Method for fabricating a non-planar nitride-based heterostructure field effect transistor

[0002] This application is related to and claims benefit of U.S. Provisional Application No. 60/411,076 filed on Sep. 16, 2002, which is incorporated by reference in its entirety.

[0003] The present invention relates to a method for fabricating a semiconductor structure useful for fabricating a non-planar heterostructure field effect transistor. More specifically, the present invention relates to a method for fabricating a semiconductor structure useful for fabricating a non-planar nitride-based heterostructure field effect transistor, wherein the non-planar region is fabricated in the group III-nitride material aluminum nitride (AlN) and the semiconductor structure is not damaged by dry etching or wet etching.

[0004] The use of group III-nitride substrates has become popular for fabricating a non-planar region in a non-planar heterostructure field effect transistor. A non-planar heterostructure field effect transistor is a field effect transistor comprising several different semiconductor layers of semiconductor material, wherein the top layer has a non-planar region. Typically a gate is then formed in the non-planar region. By forming the gate in the non-planar region, the parasitic resistance of the heterostructure field effect transistor is lowered. Furthermore, a higher breakdown voltage and transconductance, as discussed below, can be achieved. However, fabricating a non-planar heterostructure field effect transistor using group III-nitride substrates can be troublesome.

[0005] Transconductance is a measure of how the output current of the device changes with the applied voltage at the input of the device. The breakdown voltage is a threshold voltage, which, when exceeded, causes current in the gate to flow uncontrollably. This ultimately leads to the destruction of the device. The breakdown voltage is directly related to the bandgap as described above. Another benefit of having a higher breakdown voltage is improved gate modulation of the channel under a strong RF input drive, which improves power performance of the transistor.

[0006] The use of group III-nitride substrates to fabricate a non-planar region in the top layer is popular because group III-nitride substrates have much higher bandgaps than more traditional substrates such as silicon. The bandgap of a substrate refers to the degree to which it can support an applied electric field before breaking down. Thus, the applied voltage that a substrate can maintain is directly proportional to the bandgap of the substrate.

[0007] Previous attempts have been made to fabricate a non-planar heterostructure field effect transistor with a top layer comprising GaN, a group III-nitride substrate. However, using GaN has presented problems. When using a wet-etch there is no reliable or controllable method for controlling the regions in the GaN which are being etched. As a result, if the GaN layer is overetched, the layers beneath the GaN layer would be damaged by the wet etchant. There have also been attempts at fabricating a non-planar region in AlGaN where the AlGaN layer was partially wet-etched. Like GaN, using a wet-etch with AlGaN presented problems with controlling the area being etched and the depth of the etched area.

[0008] Dry etching processes have also been used in an attempt to create a non-planar region in a GaN substrate. However, dry etching introduces unrecoverable damage to the surface of the GaN substrate. Similar damage is also present when using an AlGaN substrate. The surface damage can be repaired by a post-annealing process, but removing all the surface damage is not possible. Another problem with dry-etching in GaN and AlGaN is the difficulty in controlling the etch depth. Techniques attempting to fabricate recessed gates using GaN are discussed in J. W. Burm et al., “Recessed gate GaN MODFETS,” Solid-State Electronics vol 41, pp. 247-250 (1997), and T. Egawa et al., “Recessed gate AlGaN/GaN MODFET on Sapphire grown by MOCVD,” IEDM tech Digest, pp. 401-404 (1999). These references both use dry-etching techniques to fabricate the recessed gate.

[0009] Therefore, there is a need for a method for fabricating a non-planar heterostructure field effect transistor, wherein the non-planar region is fabricated in a group III-nitride material. There is also a need for a non-planar heterostructure field effect transistor in which dry-etching and wet-etching techniques can be used to create the non-planar region which does not induce damage to the transistor and allows good control of the etching depth.

