Methods for observing crystal defects in compound semiconductors

Abstract

A device having a gate electrode (3) formed along the [2-1-10] direction on the c-plane (0001) of a wurtzite-structured compound semiconductor (1) was cut along the (10-10) plane to produce a sample (4). Using a transmission electron microscope, an electron beam (5) was incident on the sample (4) from the [-1010] direction, and edge dislocations with Burgers vectors of 1 / 3 [2-1-10] and 1 / 3 [-2110] and mixed dislocations with Burgers vectors of 1 / 3 [2-1-13] and 1 / 3 [-2113] were observed.

Description

Technical Field

[0001] The present invention relates to a method for observing crystal defects of a compound semiconductor using a transmission electron microscope. Background Art

[0002] GaN is currently being used as a material for high-output, high-efficiency, high-frequency devices and blue-light-emitting devices. Compared to Si and GaAs, GaN has more crystal defects. Therefore, efforts are underway to reduce the density of crystal defects that can negatively impact devices. While defects that severely impact quality, such as stacking defects and micropipes, have recently disappeared, linear threading dislocations, known as dislocations, still exist at high densities. Understanding the types and densities of dislocations and incorporating these into the process to reduce dislocation density is crucial.

[0003] One method for observing dislocations is the pitting method using KOH etching. However, once a device pattern is formed, observation becomes difficult, and identification of the dislocation type is also difficult. Therefore, the best method for observing dislocations is using a transmission electron microscope (TEM).

[0004] Cross-sectional TEM is a method of observing the cross section of a semiconductor device at high magnification using a TEM. Since electrons need to be injected from the cross-sectional direction, the sample for observation is usually thinned. The thinned sample is mounted on a copper or molybdenum mesh, placed in the TEM device, and irradiated with an electron beam to observe a magnified image or diffraction image. Because crystal defects such as dislocations and stacking defects have a significant impact on the characteristics and reliability of the device, analysis based on TEM is prevalent. The impact on the device varies depending on the type of crystal defect, so simply confirming the presence of the defect is not enough; detailed analysis is required.

[0005] By changing the diffraction conditions, crystal defects appear or disappear on the electron microscope screen. Therefore, it is necessary to consider the crystal structure, the diffraction vector of the electron beam, and the Burgers vector of the crystal to observe crystal defects. GaN used as a device material has a wurtzite structure. Compared with Si with a diamond structure and GaAs with a zinc blende structure, the crystal axis is not perpendicular, so the analysis is complicated. As complete dislocations in the wurtzite structure, there are edge dislocations, screw dislocations, and mixed dislocations. In order to distinguish the types of each dislocation using TEM, it is necessary to screen the diffraction (reflection) vectors (for example, refer to Patent Document 1).

[0006] Prior art literature

[0007] Patent Literature

[0008] Patent Document 1: Japanese Patent Application Laid-Open No. 2000-349338 Summary of the Invention

[0009] Slightly tilt the sample when screening the diffraction vectors. If the tilt is more than 5 degrees, the crystal zone axis will change to other crystal zone axes, and the existing analysis cannot be performed. In the case of GaN, the Burgers vectors of edge dislocations and mixed dislocations are oriented in three directions. When the inner product of the diffraction vector g and the Burgers vector b is zero (g·b=0), it disappears from the transmission electron microscope screen and cannot be observed. Therefore, there is a problem that it is impossible to observe while identifying dislocations in the wurtzite structure compound semiconductor.

[0010] The present invention has been made to solve the above-mentioned problems, and its object is to provide a method for observing crystal defects in compound semiconductors, which can observe dislocations existing in a compound semiconductor having a wurtzite structure while identifying the dislocations.

