uniformly to the interface between buffer layer B and active layer G. ing.

FIG. 2 is a schematic diagram showing an Fe concentration profile of a nitride semiconductor according to one embodiment of the present invention. In other words, the Fe concentration peak is located near the center in the thickness direction of the buffer layer B. In other words, by moving the region with high Fe concentration away from either the upper or lower interface, It can be said that the effects of the present invention are obtained.

Here, if metal organic chemical vapor deposition (MOCVD) is used, the formation of the Fe concentration profile as shown in FIG. This can be done in a timely manner by adjusting the flow rate and other gas flow rates.

Note that although the Fe concentration is evaluated by measuring the thickness direction of the nitride semiconductor substrate using SIMS, it is also possible to use a method other than SIMS.

By the way, in order to satisfy the characteristics required for a nitride semiconductor substrate (dislocation density, warpage, breakdown voltage, crystallinity, etc.), the buffer layer B should have a structure in which multiple layers with different compositions are laminated instead of a single layer as a whole. The important requirements are that the layer is 100% thick and that the layer has a certain level of thickness.

First, regarding the thickness, in the present invention, the entire thickness of the buffer layer B is 1200 nm or more, preferably 1800 nm or more and 3500 nm or less. However, if it is too thick, there is a concern that it will be difficult to control the warpage of the entire substrate, so the upper limit is set to 4500 nm or less.

The structure of the buffer layer B is preferably a structure in which a plurality of layers having different compositions are laminated, since it can also obtain properties improvement effects in addition to the effects of the present invention. In other words, even if the buffer layer B is a single layer, the effects of the present invention can be obtained, but other characteristics, especially the warpage of the substrate, become excessive, making it impractical. Note that the layer structure (layer thickness, composition, number of laminations, etc.) is not particularly limited.

More specifically, the Fe concentration peak is located within a range of ±50 nm from the midpoint of the buffer layer B in the thickness direction and at least 500 nm away from the interface between the buffer layer B and the active layer G. It is meant to be.

In the present invention, the Fe concentration peak only needs to be near the center of buffer layer B, but if it exceeds the range of ±50 nm from the exact center, buffer layer B and active layer G Either the interface between the buffer layer B and the substrate S becomes relatively close to the peak of the Fe concentration, and either of the characteristics of leakage current reduction and high-speed operation become insufficient. There are concerns and it cannot be said to be favorable.

In the present invention, the peak of the Fe concentration is separated from the interface between the buffer layer B and the active layer G by 500 nm or more, preferably about 2200 nm. This is because, as in the known technology, if there is a region with a high Fe concentration near the active layer G, during the process of manufacturing the nitride semiconductor substrate W, Fe will thermally diffuse and invade the active layer G in large amounts. This is because high-speed operability deteriorates, which is not preferable.

In the present invention, it is preferable that the Fe concentration in the vicinity of both the interface between the substrate S and the buffer layer B and the interface between the buffer layer B and the active layer G is 1×10 ¹⁷ atoms/cm ³ or less. As described above, by lowering the Fe concentration in the region in contact with the buffer layer B, the adverse effects of Fe doping can be avoided.

As described above, the nitride semiconductor substrate of the present invention enables low leakage and high-speed operation. A nitride semiconductor device manufactured using this can be utilized as a high-performance semiconductor device.

In the present invention, it is more preferable that the hydrogen concentration contained in the buffer layer B gradually increases from the interface between the buffer layer B and the active layer G toward the interface between the substrate S and the buffer layer B. .

In the operating layer G and the vicinity thereof, it is preferable that the concentration of hydrogen is not too high from the viewpoint of not hindering the operation of the device. On the other hand, if the buffer layer B is doped with Fe at a high concentration, there is a concern that the crystallinity will deteriorate; It is good if the concentration is high.

