CONTROLLED SUBSTRATE CLEAVE PROCESS AND APPARATUS

BACKGROUND OF THE INVENTION

[0001] The present invention relates to the manufacture of substrates. More particularly, the invention provides a technique including a method and a structure for forming multi- layered substrate structures using bonding techniques for the fabrication of semiconductor integrated circuit devices. Such bonding techniques include use of thermal processing to establish bonded interfaces that are substantially free of imperfections, defects, and/or other undesirable features according to a specific embodiment. In a preferred embodiment, the thermal processing causes oxygen species to be transferred from an interface region between a bonded pair to be removed to an outer region. But it will be recognized that the invention has a wider range of applicability; it can also be applied to other types of substrates for three- dimensional packaging of integrated semiconductor devices, photonic devices, piezoelectronic devices, flat panel displays, microelectromechanical systems ("MEMS"), nano-technology structures, sensors, actuators, solar cells, biological and biomedical devices, and the like.

[0002] From the very early days, human beings have been building useful articles, tools, or devices using less useful materials for numerous years. In some cases, articles are assembled by way of smaller elements or building blocks. Alternatively, less useful articles are separated into smaller pieces to improve their utility. A common example of these articles to be separated include substrate structures, such as a glass plate, a diamond, a semiconductor substrate, a flat panel display, and others. These substrate structures are often cleaved or separated using a variety of techniques. In some cases, the substrates can be separated using a saw operation. The saw operation generally relies upon a rotating blade or tool, which cuts through the substrate material to separate the substrate material into two pieces. This technique, however, is often extremely "rough" and cannot generally be used for providing precision separations in the substrate for the manufacture of fine tools and assemblies. Additionally, the saw operation often has difficulty separating or cutting extremely hard and or brittle materials, such as diamond or glass. The saw operation also cannot be used effectively for the manufacture of microelectronic devices, including integrated circuit devices, and the like.

[0003] Accordingly, techniques have been developed to fabricate microelectronic devices, commonly called semiconductor integrated circuits. Such integrated circuits are often developed using a technique called the "planar process" developed in the early days of semiconductor manufacturing. An example of one of the early semiconductor techniques is described in U.S. Patent No. 2,981,877, in the name of Robert Noyce, who has been recognized as one of the fathers of the integrated circuit. Such integrated circuits have evolved from a handful of electronic elements into millions and even billions of components fabricated on a small slice of silicon material. Such integrated circuits have been incorporated into and control many of today's devices, such as computers, cellular phones, toys, automobiles, and all types of medical equipment.

[0004] Conventional integrated circuits provide performance and complexity far beyond what was originally imagined. In order to achieve improvements in complexity and circuit density (i.e., the number of devices capable of being packed onto a given chip area), the size of the smallest device feature, also known as the device "geometry", has become smaller with each generation of integrated circuits. Increasing circuit density has not only improved the complexity and performance of integrated circuits but has also provided lower cost parts to the consumer.

[0005] Making devices smaller is very challenging, as each process used in integrated circuit fabrication has a limit. That is to say, a given process typically only works down to a certain feature size, and then either the process or the device layout needs to be changed. Additionally, as devices require faster and faster designs, process limitations exist with certain conventional processes and materials. An example of such a process is an ability to make the thickness of the substrate thin after the manufacture of the integrated circuit devices thereon. A conventional process often used to thin these device layers is often called "back grinding," which is often cumbersome, prone to cause device failures, and can only thin the device layer to a certain thickness. Although there have been significant improvements, such back grinding processes still have many limitations.

[0006] Accordingly, certain techniques have been developed to cleave a thin film of crystalline material from a larger donor substrate portion. These techniques are commonly known as "layer transfer" processes. Such layer transfer processes have been useful in the manufacture of specialized substrate structures, such as silicon on insulator or display substrates. As merely an example, a pioneering technique was developed by Francois J.

Henley and Nathan Chung to cleave films of materials. Such technique has been described in U.S. Patent No. 6,013,563 titled Controlled Cleaving Process, assigned to Silicon Genesis Corporation of San Jose, California, and hereby incorporated by reference for all purposes. Although such technique has been successful, there is still a desire for improved ways of manufacturing multilayered structures.

[0007] From the above, it is seen that a technique for manufacturing large substrates which is cost effective and efficient is desirable.