[0011] The method for observing crystal defects of a compound semiconductor of the present invention is characterized in that it comprises: a step of cutting a device having a gate electrode formed along the [2-1-10] direction on the c-plane (0001) of a wurtzite-structured compound semiconductor along the (10-10) plane to prepare a sample; and a step of observing edge dislocations with Burgers vectors of 1 / 3 [2-1-10] and 1 / 3 [-2110] and mixed dislocations with Burgers vectors of 1 / 3 [2-1-13] and 1 / 3 [-2113] by using a transmission electron microscope to allow an electron beam to be incident on the sample from the [-1010] direction.

[0012] In the present invention, a device with a gate electrode formed along the [2-1-10] direction on the c-plane (0001) of a wurtzite compound semiconductor is cut along the (10-10) plane to create a sample. This sample is then injected with an electron beam from the [-1010] direction using a transmission electron microscope. This allows for the identification and observation of dislocations in the wurtzite compound semiconductor. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 It is a diagram showing a method for observing crystal defects of a compound semiconductor according to the first embodiment.

[0014] Figure 2 A diagram showing a method for observing crystal defects in a compound semiconductor according to a comparative example.

[0015] Figure 3 This is a diagram showing a method for observing crystal defects of a compound semiconductor according to the second embodiment.

[0016] Figure 4 A diagram showing a method for observing crystal defects of a compound semiconductor according to a third embodiment.

[0017] Figure 5 is viewed from above along the c-axis Figure 4 Diagram of the crystal structure.

[0018] Figure 6 This is a diagram showing a thick film sample cut so as to include crystal defects to be observed.

[0019] Figure 7 This figure shows an example in which a thick film sample is further thinned for use in a conventional TEM.

[0020] Figure 8 A diagram showing a method for observing crystal defects of a compound semiconductor according to a fourth embodiment.

[0021] Figure 9 A diagram showing a method for observing crystal defects of a compound semiconductor according to a fourth embodiment.

[0022] Figure 10 A diagram showing a method for observing crystal defects of a compound semiconductor according to a fourth embodiment.

[0023] Figure 11 It is a diagram showing a method for observing crystal defects of a compound semiconductor according to a fifth embodiment.

[0024] Figure 12 It is a diagram showing a method for observing crystal defects of a compound semiconductor according to a fifth embodiment.

[0025] Figure 13 A diagram showing a method for observing crystal defects of a compound semiconductor according to a sixth embodiment. DETAILED DESCRIPTION

[0026] A method for observing crystal defects of a compound semiconductor according to an embodiment will be described with reference to the accompanying drawings. Identical or corresponding components are denoted by the same reference numerals and redundant descriptions may be omitted.

[0027] Implementation method 1.

[0028] Figure 1 This figure illustrates a method for observing crystal defects in a compound semiconductor according to Embodiment 1. It illustrates a unit cell of GaN1, a wurtzite-structured compound semiconductor. GaN1 has crystal axes in three directions a1, a2, and a3 perpendicular to the c-axis.

[0029] The Burgers vectors of perfect dislocations in GaN1 are roughly divided into three types. Dislocations 2a, 2b, and 2c with Burgers vectors perpendicular to the c-axis are called edge dislocations. Dislocation 2d with a Burgers vector oriented in the c-axis direction is called a screw dislocation. Dislocation 2e, a combination of an edge dislocation and a screw dislocation, is called a mixed dislocation.

[0030] Dislocation 2e and dislocation 2b are located on the same plane and tilt in the positive direction of direction a1, but there are also mixed dislocations that tilt in the opposite negative direction. Mixed dislocations also tilt in the negative direction on the plane of dislocations 2a and 2c. Therefore, there are six types of mixed dislocations.

[0031] Planes perpendicular to directions a1, a2, and a3 are collectively referred to as a-planes (2-1-10). Planes rotated 30 degrees from the a-planes (2-1-10) are collectively referred to as m-planes (10-10). Typical GaN devices, such as GaN HEMTs, are formed on a c-plane (0001), perpendicular to the c-axis. A gate electrode 3 is formed on the c-plane (0001) of GaN1 along the [2-1-10] direction. Although not shown for simplicity, the drain and source electrodes are typically also formed parallel to the gate electrode 3.

