High voltage depletion mode current source, transistor and method of manufacture - Patents.com
By integrating a depletion mode current source into an enhancement mode GaN process using a conductive electrode over a 2DEG or 2DHG channel, the challenge of high voltage startup circuits in semiconductor manufacturing is addressed, achieving efficient and cost-effective operation in devices like AC/DC power converters.
Patent Information
- Application Number
- JP2024534478
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2021-12-10
- Filing Date
- 2022-12-12
- Publication Date
- 2024-12-05
AI Technical Summary
Existing semiconductor manufacturing processes face challenges in integrating depletion mode transistors, such as junction field effect transistors (JFETs), into enhancement mode GaN processes without increasing manufacturing costs or complexity, which are necessary for high voltage startup circuits in devices like AC/DC power converters.
A depletion mode current source is integrated into an enhancement mode GaN process by using a conductive electrode over a 2DEG or 2DHG channel, allowing for high voltage operation without requiring additional masks or materials, and is fabricated alongside enhancement mode FETs on the same die.
This approach enables a cost-effective and efficient start-up circuit for power converters, capable of handling high voltages up to 650V, without the need for external depletion mode transistors, thus reducing manufacturing complexity and cost.
Smart Images

Figure 2024545164000001_ABST
Abstract
Description
[Technical field]
[0001] This description relates generally to semiconductor manufacturing, and more particularly to high voltage depletion mode current sources, high voltage depletion mode transistors, and methods of fabricating depletion mode current sources or transistors in an enhancement mode process. [Background technology]
[0002] A current source is an electronic circuit that delivers or sinks a current independent of the voltage across it, at least for a range of voltages. A simple non-ideal current source can consist of a voltage source in series with a resistor. The amount of current provided by such a current source is given by the ratio of the voltage drop across the resistor to its resistance, according to Ohm's law.
[0003] Gallium nitride (GaN) field effect transistors (FETs) offer advantages in weight, size, cost, switching speed, and energy consumption compared to silicon FETs and are used in applications such as 5G rectifiers, motor drives, and AC / DC power converters that may be used to power portable computers and charge mobile device batteries. GaN devices come in enhancement mode (e-GaN) and depletion mode (d-GaN) varieties. A depletion mode device is normally on when its gate-source voltage is zero, whereas an enhancement mode FET is normally off when its gate-source voltage is zero. An enhancement mode FET can be turned on by keeping the gate voltage greater than the source voltage in an NMOS circuit and less than the source voltage in a PMOS circuit. Enhancement mode FETs are commonly used as power switches.
[0004] In solid state physics, a two-dimensional electron gas (2DEG) is a gas of electrons tightly confined in two dimensions within a semiconductor structure. The tight confinement of the electron gas leads to quantized energy levels for motion in the third dimension. Hence, the electrons appear as a two-dimensional sheet embedded in a three-dimensional semiconductor structure. A similar structure, but with holes rather than electrons, is called a two-dimensional hole gas (2DHG). Summary of the Invention
[0005] One example is an integrated circuit (IC) that includes a two-terminal depletion mode current source. The current source includes a 2DEG or 2DHG channel coupled between a first terminal of the depletion mode current source and a second terminal of the depletion mode current source. A conductive electrode coupled to the first terminal is above and extends over at least a portion of the 2DEG or 2DHG channel.
[0006] An exemplary method of manufacturing an IC includes forming a first semiconductor layer over a substrate and forming a second semiconductor layer over the first semiconductor layer. The first and second semiconductor layers are configured to support a 2DEG or 2DHG. The method further includes forming a depletion mode current source over the IC. Forming the depletion mode current source includes patterning a 2DEG or 2DHG channel over the first semiconductor layer to provide a patterned region of the 2DEG or 2DHG channel. Forming the depletion mode current source further includes forming a dielectric layer over the second semiconductor layer and etching the dielectric layer to open first and second contact windows to the 2DEG or 2DHG channel. Forming the depletion mode current source further includes forming one or more contact layers, at least in part within the first and second contact windows (e.g., by depositing and etching one or more ohmic layers), to provide first and second contacts at respective ends of the 2DEG or 2DHG channel, at least one of the first and second contacts including a conductive electrode above the dielectric layer and extending over at least a portion of the 2DEG or 2DHG between the first and second contacts. A portion of the conductive electrode extends over a point on the 2DEG or 2DHG channel. A first distance along the 2DEG or 2DHG between the point and a first terminal is greater than a second distance along the 2DEG or 2DHG between the point and a second terminal. A first wire is provided to the first contact and a second wire is provided to the second contact to provide the depletion mode current source as a two-terminal device.
[0007] Another example is an IC that includes an enhancement mode GaN FET, a resistor having a semiconductor-to-semiconductor interface supporting a two-dimensional electron gas (2DEG) or a two-dimensional hole gas (2DHG), a metal-insulator-metal (MIM) capacitor, and a depletion mode current source. The current source includes a semiconductor-to-semiconductor interface over an area to provide a 2DEG or 2DHG channel in the patterned area. The current source further includes a metal field plate coupled to a negative terminal of the depletion mode current source, the metal field plate extending over a portion of the patterned area.
[0008] Another example is an IC that includes a depletion mode transistor that includes a 2DEG or 2DHG channel coupled between a drain terminal of the depletion mode transistor and a source terminal of the depletion mode transistor, and a conductive electrode coupled to a gate terminal of the depletion mode transistor above the 2DEG or 2DHG channel and extending over at least a portion of the 2DEG or 2DHG channel.
[0009] An exemplary method of manufacturing an IC includes forming a first GaN-based semiconductor layer on a substrate and forming a second GaN-based semiconductor layer on the first semiconductor layer. The first and second semiconductor layers can be configured to support a 2DEG or 2DHG in, for example, a 2DEG or 2DHG channel. The method further includes patterning the first and second GaN-based semiconductor layers to form a first epi structure in a depletion mode transistor area, a second epi structure in a resistor area, and a third epi structure in an enhancement mode FET area. The method further includes forming a dielectric layer on the second semiconductor layer. The method further includes etching the dielectric layer to open first and second contact windows at opposite ends of the first epi structure. The method further includes forming one or more contact layers for providing the first and second contacts at least partially within the first and second contact windows, and a conductive electrode above the dielectric layer and extending over at least a portion of the first epi structure. [Brief description of the drawings]
[0010] [Figure 1] FIG. 2 is a block diagram illustrating an example power converter system including a 2DEG or 2DHG based current source in a precharge circuit of a start-up circuit.
[0011] [Diagram 2] FIG. 1 is a block diagram of an example power converter system including a startup circuit.
[0012] [Diagram 3] 3 is a block diagram of an example startup circuit that may be used in the power converter system of FIG. 2.
[0013] [Figure 4A] 1A-1D are cross-sectional views of four exemplary devices fabricated according to an enhancement mode fabrication process. [Figure 4B] 1A-1D are cross-sectional views of four exemplary devices fabricated according to an enhancement mode fabrication process.
[0014] [Diagram 5] FIG. 13 is a top-down view of an exemplary resistive patterned area of a 2DEG or 2DHG channel.
[0015] [Figure 6] 1 is a graph of example current-voltage curves for four different 2DEG-based devices, one without a field plate and three with field plates at various distances from the 2DEG channel.
[0016] [Figure 7] 1 is a graph of example current-voltage curves for two different 2DEG-based devices, both with field plates, but with different 2DEG channel resistances.
[0017] [Figure 8A] 4A and 4B. FIG. 4C is a cross-sectional view showing an example manufacturing process leading to the device of FIG. 4A and FIG. 4B. [Figure 8B] 4A and 4B. FIG. 4C is a cross-sectional view showing an example manufacturing process leading to the device of FIG. 4A and FIG. 4B. [Figure 8C] 4A and 4B. FIG. 4C is a cross-sectional view showing an example manufacturing process leading to the device of FIG. 4A and FIG. 4B. [Figure 8D] 4A and 4B. FIG. 4C is a cross-sectional view showing an example manufacturing process leading to the device of FIG. 4A and FIG. 4B. [Figure 8E] 4A and 4B. FIG. 4C is a cross-sectional view showing an example manufacturing process leading to the device of FIG. 4A and FIG. 4B. [Figure 8F] 4A and 4B. FIG. 4C is a cross-sectional view showing an example manufacturing process leading to the device of FIG. 4A and FIG. 4B. [Figure 8G] 4A and 4B. FIG. 4C is a cross-sectional view showing an example manufacturing process leading to the device of FIG. 4A and FIG. 4B. [Figure 8H] 4A and 4B. FIG. 4C is a cross-sectional view showing an example manufacturing process leading to the device of FIG. 4A and FIG. 4B. [Figure 8I] 4A and 4B. FIG. 4C is a cross-sectional view showing an example manufacturing process leading to the device of FIG. 4A and FIG. 4B. [Figure 8J] 4A and 4B. FIG. 4C is a cross-sectional view showing an example manufacturing process leading to the device of FIG. 4A and FIG. 4B. [Figure 8K] 4A and 4B. FIG. 4C is a cross-sectional view showing an example manufacturing process leading to the device of FIG. 4A and FIG. 4B. [Figure 8L] 4A and 4B. FIG. 4C is a cross-sectional view showing an example manufacturing process leading to the device of FIG. 4A and FIG. 4B. [Figure 8M] 4A and 4B. FIG. 4C is a cross-sectional view showing an example manufacturing process leading to the device of FIG. 4A and FIG. 4B. [Figure 8N] 4A and 4B. FIG. 4C is a cross-sectional view showing an example manufacturing process leading to the device of FIG. 4A and FIG. 4B. [Figure 8O] 4A and 4B. FIG. 4C is a cross-sectional view showing an example manufacturing process leading to the device of FIG. 4A and FIG. 4B. [Figure 8P] 4A and 4B. FIG. 4C is a cross-sectional view showing an example manufacturing process leading to the device of FIG. 4A and FIG. 4B. [Figure 8Q]4A and 4B. FIG. 4C is a cross-sectional view showing an example manufacturing process leading to the device of FIG. 4A and FIG. 4B. [Figure 8R] 4A and 4B. FIG. 4C is a cross-sectional view showing an example manufacturing process leading to the device of FIG. 4A and FIG. 4B. [Figure 8S] 4A and 4B. FIG. 4C is a cross-sectional view showing an example manufacturing process leading to the device of FIG. 4A and FIG. 4B. [Figure 8T] 4A and 4B. FIG. 4C is a cross-sectional view showing an example manufacturing process leading to the device of FIG. 4A and FIG. 4B. [Figure 8U] 4A and 4B. FIG. 4C is a cross-sectional view showing an example manufacturing process leading to the device of FIG. 4A and FIG. 4B. [Figure 8V] 4A and 4B. FIG. 4C is a cross-sectional view showing an example manufacturing process leading to the device of FIG. 4A and FIG. 4B. [Figure 8W] 4A and 4B. FIG. 4C is a cross-sectional view showing an example manufacturing process leading to the device of FIG. 4A and FIG. 4B. [Figure 8X] 4A and 4B. FIG. 4C is a cross-sectional view showing an example manufacturing process leading to the device of FIG. 4A and FIG. 4B. [Figure 8Y] 4A and 4B. FIG. 4C is a cross-sectional view showing an example manufacturing process leading to the device of FIG. 4A and FIG. 4B. [Figure 8Z] 4A and 4B. FIG. 4C is a cross-sectional view showing an example manufacturing process leading to the device of FIG. 4A and FIG. 4B.
[0018] [Figure 9] 4 is a flowchart illustrating an example method of fabricating a depletion mode current source on an IC.
[0019] [Figure 10] FIG. 4B is a cross-sectional view of the depletion mode current source of FIG. 4A.