[0010] The present invention provides a transistor having a device structure that allows for the use of dry-etching and wet-etching to create a non-planar region without damaging the transistor. The present invention makes use of the group III-nitride material AlN for creating a non-planar region. AlN has not been used for this application because of the focus on GaN. Because GaN has one of the highest bandgaps of any group III-nitride material, it has been more desirable to find a compatible wet etching process that will work with GaN, than it is to attempt the process with a different group III-nitride material. However, when AlN along with the device structure of the transistor disclosed herein, is processed in conjunction with the wet-etching and dry-etching process disclosed herein, a non-planar region can be fabricated consistently and repeatedly without inducing damage to the rest of the transistor. Such results have not been attainable using GaN or other group III-nitride materials to fabricate non-planar regions in heterostructure field effect transistors.

[0011] It is an object of the present invention to provide a novel method for fabricating a non-planar nitride-based heterostructure field effect transistor. The present invention provides a substrate, whereon at least one layer of semiconductor material is deposited. A layer of AlN is deposited on the at least one layer. An active channel is created at the interface of the AlN layer and the at least one layer. Charges are induced in the channel by both spontaneous polarization and piezoelectric strain at the interface. Furthermore, the at least one layer may further consist of a plurality of layers of different semiconductor material. The interface created by the plurality of layers of semiconductor material serves as the channel of the transistor.

[0012] After depositing the AlN layer, a capping layer is preferably deposited on the AlN layer. The capping layer helps prevents oxidation from forming on the AlN layer. Ohmic metal contacts are deposited on the capping layer by metal evaporation. The ohmic metal contacts are then annealed so that they diffuse into the transistor, where they contact the channel. The ohmic metal contacts may then be used as a source and drain for the transistor.

[0013] Next, a portion of the capping layer is removed using a reactive ion etch (a dry etch) to expose a portion of the AlN layer. However, the exposed portion of the AlN layer is not removed by the dry-etch, thereby acting as an etch stop and preventing damage to the layers of semiconductor material beneath the AlN layer caused by the dry-etch. Then, by using the remaining portion of the capping layer as a mask, a portion of the AlN layer is removed with a solvent to create a non-planar region. The solvent can remove the desired portion of the AlN layer with predictable and repeatable results without reducing the performance of the transistor caused by damage to the AlN layer. Using the solvent to etch the AlN layer helps remove any surface damage on the AlN layer induced by the reaction ion etch. Also, the layers of semiconductor material beneath the AlN layer are insoluble in the solvent. As a result, the layers of semiconductor material work as a controllable etch stop for etching AlN, thereby preventing damage to the semiconductor layers beneath the AlN layer.

[0014] First Embodiment

[0015] FIG. 1ashows the substrate according to a first embodiment;

[0016] FIG. 1bshows the first layer deposited on the substrate;

[0017] FIG. 1cshows the AlN layer deposited on the first layer and the interface;

[0018] FIG. 1dshows the capping layer and photoresist deposited on the AlN layer;

[0019] FIG. 1eshows a portion of the photoresist layer removed;

[0020] FIG. 1fshows the ohmic metal contacts deposited on the capping layer;

[0021] FIG. 1gshows the ohmic contact regions;

[0022] FIG. 1hshows the second window to expose a portion of the capping layer;

[0023] FIG. 1ishows the removal of a portion of the capping layer;

[0024] FIG. 1jshows the non-planar gate region; and

[0025] FIG. 1kshows the gate deposited in the non-planar gate region.

[0026] Second Embodiment

[0027] FIG. 2ashows a substrate of the second embodiment;

[0028] FIG. 2bshows a first layer deposited on the substrate;

[0029] FIG. 2cshows a second layer deposited on the first layer, and the interface;

[0030] FIG. 2dshows the third layer deposited on the second layer;

[0031] FIG. 2eshows the AlN layer deposited on the third layer;

[0032] FIG. 2fshows the capping layer and photoresist layer deposited on the AlN layer;

[0033] FIG. 2gshows the removal of a portion of the photoresist layer;

[0034] FIG. 2hshows the ohmic metal contacts deposited on the capping layer;

[0035] FIG. 2ishows the ohmic metal regions in the capping layer, AlN layer, second layer, and third layer;

[0036] FIG. 2jshows a portion of the photoresist layer removed;

[0037] FIG. 2kshows a portion of the capping layer removed;

[0038] FIG. 2lshows the non-planar gate region; and

[0039] FIG. 2mshows the gate deposited in the non-planar gate region.

