The present disclosure relates to a bonding system and a bonding method.
Advanced packaging technologies demand high throughput and precise placement of chips. Hybrid bonding can be particularly challenging with small misalignment tolerances.
In a flip chip to wafer bonding process, alignment metrology is often carried out through an off-axis measurement scheme where the die position on the chuck is obtained before the bonding step begins and then the chip is bonded to the substrate in passive alignment. Further, the achievable accuracy with existing flip chip bonders is not very high (>200 nm). Aligning die with wafer in a heterogeneous 3D packaging equipment within <100 nm overlay accuracy is difficult.
One of the key challenges in the process is the placement of the dies from the die chuck to the product wafer with high overlay accuracy so that the active side of dies are placed on the transfer wafer within an acceptable error bound. To enable good placement accuracy, a metrology system is needed that can provide die placement feedback to the actuators on the die chuck and adjust its position and angular orientation when it is being transferred to the product wafer or substrate.
A need exists for a high throughput placement while still meeting specifications for chip placement.
According to an aspect of the present disclosure, a bonding method includes providing a source substrate, the source substrate includes active side alignment marks on an active side of the source substrate, first non-active side alignment marks and second non-active side alignment marks on a non-active side of the source substrate, and obtaining positions of the active side alignment marks and positions of the first non-active side alignment marks of the source substrate, and determining alignment marks placement errors based on offsets between the positions of the active-side alignment marks and the positions of the first non-active side alignment marks in the X, Y, and 0 directions.
Further features of the present disclosure will become apparent from the following description of exemplary embodiments with reference to the attached drawings.
It shall be understood that elements in the figures are illustrated for simplicity and clarity and have not necessarily been drawn to scale. For example, the dimensions of some of the elements in the figures can be exaggerated relative to other elements to help improve understanding of implementations of the disclosure.
Hereinafter, exemplary embodiments of the present disclosure will be described in detail with reference to the accompanying drawings. The following exemplary embodiments are not intended to limit the present disclosure set forth in the appended claims, and not all combinations of features described in the exemplary embodiments are necessarily indispensable to solving means of the present disclosure. The same reference numerals denote similar constituent elements, and a description thereof will be omitted.
The bonding method is not limited to a specific bonding method. For example, the bonding method may be bonding using an adhesive, bonding by hybrid bonding, atomic diffusion bonding, vacuum bonding, bump bonding or the combination of the like.
As shown in
Based on the positions of the set of active-side alignment marks 210, a set of corresponding non-active side alignment marks 260 are fabricated on the non-active side (backside) of the source substrate 1030 as illustrated in
Before the non-active side alignment marks 260 are made and the metrology information is obtained and stored in memory 162, the non-singulated source substrate 1030 can be thinned down, if necessary, to a desired thickness of for example less than 0.1 mm. The thinning techniques may be performed using one or more back grinding processes such as mechanical grinding, chemical mechanical planarization (CMP), wet etching, dry etching, plasma etching, or any other process of removing material in a controlled manner. The thinning process may, for example, be performed by a DGP8761 fully automatic Grinder/Polisher from DISCO Corporation of Tokyo, Japan.
Alignment marks may then be formed on the non-active side of the thinned source substrate 1030. In an alternative embodiment, the alignment marks 260 are formed on the non-active side of the source substrate 1030 before thinning to a final thickness. In yet another embodiment, the alignment marks 260 are formed on the non-active side of the source substrate 1030 and the thinning process is not performed on the source substrate 1030. Formation of the alignment marks can be done using standard lithography techniques, which may include coating the non-active side with the alignment mark material such as coating the alignment mark material with a resist, patterning the resist, etching the resist, and removing the remaining resist. The alignment mark material may be any material that forms a contrast with the substrate material such as aluminum, chrome, gold, silver, etc. The patterning of the resist may be done with, for example, using a FPA-5520iV i-line stepper from Canon Inc., of Tokyo, Japan. When patterning a resist, a mask may be used which includes non-active side alignment marks in the unused area that coincide with the active side alignment marks and a set of die alignment marks that are within periphery of each of the die. In an alternative embodiment, maskless patterning is used to pattern the resist. In another embodiment a resist is not used, and laser etching is used to pattern the alignment mark material. In yet another embodiment, the alignment marks are formed using a direct printing technique in which a contrasting material is printed directly onto the first substrate. In yet another embodiment, the alignment marks are formed using a direct marking technique in which the substrate material is modified directly via optical or chemical means. Once the non-active side alignment marks are fabricated in both the unused areas and the die areas, the placement error of the marks in the unused area may be measured.