[0020] [Figure 11] 1A-1D are cross-sectional views of different exemplary depletion mode transistors. [Figure 12] 1A-1D are cross-sectional views of different exemplary depletion mode transistors. [Figure 13] 1A-1D are cross-sectional views of different exemplary depletion mode transistors. [Figure 14]1A-1D are cross-sectional views of different exemplary depletion mode transistors. [Figure 15] 1A-1D are cross-sectional views of different exemplary depletion mode transistors. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0021] Certain electrical circuits are designed to start when power is applied, such as a power supply for a device charging system. In some cases, the applied power takes the form of a high voltage, and the start-up circuitry needs to have high voltage withstand capability. The high voltage start-up circuitry may include a high voltage normally-on transistor, such as a junction field effect transistor (JFET) or other depletion mode transistor. When power is initially applied, the normally-on transistor can turn on without an auxiliary supply voltage. The normally-on transistor is then used to charge a capacitor until an acceptable value for the supply voltage is reached. Thus, the normally-on transistor acts as a current source at start-up, and other circuitry can disable the normally-on transistor after start-up for low leakage post-start-up operation. However, using an external (off-die) JFET or other external depletion mode transistor is expensive in terms of manufacturing cost and circuit area, and it has not yet been possible to integrate a JFET or other depletion mode transistor into an enhancement mode GaN manufacturing process without increasing the mask count and other changes to the manufacturing process flow, thus increasing the manufacturing cost of the die. The depletion mode current source devices described herein can be fabricated in processes designed for the construction of enhancement mode FETs without requiring modifications to the enhancement mode process, such as requiring additional masks or materials. Thus, the current source devices and associated fabrication methods described herein solve the problem of providing a start-up circuit useful in, for example, AC / DC power converters, without requiring off-die depletion mode FETs, or expensive modifications to the enhancement mode process that would be required to fabricate enhancement mode and depletion mode FETs on the same die.
[0022] The block diagram of FIG. 1 illustrates a system 100 for power conversion that includes a current source 108 that can form all or part of a pre-charge circuit 106 in a start-up circuit 104 in a power converter 102. The power converter 102 can be an AC / DC power converter, such as may be used to charge a battery in a cell phone or other mobile device. The power converter 102 can start up by itself when plugged into a grid voltage. The current source 108 includes a patterned resistive 2DEG or 2DHG channel 110 and one or more conductive electrodes 112 (e.g., metal field plates) disposed over the channel 110 such that the conductive electrodes provide coverage over the channel 110 in a top-down view of the channel. Examples of such electrodes are described below with reference to an example current source 426 shown in FIG. 4A, in which electrodes 434, 436, 438 are illustrated.
[0023] The channel 110 may be formed by an interface between two or more semiconductor layers capable of supporting a 2DEG or 2DHG, and all or a portion of the channel may be defined into a channel pattern, e.g., a serpentine channel pattern, by, for example, damage implantation or mesa etching, within a patterned area as seen in a top-down view of the channel. An exemplary resistive channel patterning is illustrated in the top-down view of FIG. 5. One or more conductive electrodes 112 may extend over or may be on all or a substantial portion of the resistive patterned area of the channel 110. As used herein, "on" and variations thereof refer to a substantial spatial intersection in a top-down view and do not require direct contact between the metal of one of the conductive electrodes 112 and the semiconductive layer that forms the channel 110. For example, there may be one or more dielectric (e.g., nitride) layers between the channel 110 and the one or more conductive electrodes 112. For example, at least one of the one or more conductive electrodes 112 may overlie at least 10% of the patterned area of the channel 110 in the current source 108. For example, at least one of the one or more conductive electrodes 112 may overlie at least 20% of the patterned area of the channel 110. For example, at least one of the one or more conductive electrodes 112 may overlie at least 50% of the patterned area of the channel 110. For example, at least one of the one or more conductive electrodes 112 may overlie at least 80% of the patterned area of the channel 110. In some examples, the one or more conductive electrodes 112 overlie the entire area of the resistive patterned portion of the channel 110. The exact percentage of spatial intersection between at least one of the one or more conductive electrodes 112 and the patterned area of the channel 110 in a top-down view may depend on the configuration of the channel, the desired saturation current, and the pinch-off voltage of the current source 108.
[0024] The start-up circuit 104 may also include one or more enhancement mode FETs 114, at least one resistor including a 2DEG or 2DHG channel, and a metal-insulator-metal (MIM) capacitor fabricated on the same die as the current source 108. The start-up circuit 104 in the power converter 102 may include a high voltage, low current DC path (not shown). When the power converter is coupled to an AC or DC power source (e.g., plugged into a wall outlet that provides AC power), the DC path provides DC current from the start-up circuit 104 without the application of a control voltage. The DC current may need to be small enough to meet standby power requirements. The DC current may be used to boost the output of the supply voltage for the power converter 102 to start up.
[0025] In contrast to resistor 116, current source 108 has a sufficiently low saturation current so that it can supply the DC path with a current of, for example, less than 10 μA from 400 V. If resistor 116 is used for such purposes (as a current source), a resistor of 1 mm is required. 2 A current source 108 may require a die area of more than 100 volts, which may significantly increase the cost of the die. The system 100 with the current source 108 allows for an integrated startup circuit 104 in a power converter IC fabricated using an all-enhancement mode process, such as an enhancement mode GaN process, in which depletion mode transistor devices may not be available or may require additional photolithographic masks. The current source 108 can be fabricated on the same die as the enhancement mode FET 114 (e.g., a GaN FET) without the need to modify the enhancement mode fabrication process used to create the FET 114 with additional mask layers or materials, providing a depletion mode device with current source characteristics that can handle high voltages, e.g., up to at least 650V. The depletion mode current source 108 may be used as a precharge device to turn on the enhancement mode transistor (e.g., FET 114) to increase the supply voltage for the power converter 102.
[0026] As an example, a depletion mode current source 108 can be formed in an enhancement mode GaN fabrication process by placing a conductive material (e.g., metal), sometimes referred to herein as a field plate, as one or more conductive electrodes 112 on top of a patterned 2DEG or 2DHG channel 110. The field plate 112 can be connected to the low voltage ("source") terminal of the current source 108. A negative potential difference between the 2DEG or 2DHG channel 110 and the overlying field plate metal partially depletes the 2DEG or 2DHG channel 110, reducing the saturation current of the device compared to a resistor 116 that may be similarly configured without the field plate 112. The resulting current can be set by the nominal resistance of the patterned 2DEG or 2DHG channel 110. The operation of the current source 108 is similar to that of a depletion mode device, with the high voltage terminal of the resistor being the drain, the metal field plate being the gate, and the low voltage terminal being the degenerate source. Current source 108 can be implemented in an enhancement-mode only process without additional masks beyond those used to fabricate enhancement-mode devices. For start-up circuit applications that do not require high precision in the output current of the fabricated current source, fabrication of a current source having an output current of about 100 nA to about 10 pA is sufficient.
[0027] The block diagram in Figure 2 shows the AC voltage V LINE2 shows an example power converter system 200 that can be coupled to an AC power source (e.g., by plugging the converter or a device incorporating the converter into a wall outlet) to provide a buck converter. As an example, the power converter can be implemented as a pluggable power source for a mobile device (e.g., a cell phone or a tablet or notebook computer). The power converter system 200 includes an AC / DC power converter having a high voltage portion 202 that operates in a high voltage domain (e.g., about a 400V domain) and a low voltage portion 204 that operates in a low voltage domain (e.g., about a 5-6V domain). The high voltage portion 202 can be configured to implement the AC / DC conversion. A step-down IC 206, an auxiliary capacitor C AUX , and shunt resistor R SHUNT A step-down regulator, which may include a IN -GROUND is set to a lower voltage V OU The step-down IC 206 is configured to step down the voltage from the switch voltage node V SW and the drain voltage node V DD and a pulse width modulator (PWM) 210, which may include a start-up circuit 208 coupled between the FET Q S and controls a switch whose gate is coupled to the control output of PWM 210. AUX is the drain voltage node V DD and ground. Node V SW The switch voltage at S and hence the power switch Q S The drain of the node V SW FET Q S The source of is the shunt resistor R SHUNT The circuit elements in the high voltage domain 202 include a diode bridge rectifier 212, an electromagnetic interference (EMI) filter 214, and a freewheeling diode D FWand the output capacitor C OUT and the output inductor L OUT The diode bridge rectifier 212 may include a high voltage AC source V LINE , and also coupled between ground and an EMI filter 214. The EMI filter 214 is coupled between the diode bridge rectifier 212 and the input voltage node V IN The freewheeling diode D FW is the switch voltage node V SW and the input voltage node V IN The output capacitor C OUT is the voltage difference V between the negative and positive terminals of the output. OUT and the positive output terminal is coupled to the input voltage node V IN The output inductor L OUT is the negative terminal of the output and the switch voltage node V SW The output voltage V of the converter system 200 is coupled between OUT is the output capacitor C OUT The converter system 200 is connected to the output of the PWM 210 and the FET Q S and the gate of the FET Q S and a feedback connection 216 between the source of PWM 210 and the switching component PWM 210 and the FET Q S may require a certain minimum supply voltage in order to operate.
[0028] System start-up for power converter system 200 may occur, for example, when power converter system 200 is first plugged into a wall outlet or is otherwise powered by AC voltage V LINE This occurs when the power supply is first applied. The controller, including the PWM 210, starts its control operation by applying a startup circuit output node V DD The startup circuit may require the potential at the output node V to be at a sufficiently low voltage (e.g., 5V, 10V, or 15V). DDThe controller cannot operate and control start-up until the required voltage level is available at . An enhancement-mode FET is normally off when its gate-to-source voltage is zero, and at start-up, there is no powered controller available to control the FET to turn on, so the start-up circuit output node V DD It may not be possible to turn on an enhancement mode FET to generate the required voltage at
[0029] Using a depletion mode device that is normally on (the channel under the gate is partially or fully on at 0V gate bias), DD The auxiliary capacitor C is turned on until the voltage at AUX A small current I slowly charges SMALL can be provided at startup. SMALL is the AC voltage V LINE is provided (e.g., as long as the power converter system 200 is plugged into a wall outlet). To meet standby power requirements, a low power I SMALL The small current I SMALL The use of off-chip devices, such as discrete JFETs or other discrete depletion mode transistors on a PCB, to provide this increases the cost and complexity of the converter system 200. As described herein, a 2DEG-based current source can be provided as an on-chip device (on the step-down IC 206) that can be fabricated using the same process flow used to fabricate high voltage (e.g., 650V) e-GaN FETs.
[0030] 3A is a block diagram of an example startup circuit 300 that can be used to implement the startup circuit 208 of FIG. 2. The startup circuit 300 includes an enhancement mode transistor Q1, a current limiting transistor Q2, and a current limiting resistor R LIMa capacitor C1 and a voltage regulation circuit element including a comparator (e.g., an operational amplifier) 304 that provides a non-zero reference voltage V REF The positive input terminal has a drain voltage V DD is provided, which in turn produces a voltage regulated output voltage V SD The drain of enhancement mode transistor Q1 is coupled to the switch voltage node V, as is the positive terminal of precharge circuit 302. SW The gate of enhancement mode transistor Q1, the drain of current limiting transistor Q2, and the positive terminal of capacitor C1 are coupled to a voltage node V GS1 The drain of enhancement mode transistor Q1 is coupled to the negative terminal of precharge circuit 303 at S1 In the current limiting transistor Q2, the gate of the current limiting resistor R LIM The source of the current limiting transistor Q2 is coupled to the drain voltage node V DD In the negative terminal of the capacitor C1 and the current limiting resistor R LIM and the second terminal of the drain voltage node V DD 3. An auxiliary capacitor C of FIG. 3 is coupled between AUX The auxiliary capacitor C in Figure 2 AUX During system startup, the precharge circuit 302 switches the switch voltage node V SW Start-up circuit input voltage corresponding to node V SW The precharge circuit 302 provides a small constant current from node V at the gate of enhancement mode transistor Q1. GS1 2. The start-up circuit provides charge to turn on enhancement mode transistor Q1. The gate of enhancement mode transistor Q1 and the start-up circuit output node V DD A capacitor C1 is coupled between the
[0031] Start-up circuit input voltage V SWThe start-up circuit input voltage V at the input of the precharge circuit 302 increases even if the rate of increase in V is insufficient to turn on the enhancement mode transistor Q1. SW The rising slope of V turns on precharge circuit 302, which charges capacitor C1 and the gate-to-source capacitance of enhancement mode transistor Q1, which turns on enhancement mode transistor Q1. Figures 3B-3E illustrate example start-up voltage and current levels relative to each other and with respect to time. Figure 3B illustrates the voltage V at the start-up circuit input node V SW FIG. 3C shows the rise of node V at the output of precharge circuit 302 and at the gate of enhancement mode transistor Q1. GS1 FIG. 3D shows a slight rise in voltage at the start-up circuit input node V SW and the drain of the enhancement mode transistor Q1. D1 FIG. 3E shows the rise of the startup circuit input node V DD 4 shows a further increase in voltage at
[0032] A normally-off enhancement mode FET is not a suitable device for implementing the precharge circuit 302 because an enhancement mode device cannot be turned on at the start-up of the circuit when its gate-to-source voltage is zero. The precharge circuit 302 may be implemented as a normally-on depletion mode FET. However, enhancement mode transistors, such as enhancement mode FET Q1, as may be used to implement the precharge circuit 302, and depletion mode transistors cannot be easily integrated on the same wafer because enhancement mode and depletion mode transistors may require different masks in the fabrication process. Therefore, implementing the precharge circuit 302 as a depletion mode FET may require the depletion mode FET to be an off-chip component or the process used to fabricate the enhancement mode FET Q1 to be adapted to fabricate a depletion mode device, either of these options increasing the cost and complexity of fabricating the start-up circuit 300.