[0040] Third Embodiment

[0041] FIG. 3ashows the substrate;

[0042] FIG. 3bshows the first layer deposited on the substrate;

[0043] FIG. 3cshows the second layer deposited on the first layer;

[0044] FIG. 3dshows the AlN layer deposited on the second layer

[0045] FIG. 3eshows the capping layer and photoresist layer deposited on the AlN layer;

[0046] FIG. 3fshows a portion of the photoresist removed;

[0047] FIG. 3gshows ohmic metal contacts deposited on the capping layer;

[0048] FIG. 3hshows the ohmic metal regions in the capping layer, AlN layer, and second layer;

[0049] FIG. 3ishows a portion of the photoresist layer on the capping layer removed;

[0050] FIG. 3jshows a portion of the capping layer removed

[0051] FIG. 3kshows the non-planar gate region; and

[0052] FIG. 3lshows the gate deposited in the non-planar gate region.

[0053] The present invention will now be described more fully hereinafter with reference to the accompanying drawings in which preferred embodiments of the invention are shown. This invention may be embodied in many different forms and should not be construed as limited to the embodiments set forth herein.

[0054] First Embodiment

[0055] A method useful for fabricating a non-planar nitride-based heterostructure field effect transistor according to a first embodiment of the present invention is described with reference to FIGS. 1a-1k. In this first embodiment, a substrate 102 is provided as shown in FIG. 1a. The substrate 102 preferably comprises sapphire, silicon carbide, or GaN. Next, a first layer 104 is provided as shown in FIG. 1b. The first layer 104 is deposited, preferably epitaxially, on the substrate 102. The first layer 104 preferably comprises GaN, however other materials such as InP or InGaN can be used as well. Next, an AlN layer 108 is provided as shown in FIG. 1c. The AlN layer 108 is preferably deposited epitaxially on the first layer 104. The AlN layer 108 has a thickness of preferably not more than 10 nm. When the AlN layer 108 is deposited on the first layer 104, an interface 106 is created as shown in FIG. 1c. The interface 106 serves as the channel of the transistor, which will be discussed later. After the AlN layer 108 is deposited, a capping layer 109 is preferably deposited on the AlN layer 108 as shown in FIG. 1d, followed by a layer of photoresist 110. The capping layer 109 preferably comprises GaN and helps prevent oxidation from forming on the AlN layer 108 during subsequent processing steps.

[0056] Next, a portion of the photoresist layer 110 is patterned and removed using techniques well-known in the art, to create first windows 112, which expose part of the surface of the capping layer 109 as shown in FIG. 1e. Ohmic metal contacts 118 are deposited in the first windows 112 on the surface of the capping layer 109 using metal evaporation as shown in FIG. 1f. The ohmic metal contacts 118 can be comprised of a combination of Ti/Al or Ta/Ti/Al, which are deposited in that order and are about 320 nm thick. The ohmic metal contacts 118 are then annealed at a temperature in the range of about 600-800° C. for about a minute. This allows the ohmic metal contacts 118 to diffuse into the capping layer 109 and AlN layer 108, thereby creating an ohmic contact region 119 as shown in FIG. 1g. The ohmic metal contact region 119 can then be used as a source and a drain.