[0032] This device was cut with the m-plane (10-10) tilted relative to the gate electrode 3 to produce sample 4. Using a transmission electron microscope (TEM), an electron beam 5 was directed into sample 4 from the [-1010] direction. After accurately aligning the incident direction with the crystal's zone axis, the c-axis was slightly tilted a few degrees in the ±a1 direction, adjusting the excitation diffraction vector to the (-12-10) point. This allowed observation of edge dislocations with Burgers vectors of 1 / 3 [2-1-10] and 1 / 3 [-2110], and mixed dislocations with Burgers vectors of 1 / 3 [2-1-13] and 1 / 3 [-2113]. It should be noted that dislocations with equivalent Burgers vectors can also be observed in the same manner.

[0033] Next, the effects of this embodiment will be described in comparison with a comparative example. Figure 2 This is a diagram showing a method for observing crystal defects in a compound semiconductor according to a comparative example. In the comparative example, the device is cut with the a-plane (2-1-10) perpendicular to the gate electrode 3 to produce a sample 4. Then, an electron beam is incident on the sample 4 from the [-2110] direction perpendicular to the cross section using a TEM. After correctly aligning the incident direction with the crystal zone axis, the c-axis is slightly tilted a few degrees in the ±a1 direction and adjusted to the (01-10) point of the excitation diffraction vector. This allows for the observation of some edge dislocations and mixed dislocations. The inner product of the Burgers vector b and the diffraction vector g of the screw dislocation becomes zero (g·b=0), so it disappears on the screen of the transmission electron microscope and cannot be observed. In the comparative example, the inner product of the dislocations 2b and 2e, which are edge dislocations, also becomes zero (g·b=0), and cannot be observed.

[0034] In contrast, in this embodiment, the cutting tool is tilted 30 degrees relative to the gate electrode 3 to produce the sample 4, and the electron beam is incident on the sample 4 from the [-1010] direction to observe dislocations. This avoids the extinction rule that the inner product of the diffraction vector g and the Burgers vector b is zero (g·b=0), and allows for the identification and observation of all edge and mixed dislocations present in a wurtzite-structured compound semiconductor.

[0035] It should be noted that in the comparative example, dislocations could be observed if the electron beam was incident at an angle of 30 degrees. However, recent TEMs prioritize observation at high magnifications and are often unable to tilt the sample 4 at high angles. In contrast, the crystal defect observation method of this embodiment is applicable to all TEMs.

[0036] Implementation method 2.

[0037] Figure 3 This figure illustrates a method for observing crystal defects in a compound semiconductor according to Embodiment 2. As in Embodiment 1, a gate electrode 3 is formed along the [2-1-10] direction on the c-plane (0001) of GaN 1, a wurtzite-structured compound semiconductor. In this embodiment, as in the comparative example, the device is cut along the (2-1-10) plane perpendicular to the gate electrode 3 to produce a sample 4.

[0038] TEM observation was performed using an electron beam 5 incident on sample 4 from the [-2110] direction, similar to conventional observation methods. Furthermore, the electron beam 5 was also incident on sample 4 from the [-1-120] and [-12-10] directions, with the sample 4 rotated and tilted 60 degrees within the microscope. If the electron beam 5 is incident from the [-2110] direction, edge dislocations 2b satisfy g·b = 0, and therefore cannot be observed.

[0039] On the other hand, by directing the electron beam 5 into the sample 4 from the [-1-120] direction, dislocations with Burgers vectors of 1 / 3 [11-20], 1 / 3 [-1-120], 1 / 3 [11-23], and 1 / 3 [-1-123] disappear on the transmission electron microscope screen, while dislocations with Burgers vectors of 1 / 3 [-2110], 1 / 3 [2-1-10], 1 / 3 [-2113], and 1 / 3 [2-1-13] appear. Furthermore, by directing the electron beam 5 into the sample 4 from the [-12-10] direction, dislocations with Burgers vectors of 1 / 3 [1-210], 1 / 3 [-12-10], 1 / 3 [1-213], and 1 / 3 [-12-13] disappear. This allows the directions of the Burgers vectors originating from dislocations a1, a2, and a3 to be distinguished.