In order to obtain the above effects, it is ideal to be able to set the hydrogen concentration profile shape in accordance with the Fe concentration profile shape. However, it is not easy to control the hydrogen concentration profile in nitride using the vapor phase growth method as compared to controlling dopants such as Fe, carbon, and Mg. Furthermore, it is even more difficult to control the hydrogen concentration and Fe concentration at the same time.

Therefore, in the present invention, as a method of easily forming a hydrogen concentration profile in accordance with the Fe concentration profile shape of the present invention and obtaining the above-mentioned effects, the hydrogen concentration profile is changed between the buffer layer B and the active layer G. The thickness gradually increases from the interface between the substrate S and the buffer layer B to the interface between the substrate S and the buffer layer B.

Specifically, in the active layer G, the hydrogen concentration is 5×10 ¹⁷ atoms/cm ³ or more and 2×10 ¹⁸ atoms/cm ³ or less, and in the buffer layer B, it gradually increases to 4×10 ¹⁸ atoms/cm ³ Examples include modes in which the concentration increases to 3×10 ¹⁹ atoms/cm ³ or less. Note that it is not necessary to strictly match the hydrogen concentration with the Fe concentration.

Furthermore, the hydrogen concentration may gradually increase up to the region where the Fe concentration peak exists, and then decrease toward the substrate S. While the effect obtained by increasing the hydrogen concentration in the latter half is not so great, increasing the hydrogen concentration to this region has little technical or cost advantage.

EXAMPLES Hereinafter, the present invention will be specifically explained based on Examples, but the present invention is not limited to the Examples below.

[Common growth conditions]
As the substrate S, a 6-inch Si single crystal substrate with crystal plane orientation (111), p type, was cleaned by a known substrate cleaning method, and then set in an MOCVD apparatus, and after replacing the inside of the apparatus with a carrier gas, the temperature was raised. Then, heat treatment was performed at 1000° C. for 15 minutes in a 100% hydrogen atmosphere to remove the natural oxide film on the surface of the silicon single crystal.

Next, using trimethylaluminum (TMAl), trimethylgallium (TMGa), and ammonia (NH ₃ ) as raw materials, an initial layer of 150 nm of Al _0.2 Ga _0.8 N was stacked on 100 nm of AlN, and two layers of 5 nm of AlN and 30 nm of Al _0.2 Ga _0.8 N were formed. Buffer layer B was obtained by stacking a multilayer of 80 repetitions of stacking and a single layer of GaN of 1500 nm by vapor phase growth in the above-described order. Here, the layer thickness of buffer layer B is 4550 nm.

Next, as the active layer G, the first layer 1 was made of GaN of 100 nm, the second layer 2 was made of Al _0.22 Ga _0.78 N of 20 nm, and each layer was laminated in this order. Here, the growth conditions of the buffer layer B and the operating layer G were changed as appropriate when growing each layer, with the growth temperature of 1050° C. and the growth pressure of 60 hPa as approximate reference values.

(Example 1)
In order to dope Fe, Cp ₂ Fe (dicyclopentadienyl iron, commonly known as ferrocene) was used as the Fe raw material. In the process of repeatedly laminating multiple layers, Cp ₂ Fe was introduced after repeating two layers of AlN 5 nm and Al _0.2 Ga _0.8 N 30 nm 30 times, and the introduction was stopped when the layers were repeated 80 times. The Fe concentration is 1×10 ²⁰ atoms/cm ³ at the peak position, and the distances from the peak position to the active layer G and from the peak position to the substrate S are both approximately 2275 nm. It was within the range of ±50 nm from approximately the middle of . An evaluation sample of Example 1 was prepared in the manner described above.

The above Fe concentration profile is due to the fact that Fe existing in the high concentration layer uniformly diffuses to the substrate S side and the active layer G side in the vapor phase growth after Fe introduction is stopped. It is formed. Utilizing this, a peak of Fe concentration can be formed in a predetermined region by adjusting the flow rate, introduction time, minute growth temperature, etc. when introducing Cp ₂ Fe.