[0057] Next, the remaining portion of the photoresist layer 110 on the AlN layer 108 is pattern to create a second window 122 as show in FIG. 1h. The second window 122 exposes part of the capping layer 109. The exposed portion of the capping layer 109 is etched away using a reactive ion etch preferably with chlorine gas at an etch rate of about 72 nm/min. Etching away a portion of the capping layer 109 exposes a portion of the surface of the AlN layer 108 as shown in FIG. 1i, however the reactive ion etching does not remove any portion of the AlN layer 108. The AlN layer 108 effectively acts as an etch stop, thereby preventing the reactive ion etch from damaging the first layer 104 beneath the AlN layer 108. It is possible though, for the exposed portion of the surface of the AlN layer 108 to incur damage caused by the reactive ion etch. Also, the remaining portion of the photoresist layer 110 is removed using techniques known in the art. Next, using the remaining portion of the capping layer 109 as a mask, the exposed portion of the AlN layer 108 is etched away at room temperature with a solvent to create a non-planar gate region 124 as shown in FIG. 1j. The solvent preferably comprises potassium hydroxide (KOH), water, and potassium borates, and is sold under the trade name AZ-400 by the Clariant Corporation of Somerville, N.J. Etching the AlN layer 108 helps removes any surface damage on the AlN layer 108 caused by the reactive ion etching of the capping layer 109. AZ-400 has an etch rate of approximately 100 Å/min.

[0058] Finally, a gate 126 is deposited in the non-planar gate region 124 as shown in FIG. 1k. The gate 126 is preferably T-shaped to help reduce intrinsic resistance. Fabricating a T-shaped structure is a technique well known in the art.

[0059] Because the AlN layer 108 and first layer 104 are comprised of group III-V materials, the interface 106 between the AlN layer 108 and first layer 104 already contains carrier charges due to the well-known effects of spontaneous polarization. In this way, the interface 106 acts as a channel for the transistor without requiring any additional doping. However, additional doping can be provided, if desired. When the transistor is biased with a voltage at the gate 126, and at either of the ohmic contact regions 119, the carrier charges at the interface 106 flow between the ohmic contact regions 119 allowing operation of the non-planar heterostructure field effect transistor.

[0060] Second Embodiment

[0061] A method for fabricating a non-planar heterostructure field effect transistor according to a second embodiment of the present invention is described with reference to FIGS. 2a-2m. In this embodiment a substrate 202 as shown in FIG. 2ais provided. The substrate 202 preferably comprises sapphire, silicon carbide, or GaN. A first layer 204 is deposited, preferably epitaxially, on the substrate 202 as shown in FIG. 2b. The first layer 204 preferably comprises GaN, however other materials such as InN or InGaN can be used as well. A second layer 206 is deposited, preferably epitaxially, on the first layer 204. The second layer 206, as shown in FIG. 2c, preferably comprises AlGaN. By depositing the second layer 206 on top of the first layer 204, an interface 208 is created. The interface 208 is located where the first layer 204 contacts the second layer 206 and is further discussed later. A third layer 210 is deposited, preferably epitaxially, on the surface of the second layer 206 as shown in FIG. 2d. This third layer 210 preferably comprises GaN. The purpose of the third layer 210 will be discussed later. After the third layer 210 is deposited, an AlN layer 212 is deposited preferably epitaxially as shown in FIG. 2e. The AlN layer 212 is preferably no greater than 10 nm thick. Finally, a capping layer 213 as shown in FIG. 2fis preferably deposited on the AlN layer 212, followed by a photoresist layer 214. The purpose of the capping layer 213 is to prevent oxidation from forming on the surface of the AlN layer 212 during subsequent processing steps.

[0062] After depositing the photoresist layer 214, a portion of the photoresist layer 214 is patterned, and removed using techniques known in the art to create first windows 216, which expose part of the surface of the capping layer 213 as shown in FIG. 2g.

[0063] Next, ohmic metal contacts 222 are deposited in the first windows 216 as shown in FIG. 2h. The ohmic metal contacts 222 can be comprised of a combination of Ti/Al or Ta/Ti/Al, which are deposited in that order and are about 320 nm thick. The ohmic metal contacts 222 are annealed at a temperature in the range of about 600-800° C. for about a minute. This allows the ohmic metal contacts 222 to diffuse into the capping layer 213, the AlN layer 212, the third layer 210, and the second layer 206, creating an ohmic contact region 223 as shown in FIG. 2i. The ohmic contact region 223 can then be used as a source and a drain.