[0033] 4A and 4B are cross-sectional views of four exemplary devices fabricated according to an enhancement mode fabrication process. Thus, the four devices 402, 412, 418, 426 can be fabricated on the same wafer and die, and multiple instances of these devices can be fabricated and wired together to form electronic circuits. FIG. 4A is a cross-sectional view of an exemplary high voltage (e.g., 650V) enhancement mode FET 402 having a metal source terminal 404, a gate 406, and a metal drain terminal 408. The channel of the enhancement mode FET 402 accommodates a 2DEG 410 having a very thin sheet of electrons. The 2DEG is shown in FIGS. 4A and 4B as a horizontal dashed line. FIG. 4A further shows a depletion mode current source 426 having a first metal terminal 428 and a second metal terminal 430. The depletion mode current source 426 can use the same channel layer as that used by the enhancement mode FET 402. The channel of the depletion mode current source 426 accommodates the 2DEG 432. The depletion mode current source 426 includes field plates 434, 436, 438 that overlie the 2DEG 432. The field plates 434, 436, 438 allow the current source 426 to operate at high input voltages, for example up to 650V or 1000V. As an example, the middle field plate 436 can be completely over the lower field plate 434 and the upper field plate 438 can be completely over the middle field plate 436. FIG. 4B is a cross-section of a MIM nitride capacitor 412 having a first metal terminal 414 and a second metal terminal 416. The nitride capacitor 412 can be fabricated using the same process flow as the enhancement mode FET 402, with no additional materials or masks in some examples or with only one additional mask layer in other examples. 4B further shows a 2DEG-based resistor 418 having a first metal terminal 420 and a second metal terminal 422. The 2DEG-based resistor 418 may use the same channel layer as that used by the enhancement-mode FET 402. The channel of the resistor 418 accommodates a 2DEG 424.The 2DEG-based resistor 418 can be similarly fabricated using the same process flow as the enhancement-mode FET 402, without any additional materials or masks.
[0034] Since all four devices 402, 412, 418, 426 can use the same materials and masks, all four devices can be fabricated on the same wafer and die as part of a single IC. Several layers of materials can be used in forming the four different devices 402, 412, 418, 426 of Figures 4A and 4B. The epi structure used to create the 2DEGs 410, 424, 432 can include a first semiconductor layer 440, e.g., GaN, and a second semiconductor (barrier) layer 442, e.g., Aluminum Gallium Nitride (AlGaN), at the interface where the 2DEGs 410, 424, 432 are. A third layer 444 of doped semiconductor in the epi structure can be used to create the gate 406 of the FET 402. This doped semiconductor can be, e.g., GaN doped p-type (p-GaN). This third layer 444 can have a thin passivation layer 446, e.g., Silicon Nitride (SiN), provided thereon. At certain locations 448, the epi structure may be damaged or layer 442 may be etched into mesas to separate devices 402, 412, 418, 426 from one another and / or to create patterns in the epi structure, such as a serpentine pattern used in forming 2DEG resistor 418 and current source 426, as described below with respect to Figure 5. Alignment marks 450, 452 may be etched to allow a photolithography tool to align to the wafer.
[0035] The first dielectric layer 454 may be, for example, a nitride film. The first contact metal layer 456, which forms part of the ohmic contacts 404, 408, 420, 422, 428, 430 to the 2DEGs 410, 424, 432, may be a multi-layer structure including a layer of titanium (Ti), a layer of aluminum copper alloy (AlCu) above the Ti layer, a layer of titanium nitride (TiN) above the AlCu layer, and a layer of titanium tungsten alloy (TiW) above the AlCu layer. In the illustrated example, the first contact metal layer 456 forms the lower field plate 434 of the current source 426. The second dielectric layer 460 prevents the first contact metal layer 456 from shorting to the second contact metal layer 472. The second dielectric layer 460 may be, for example, a nitride film. A gate metal layer 464 forms the gate of the enhancement-mode FET 402 and may be, for example, TiW.
[0036] The third dielectric layer 468 may be, for example, a nitride layer. The first glue layer 472 may be, for example, TiW. The second contact metal layer 476 may be, for example, AlCu. In the illustrated example, the first glue layer 472 and the second contact metal layer 476 together form a bottom plate 478 of the MIM capacitor 412 and a middle field plate 436 of the current source 426. The fourth dielectric layer 480, which may form an interplate dielectric of the MIM capacitor 412, may be, for example, a nitride layer. The second glue layer 484 may be, for example, TiW. The third contact metal layer 486 may be, for example, AlCu. In the illustrated example, the second glue layer 484 and the third contact metal layer 486 together form a top plate 488 of the MIM capacitor 412.
[0037] The fifth dielectric layer 490 may be, for example, a PECVD nitride layer. The third adhesion layer 494 may be, for example, TiW. The fourth contact metal layer 496 may be, for example, AlCu. In the illustrated example, the third adhesion layer 494 and the fourth contact metal layer 496 together form the upper field plate 438 of the current source 426. The sixth dielectric layer 498 may form a protective overcoat (PO) and may include, for example, a lower oxide layer and an upper nitride layer. By way of example, the lower oxide layer may be silicon dioxide (SiO2) formed using tetraethyl orthosilicate (TEOS) and the upper nitride layer may be SiN. In another example, not shown, the PO may be omitted and the devices 402, 412, 418, 426 may be directly coated with one or more polyimide coatings or other polymer-based films.
[0038] During operation of any of the devices 402, 418, 426, a corresponding 2DEG 410, 424, 432 is formed just below the interface between the first semiconductor layer 440 and the second semiconductor layer 442 (e.g., just below the GaN / AlGaN interface if the first semiconductor layer 440 is a relatively thick layer of GaN and the second semiconductor layer 442 is a relatively thin layer of AlGaN). Although the 2DEGs 410, 424, 432 are shown with dashed lines, in three dimensions the 2DEG is an electron cloud with a carrier distribution in the vertical dimension with a centroid that may be, for example, about 2 nm below the interface between the first semiconductor layer 440 and the second semiconductor layer 442. In some examples, instead of having an epi structure that supports a 2DEG, the devices 402, 418, 426 can have an epi structure that supports (allows hole gas to exist in) a 2DHG, for example, by providing a second thick GaN layer on top of a thin AlGaN layer such that hole gas can form just above the upper GaN / AlGaN interface. References herein to a 2DEG should be understood to include a 2DHG in such alternative examples.
[0039] 5 is a top-down view of an example resistively configured 2DEG channel pattern 502 that can be used to confine all or a portion of the 2DEG 424 in resistor 418 or the 2DEG 432 in current source 426 (or the corresponding 2DHG in examples using 2DHGs as an alternative to 2DEGs). Although the resistively configured 2DEG pattern 502 is shown as being confined within the area shown in FIG. 5 and including a non-serpentine portion in region 510 and a regular serpentine portion in region 512, in other examples (not shown) it can take other two-dimensional forms that achieve the same resistive effect in which a voltage drop is observed between a first end 504 of the pattern 502 (e.g., at the bottom left of FIG. 5) and a second end 506 of the channel pattern 502 (e.g., at the top right of FIG. 5). Ohmic contacts can be made to the first end 504 and the second end 506 of the channel pattern 502 (e.g., by metal terminals 420 and 422, respectively, of resistor 418, or by metal terminals 428 and 430, respectively, of current source 426 in FIG. 4). Area 508, which covers all the area outside the channel 502, does not support the 2DEG, and defines the resistively configured 2DEG pattern 502, can be created, for example, by the same epi-damage or mesa-etch process used to create device isolation 448 shown in FIGS. 4A and 4B. As shown in FIG. 4, field plates 434, 436, 438 above resistively configured 2DEG 432 in current source 426 can be conductively connected to terminal 430, which can serve as the negative port (lower voltage port or "source" port) of current source 426. When a voltage is applied across terminals 428, 430 of current source 426, specifically in area 506, but anywhere other than under metal field plates 434, 436, 438, a negative control voltage less than the voltage at positive terminal 428 will partially deplete the 2DEG 432 in resistively configured 2DEG pattern 502 as long as there is current flow and a voltage drop across the resistively configured 2DEG pattern 502.This reduces the saturation current of the current source 426 and can be set, for example, by configuring the area of the patterned portion of the resistively configured 2DEG pattern 502. The field plates 434, 436, 438 therefore allow a high voltage (e.g., 650V or more) to be applied to one end (e.g., end 504) of the resistively configured 2DEG pattern 502 in the current source 426 (e.g., at terminal 428) while maintaining a very high resistance across the 2DEG pattern 502 in the current source 426. In current supply applications, the high precision threshold voltage V may be required in transistors, such as FET 402, used for power switching. th is not required for the power source 426.
[0040] In the illustrated example, area 504 in Figure 5 corresponds to the overlap of the lower electrode 434 in Figure 4A, area 506 in Figure 5 corresponds to the overlap of the middle electrode 436 in Figure 4A, and area 508 in Figure 5 corresponds to the overlap of the upper electrode 476 in Figure 4A. Vias 520 connecting the fourth contact metal layer 496 (used to form the upper electrode 438) to the second contact metal layer 476 (used to form the middle electrode 436) are represented by the 15 small squares in the lower right of Figure 5. Contacts connecting the second contact metal layer 476 to the first contact metal layer 456 (used to form the lower electrode 434), and thus to the 2DEG channel 432, are represented by the two rectangles in the lower left of Figure 5. The contacts represented by the four rectangles on the left side of FIG. 5 that connect the second contact metal layer 476 to the gate metal layer 464 can be used in instances where the gate metal layer 464 is used to form a bottom electrode 434 (not shown) or to form a contact-discontinuous gate electrode 1203 of a depletion mode transistor or current source as shown in FIG. 12.