(Example 2)
Evaluation sample of Example 2, in which the Cp ₂ Fe flow rate and growth temperature were adjusted so that the Fe concentration at the peak position was 2×10 ¹⁷ atoms/cm ³ , and the rest was the same as Example 1. I got it.

(Example 3)
In Example 3, the Cp ₂ Fe flow rate and growth temperature were adjusted so that the Fe concentration at the peak position was 1.1×10 ²⁰ atoms/cm ³ , and the other conditions were the same as in Example 1. An evaluation sample was obtained.

Example 4
An evaluation sample of Example 4 was obtained that was the same as Example 1 except that the flow rate of Cp ₂ Fe and the growth temperature were adjusted so that the peak position was 50 nm from the center of the buffer layer B toward the active layer G side.

(Example 5)
An evaluation sample of Example 5 was obtained by adjusting the Cp ₂ Fe flow rate and growth temperature so that the peak position was 50 nm from the center of the buffer layer B toward the substrate S side, and otherwise the same as Example 1. .

(Example 6)
An evaluation sample of Example 6 was obtained, which was the same as Example 1 except that the buffer layer B was made of GaN with a thickness of 700 nm and the layer thickness was 700 nm. At this time, the peak position of the Fe concentration was 1875 nm from the interface of the active layer G.

In all of the evaluation samples in Examples, the Fe concentration was set to approximately 1×10 ¹⁷ atoms/cm ³ both near the interface between the substrate S and the buffer layer B and near the interface between the buffer layer B and the active layer G. Growth conditions were adjusted. The Fe concentration was determined by cleaving each evaluation sample at its diameter and evaluating the cross section using SIMS.

(Comparative example 1)
In the process of repeatedly stacking multiple layers of Cp ₂ Fe, Cp ₂ Fe was introduced from the first stacking of two layers of 5 nm of AlN and 30 nm of Al _0.2 Ga _0.8 N, and the introduction was stopped after 80 repetitions. The Fe concentration did not have a peak, was flat over almost the entire buffer layer B, and was approximately 1×10 ¹⁷ atoms/cm ³ . An evaluation sample of Comparative Example 1 was prepared in the manner described above.

(Comparative example 2)
In the process of repeatedly laminating multiple layers of Cp ₂ Fe, Cp ₂ Fe is introduced from the first lamination of two layers of 5 nm of AlN and 30 nm of Al _0.2 Ga _0.8 N, repeated 80 times, and then a layer of 1500 nm of GaN is formed. The introduction was stopped when the thickness reached 1000 nm. The Fe concentration did not have a peak, was flat over almost the entire buffer layer B, and was approximately 1×10 ²⁰ atoms/cm ³ . An evaluation sample of Comparative Example 2 was prepared in the manner described above.

(Comparative Example 3)
In the process of repeatedly stacking multiple layers of Cp ₂ Fe, Cp ₂ Fe was introduced from the initial stacking of two layers of AlN 5 nm and Al _0.2 Ga _0.8 N 30 nm, and this was repeated 80 times. In the process of forming a GaN layer of 1500 nm, the introduction was stopped at the stage where the layer thickness reached 250 nm. The Fe concentration did not have a peak, was 1×10 ²⁰ atoms/cm ³ at the interface with the substrate S, and gradually decreased toward the interface with the operating layer G, reaching 1×10 ¹⁷ atoms/cm ^3. By following the above, an evaluation sample of Comparative Example 3 was produced.

(Comparative Example 4)
In the process of forming a 1500 nm GaN layer, Cp ₂ Fe was introduced when the layer thickness reached 450 nm, and the introduction was stopped when the _layer thickness reached 1500 nm. The Fe concentration did not have a peak, was 1×10 ¹⁷ atoms/cm ³ at the interface with the substrate S, and gradually increased toward the interface with the operating layer G, reaching 1×10 ²⁰ atoms/cm ^3. By following the above procedure, an evaluation sample for Comparative Example 4 was produced.