[0040] Implementation method 3.

[0041] Figure 4 This figure illustrates a method for observing crystal defects in a compound semiconductor according to Embodiment 3. A gate electrode 3 is formed along the [10-10] direction on the c-plane (0001) of GaN 1, a wurtzite-structured compound semiconductor. The formation direction of gate electrode 3 differs from that of Embodiment 1, but generally varies depending on the manufacturer and manufacturing method. The device is cut along the (10-10) plane perpendicular to gate electrode 3 to produce sample 4.

[0042] The electron beam 5 was incident on the sample 4 from the [-1010] direction perpendicular to the sample 4 and also from the [-1-120] and [-2110] directions by rotating and tilting the sample 4 30 degrees within the microscope for observation using the TEM.

[0043] By directing the electron beam 5 from the direction [-1-120] to sample 4, g·b = 0 is satisfied, and dislocations with Burgers vectors of 1 / 3 [11-20], 1 / 3 [-1-120], 1 / 3 [11-23], and 1 / 3 [-1-123] disappear from the transmission electron microscope screen. By directing the electron beam from the direction [-2110] to sample 4, dislocations with Burgers vectors of 1 / 3 [-2110], 1 / 3 [2-1-10], 1 / 3 [-2113], and 1 / 3 [2-1-10] disappear. This allows the Burgers vector directions of dislocations originating from a1, a2, and a3 to be distinguished.

[0044] Implementation method 4.

[0045] Figure 5 is viewed from above along the c-axis Figure 4 A diagram of the crystal structure. Conventionally, the location of the crystal defect 6 to be observed was determined by performing electrical operation, electroluminescence (EL), optical beam heating resistance change (OBIRCH), or thermal analysis under the device state. Then, the area is cut using a FIB or the like to make the sample 4. In the case of a normal TEM, the acceleration voltage of the electron beam is 70kV to 300kV, so the electron beam needs to be thinned to 0.02μm to 0.3μm in order to transmit the sample 4. Since the alignment accuracy of EL, OBIRCH, luminescence observation and FIB processing is about 1μm, the area to be observed is sometimes cut out.

[0046] To reliably observe crystal defects 6, an ultrahigh-voltage electron microscope is effective. An ultrahigh-voltage electron microscope uses electrons with an accelerating voltage of 1 MV or higher. Existing ultrahigh-voltage electron microscopes can achieve a maximum acceleration of 3 MV. Because electrons have high penetration power, even a 3μm-thick sample 4 can be observed in GaN. This allows for reliable detection of the desired area.

[0047] Figure 6 This diagram shows a thick-film sample cut to include the crystal defects to be observed. For example, observation is performed by injecting an electron beam of 1 MV or greater from the [-1010] direction into the thick-film sample 4a. The location of the crystal defect 6 can be determined using stereoscopic techniques or other methods. Once the location is determined, the thick-film sample 4a is thinned using FIB. Figure 7 This figure shows an example of a thick film sample 4a being thinned further for use in a normal TEM. Thinning enables ultra-high-resolution observation, enabling observation of atomic images, detailed analysis of Burgers vectors, and analysis using energy loss spectroscopy.

[0048] Figures 8 to 10 : is a diagram showing a method for observing crystal defects of a compound semiconductor according to a fourth embodiment. Figure 8 As shown in FIG. 1 , a device is manufactured in which a gate electrode 3 is formed along the [10-10] direction on the c-plane (0001) of GaN having a wurtzite structure. Next, as shown in FIG. Figure 9 As shown, a thick film sample 4a with a thickness of 0.1 μm to 5 μm was produced using a (2-1-10) surface cutting device.