Based on the measurement result, the controller 160 determines the alignment marks placement errors in the X, Y, and θ (rotational) directions based on the offset between the measured positions of the active-side alignment marks 210 and non-active-side alignment marks 260 on the source substrate 1030, where
The ErrorAlignmentMarks(i) is a measured placement error for each alignment mark (i) for a plurality of alignment marks 210 and 260 on the source substrate 1030 at a specific position (x, y). In one embodiment, the alignment marks may be thin marks that only allow measurement of the placement error in one or two of the dimensions {exfronttoback, eyfronttoback, eθfronttoback}. The full set of ErrorAlignmentMarks may be fitted to a model f that describes the placement error (ex,ey,eθ) of the alignment marks across the source substrate 1030 as a function of the mark location on the source substrate. The model f may also take into account the type of errors associated with the fabrication technique used to fabricate the marks. The model f may then be used to estimate the second alignment marks placement error 330 of
In particular, the source substrate 1030 has a set of active side alignment marks 210 (A). There are N active side alignment marks Aj in the set of active side alignment marks 210 (A={A1, . . . Aj, . . . AN}). Each active side alignment mark j is located at an active side alignment mark position Aj on the active side of the source substrate 1030. Active side alignment mark position Aj includes a position {Aj,x, Aj,y} in the coordinate system of the substrate (Ai={Aj,x, Aj,y}).
The non-active side of the source substrate is patterned with a set of corresponding non-active side alignment marks 260 (B) and set of die alignment marks 330 (C). There are N non-active side alignment marks Bj in the set of non-active side alignment marks 260 (B={B1, . . . Bj, . . . BN}). Each non-active side alignment mark j is located at non-active side alignment mark position Bj on the non-active side of the source substrate 1030. Ideally, the non-active side alignment mark positions Bj should be collinearly located with active side alignment mark positions Aj but on the back side of the substrate and they may have placement errors relative to active side alignment mark positions Aj. The non-active side alignment mark position Bj includes a position {Bj,x, Bj,y} in the coordinate system of the substrate (Bj={Bj,x, Bj,y}).
Referring to
A set of positioning errors ΔP is obtained based on measuring the sets A and B or measuring the difference between the sets A and B analyzing crossed gratings of the corresponding alignment marks. There are N positioning errors ΔPj in the set of positioning errors (ΔP={ΔP1, . . . ΔPj, . . . ΔPN}). Each positioning error includes the measurement error in each axes (ΔPj={Aj,x−Bj,x,Aj,y−Bj,y}). If crossed gratings are used to obtain the positioning error then each positioning error may also include rotation (ΔPj={Aj,x−Bj,x,Aj,y−Bj,yθj}).
A model f is created that describes how ΔP is related to active side alignment marks A nominal position (ΔP=f(k,A)+ε) based on a set of fitting parameters k that minimize an error ¿. The model f may be a series of steps in a process, a formula, a series of formulas. The set of fitting parameters k are calculated using standard model fitting techniques such as least squares fitting. Once a set of fitting parameters k is known then the model f may be used to calculate the die mark (second alignment mark) placement error for die alignment mark i using the nominal die mark location Ci (ErrorAlignmentMarks2 (i)=f(k,Ci)), it may also be used to calculate a set of die mark placement errors for the set of die alignment marks 330 (ErrorAlignmentMarks2=f(k,C)).
The source substrate 1030 is then diced (singulated) into individual dies after the measurement. Prior to singulation, the support material of source substrate 1030 may be removed, and a dicing tape is added to the source substrate 1030. In an embodiment, the dicing tape is attached to the non-active side of the source substrate 1030. In an alternative embodiment, the support material is the dicing tape. The source substrate 1030 may then be subjected to a singulation process (singulating) in which the source substrate 1030 is diced. Singulating the source substrate 1030 will destroy, make unusable for alignment, or make unreadable for purposes of alignment the set of active side alignment marks 210 and the set of corresponding non-active side alignment marks 260. However, singulating the source substrate 1030 will have minimal impact on the set of die alignment marks 330 for purposes of alignment.
One or more machines may be involved in the dicing process in which the unused areas are removed leaving an array of dies on the tape frame. The singulation process may be performed with one or more of: a mechanical dicing saw; a laser saw; a waterjet saw; a scriber; and an etcher. After the singulation process, a die separator may be used to stretch the tape frame so that there is a larger separation between individual die on the tape frame. The die separator may heat up the tape frame and stretch it by 1%-20%. After the dies have been separated, the dies may be transferred to a holding substrate 1032 or the tape frame may act as the holding substrate 1032. After this the dies have been singulated, the dies may be cleaned and activated for the hybrid bonding process. Activating the dies may include, for example, treating the die with deionized water and a plasma. In an alternative embodiment, the dies are activated while on the on the bonding head 124 or the pick-up head 144.