[0041] The 2DEG (or 2DHG) channel in a GaN device is likened to an n-type (or p-type for 2DHG implementation) doped region in a silicon-based device in that the channel is conductive with limited conductivity or finite resistance (conductivity is defined as current divided by voltage, and resistance is defined as voltage divided by current). As described above with respect to the 2DEG (or 2DHG) resistor 418 shown in FIG. 4B, the 2DEG (or 2DHG) channel 424 can be used to form a resistor in GaN in the same way that an n-type (or p-type) doped region can be used to form a resistive channel in silicon. If a relatively large resistance (high voltage drop for low current) is desired, the 2DEG (or 2DHG) channel can be patterned to be relatively thin and long for the current path. Such a relatively elongated pattern 502 is shown in FIG. 5. The density of carriers (electrons in the 2DEG and n-type silicon, or holes or defect electrons in the 2DHG and p-type silicon), and therefore the resistance of the channel, can be modulated by an electric field perpendicular to the direction of current flow, forming a current source (e.g., depletion mode current source 426) or transistor (as described below with respect to Figures 11-15). The electric field can be generated, for example, from a 2DEG (e.g., 2DEG 432 in Figure 4A) and an overlying conductive electrode (e.g., contact continuous electrodes 434, 436, 438 in Figure 4A, which show examples of current sources, or contact non-continuous electrodes 1103, 1203, 1303, 1403, 1503, 1507 in Figures 11-15, which show examples of transistors or current sources). The strength of the electric field is defined as the potential difference between the 2DEG (or 2DHG) and the electrode divided by the distance between the 2DEG (or 2DHG) and the electrode. At least a minimum threshold field strength must be provided to modulate the carrier density and therefore the conductivity of the 2DEG (or 2DHG).
[0042] Since the minimum threshold electric field strength depends on the distance between the channel and the electrode, there may be different threshold voltages for electrodes at different separation distances (heights) relative to the 2DEG (or 2DHG) and may be implemented using different metal layers in the current source or transistor device fabrication process. For example, the threshold voltage for the top electrode 436 of the depletion mode current source 426 in FIG. 4A may be −200V. As an example, then, a high 400V potential at the end 506 of the channel 502 and a low potential of 0V at the electrode-contact connection end 504 of the channel 502 results in a −400V potential difference between the top electrode 436 and the underlying 2DEG 432. The electric field created by this potential difference reduces the electron density in the 2DEG 432, increasing its resistance. The reduction in electron density settles into an equilibrium state where the resistance of the 2DEG 432, modulated by the top electrode 436, is high enough that the current flowing through the 2DEG 432 causes a voltage drop of 200V across the 2DEG 432. This resistance modulation is most visible in the region of the 2DEG 432 where there is no intervening conductive layer between the top electrode 436 and the 2DEG 432 and where the top electrode 436 is above and closest to the 2DEG 432. In the view and orientation of FIG. 5, this particular region of resistance modulation is located above the region labeled 510. Thus, in this example, a drop of at least 200V, from 400V to 200V, occurs between the end 506 of the patterned portion of the channel 502 and a point short of the top edge of region 506 (corresponding to the overlap of the middle electrodes in FIG. 4A). For portions of the 2DEG where the potential is below 200V, the top electrode 438 (formed by the fourth contact metal layer 496) has substantially no effect on the electrons in the 2DEG.
[0043] The middle electrode 436, shown in the diagram of FIG. 5 as area 506, affects the channel 502 that is essentially restricted to the region between the top of area 506 and the top of area 504. Between these points in the channel 502, the potential of the 2DEG drops from 200V, for example to 80V. The bottom electrode 434, shown in the diagram of FIG. 5 as area 506, shown in FIG. 4A as being made from the first contact metal layer 456, and in some examples, such as that of FIG. 12, can be made from the gate metal layer 464, affects the channel 502 that is essentially restricted to the upper portion of the region 510, where the potential of the underlying 2DEG drops from 80V, for example to 2V. Thus, after a very short distance of the length of the channel 502 below the bottom electrode 434, a maximum of about 100 nm total in the illustrated example, the potential of the 2DEG 432 below the bottom metal electrode 434 is less than the threshold voltage of the bottom electrode 434. From an electrical point of view, it is irrelevant to the operation of the depletion mode current source 426 (or, for example, the depletion mode transistor illustrated in any of the Figures 11-15) whether, on the one hand, the metal of the electrodes 434, 436, 438 extends across the serpentine portion 512 of the patterned area of the 2DEG, or, on the other hand, whether the metal of the electrodes 434, 436, 438 extends only within the region 510, because there is no effect of the electrodes 434, 436, 438 on the channel 502 in the portion between the region 510 and the channel end 504, and the voltage difference between the 2DEG and the gate electrode is small in the portion of the channel 502 between the region 510 and the channel end 504. Within the serpentine patterned region 512, the voltage of the channel 502 can further drop, for example, from 2V to the 0V potential of the end 504 of the channel 502, which is electrically coupled to the 0V contact 430.
[0044] The serpentine patterning in region 512 adds resistance to channel 502 using only a relatively small, near-square area. The resistance of the serpentine patterned region 512 sets the current of current source 426. If the voltage of the 2DEG in region 510 is 2V, the current generated by current source 426 is equal to the 2V drop across the serpentine patterned portion 512 of channel 502 divided by the resistance of the serpentine patterned portion 512 of channel 502. This current is substantially independent of the voltage between channel ends 506 and 504 (between contacts 428 and 430 of current source 426), provided the potential difference between contacts 428 and 430 of current source 426 exceeds 2V. The inclusion of one or more metal electrodes (e.g., electrodes 434, 436, 438 in FIG. 4A or electrodes 1103, 1203, 1303, 1403, 1507 in FIGS. 11-15, etc.) having a potential of about 0 V above the patterned area of the 2DEG (or 2DHG) channel near the high voltage terminal (contact 428 in FIG. 4A) provides the above-mentioned channel depletion functionality and utility of the device as a depletion mode current source (or, in some examples, a depletion mode transistor).
[0045] Thus, to provide the intended depletion mode current source or transistor functionality, one or more metal electrodes of the current source or transistor may be made to overlap at least a region 510 near the channel end 506, regardless of whether the metal electrodes are over the serpentine patterned region 512. Conductively coupling one or more metal electrodes to the channel end 504 (e.g., at contact 430 in FIG. 4A) can cause the depletion mode device to act as a current source rather than a transistor. For example, in a depletion mode device configured as a two-terminal current source having a first contact configured at a first voltage and a second contact configured at a second voltage higher than the first voltage (e.g., at least 400V higher than the first voltage, e.g., at least 600V higher than the first voltage), the electrode conductively coupled to the first contact may be configured to overlap a 2DEG or 2DHG channel running between the first and second contacts in at least an area closer to the second (high voltage) contact than the first (low voltage) contact. As another example, in a depletion mode device configured as a two-terminal current source having a first contact set at a first voltage and a second contact set at a second voltage higher than the first voltage (e.g., at least 400V higher than the first voltage, e.g., at least 600V higher than the first voltage), an electrode conductively coupled to the first contact can be configured to be above a point on a 2DEG or 2DHG channel running between the first and second contacts (this point may, for example, correspond to any point on channel 502 in region 510 of FIG. 5 ) such that a first distance along the 2DEG or 2DHG channel between that point (in region 510 in FIG. 5 ) and the first (low voltage) contact (e.g., corresponding to channel end 504 of FIG. 5 ) is greater than a second distance along the 2DEG or 2DHG channel between that point (in region 510 of FIG. 5 ) and the second (high voltage) contact (corresponding to channel end 506 of FIG. 5 ).
[0046] The graph of FIG. 6 shows current-voltage (IV) plots for four different example current sources fabricated like the depletion mode current source 426, with various configurations of field plates. The IV plot 602 corresponds to a device without a field plate, structurally similar to the 2DEG resistor 418 of FIG. 4, that does not exhibit the current saturation behavior desired from a current source device. The IV plot 604 represents a depletion mode current source with only the top and middle field plates 438, 436, where the first contact metal layer 456 is not configured as a field plate and does not substantially overlie the 2DEG 432. As shown in FIG. 6, the device represented by the IV plot 604 exhibits a current saturation of about 12 μA at a potential difference of more than about 15 V. The IV plot 606 represents a depletion mode current source with the top and middle field plates 438, 436, but where the lower field plate is higher than that shown in FIG. 4 and is formed by the gate metal layer 464 rather than by the first contact metal layer 456. As shown in Figure 6, the device represented by IV plot 606 exhibits a current saturation of about 8 μA at a potential difference of greater than about 10 V. IV plot 608 represents a depletion mode current source having upper, middle, and lower field plates 438, 436, 434 substantially as shown in Figure 4. As shown in Figure 6, the device represented by IV plot 608 exhibits a current saturation of about 4 μA at a potential difference of greater than about 5 V. According to Figure 6, generally, the lower the metal is relative to the field plates relative to the 2DEG, the lower the saturation current and the lower the potential difference at which the saturation current is reached.
[0047] The graph of FIG. 7 shows IV plots for two different example current sources fabricated like the depletion mode current source 426 and with various 2DEG resistances. IV plot 702 corresponds to a depletion mode current source having upper, middle, and lower field plates 438, 426, 434 substantially as shown in FIG. 4 and with a 2DEG pattern configured to provide a resistance of 500 kΩ. As shown in FIG. 7, the device represented by IV plot 702 exhibits a current saturation of about 8.5 μA at a potential difference of greater than about 10 V. IV plot 704, which is identical to IV plot 608 of FIG. 6, corresponds to a depletion mode current source having upper, middle, and lower field plates 438, 426, 434 substantially as shown in FIG. 4 and with a 2DEG pattern configured to provide a resistance of 1 MΩ. As shown in FIG. 7, the device represented by IV plot 704 exhibits a current saturation of about 4 μA at a potential difference of greater than about 5 V. According to FIG. 7, generally, the higher the resistance to which the 2DEG pattern is constructed, the lower the saturation current.
[0048] Figures 8A-8Z collectively illustrate an example fabrication process by which the enhancement-mode FET 402, MIM capacitor 412, 2DEG resistor 418, and depletion-mode current source 426 of Figures 4A-4B may be fabricated. The example fabrication process uses 14 mask layers, as shown in Table 1. [Table 1]
[0049] 8A and 8B show a 2DEG supporting epi structure that can be grown or deposited on a substrate (not shown), which can be, for example, a silicon substrate. A thick buffer layer (also not shown) can be grown or deposited to allow for the growth of high quality material on top of the silicon substrate. The buffer layer can be composed of, for example, aluminum nitride layers and various compositions of AlGaN, or superlattice structures starting with aluminum nitride (AlN) and having a repeating periodic structure of GaN and AlN or AlGaN and AlN, or a combination of the two.
[0050] Above the buffer layer, a 2DEG supporting epi structure is grown or deposited, including at least a first semiconductor layer 440, e.g., GaN, and a second semiconductor (barrier) layer 442, e.g., AlGaN. The thickness of the AlGaN layer 442 and the percentage of aluminum in the AlGaN layer can be selected to obtain a positive threshold voltage of the enhancement mode FET 402, taking into account the selected thickness of the provided p-type (e.g., p-GaN) layer 444. The AlGaN layer 442 can be a single AlGaN layer, or a combination of a thin AlN layer and an AlGaN layer, where the AlN layer can be below or above the AlGaN layer. In some examples, the AlGaN layer 442 can also be replaced with a layer of aluminum indium gallium nitride (AlInGaN), aluminum indium nitride (AlInN), or AlN. The GaN layer 440 can also contain aluminum. The epi structure may be 2DEG supporting, with no p-type layer 444 (eg, p-GaN) on top, and with different materials having different compositions between the lower layer 440 and the layer 442 above it.
[0051] A doped semiconductor layer 444, for example p-GaN, p-type indium gallium nitride (InGaN), or a combination of InGaN and AlN layers, can be grown or deposited on top of the 2DEG supporting epi structure. The doped semiconductor layer 444 can be, for example, magnesium (Mg) at a concentration of about 1×10 17 ions / cm 3 ~Approx. 1×10 20 ions / cm 3 , for example, about 1 x 10 19 ions / cm 3 ~Approx. 3×10 19 ions / cm 3 The p-type doped semiconductor layer 444 may be doped with a dopant concentration of about 100 Angstroms to 3,000 Angstroms, for example, a thickness of about 700 Angstroms. The p-type doped semiconductor layer 444 depletes the channel and prevents the formation of a 2DEG channel where the doped semiconductor layer 444 is present.
[0052] A very thin passivation layer 446 of any dielectric material, for example SiN, can be applied to protect the surface. The passivation layer 446 can have a thickness of, for example, 100 Angstroms. The passivation layer 446 can be applied, for example, by low pressure chemical vapor deposition (LPCVD). A first photolithography mask POLE0 can be applied and etched to create alignment marks, such as mark 452, for the photolithography tool to align to the wafer.