[0064] Next, a portion of the remaining photoresist layer 214 is patterned and removed using techniques well-known in the art, creating a second window 226, as shown in FIG. 2j. The second window 226 exposes part of the capping layer 213. The exposed portion of the capping layer 213 is etched away using a reactive ion etch preferably with chlorine gas at an etch rate of about 72 nm/min. Etching away a portion of the capping layer 213 exposes a portion of the surface of the AlN layer 212 as shown in FIG. 2k, however the reactive ion etching does not remove any portion of the AlN layer 212. The AlN layer 212 effectively acts as an etch stop, thereby preventing the reactive ion etch from removing portions of the first, second, or third layers 204, 206, 210 beneath the AlN layer 212. It is possible though, for the exposed portion of the surface of the AlN layer 212 to incur damage caused by the reactive ion etch. Also, the remaining portion of the photoresist layer 214 is removed using techniques known in the art. Next, using the remaining portion of the capping layer 213 as a mask, a portion of the exposed surface of the AlN layer 212 is removed with a solvent to create a non-planar gate region 228 as shown FIG. 21. The solvent preferably comprises potassium hydroxide (KOH), water, and potassium borates, and is sold under the trade name AZ-400 by the Clariant Corporation of Somerville, N.J. Etching the AlN layer 212 helps removes any surface damage on the AlN layer 212 caused by the reactive ion etching of the capping layer 213. As shown in FIG. 21, the non-planar gate region 228 is created by completely etching away a portion of the AlN layer 212. However, this embodiment would work if the AlN layer 212 was not completely etched away as shown in FIG. 1jof the first embodiment. In the first embodiment, the AlN layer 108 was not completely etched away. However, should a portion of the AlN layer 212 be completely etched through to the third layer 210 as shown in FIG. 21, the substrate layers located beneath the AlN layer 212 will not be affected by the AZ-400 solution. In this second embodiment, the third layer 210 preferably comprises GaN. GaN is insoluble in AZ-400 and effectively acts as an etch stop, preventing the AZ-400 from damaging the layers located under the AlN layer 212.

[0065] A gate 230 is then deposited in the non-planar gate region 228 as shown in FIG. 2m. The gate 230 is preferably T-shaped to help reduce intrinsic resistance and capacitance. Fabricating a T-shaped structure is a technique well-known in the art.

[0066] Because the first layer 204 and the second layer 206 are comprised of group III-V materials, the interface 208 between the first layer 204 and the second layer 206 already contains carrier charges due to the well-known effects of spontaneous polarization. In this way, the interface 208 acts as a channel for the transistor without requiring any additional doping. However, additional doping can be provided, if desired. When the transistor is biased with a voltage at the gate 230 and at either of the ohmic contact regions 223, the charges at the interface 208 flow between the ohmic contact regions 223, allowing operation of the non-planar heterostructure field effect transistor.

[0067] Third Embodiment

[0068] A method for fabricating a non-planar heterostructure field effect transistor according to a third embodiment is described with reference to FIGS. 3a-3l. In this embodiment, a substrate 302 is provided as shown in FIG. 3a. The substrate 302 preferably comprises sapphire, silicon carbide, or GaN. Next, a first layer 304 preferably comprising GaN is provided as shown in FIG. 3b, however, other materials such as InN or InGaN could work equally as well. The first layer 304 is deposited, preferably epitaxially, on the substrate 302. A second layer 306 is provided as shown in FIG. 3c. The second layer 306 preferably comprises AlGaN and is deposited, preferably epitaxially, on the first layer 304. By depositing the second layer 306 on the first layer 304, an interface 305 is created. The interface 305 is located where the first layer 304 contacts the second layer 306 and is further discussed later. Next, an AlN layer 308 is deposited, preferably epitaxially on the second layer 306 as shown in FIG. 3d. The AlN layer 308 is preferably no greater than 10 nm thick. Finally, a capping layer 309 preferably comprising GaN, is preferably deposited on the AlN layer 308, followed by a photoresist layer 310 as shown in FIG. 3e. The purpose of the capping layer 309 is to prevent oxidation from forming on the surface of the AlN layer 308 during subsequent processing steps.