Returning to
Components within the bonding apparatus 100 will be generally described in the order in which a set of dies will be transferred from a holding substrate 1032 coupled to the source chuck 122 to a destination substrate 1048 coupled to the destination chuck 148. Due to similarities in operation, the pick-up head carriage 142 and the destination carriage 146 are described in the same passage later in this specification.
The terms “transfer operation” and “transfer cycle” are addressed to aid in understanding embodiments as described herein. A transfer operation starts no later than picking up a set of dies from the source substrate, where the set of dies will be the first set of dies associated with the source substrate bonded to the destination substrate and ends when the last set of dies is transferred from the source substrate to the destination substrate. A transfer cycle starts no later than with picking up a particular set of dies from the source substrate until that same particular set of dies is transferred to the destination substrate. A transfer operation can include one or more transfer cycles.
The source chuck 122 can be a vacuum chuck, a pin-type chuck, a groove-type chuck, an electrostatic chuck, an electromagnetic chuck, or the like. The source chuck 122 can be coupled to the bridge 120 by being attached to the bridge directly or can be coupled to the bridge via a carriage (not illustrated). The carriage may be able to provide translation and/or rotation motion as described in more detail below with respect to the pick-up head carriage 142 and the destination carriage 146. The source chuck 122 has a chucking surface 123 that faces the base 140 or a component coupled to the base 140.
In
The array of pick-up heads 144 are coupled to the pick-up head carriage 142 and have pick-up chucks (not shown) that face the bridge 120 or a component coupled to the bridge 120. At least one of the pick-up heads 144 has a body where the body is disposed between such pick-up head carriage 142. In an alternative embodiment, the pick-up head carriage 142 and the destination carriage 146 may be unified into a single carriage.
The array of pick-up heads 144 can be configured as a vector (a row or a column of pick-up heads) or as a matrix (at least two rows and at least two columns of pick-up heads). Regarding the matrix, the number of pick-up heads within the array of pick-up heads 144 may be different between rows, between columns, or between rows and columns. Some array configurations can be 3×1, 6×1, 2×2, 2×3, 2×4, 10×10, or another rectangular shape, where the first number corresponds to the number of pick-up heads along a row or column, and the second number corresponds to the number of pick-up heads along the other of the row or column. In theory, dies from an entire holding substrate 1032 could be transferred all at once. For such a configuration, the array of pick-up heads 144 may have fewer pick-up heads along rows closer to the top and bottom of the array as compared to the row or the pair of rows closest to the center of the array, and the array of pick-up heads 144 may have fewer pick-up heads along columns closer to the left-hand side and right-hand side of the array as compared to the column or the pair of columns closest to the center of the array.
In an embodiment, a signal from the bridge 120, base 140, or any one or more components coupled to the bridge 120 or base 140 can be transmitted to the controller 160. For example, after the array of pick-up heads 144 have picked up a set of dies from the source substrate, a signal can be transmitted to the controller 160 that the picking up of the set of dies has been completed. After the array of pick-up heads 144 have transferred the set of dies 1522 to the array of bonding heads 124, a signal can be transmitted to the controller 160 that the transfer from the pick-up heads to the bonding heads has been completed.
The array of bonding heads 124 are coupled to the bridge 120. Each of the bonding heads within the array of bonding heads 124 can include a die chuck (not shown) and a body disposed between the die chuck and the bridge 120. The die chuck faces the base 140 or a component coupled to the base 140. Different design considerations may be used for the array of bonding heads 124 as compared to array of pick-up heads 144. Between the array of pick-up heads 144 and the array of bonding heads 124, the bonding heads can have the same design or a different design as compared to the pick-up heads 142.
Similar to the array of pick-up heads 144, the array of bonding heads 124 can be configured as a vector (a row or a column of bonding heads 124) or as a matrix (at least two rows and at least two columns of bonding heads 124). Regarding the matrix, the number of bonding heads within the array of bonding heads 124 may be different between rows, between columns, or between rows and columns. Some array configurations can be 3×1, 6×1, 2×2, 2×3, 2×4, 10×10, or other rectangular shape, where the first number corresponds to the number of bonding heads along a row or column, and the second number corresponds to the number of bonding heads along the other of the row or column. For such a configuration, the array of bonding heads 124 may have fewer bonding heads along rows closer to the top and bottom of the array as compared to the row closest to the center of the array, and the array of bonding heads 124 may have fewer bonding heads along columns closer to the left-hand side and right-hand side of the array as compared to the column closest to the center of the array. In an embodiment, the array of bonding heads 124 has the same number of rows and columns as compared to the array of pick-up heads 144.