[0053] Figures 8C and 8D show the structures of Figures 8A and 8B following the application of a second photolithography mask, PBGT, which is used to form an enhancement-mode FET gate at location 406 by removing (e.g., dry etching) doped layer 444 and passivation layer 446 everywhere except at gate location 406. A 2DEG channel 447, shown in dashed lines in Figures 8C and 8D, is then formed everywhere where doped layer 444 (e.g., p-GaN) was removed.
[0054] 8E and 8F show the structure of FIGS. 8C and 8D following application of a third photolithography mask ISO1, which is used for damage implantation to separate the devices by separating the connected 2DEG 447 into regions 410, 424, 432. The damage implantation can be used to damage layers 440 and / or 442 in areas where it is not desired that a 2DEG be present. The damage implantation can be, for example, argon (Ar), implanted with sufficient surface flux and energy to create damage to disable the 2DEG 447 in the damaged area. Other ion species that can be used for damage implantation include magnesium (Mg) and fluorine (F). This damage process can also be used to pattern 2DEG channels for resistor and current source devices 418, 426 in pattern-defined regions 508 outside the serpentine pattern 502, as shown in FIG. 5, by damaging to preclude the formation of 2DEG in these regions 508. In another example (not shown), instead of using a damage injection process, a mesa etch can be performed to etch away layer 442 and / or layer 440 (e.g., the AlGaN and GaN layers) sufficient to prevent the formation of a 2DEG in the etched regions.
[0055] 8G and 8H show the structure of FIG. 8E and 8F following the growth or deposition of a passivation film 454 necessary to prevent any trap surface effects, application of a fourth photolithography mask CONTS for source and drain contacts, and etching, except in the areas where the source / drain contacts 408, 404, 2DEG resistor head contacts 420, 422, and current source head contacts 428, 430 will be formed. The passivation layer 454 can be, for example, SiN, SiO2, silicon oxynitride (SiON), AlN, a combination of any of these dielectrics, or any dielectric film that can potentially passivate surface traps and also prevent leakage through the film 454 after placement of the terminals 404, 408, 420, 422, 428, 430. As an example, the passivation film 454 can be a nitride deposited using LPCVD to a thickness of about 1,525 angstroms. The source and drain contact mask CONTSD can be patterned to create opening windows, e.g., windows 802 and 804, at locations where ohmic contacts to the 2DEG will later be formed. Although the etching process is shown in the cross-sections of Figures 8G and 8H as having portions of layer 442 (e.g., AlGaN) etched away, depending on the metal process used to form the ohmic contacts, in some examples layer 442 can be retained and only passivation layer 454 (e.g., SiN) is etched away for later formation of the ohmic contacts. In still other examples, an ohmic contact is made to the 2DEG by etching through layer 442. In some examples, the resistance of 2DEG resistor 418 and current source 426 can be adjusted by selection of ohmic contact formation parameters. In some examples, a depletion mode device, such as current source 426, uses only a partial etch into layer 442 for an ohmic contact, and an enhancement mode device, such as FET 402, etches only to the edge of layer 442 without needing to etch into layer 442 for an ohmic contact.
[0056] Figures 8I and 8J show the structure of Figures 8G and 8H following metal deposition, a fifth photolithographic mask METSD, etching, and an ohmic anneal used to form metal layer 456. Metal layer 456 provides metal contact to the source 404 and drain 408 of enhancement mode transistor 402, the 2DEG resistor heads 420, 422, and the current source heads 428, 430 and lower field plate 434 of current source 426. The first contact metal layer 456 forms part of the ohmic contacts 404, 408, 420, 422, 428, 430 to the 2DEGs 410, 424, 432 and may be titanium-based and aluminum-based contacts. For example, the first contact metal layer 456 may be a multilayer including a layer of titanium (Ti) having a thickness of, for example, about 300 Å to about 500 Å, e.g., about 400 Å, a layer of aluminum copper alloy (AlCu) over the Ti layer having a thickness of, for example, about 750 Å to about 1,250 Å, e.g., about 1,000 Å, a layer of titanium nitride (TiN) over the AlCu layer having a thickness of, for example, about 400 Å to about 600 Å, e.g., about 500 Å, and a layer of TiW having a thickness of, for example, about 1,000 Å. The first metal contact layer 456 may be formed of any of several other metals, including TiAl, tantalum (Ta), nickel (Ni), or gold (Au), in other examples. In some examples, the lower surface of lower field plate 434 can be about 1,500 angstroms to about 1,700 angstroms above 2DEG channel 424 .
[0057] A fifth photolithographic mask and subsequent etch can remove the ohmic contact layer 456 except where it is desired to provide contact to the 2DEG and field plate 434. The etch may be followed by an ohmic anneal.
[0058] 8K and 8L show the structure of FIG. 8I and 8J following growth or deposition of a second dielectric layer 460, application of a sixth photolithography mask PCONT, etching, and annealing. The second dielectric layer 460 can be, for example, a thin UV nitride stack deposited using PECVD, for example, to a thickness of about 740 Angstroms. The second dielectric layer 460 can be, for example, SiN, or in other examples, SiO2, SiON, aluminum oxide (Al2O3), or combinations thereof. The second dielectric layer 460 prevents the ohmic contact layer 456 from shorting to a subsequently deposited metal layer. The sixth photolithography mask can be allowed to be etched down to the FET gate doped semiconductor (e.g., p-GaN) layer 444, thereby creating a window that allows the gate metal layer 464 to contact the doped semiconductor layer 444 in a subsequent metal layer deposition.
[0059] After opening the contact at the FET gate location 406, an activation anneal can be performed. In the example using p-GaN as the FET gate doped semiconductor layer 444, depending on the film deposition conditions of the SiN films 454, 460, the p-GaN can become inactive. Once the window to the p-GaN is opened, an activation anneal is performed to reactivate the p-GaN. The term "activation" is used differently here than with respect to doped silicon based device fabrication processes, and the activation anneal is used to ensure that the activated implanted seed material is incorporated into the semiconductor crystal structure. Here, the magnesium in the p-GaN is already in the crystal structure of the p-GaN layer 444, but there may be some hydrogen complexes in the p-GaN layer 444 that may make the p-GaN layer 444 electrically inactive. Rather than forming holes that are required for the conductive p-GaN material, the presence of hydrogen from the growth of the p-GaN layer 444 or the deposition of the SiN layers 454, 460 may make the p-GaN layer 444 inactive. Therefore, an activation anneal can be performed to decompose the hydrogen magnesium complexes.
[0060] Figures 8M and 8N show the structures of Figures 8K and 8L, followed by gate metal layer 464, which can be Ti, TiN, Ni, or Al. Gate metal layer 464 can be a multi-stack composed of different materials in some examples. A seventh photolithography mask, and accompanying etch, removes gate metal layer 464 except over FET gate locations 406.
[0061] 8O and 8P show the structure of FIG. 8K and 8L following the growth or deposition of a third dielectric layer 468, the application of an eighth photolithography mask CONT, and an etch. The third dielectric layer 468 may be, for example, a nitride film deposited using PECVD. The third dielectric layer 468 may be, for example, p-SiN deposited to a thickness of about 1,700 Angstroms. The eighth photolithography mask and associated etch may be used to open windows, such as windows 806 and 808, through the third dielectric layer 468 to the previously deposited ohmic contacts, to the source 404 and drain 408 of the enhancement mode transistor 402, to the 2DEG resistor heads 420, 422, and to the current source heads 428, 430 (as shown in FIG. 4A and FIG. 4B). Conductive contacts to the ohmic contacts may later be formed at the locations of these windows. Although not shown in FIG. 8O, the FET gate 464 may also be opened, by the same etch, to a contact in the area where the FET gate 464 connects at the finger edges.
[0062] 8Q and 8R show the structure of FIGS. 8O and 8P following deposition of a first adhesion layer 472, deposition of a second contact metal layer 476, and growth or deposition of a fourth dielectric layer 480. The first adhesion layer 472 can be, for example, TiW deposited to a thickness of about 1100 Angstroms. The second contact metal layer 476 can be, for example, AlCu deposited to a thickness of about 3,000 Angstroms. The second metal contact layer 476 can be copper (Cu) or Au in other examples. A very thin layer of TiN (not shown) can be disposed on top of the second contact metal layer 476, for example to a thickness of about 300 Angstroms. In some examples, this TiN layer can be omitted or eliminated due to subsequent deposition or etching of, for example, a fourth dielectric layer 480, as shown in, for example, FIGS. 8U and 8V. A thin fourth dielectric layer 480 may be grown or deposited on top of the metal layers 472, 476 to form a dielectric for the MIM capacitor 412. This fourth dielectric layer 480 may be, for example, a nitride layer deposited using PECVD to a thickness of about 740 angstroms. In some examples, the lower surface of the middle field plate 436 may be about 3,900 angstroms to about 4,200 angstroms above the 2DEG channel 424.
[0063] 8S and 8T show the structure of FIG. 8Q and 8R following deposition of a second adhesion layer 484, deposition of a third contact metal layer 486, application of a ninth photolithography mask CAPD, and etching. The second adhesion layer 484 can be, for example, TiW deposited to a thickness of about 500 Angstroms. The third contact metal layer 486 can be, for example, AlCu deposited to a thickness of about 1,000 Angstroms. The third metal contact layer 486 can be Cu or Au in other examples. The second adhesion layer 484 and the third contact metal layer 486 together form the upper plate 488 of the MIM capacitor 412. Another very thin layer of TiN (not shown) can be placed on top of the third contact metal layer 486, for example to a thickness of about 300 Angstroms. In some examples, this TiN layer may be omitted or may be eliminated due to subsequent etching, for example as shown in Figures 8U and 8V. A ninth photolithographic mask is used to pattern and etch metal layers 484, 486 to form a capacitor top plate 488.
[0064] Figures 8U and 8V show the structure of Figures 8S and 8T following the application and etching of a tenth photolithographic mask MET1. Metal layers 472, 476 and fourth dielectric layer 480 are selectively etched away to form the middle field plate 436 of current source 426 and the lower plate 478 of capacitor 412, as well as to continue the ohmic connections to FET 402, resistor 418, and current source 426.
[0065] 8W and 8X show the structure of FIGS. 8U and 8V following the growth or deposition of a fifth dielectric layer 490, application of an eleventh photolithography mask VIA1, and etching. The fifth dielectric layer 490 may be, for example, a nitride deposited using PECVD to a thickness of about 5,500 angstroms. In the area where the upper capacitor plate 488 is present, the fifth dielectric layer 490 resides above the upper capacitor plate 488. An eleventh photolithography mask may be used to pattern and open windows, for example, windows 810 and 812, to contacts to a metal layer, which may then extend vertically through the fifth dielectric layer 490.
[0066] 8Y and 8Z show the structure of FIG. 8W and 8X following deposition of a third adhesion layer 494, deposition of a fourth contact metal layer 496, application of a twelfth photolithography mask MET2, and etching. The third adhesion layer 494 can be, for example, TiW deposited to a thickness of about 3,000 Angstroms. The fourth contact metal layer 496 can be, for example, AlCu deposited to a thickness of about 30.4 kAngstroms. The fourth contact metal layer 496 can be Cu or Au in other examples. The third adhesion layer 494 and the fourth contact metal layer 496 together form the upper field plate 438 of the current source 426. Another very thin layer of TiN (not shown) can be disposed on top of the third contact metal layer 486 to a thickness of about 300 Angstroms. In some examples, this TiN layer may be omitted or may be eliminated due to a subsequent etch. A twelfth photolithographic mask is used to pattern and etch metal layers 494, 496 to form terminals 404, 408, 420, 422, 428, 430 and upper field plate 438. In some examples, the lower surface of upper field plate 438 may be above 2DEG channel 424 by about 9 k Angstroms to about 10 k Angstroms, for example, about 9,450 Angstroms to about 9,650 Angstroms.