BRIEF SUMMARY OF THE INVENTION [0008] Embodiments in accordance with the present invention describe an apparatus and method for performing a controlled cleaving process to separate bonded substrates. Particular embodiments relate to a process and apparatus for cleaving bonded substrates utilizes tensile force applied to the bonded substrates from an arm, in combination with a blade to initiate local cleaving. Upon sensing a drop in the load on the arm following initiation of the local cleaving, movement of the arm by a motor is carefully controlled to reduce the load, such that a speed of subsequent global cleaving substantially matches the speed of the initial local cleaving. A tool for performing this cleaving may include a suction member configured to secure a first bonded substrate to a base, an arm having a first end configured to pivot about the base, and a second end having a suction member configured to contact the second bonded substrate. A motor applies force to the arm that is translated into tensile force applied across the bonded wafer pair. The motor is configured to move the arm in a carefully controlled manner to reduce the applied tensile force, once cleaving has been initiated with a blade.

[0009] Depending upon the embodiment, one or more of these benefits may be achieved. These and other benefits may be described throughout the present specification and more particularly below.

BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 is a simplified cross-sectional view showing an embodiment of a tool for performing cleaving in accordance with the present invention.

[0011] Figure 2 is a schematic diagram showing initiation of the cleaving process and propagation of the cleaving.

[0012] Figure 3 plots force versus time for an embodiment of a cleaving process in accordance with the present invention.

[0013] Figure 4 is a simplified flow diagram showing steps in the cleaving process in accordance with an embodiment of the present invention.

[0014] Figure 5 plots counts from a load cell of the tool of Figure 1 , versus time, during an actual cleaving process.

[0015] Figure 6 plots actual and programmed angular position of the motor of the tool of Figure 1 as a function of time.

[0016] Figure 7A is a Atomic Force Microscopy roughness result showing substrates cleaved in accordance with an embodiment of the present invention.

[0017] Figure 7B is a Atomic Force Microscopy section analysis showing the substrates cleaved in accordance with an embodiment of the present invention.

DETAILED DESCRIPTION OF THE INVENTION

[0018] A method and apparatus for cleaving bonded substrates utilizes tensile force applied to the bonded substrates from an arm, in combination with a blade to initiate local cleaving. Upon sensing a drop in the load on the arm following initiation of the local cleaving, movement of the arm by a motor is carefully controlled to reduce the load, such that a speed of subsequent global cleaving substantially matches the speed of the initial local cleaving. A tool for performing this cleaving may include a suction member configured to secure a first bonded substrate to a base, an arm having a first end configured to pivot about the base, and a second end having a suction member configured to contact the second bonded substrate. A motor applies force to the arm that is translated into tensile force applied across the bonded wafer pair. The motor is configured to move the arm in a carefully controlled manner to reduce the applied tensile force once cleaving has been initiated with a blade.

[0019] A cleaving process in accordance with an embodiment of the present invention may be performed utilizing the tool 100 shown in cross-section in the view of Figure 1. Specifically, bonded wafer pair 102 is secured in place by suction cups 104 and 106. Bottom

suction cup 104 is fixed to the tool base 108. Top suction cup 106 is secured to extension arm 110 that is in pivotal communication with a motor 111 along axis Z. A processor 113 is configured to control the operation of motor 111. A load cell 115 is configured to detect the force present on the arm.

[0020] Linear Variable Differential Transducer (LVDT) 119 is shown positioned on the base 108 with the pointed tip shown in the retracted state. LVDT 119 is extended to detect the vertical position of the bonded wafer pair. Blade 112 is configured to be applied to a precise point on the side of the bonded wafer pair 102, in order to initiate the cleaving process.

[0021] An embodiment of a tool in accordance with the present invention exhibits the following dimensions: 40" W x 40" D x 94" H (178 cm W x 152 cm D x 203 cm H). The tool utilizes the following electric hookup: 208 V 3φ; 20 Amps; 60 Hz. The tool utilizes stepper motors, and employs a vacuum of 25-28" Hg / 8 cfm / 224 liters/min. The tool includes an exhaust pressure manifold to house exhaust. The pneumatic pressure of the system is 60-100 psi / 4.1 - 6.9 bar. The tool is configured to cleave about 30 8"/200 mm bonded wafer pairs/hr. The tool utilizes a robot handling system where the cassette configuration is 2 cassette stations (1 send, 2 receive after separation), the robot is an ASYST Model 21, the pre- aligner is a non contact integrated system, the sequence and speed is recipe programmable, and the aligner angles are recipe programmable.

[0022] Figure 2 is a schematic diagram showing initiation of the cleaving process and propagation of the cleaving. Specifically, Figure 2 shows wafer pair 102 in the form of donor substrate 102a bonded to handle wafer 102b along bonding interface 130. Application of blade 112 in the manner indicated during the application of tensile force from extension arm 110 pivoting about axis Z, causes cleavage of the donor substrate 102a along a cleave plane 132 to yield the handle wafer bearing a thermal oxide and an overlying semiconductor layer, in the manner desired by silicon-on-insulator (SOI) technology.