(Comparative example 5)
An evaluation sample of Comparative Example 5 was obtained by adjusting the Cp ₂ Fe flow rate and growth temperature so that the peak position was 70 nm from the center of the buffer layer B toward the active layer G side, and the other conditions were the same as in Example 1. Ta.

(Comparative example 6)
An evaluation sample of Comparative Example 6 was obtained by adjusting the Cp ₂ Fe flow rate and growth temperature so that the peak position was 70 nm from the center of the buffer layer B toward the substrate S side, and otherwise the same as Example 1. .

(Comparative Example 7)
An evaluation sample of Comparative Example 7 was obtained that was equivalent to Example 1, except that the buffer layer B was changed to a 1,500 nm thick GaN layer, the layer thickness was changed to 300 nm, and the flow rate of _Cp2Fe and the growth temperature were adjusted so that the peak position was significantly shifted from the center of the buffer layer B. That is, the peak position of the Fe concentration was 450 nm from the interface with the active layer G.

[Rating 1 - Warpage]
The bow of each evaluation sample was measured using a general-purpose warpage measurement device that is commonly used for measuring the shape of semiconductor substrates. A bow value of -50 μm or more and 20 μm or less was determined as pass (◯).

[Evaluation 2~Pressure pressure]
From each evaluation sample, a strip-shaped test piece with a width of 2 mm was cleaved and cut from the center of the main surface of the substrate to the edge of the substrate. Next, part of the electron supply layer 2 and electron transit layer 1 of this test piece was removed by dry etching. In this state, a 10 mm ² Au electrode was vacuum-deposited on the surface exposed by dry etching to form a Schottky electrode, and current was applied to the Si single crystal substrate side using a commercially available curve tracer to measure the IV characteristics. The current values at 600V were compared. A value of 1×10 ^-8 (A) or less was determined to be a pass (〇).

[Evaluation 3 ~ High-speed operation characteristics]
For each evaluation sample, Hall effect measurement was performed using the Van Der Pauw method to evaluate electron mobility. First, the evaluation sample was diced into 7 mm square chips, and Ti/Al electrodes with a diameter of 0.25 mm were formed at the four corners of the electron supply layer 2 of each chip by vacuum evaporation. Next, alloying heat treatment was performed at 600° C. for 5 minutes in a N ₂ atmosphere. Then, Hall effect measurement was performed using ACCENT's HL5500PC. The evaluation expressed the level of mobility as a ratio with Comparative Example 1. 1 or less was evaluated as ×, more than 1 and 1.1 or less was evaluated as △, 1.2 or more was evaluated as ○, and △ and 〇 were evaluated as pass.

The data and evaluation results of each evaluation sample described above are summarized in Table 1 below.

As is clear from the results in Table 1, it can be said that those within the scope of the present invention are good in all characteristics of warpage, withstand voltage, and high-speed operation.

W: nitride semiconductor substrate S: substrate (single crystal silicon)
B Buffer layer G Operating layer 1 First layer (electron transport layer)
2. Second layer (electron supply layer)

Claims (2)

comprising a substrate, a buffer layer made of a nitride semiconductor on the substrate, and an active layer made of a nitride semiconductor layer on the buffer layer,
The buffer layer contains Fe,
The concentration of Fe gradually increases uniformly from the interface between the substrate and the buffer layer in the thickness direction of the buffer layer, and is 2×10 ¹⁷ atoms/cm ³ or more and 1.1×10 ²⁰ atoms/cm ^{3 .} After taking a maximum value in the following range, a concentration profile is formed that gradually decreases uniformly to the interface between the buffer layer and the operating layer,
The position where the maximum value is obtained is within a range of ±50 nm from the midpoint in the thickness direction of the buffer layer, and is 1875 nm or more away from the interface between the buffer layer and the operating layer. Nitride semiconductor substrate. 2. The nitride semiconductor substrate according to claim 1, wherein the buffer layer has a thickness of 1200 nm or more and 4500 nm or less.

Images