[0067] 4A and 4B show the structure of FIG. 8Y and FIG. 8Z following the growth or deposition of a sixth dielectric layer 498. The sixth dielectric layer 498 may be a PO film, which may include an oxide (e.g., SiO2), a nitride (e.g., SiN), an oxynitride (e.g., SiON), or a combination thereof. As an example, the sixth dielectric layer 498 may include a lower oxide layer and an upper nitride layer. In some examples, the lower oxide layer may be a layer of SiO2 fabricated with TEOS deposited using PECVD to a thickness of about 10 kA, and the upper nitride layer may be SiN deposited using PECVD to a thickness of about 20 kA. In some examples, the lower oxide layer of the sixth dielectric layer 498 may be followed by sintering, application of a thirteenth photolithography mask TRENCH, and etching to form trenches for metal wire layers for connecting devices. In some examples, deposition of the upper nitride layer of the sixth dielectric layer 498 may be followed by application of a fourteenth photolithography mask POR and etching to form pad openings in the PO to open windows for accessing contacts to the terminals 404, 408, 420, 422, 428, 430. This may then be followed by extended sintering, for example at about 400° C. for about 6 hours and an additional 6 hours. In other examples, not shown, the PO may be omitted and the devices 402, 412, 418, 426 may be directly coated with one or more polyimide coatings or other polymer-based films. Table 2 summarizes example materials and thicknesses for the various layers shown in FIGS. 4A and 4B. [Table 2]
[0068] The flow chart of FIG. 9 illustrates an example method 900 for fabricating a depletion mode current source on an IC, such as the current source 426 shown in FIG. 4A. A first semiconductor layer and a second semiconductor layer are grown or deposited on a substrate (902). The first semiconductor layer may correspond, for example, to layer 440 in FIG. 4A and FIG. 8A and may be, for example, GaN. The second semiconductor layer over the first semiconductor layer may correspond, for example, to layer 442 in FIG. 4A and FIG. 8A and may be, for example, AlGaN. The first and second semiconductor layers are configured to support a 2DEG or 2DHG. One or both of the first and / or second semiconductor layers are implanted (e.g., by damage implant) or etched (e.g., by mesa etch) to pattern a 2DEG or 2DHG channel over the area of the channel (904). An example isolation damage implant is shown in FIG. 8E to create isolation region 448. The pattern may be, for example, a serpentine pattern, shown as example channel pattern 502 in FIG.
[0069] A first dielectric layer is grown or deposited (906) on the second semiconductor layer. The first dielectric layer may correspond to layer 454 in FIG. 4A and FIG. 8G and may be, for example, SiN. The first dielectric layer is photolithographically etched (908) to open first and second ohmic contact windows to the 2DEG or 2DHG channel, for example at the respective ends of the channel. Exemplary windows 802 and 805 are shown in FIG. 8G. At the locations of the ohmic contact windows, one or more ohmic contact layers are deposited and etched (910) to form, for example, first and second ohmic contacts to the respective ends of the 2DEG or 2DHG channel. At least one of the first and second ohmic contacts includes a conductive electrode (e.g., a field plate), for example the lower field plate 434 in FIG. 4A, above the first dielectric layer. In some instances, the conductive electrode may extend over at least a portion of the resistive patterned area of the 2DEG or 2DHG, and in some instances, may extend over the entire resistive patterned area of the 2DEG or 2DHG. As an example, the lower surface of the conductive electrode may be about 1,500 Angstroms to about 1,700 Angstroms above the 2DEG or 2DHG channel. The conductive electrode may be the only field plate fabricated in the current source device, or may be the lower field plate in a current source device structure having multiple field plates where method 900 continues to fabricate one or more additional field plates.
[0070] The method 900 of FIG. 9 may continue with the growth or deposition of a second dielectric layer over the first dielectric layer, the first and second ohmic contacts, and the second conductive electrode, which may be an intermediate field plate (912). This second dielectric layer may be, for example, either the dielectric layer 460 or the dielectric layer 468 shown in FIG. 4A, FIG. 8K, and FIG. 8O, or a combination of the two, and may be, for example, SiN or p-SiN, or a combination of the two. The second dielectric layer is etched (914) to open first and second conductive contact windows, for example windows 806, 808 in FIG. 8O, through the second dielectric layer to the first and second ohmic contacts. A metal layer is deposited and etched (916) to form first and second conductive contacts to the first and second ohmic contacts, respectively, and a second conductive electrode extending from one of the first or second conductive contacts above the second dielectric layer onto the lower conductive electrode. The deposition may be as shown in Figure 8Q. The etching may be as shown in Figure 8U. The second field plate may be, for example, the middle field plate 436 shown in Figure 4A. The method 900 may end at this point or may continue to create a third, upper field plate.
[0071] The method 900 of FIG. 9 may continue with the growth or deposition of a third dielectric layer over the second dielectric layer, the first and second conductive contacts, and the second conductive electrode. The third dielectric layer may be, for example, layer 590 as shown in FIG. 4A and FIG. 8W. The third dielectric layer is etched (920) to open first and second windows, for example windows 810 and 812 in FIG. 8W, through the third dielectric layer to the first and second conductive contacts. An upper metal layer is deposited and etched (922) to form a third conductive electrode (e.g., an upper field plate) that extends the first and second conductive contacts up through the first and second windows through the third dielectric layer and from one of the first or second conductive contacts over the middle field plate above the third dielectric layer. The result of the deposition and etching may be as shown in FIG. 8Y, resulting in current source terminals 428, 430 and upper field plate 438. Method 900 may end at this point or may continue to add PO dielectric layers, wire layers, and / or additional field plates (not shown in FIG. 9). Method 900 may also include various other etches, anneals, and / or sintering (also not shown in FIG. 9). Method 900 may also include additional metal layers to create additional, more highly conductive electrodes (e.g., field plates). Method 900 may also include providing traces on one or more metal layers to couple to the terminals of the depletion mode current source, for example, as shown in FIGS. 10-15. For example, method 900 may include a first trace to a first ohmic contact and a second trace to a second ohmic contact to provide the depletion mode current source as a two-terminal device.
[0072] The devices of FIGS. 4A and 4B may be wired together using wiring layers to form a startup circuit such as startup circuit 300 of FIG. 3A, in which depletion mode current source 426 of FIG. 4A is used to implement precharge circuit 302 of FIG. 3A, enhancement mode FET 402 of FIG. 4A is used to implement transistors Q1 and Q2 of FIG. 3A, MIM capacitor 412 of FIG. 4B is used to implement capacitor C1 of FIG. 3A, and 2DEG resistor 418 of FIG. 4B is used to implement current-limiting resistor R LIM The drain terminal 408 of the FET Q1 may be coupled by wiring to the positive terminal 428 of the current source in one or more wiring layers (not shown), such coupling being represented by node V in FIG. SW The gate of the FET Q1 may be coupled to the negative terminal 430 of the current source, such coupling forming an electrical node corresponding to node V in FIG. GS1 A first (e.g., positive) terminal 414 of the capacitor C1 similarly forms an electrical node corresponding to node V GS1 A second (e.g., negative) terminal 416 of the capacitor C1 may be coupled to the 2DEG resistor R LIM The 2DEG resistor R LIM The first terminal 420 of the S1 FET Q2 of FIG. 3A can be fabricated on the same die using the same process used to create FET Q1 and can be connected using wiring layers similar to those shown in accordance with the diagram of FIG. 3A, with the drain terminal of FET Q2 being coupled to the drain terminal 404 of FET Q1 at an electrical node corresponding to node V. GS1 and the source terminal of FET Q2 is coupled to the node V DD 2DEG resistor R LIM and the second terminal 416 of the capacitor C1, and the gate of the FET Q2 is coupled to the node V S1 The source terminal 404 of the FET Q1 and the 2DEG resistor R LIM and terminal 420.
[0073] The process shown in Figures 8A-8Z and 4A-4B allows the depletion mode current source 426 to be fabricated together with the high voltage p-GaN FET 402, the 2DEG resistor 418, and the MIM nitride capacitor 412 on a single substrate, and allows the depletion mode current source 426 to be used as a pre-charge circuit to build a start-up circuit without requiring different masks or materials to form the depletion mode current source 426 and the high voltage p-GaN FET 402. The 2DEG resistor 418 does not need to use a thin metal film such as silicon chrome (SiChrome, SiCr) or nickel chrome (NiChrome, NiCr) as a resistive element. Providing a lower field plate (spatially closer to the 2DEG 432) in the depletion mode current source 426 results in a lower saturation current of the current source 426, about 5 μA in some examples, and does not break down up to 1,000 V over a temperature range of about -40°C to 125°C. At higher temperatures, a lower saturation current may be expected for the current source 426. In some examples, the upper field plate 438 may be omitted from the depletion mode current source 426. In some examples, the middle field plate 436 and the upper field plate 438 may be omitted from the depletion mode current source 426. The proximity of the bottom field plate (e.g., field plate 434) to the channel of the 2DEG 432 may be determined by the current requirements of the start-up circuit 300. The width of the 2DEG current source 426 may also be adjusted to tune the current density out of the current source 426, and the parameters may have an approximately linear relationship to each other.
[0074] In contrast to the coverage that the field plates 434, 436, 438 of the depletion mode current source 426 provide above the channel of the 2DEG 432 in a top-down view, the metal layer of the 2DEG resistor 418 does not extend over the resistive patterned area of the 2DEG channel. The coverage of the field plates in the current source 426 may allow the potential difference between the lower voltage field plate and the highest voltage portion of the patterned 2DEG channel below the field plate to be greater than the pinch-off voltage of the capacitor formed with the field plate above the 2DEG channel. The extension of the coverage of the field plates in the current source 426 may allow pinch-off to occur before the intrinsic saturation of the semiconducting 2DEG channel. As an example, if the pinch-off voltage is 20V, the field plate coverage (in a top-down view) covers more of the field plate length than is necessary to drop 20V across the patterned 2DEG channel, and the extension of the field plate coverage is shorter than the patterned channel required to drop 20V, the clamping effect and resulting current saturation will be weaker or not observed. The widths (in the left-right dimension) of the metal layers of 2DEG resistor 418 are exaggerated in FIG. 4B for purposes of illustrating the design rule that the upper metal layers (including layer 494) are wider in resistor 418 than the lower metal layers (including layer 472). FIG. 4B should not be interpreted as showing a field plate in resistor 418 or as showing the metal layers extending over the resistive patterned areas of the 2DEG channel in resistor 418.
[0075] FIG. 10 illustrates an example depletion mode current source 1000, which may correspond to the depletion mode current source 426 of FIG. 4A, with wires 1008, 1010 fabricated to conductively couple the terminals of the depletion mode current source 1000. The drain region 1002 in FIG. 10 is conductively coupled to the wire 1008 via what corresponds to the first metal terminal 428 of FIG. 4A. The source region 1004 in FIG. 10 is conductively coupled to the wire 1010 via what corresponds to the second metal terminal 430 of FIG. 4A. The drain region 1002 and the source region 1004 are electrically coupled through a resistive 2DEG channel 1006, which may be a 2DHG channel in some examples, as described above. The field plates 1016, 1018 in FIG. 10 may correspond to the field plates 436, 438, respectively, of FIG. 4A. In some examples, the upper dielectric layer 1020, corresponding to the sixth dielectric layer 498 of FIG. 4A, may be opened to provide conductive coupling of the regions 1002, 1004 to the traces 1008, 1010, which may be laid down on an additional metal layer. In other examples (not shown), the traces 1008, 1010 may be provided on one of the metal layers already present in the current source device 1000, such as the second contact metal layer 476 used to fabricate the field plate 1016, or the fourth contact metal layer 496 used to fabricate the field plate 1018. For example, such traces may be located in front of or behind the drawing plane of FIG.