[0023] An embodiment of the cleaving process is described in the simplified plot of force versus time of Figure 3, and in the simplified flow diagram of Figure 4. Specifically, after processing through bond treatment, the bonded wafer pairs are ready to be mechanically separated, leaving a film transferred from the donor wafer on the handle wafer.

[0024] In a first step 402 of process flow 400, the bonded wafer pair is positioned within the tool. A vacuum is applied to the suction cups to secure the bonded wafer pair in place.

[0025] In a second step 404, at time T ₀ a prescribed tensile force sufficient to cause cleave initiation is applied to the bonded wafer pair by rotating the extension arm to achieve an initial tensile force setpoint A. In certain embodiments this tensile force setpoint A may be around 5 pounds.

[0026] In a third step 406, once the tensile force setpoint A has been achieved, at time Ti the blade is inserted into the bonded pair interface at precisely prescribed distance and location. Accurate positioning of the blade is achieved by computer control.

[0027] Application of the blade to the interface causes initiation of the cleaving process in a local area. As shown in detail in Figure 3, as the cleave initiation proceeds the force on the extension arm drops at time T ₂ to a second level B. In step 408, the motor of the tool is programmed to undergo a specific motion once the monitored tensile force decreases by a prescribed number of counts below the initial setting, for example to a level B. As shown in step 410, the motor continues to turn as prescribed by the program until separation of the formerly bonded wafers is complete.

[0028] In the specific example shown in Figure 5, no strain on the extension arm corresponds to a load cell count of F~30,000. A load cell count of F= -19,000 counts represents the tensile setpoint A, and corresponds to an equivalent of -5.5 pounds of tensile force. The plot of Figure 5 shows the basic form of operation of the invention: a rapid decrease in strain occurs during the initiation phase (t = 0.4), followed by a slower main cleave that is controlled or restrained by the motion of the motor

[0029] As shown above, in order to achieve a uniform cleaved surface, it is important to properly integrate the initiation phase of the cleave, with the overall cleave. This can be done with a user-prescribed programmed motor motion. Figure 6 plots actual and programmed angular position of the motor of the tool of Figure 1 as a function of time in arbitrary units (AU).

[0030] The desired integration of the initiation cleave phase with the overall cleave can be achieved by initiating the programmed motor motion before the initiation cleave phase is completed, i.e. while the tensile force is still falling due to the beginning of wafer separation. To initiate the programmed motor motion before the initiation cleave phase is completed, the following steps may be performed.

1) Set the tensile force for triggering the programmed motor motion (Strain Delta parameter) to be less than the initial setpoint by ~10%, (i.e. not at the minimum to avoid starting the motor after the initiation is complete); and

2) Apply a sufficiently high angular acceleration (Acel parameter) so the cleaving will continue after the initiation phase is complete with a speed that is comparable to the very fast initiation phase, thus ensuring a seamless transition between the initiation and main cleave phases.

As a result, roughness of cleaved surfaces in the two regions will be comparable.

[0031] As mentioned previously, the angular acceleration, Acel (vertical separation of the handle and donor), may achieve a seamless transition between the initiation and main cleave stages. As a consequence, the acceleration parameter is an important input cleave process parameter.

[0032] Other parameters can also be adjusted to control the cleaving process. These are listed in the following TABLE 1 as they appear in the software controlling operation of the tool. Certain parameters are of particular relevance and are discussed further. Typical settings are described for cleaving a 200mm wafer pair.

[0033] TABLE 1

[0034] The Strain Delta and Acel parameters are set to achieve the proper integration between the initiation and main cleaves. The acceleration ends either when the Split Speed setting is reached, or the maximum speed achievable given the constraints imposed by the Acel, Decel, and Move_After_Split parameter settings.

[0035] The Split Speed is the principal process parameter to be adjusted to minimize cleave lines during the main cleave. Typically, the Split Speed parameter is adjusted to be fast enough to be above the threshold to achieve minimum cleave lines. If the Split Speed parameter is adjusted to be too fast, the wafers may fracture. Between these two boundary settings is a process window in which the resulting cleaves are minimized without the risk of wafer breakage.

[0036] The Decel parameter slows motion of the extension arm, so that at the Move_After_Split command, the motor will stop with the pair completely separated. The Decel parameter should be chosen carefully, as slowing the extension arm too much will yield heavy cleave lines near the end of the cleave.

[0037] As noted above in Figure 6, since the motor motion voltage is via a closed loop program, the actual motion of the motor may be different. Therefore, there is a need to distinguish between the programmed motion and the actual motion. PID (Proportional- Integral-Derivative) circuitry attempts to minimize the difference (ferr = following error) between the programmed location of the arm and the actual position of the arm at any instant in time after TO.