[0069] Next, the photoresist layer 310 is patterned and removed using techniques well known in the art to create first windows 312, which expose part of the surface of the capping layer 309 as shown in FIG. 3f. Ohmic metal contacts 318 are deposited in the first windows 312 using metal evaporation as shown in FIG. 3g. The ohmic metal contacts 318 can be comprised of a combination of Ti/Al or Ta/Ti/Al, which are deposited in that order and are about 320 nm thick. The ohmic metal contacts 318 are annealed at a temperature in the range of about 600-800° C. for about a minute. This allows the ohmic metal contacts 318 to diffuse into the capping layer 309, the AlN layer 308, and the second layer 306, creating an ohmic contact region 319 as shown in FIG. 3h. The ohmic contact region 319 can then be used as a source and a drain.

[0070] Next, a portion of the remaining photoresist layer 310 is patterned and removed as shown in FIG. 3iusing techniques well-known in the art, to create a second window 322. The second window 322 exposes part of the capping layer 309 as shown in FIG. 3iThe exposed portion of the capping layer 309 is removed using a reactive ion etch preferably with chlorine gas at an etch rate of about 72 nm/min. Etching away a portion of the capping layer 309 exposes a portion of the surface of the AlN layer 308 as shown in FIG. 3j, however the reactive ion etching does not remove any portion of the AlN layer 308. The AlN layer 308 effectively acts as an etch stop to the reactive ion etching, thereby preventing the reactive ion etch from damaging the first or second layers 304, 306 beneath the AlN layer 308. It is possible though, for the exposed portion of the surface of the AlN layer 308 to incur damage caused by the reactive ion etch. Also, the remaining portion of the photoresist layer 310 is removed using techniques known in the art. Next, using the remaining portion of the capping layer 309 as a mask, a portion of the exposed AlN layer 308 is etched away with a solvent to create a non-planar gate region 324 as shown in FIG. 3k. The solvent preferably comprises potassium hydroxide (KOH), water, and potassium borates, and is sold under the tradename AZ-400 by the Clariant Corporation of Somerville, N.J. Etching the AlN layer 308 also helps removes any surface damage on the AlN layer 308 caused by the reactive ion etching of the capping layer 309. As shown in FIG. 3kthe non-planar gate region 324 is created by completely etching a portion of the AlN layer 308 through to the second layer 306. However, this embodiment would work if the AlN layer 308 was not completely etched away as shown in FIG. 1jof the first embodiment. In the first embodiment the AlN layer 108 was not completely etched away. However, should the AlN layer 308 be completely etched through to the second layer 306 as shown in FIG. 3k, the layers located beneath the AlN layer 308 will not be affected by the AZ-400 solution. In this third embodiment, the second layer 306 preferably comprises AlGaN. AlGaN is insoluble in AZ-400 and effectively acts as an etch stop, preventing the AZ-400 from damaging the layers located under the AlN layer 308 should a portion of the AlN layer 308 be completely etched through to the second layer 306.

[0071] A gate 326 is deposited in the non-planar gate region 324 as shown in FIG. 31. The gate 326 is preferably T-shaped in order to help reduce intrinsic resistance and capacitance. Fabricating a T-shaped structure is a technique well-known in the art.

[0072] Because the first layer 304 and second layer 306 are comprised of group III-V materials, the interface 305 between the first layer 304 and second layer 306 already contains carrier charges due to the well-known effects of spontaneous polarization. In this way, the interface 305 acts as a channel for the transistor without requiring any additional doping. However, additional doping of the transistor can be provided, if desired. When the transistor is biased with a voltage at the gate 326 and at either of the ohmic contact regions 319, the charges at the interface 305 flow between the ohmic contact regions 319, allowing operation of the non-planar heterostructure field effect transistor.

[0073] Let it be understood that the foregoing description is only illustrative of the invention. Various alternatives and modifications can be devised by those skilled in the art without departing from the spirit of the invention. Accordingly, the present invention is intended to embrace all such alternatives, modifications, and variances which fall within the scope of the appended claims.