[0076] 10 is a two-terminal device. As an example, in the example start-up circuit 300 of FIG. 3A, the depletion mode current source 1000 is used as part or all of the pre-charge circuit 302, with line 1008 connected to the switch voltage node V SW 10, and the line 1010 is coupled to the node V GS1 The depletion mode current source 1000 can be coupled to approximately I=V th / R source Acts as a current source to provide a current at V th is the threshold voltage of the depletion mode current source 1000, and R sourceis the source feedback resistance of the 2DEG or 2DHG channel between the drain region 1002 and the source region 1004, and is only very weakly dependent on the voltage applied across the depletion mode current source 1000. The source feedback resistance may be provided primarily by resistive patterning of the 2DEG or 2DHG channel in the area labeled 1005. That is, in such an example, only in the area 1005 of FIG. 10, the 2DEG or 2DHG channel 1006 is resistively patterned, similar to area 512 in FIG. 5. FIG. 10 is not necessarily to scale, and area 1005 may actually be longer than other illustrated features shown in FIG. 10 and related figures. Higher level field plates 1016, 1018 may be provided to obtain the desired voltage rating of the depletion mode current source 1000 (e.g., 650V). The distal end 1003 of the bottom most field plate can be thought of as the active gate region of the field plate in that only this distal end 1003 affects the depletion of the channel 1006. The portion of the channel 1006 below the active gate region 1003 can correspond, for example, to region 510 in FIG. 5 where that area of the channel is not serpentine patterned.
[0077] 10, when the drain wiring 1008 is at a potential of approximately 400V and the source wiring 1010 is at a potential of approximately 5V, the potential of the channel 1006 at the drain region 1002 is also at approximately 400V and the potential of the channel 1006 at the source region 1104 is also at approximately 5V. The channel 1106, being conductive, carries a current that flows in a direction from the drain region 1102 to the source region 1004. The active gate region 1003, which is conductively coupled to the source region 1004 and therefore at approximately the same potential as the source region 1004, has a negative potential relative to the potential in the region of the channel 1006 underlying the active gate region 1003. The channel 1006 at this point beneath the active gate region 1003 may experience, for example, a potential drop of about 95 V from the point beneath the drain region 1002, resulting in a potential of the channel 1006 beneath the active gate region 1003 of, for example, about 305 V. This approximately 300 V potential difference between the active gate region 1003 (about 5 V) and the area beneath the channel 1006 (about 305 V) creates an electric field between the area beneath the channel 1006 and the active gate region 1003, which depletes the channel 1006 and reduces the conductivity between the drain wiring 1008 and the source wiring 1010. The active gate region 1003 is so represented herein because only the distalmost few nanometers from the contact 430 through which the field plate 434 extends, which corresponds to the lower field plate 434 of FIG. 4A, affect the depletion of the channel 1006.
[0078] FIG. 11 shows an example depletion mode transistor 1100 that is fabricated similarly to the depletion mode current source 1000, but where the corresponding lower field plate 434 of FIG. 4A is fabricated discontinuously as two separate electrodes 1103, 1004 to provide a three terminal device. Wiring 1008, 1010, 1112 are fabricated to conductively couple to the three terminals of the depletion mode transistor 1100, the drain, the source, and the gate, respectively. The upper dielectric layer 1020 may be opened to provide conductive coupling of the drain, gate, and source regions 1002, 1102, 1004 to the wiring 1008, 1112, 1010, respectively. The drain region 1002 is conductively coupled to the drain wiring 1008. The source region 1004 is conductively coupled to the source wiring 1010. The active gate region 1103 influences the underlying channel 1006. The active gate region 1103 is conductively coupled to a gate wiring 1112 that is positioned so as not to physically intersect in three-dimensional space with the field plates 1016, 1018 above the active gate region 1103 (e.g., along the dotted line 1114, but behind or in front of the plane of the cross section shown in FIG. 11). The drain region 1002 and the source region 1004 are electrically coupled to each other through a 2DEG channel 1006, which in some instances may be a 2DHG channel, as described above. In the depletion mode transistor 1100, compared to the depletion mode current source 1000, the source return resistance is reduced to near zero (e.g., by reducing the length 1005 of the resistive patterned portion 512 of the 2DEG channel to near zero) and the gate contact is separated from the source contact. 4A, the depletion mode transistor 1100 does not include p-type doped GaN 444 in the gate stack, resulting in a negative threshold voltage and therefore the characteristics of a depletion mode transistor. Therefore, the depletion mode transistor 1100 may be formed without introducing additional masks or process steps into the enhancement mode GaN process.
[0079] The potential on the gate wiring 1112 can be used to control the resistivity of the depletion mode transistor 1100 between the drain wiring 1008 and the source wiring 1010 by adjusting the degree to which the active gate region 1103 depletes the channel 1006. The distance between the channel 1006 and the gate electrode in the active gate region 1103 defines the minimum negative voltage required to fully deplete the underlying channel 1006.
[0080] In the above examples, any subsequent layers used in the enhancement mode process flow that can be used to fabricate the enhancement mode FET 402, the capacitor 412, the resistor 418, and the depletion mode current source 426 or 1000 and / or the depletion mode transistor 1100, namely the first contact metal layer 456, the gate metal layer 464, the second contact metal layer 476, and the fourth contact metal layer 496, can be configured as individual gate electrodes of the depletion mode transistor. In the example depletion mode transistor 1100 of FIG. 11, it is the first contact metal layer 456 that is configured as the gate electrode 1103. FIG. 12 shows another example depletion mode transistor 1200 in which the gate metal layer 464 is the layer configured as the gate electrode 1203. FIG. 13 is yet another example depletion mode transistor 1300 in which the second contact metal layer 476 is the layer configured as the gate electrode 1303. FIG. 14 illustrates yet another example depletion mode transistor 1400 in which the fourth contact metal layer 496 is a layer configured as a gate electrode 1403 .
[0081] In the example depletion mode transistor 1100 of Figure 11, the first contact metal layer 456 can be formed without a prior step of etching through the first dielectric layer 454 in relevant regions of the depletion mode transistor gate, resulting in a depletion mode transistor gate having a threshold voltage of about -2V to about -3V, corresponding to the measurement curve 608 of Figure 6. In the example of Figure 11, the channel 1006 is depleted when the potential difference between the gate electrode 1103 and the underlying portion of the channel 1006 is at or below this threshold voltage of about -2V to about -3V.
[0082] In the example depletion mode transistor 1200 of FIG. 12, the gate metal layer 464, which can be used to form the gate of the enhancement mode FET 402, can be formed as a separate gate electrode 1203 on top of the second dielectric layer 460. As a result, this depletion mode transistor gate electrode 1203 has a slightly larger distance to the 2DEG channel 1006 than does the gate electrode 1103 of the transistor 1100, providing a more negative threshold of about -4V to about -5V, corresponding to the measurement curve 606 of FIG. 6. In the example of FIG. 12, the channel 1006 is depleted when the potential difference between the gate electrode 1203 and the underlying portion of the channel 1006 is at or below this threshold voltage of about -4V to about -5V. The gate electrode 1203 can be conductively coupled to the wiring 1112, or the depletion mode transistor 1200 can function in other manners as described above with respect to the depletion mode transistor 1100.
[0083] The second contact metal layer 476 and the fourth contact metal layer 496, which can be used for electrical interconnection between devices on a semiconductor chip, can be configured as field plates 436, 438, respectively, in the depletion mode current source 426 or 1000, or as separate gate electrodes 1303 and 1403, respectively, in the depletion mode transistors 1300 and 1400 of FIG. 13 or FIG. 14. In the example depletion mode transistor 1300 of FIG. 13, the second contact metal layer 476 can be formed as a separate gate electrode 1303 on top of the third dielectric layer 468. As a result, the depletion mode transistor gate electrode 1303 has a greater distance to the 2DEG channel 1006 than does the gate electrode 1203 of the transistor 1200, providing a more negative threshold voltage of about -15V to about -40V. 14, the fourth contact metal layer 496 can be formed as a separate gate electrode 1403 on top of the fifth dielectric layer 490. As a result, the depletion mode transistor gate electrode 1403 has a greater distance to the 2DEG channel than does the gate electrode 1303 of transistor 1300, providing a more negative threshold voltage of about -200V to about -300V.
[0084] Although not shown exhaustively, other examples can be based on each of the example depletion mode transistors 1100, 1200, 1300, and 1400 of Figures 11, 12, 13, and 14, and can include one or more overlapping field plates 1016, 1018 fabricated as separate gate electrodes or additional metal layers to provide more and / or higher field plates or separate gate electrodes to enable even higher voltage applications. In some examples (not shown), additional conductive (e.g., metal) layers above the fourth contact metal layer 496 can be provided to enable additional, higher voltage applications, for example, operating in the range of about 900V to about 1200V or about 2000V.
[0085] As just one example of separate electrodes formed on multiple layers, FIG. 15 shows a depletion mode transistor 1500 in which the lower field plate 434 and the middle field plate 436 are fabricated as separate gate electrodes 1503, 1507. In any such example, the electrode acting as the gate is the lowest electrode present because it has the smallest distance to the 2DEG channel 1006 and therefore exhibits the lowest threshold voltage (threshold voltage closest to 0V). In operation of the depletion mode transistor 1500, the gate electrode 1503 fabricated using the first contact metal layer 456 depletes the 2DEG channel 1006 when its gate voltage is between about -2V and about -3V, while the gate electrode 1507 fabricated using the second contact metal layer 476 depletes the 2DEG channel 1006 when its gate voltage is between about -40V and about -50V. In the latter case, the transistor channel 1006 connecting the drain 1002 and source 1004 is already open (non-conductive) because the gate electrode 1503 has already depleted the 2DEG channel 1006. However, at or above (below) a more negative threshold voltage of the gate electrode 1507, the gate electrode 1507 depletes the 2DEG channel 1006 further away from the gate edge facing the drain 1002, as shown by dashed line 1511, in comparison to dashed line 1509, reducing the electric field at this gate edge, thereby improving the maximum drain-to-source / drain-to-gate voltage capability of the depletion-mode transistor 1500, as described above with respect to FIG. 6, consistent with the action of the field plates 436, 438 of the depletion-mode current source 426 of FIG. 4A. For such effects to be observed, it is substantially irrelevant whether the second contact metal layer 476 is configured as a separate gate electrode 1507 as shown in FIG. 15 or as a field plate connected to the source 1004 of the transistor 1500, so long as it is at a lower voltage potential compared to the drain 1002 of the transistor 1500.In examples such as FIG. 15 having separate gate electrodes on multiple layers, only the bottommost conductive (e.g., metal) layer functions as the transistor gate, and all conductive layers above it act as field plates, as long as they are conductively coupled to a low voltage potential, such as the gate line 1112 or the source line 1010, or to another potential substantially lower than the potential of the drain 1002.
[0086] Thus, for example, if a depletion mode transistor according to the above example is desired to have a threshold voltage of about -5V, and the gate-channel distance resulting in the desired threshold voltage, the depletion mode transistor gate can be fabricated using gate metal layer 464, as shown in FIG. 12, omitting any field plate or gate electrode fabricated using first contact metal layer 456. Failure to omit first contact metal layer 456 in such a depletion mode transistor provides an even lower threshold voltage of about -2V to about -3V. Similarly, if a depletion mode transistor according to the above example is desired to have a threshold voltage of about -40V, and the gate-channel distance resulting in the desired threshold voltage, the depletion mode transistor gate can be fabricated using second contact metal layer 476, as shown in FIG. 13, omitting any field plate or gate electrode fabricated using first contact metal layer 456 or gate metal layer 464.
[0087] A depletion mode transistor consistent with any of the above examples 1100, 1200, 1300, 1400, 1500, or extensions thereof, can be fabricated in a manner substantially similar to that described above with respect to the flowchart of Figure 9, incorporating conductive electrode isolation and channel resistance patterning reduction (e.g., near zero) differentiation as described above. A depletion mode transistor consistent with any of the above examples 1100, 1200, 1300, 1400, 1500, or extensions thereof can be used as the current source 108 in Figure 1, in the start-up circuit 208 in Figure 2, or in the pre-charge circuit 302 in Figure 3A. When a three-terminal depletion mode transistor is used as a two-terminal device, such as the pre-charge circuit 302 in Figure 3A, the three-terminal depletion mode transistor can be effectively made into a two-terminal current source, for example, by conductively coupling the gate of the depletion mode transistor to the source of the depletion mode transistor by a wire in a wiring layer. For example, in the depletion mode transistor 1100 of FIG. 11, the gate line 1112 can be connected to the source line 1010 to create a two-terminal device.