[0038] The following steps show the sequence of program operation of the motor. Initially, the cleave sequence is started and then the flipper closes, and vacuum is applied to the suction cups to secure the wafer pair.

[0039] The motor starts pulling on the wafer by moving one angular increment (e.g. 0.001 radian) and then the load cell is checked to determine strain. If the strain indicated at the load cell is not at the Strain counts (e.g. -5000 counts), this motor is programmed to move another increment and to then recheck the strain. This cycle continues until the Strain setpoint is reached, or a Pull Count (# of times the motor is stepped by the Pull Increment) is reached.

[0040] Next, the LVDT moves up to touch the bottom of the wafer and retracts. Blade initiation is activated once the tensile strain setpoint (e.g. -5000 counts) is reached. This time may be ~ 0.35 seconds after TO. Application of the blade to the interface initiates the cleaving process.

[0041] When a strain gauge indicates that force has decreased by a preset number of counts (the Strain_Window parameter level ( e.g. 2000 count delta or -8000 counts)), the programmed motor sequence is initiated (0— >1). A signal is sent to the motor to go to a predetermined relative angular position (Instantaneous_Move or Boost parameter, e.g. 0.01 radians). The motor moves as fast it can to that point, i.e. overrides any accel setting. This prevents the encoder from driving the motor backwards to correct for a sudden springing of the extension arm ahead due to the sudden release in tension at wafer initiation.

[0042] Once at this angular position, the motor is now commanded to go to the move_after_split position (e.g. 0.099314 radians) with a prescribed acceleration (split_acel, e.g. 400 rad/sec ) and will try to reach a prescribed constant split_speed (e.g. 4 radians/sec) while approaching the angular position. The programmed motion already includes or excludes the constant velocity phase, since the total distance, x, = a ₁ t ₁ ² /2+vt ₂ -a ₂ t ₃ ² /2 = move_after_split distance. Since the overriding setpoint is a relative angular distance ( e.g. 0.099314 radians) rather than speed, the motor may never reach the constant speed if it

reaches the move_after_split position first. Again, the programmed motion given the move_after_split, split_acel, and the prescribed deceleration (Decel, e.g. 140 rad/sec ² ) already has determined how long the velocity will be constant, t ₂ . The actual motion could be significantly different.

[0043] For example, as can be the case with certain waxed wafers, if the wafer pairs start separating before blade initiation so that the strain drops by the Strain_Window setting, the software will cause vacuum to be turned off, the flipper to stop, and an alarm will sound. Therefore, for waxed pairs, the Strain_Window with Strain_Pull_fixed is made larger (e.g. 2000→5000 counts for waxed 300 mm pairs).

[0044] Aspects of the cleaving process according to an embodiment of the present invention may exhibit certain interdependencies. For example, cleaving process conditions are dependent on the effective surface energy in the cleave plane. In turn, this cleave plane surface energy is a function of the implant conditions and bond treatment. Variations in cleaving results are principally accommodated by adjustments to initial tensile force (the Strain cleaving parameter).

[0045] Once the cleaving process has been performed according to an embodiment of the present invention, engineering characterization and inspection of wafer after cleave can be achieved in a variety of ways. One approach is to inspect the wafers with the naked eye under a bright light. Observations are then recorded on an inspection sheet. Still another approach is to examine the wafers under a microscope. The surface of the wafers may be inspected for silicon defects and microscopic voids. The following TABLE 2 summarizes the types of defects that may be noted.

TABLE 2

[0046] Yet another approach is to utilize atomic force microscopy (AFM) to inspect the wafers. Figure 7A is a Atomic Force Microscopy roughness result showing substrates cleaved in accordance with an embodiment of the present invention. RMS surface roughness for as-cleaved wafer surfaces is typically 3O-6θA. Figure 7B is a Atomic Force Microscopy section analysis showing the substrates cleaved in accordance with an embodiment of the present invention.

[0047] While the above is a full description of the specific embodiments, various modifications, alternative constructions and equivalents may be used. Although the above has been described using a selected sequence of steps, any combination of any elements of steps described as well as others may be used. Additionally, certain steps may be combined and/or eliminated depending upon the embodiment. Furthermore, the particles of hydrogen can be replaced using co-implantation of helium and hydrogen ions to allow for formation of the cleave plane with a modified dose and/or cleaving properties according to alternative embodiments. Of course there can be other variations , modifications, and alternatives. Therefore, the above description and illustrations should not be taken as limiting the scope of the present invention which is defined by the appended claims.