[0049] The thick film sample 4a is observed by ultra-high voltage electron microscope, and the position of the crystal defect 6 is determined by stereo method or the like. Then, the thick film sample 4a is thinned obliquely so that the (-2110) plane appears. Figure 10 Film sample 4b was prepared as shown.

[0050] The electron beam is incident on the thin film sample 4b from three directions: [-12-10], [-2110], and [-1-120] by TEM to observe the crystal defects 6. Figures 5 to 7 The dislocations from a2 that cannot be eliminated by the method of also disappear. Therefore, the dislocations from a1, a2, and a3 can be eliminated separately, and the directions of the Burgers vectors of all edge dislocations and mixed dislocations can be distinguished and identified.

[0051] Implementation method 5.

[0052] Figure 11 and Figure 12 : is a diagram showing a method for observing crystal defects of a compound semiconductor according to Embodiment 5. Figure 11 As shown, the device formed on the c-plane (0001) of GaN1, a compound semiconductor having a wurtzite structure, is cut parallel to the gate electrode 3 to produce a sample 4 having a thickness of 0.1 μm to 5 μm. In a conventional thin-filming method, the cut sample 4 is thinned parallel to the gate electrode 3. On the other hand, in this embodiment, as shown in FIG. Figure 12As shown, multiple thin slices with three or more different plane orientations are formed in a single sample 4. Here, thin slices with plane orientations of (11-20), (-12-10), and (-2110) are formed in a single sample 4. Crystal defects in the multiple thin slices are observed using a TEM, and the Burgers vectors are analyzed. This reduces the number of sample preparation steps and sample placement operations compared to preparing multiple samples 4.

[0053] Implementation method 6.

[0054] Figure 13 This figure shows a method for observing crystal defects in a compound semiconductor according to Embodiment 6. A gate electrode 3 is formed on the c-plane (0001) of GaN1, a wurtzite-structured compound semiconductor, in the [10-10] direction perpendicular to the m-plane (10-10).

[0055] In the wurtzite structure, partial dislocations exist separately from the perfect dislocations described in the above embodiments. Representative examples are partial dislocations 7a to 7d with Burgers vectors b of 1 / 3 [10-10], 1 / 3 [02-23], 1 / 3 [20-23], and 1 / 2

[0001] . When sample 4 is cut perpendicular to gate electrode 3 and subjected to vertical electron beam incidence, partial dislocations 7a and 7c disappear and become unobservable.

[0056] In contrast, in this embodiment, the device is cut along the a-plane (2-1-10) tilted 30 degrees relative to the gate electrode 3 to produce sample 4. This allows TEM observation of partial dislocations 7a and 7c in sample 4 with Burgers vectors of 1 / 3 [10-10] and 1 / 3 [20-23].

[0057] It should be noted that while the analysis method for representative Burgers vectors has been described in Embodiments 1-6, other equivalent Burgers vectors can also be analyzed using the same method. Furthermore, while the analysis method for Burgers vectors of the wurtzite structure has been described, the same method can also be used for analysis of other crystal structures.

[0058] Explanation of symbols

[0059] 1GaN, 2a, 2b, 2c, 2d, 2e dislocations, 3 gate electrode, 4 sample, 4a thick film sample, 4b thin film sample, 5 electron beam, 6 crystal defects, 7a~7d partial dislocations.

Claims

1. A method for observing crystal defects of a compound semiconductor, characterized in that: have: A step of preparing a sample by cutting a device having a gate electrode formed along a [2-1-10] direction on a c-plane (0001) of a wurtzite compound semiconductor at a (10-10) plane; and The process involves using a transmission electron microscope to allow an electron beam to be incident on the sample from the [-1010] direction, and observing edge dislocations with Burgers vectors of 1 / 3 [2-1-10] and 1 / 3 [-2110] and mixed dislocations with Burgers vectors of 1 / 3 [2-1-13] and 1 / 3 [-2113].

Images