[0088] A depletion mode current source or transistor as described herein, including an electrode (e.g., a metal field plate) for depleting the underlying 2DEG or 2DHG channel, can be fabricated in an enhancement mode only process without additional masks beyond those used to fabricate the enhancement mode transistor. Thus, a depletion mode current source or transistor as described herein allows for the implementation of a robust start-up circuit on a single IC in an enhancement mode only GaN fabrication process, without the need for a separately fabricated depletion mode transistor (e.g., fabricated on a different IC) to obtain the same functionality, and without the need for additional masks in the enhancement mode fabrication process flow. Thus, process costs are kept low. High quality, well-controlled insulators are not required to obtain the depletion mode current source characteristics. A lower metal field plate in the current source allows for a lower saturation voltage and threshold voltage V of the current source. th Multiple field plates may be used to obtain higher voltage capability, thereby allowing the start-up circuit to operate at high voltages, such as greater than about 400 V, e.g., greater than about 650 V, e.g., greater than about 1,000 V, without breakdown.
[0089] In this description, the term "based on" means at least partially based on. Also, in this description, the term "coupled" means either indirectly or directly, wired or wirelessly connected. Thus, when a first device, element, or component couples to a second device, element, or component, the coupling can be through a direct coupling or through an indirect coupling via other devices, elements, or components and connections. Similarly, a device, element, or component coupled between a first component or location and a second component or location may be through a direct connection or through an indirect connection via other devices, elements, or components and / or couplings. Modifications in the described embodiments are possible and other embodiments are possible within the scope of the claims.
Claims
1. 1. An integrated circuit (IC), comprising: A two-terminal depletion mode current source, The terminal depletion mode current source a two-dimensional electron gas (2DEG) or two-dimensional hole gas (2DHG) channel coupled between a first terminal of the depletion mode current source and a second terminal of the depletion mode current source; a conductive electrode coupled to the first terminal; Including, the conductive electrode is above and extends over at least a portion of the 2DEG or 2DHG channel; IC.
2. 2. The IC of claim 1 , wherein a portion of the conductive electrode extends over a point of a 2DEG or 2DHG channel, and a first distance along the 2DEG or 2DHG channel between the point and the first terminal is greater than a second distance along the 2DEG or 2DHG channel between the point and the second terminal.
3. 2. The IC of claim 1, wherein the lower surface of the conductive electrode is about 1,500 Angstroms to about 1,700 Angstroms above the 2DEG or 2DHG channel.
4. 2. The IC of claim 1 , wherein the conductive electrode is a lower conductive electrode and the depletion mode current source further includes a second conductive electrode, the second conductive electrode being above and extending over the lower conductive electrode.
5. 5. The IC of claim 4, wherein the second conductive electrode is an intermediate conductive electrode and the depletion mode current source further includes a third conductive electrode, the third conductive electrode being above and extending over the intermediate conductive electrode.
6. 6. An IC according to claim 5, the third conductive electrode is an upper conductive electrode; a lower surface of the lower conductive electrode is about 1,500 angstroms to about 1,700 angstroms above the 2DEG or 2DHG channel; a lower surface of the intermediate conductive electrode is about 3,900 angstroms to about 4,200 angstroms above the 2DEG or 2DHG channel; An IC, wherein a lower surface of the upper conductive electrode is about 9 kA to about 10 kA above the 2DEG or 2DHG channel.
7. 10. The IC of claim 1, further comprising an enhancement-mode field effect transistor (FET).
8. 8. The IC of claim 7, wherein the enhancement mode FET is a p-type doped gallium nitride (p-GaN) FET.
9. 9. The IC of claim 8, wherein the current source is coupled to the p-GaN FET and to a metal-insulator-metal (MIM) capacitor fabricated on the IC in a start-up circuit in an AC / DC power converter.
10. 10. The IC of claim 1 further comprising an interface between a gallium nitride (GaN) semiconductor layer and an aluminum gallium nitride (AlGaN) semiconductor layer for supporting the 2DEG or 2DHG channel.
11. 2. The IC of claim 1, wherein the conductive electrode comprises a multi-layer structure; The multilayer structure comprises: a lower layer of titanium (Ti) having a thickness of about 300 Angstroms to about 500 Angstroms; an intermediate layer of aluminum copper (AlCu) having a thickness of about 750 angstroms to about 1,250 angstroms on the layer of Ti; an upper layer of titanium nitride (TiN) having a thickness of about 400 angstroms to about 600 angstroms over the layer of AlCu; Including, IC.
12. 2. The IC of claim 1, further comprising an enhancement mode field effect transistor (FET), a 2DEG-based or 2DHG-based resistor, and a metal-insulator-metal (MIM) capacitor on the IC with the depletion mode current source; an enhancement mode FET and a resistor both using the first and second semiconductor layers to support respective 2DEG or 2DHG channels between a first terminal and a second terminal of the resistor, between a drain and a gate of the enhancement mode FET, and between a gate and a source of the enhancement mode FET, respectively.
13. 1. A method for manufacturing an integrated circuit (IC), the method comprising: forming a first semiconductor layer over a substrate; forming a second semiconductor layer on the first semiconductor layer; forming a depletion mode current source; Including, the first and second semiconductor layers being configured to support a two-dimensional electron gas (2DEG) or a two-dimensional hole gas (2DHG); forming the depletion mode current source patterning a 2DEG or 2DHG channel on the first semiconductor layer to provide a patterned area of the 2DEG or 2DHG channel; forming a dielectric layer over the second semiconductor layer; Etching the dielectric layer to open first and second contact windows to the 2DEG or 2DHG channel; forming one or more contact layers at least partially within the first and second contact windows to provide first and second contacts to respective ends of the 2DEG or 2DHG channel; providing a first wire at the first contact and a second wire at the second contact to provide the depletion mode current source as a two terminal device; The method of claim 1, wherein the first contact comprises a conductive electrode above the dielectric layer and extending over at least a portion of the 2DEG or 2DHG channel between the first contact and the second contact, a portion of the conductive electrode extends over a point of the 2DEG or 2DHG channel, and a first distance along the 2DEG or 2DHG channel between the point and the first terminal is greater than a second distance along the 2DEG or 2DHG channel between the point and the second terminal.
14. 14. The method of claim 13, wherein the dielectric layer is a first dielectric layer and the conductive electrode is a lower conductive electrode, the method further comprising: forming a second dielectric layer over the first dielectric layer, the first and second contacts, and the lower conductive electrode; etching the second dielectric layer to open the first and second conductive contact windows through the second dielectric layer to the first and second contacts; forming a metal layer at least partially within the first and second conductive contact windows to provide first and second conductive contacts to the first or second contacts, respectively, and a second conductive electrode extending from one of the first or second conductive contacts overlying the lower conductive electrode above the second dielectric layer; A method comprising:
15. 15. The method of claim 14, wherein the metal layer is an intermediate metal layer and the second conductive electrode is an intermediate conductive electrode, the method further comprising: forming a third dielectric layer over the second dielectric layer, the first and second conductive contacts, and the intermediate conductive electrode; etching the third dielectric layer to open first and second windows through the third dielectric layer to the first and second contacts; forming an upper metal layer at least partially within the first and second windows, to extend the first or second conductive contacts through the first and second windows and through the third dielectric layer, and to form an upper conductive electrode extending from one of the first or second conductive contacts over the intermediate conductive electrode above the third dielectric layer; A method comprising:
16. 14. The method of claim 13, wherein the lower surface of the conductive electrode is about 1,500 Angstroms to about 1,700 Angstroms above the 2DEG or 2DHG channel.
17. 14. The method of claim 13, the first semiconductor layer is gallium nitride (GaN); the second semiconductor layer is aluminum gallium nitride (AlGaN); the dielectric layer is silicon nitride (SiN); the conductive electrode comprises a multi-layer structure; The multilayer structure comprises: a lower layer of titanium (Ti); an intermediate layer of aluminum copper (AlCu) above the layer of TI; an upper layer of titanium nitride (TiN) over said layer of AlCu; A method comprising:
18. 14. The method of claim 13, further comprising fabricating an enhancement mode field effect transistor (FET) on the IC comprising the depletion mode current source, the method further comprising: forming a third layer of a doped semiconductor over the second layer prior to forming the dielectric layer; forming a passivation layer over the third layer; selectively removing the third layer and the passivation layer to leave the third layer and the passivation layer in an area of a gate of the enhancement mode FET; selectively etching through the dielectric layer and the passivation layer to the third layer after forming the contact layer to create a window in the area of the gate of the enhancement mode FET; forming a metal gate in the area of the gate of the enhancement mode FET; A method comprising:
19. 20. The method of claim 18, wherein the third layer is gallium nitride doped p-type (p-GaN), the passivation layer is silicon nitride (SiN), and the metal gate comprises titanium tungsten (TiW).
20. 20. The method of claim 18, fabricating a 2DEG-based or 2DHG-based resistor and a metal-insulator-metal (MIM) capacitor on the IC comprising the depletion-mode current source and the enhancement-mode FET; the resistor uses the first and second semiconductor layers to support a resistive patterned 2DEG or 2DHG channel between a first terminal of the resistor and a second terminal of the resistor; The method further comprising forming metal layers as capacitive plates separated by an insulating layer to form the capacitor.
21. 1. An integrated circuit (IC), comprising: an enhancement mode gallium nitride (GaN) field effect transistor (FET); a resistor including a semiconductor-to-semiconductor interface supporting a two-dimensional electron gas (2DEG) or a two-dimensional hole gas (2DHG); a metal-insulator-metal (MIM) capacitor; A depletion mode current source; Including, the depletion mode current source the semiconductor-semiconductor interface being patterned over an area to provide a 2DEG or 2DHG channel within the patterned area; a metal field plate coupled to a negative terminal of the depletion mode current source, the metal field plate extending over a portion of the patterned area; Including, Integrated circuit (IC).
22. 22. The IC of claim 21, a positive terminal of the current source is coupled to the drain of the GaN FET; the negative terminal of the current source is coupled to a gate of the GaN FET and to a positive terminal of the MIM nitride capacitor; a negative terminal of the MIM nitride capacitor coupled to a negative terminal of the resistor; the source of the GaN FET is coupled to the positive terminal of the resistor.
23. 23. The IC of claim 22, wherein the GaN FET is a first GaN FET, the IC further comprising a second GaN FET; the second GaN FET comprising: a drain coupled to the gate of the first GaN FET, to the negative terminal of the current source, and to the negative terminal of the MIM nitride capacitor; a gate coupled to the source of the first GaN FET and to a positive terminal of the resistor; and a source coupled to the negative terminal of the MIM nitride capacitor and to the negative terminal of the resistor; Including, IC.
24. 22. The IC of claim 21, wherein the current source has a saturation current of about 100 nA to about 10 μA at a voltage greater than about 10 V across the current source.
25. 22. The IC of claim 21, wherein the metal field plate of the current source is about 1,500 Angstroms to about 1,700 Angstroms above the 2DEG or 2DHG channel.
26. 1. A method for manufacturing an integrated circuit (IC), the method comprising: forming a first gallium nitride (GaN) based semiconductor layer over a substrate; forming a second GaN-based semiconductor layer over the first semiconductor layer; patterning the first and second GaN-based semiconductor layers to form a first epi structure in a depletion mode transistor area, a second epi structure in a resistor area, and a third epi structure in an enhancement mode FET area; forming a dielectric layer over the second semiconductor layer; etching the dielectric layer to open first and second contact windows at opposite ends of the first epi structure; forming one or more contact layers at least partially within the first and second contact windows; Including, forming the one or more contact layers First and second contacts; a conductive gate electrode above the dielectric layer and extending over at least a portion of the first epi structure; In order to provide method.
27. 27. The method of claim 26, wherein the conductive gate electrode is not contiguous with the first and second contacts.