The alignment component 1271 may include one or more alignment hardware such as an X,Y, θ stage and z-axis actuator such that of the each bonding head 124 can be moved in 1-6 axes (i.e., X, Y, Z, Pitch, Roll, Yaw) independently relative to the bridge 120 and the base 140. As shown in
Returning to
In an embodiment, the downward facing metrology and alignment hardware 127 may be used for inspecting the surface of the destination substrate 1048 to obtain the bonding positions of the die prior to bonding. In an embodiment, the upward facing metrology and alignment hardware 147 may be used for measuring and inspecting the dies 1022 on the holding substrate 1032 and the dies on the bonding head 124.
The bonding apparatus 100 can be controlled by a controller 160 in communication with the bridge 120, any component coupled to the bridge 120, the base 140, any component coupled to the base 140, or a combination thereof. The controller 160 can execute computer readable program or instructions, optionally stored in memory 162. The controller 160 can include a processor (for example, a central processing unit of a microprocessor or microcontroller), a field-programmable gate array (FPGA), an application specific integrated circuit (ASIC), or the like. The controller 160 can be within the apparatus 100. In another implementation (not illustrated), the controller 160 can be at least part of a computer external to the apparatus 100, where such computer is bidirectionally coupled to the apparatus 100. The memory 162 can include a non-transitory computer readable medium that includes instructions to carry out the actions associated with the transfer operation. In another embodiment, the bridge 120, a component coupled to the bridge 120, the base 140, or a component coupled to the base 140 can include a local controller that provides some of the functionality that would otherwise be provided by the controller 160.
The bonding apparatus 100 can be configured with some or all of the components in a different configuration. For example, all components that are illustrated in
Attention is directed to methods of using the bonding apparatus 100 when transferring a set of dies or die structures from a source substrate and bonding the set of dies to destination bonding sites on a destination substrate. The methods described in more detail below can be used for dies and die structures. Die structures may include a thin die, a backing plate, and an adhesive layer between the thin die and the backing plate. For simplicity, the methods are described with respect to dies. The methods can also be used for die structures although much of the description below focuses on dies.
A method of transferring and bonding dies to bonding sites of a destination substrate according to an embodiment of the present disclosure will be described below with reference to
In step S502, metrology is performed on source substrate 1030 as described above in connection with
Upon determining the non-active side alignment marks placement error in step S503, the first substrate is singulated into a set of dies that are mounted on holding substrate 1032 without the active side alignment marks 210 and the set of corresponding non-active side alignment marks 260 in step S504. More specifically, a die singulation process is performed for dicing the source substrate 1030 into dies 1022. The source substrate 1030 including the dies can be attached to a dicing tape, and singulated into dies as previously described so as to form a holding substrate 1032 as described above. The holding substrate 1032 including the dies may then be loaded on to the source chuck 122.
Next, as illustrated in
The array of pick-up heads 144 can be extended in the Z-direction and pick up the dies 1522 from holding substrate 1032 as illustrated in
Next, after the dies 1522 are transferred to the array of bonding heads 124, the pick-up head carriage 142 and the destination carriage 146 are moved to the left.
In step S506, upon obtaining the measured information, the metrology hardware 127 provides the information to the controller 160 to determine the offset between the die alignment marks 330 on the non-active side of the dies and the bonding head alignment marks 710 (die placement error) as described in the following equation (S507):
In another embodiment, the two sets of marks, 810 and 820 can be simultaneously imaged using the metrology and alignment hardware 127 within the same field of view to obtain the alignment error information of the bonding head relative to the substrate.
The metrology and alignment hardware 127 may be positioned or setup in a manner to accommodate both sets of fiducials, 810 and 820 so that they are within the depth of focus of the focal plane of the metrology and alignment hardware 127.
Based on the obtained information, a bonding head alignment error is determined by controller 160 based on the offset between the bonding head fiducials 810 and destination fiducials 820 in the X, Y, and 0 directions as described in the equation below:
In another embodiment, as illustrated in
The destination chuck 148 can be controlled in closed loop feedback using interferometers or any other sensors (not shown) monitoring its translations and rotations of the destination chuck 148.
Thus, when the destination substrate 1048 is moved to a desired position, together with destination chuck 148 and destination carrier 146, bonding head alignment error can be determined based on the positions of bonding head fiducials 910 and the destination chuck reference marks 920.
Next, in step S510 in
Upon obtaining the total correction information in the X, Y, and 0 directions, the corrections can be applied to destination chuck 148, destination carriage 146 or bonding heads 124 (step S511) prior to or in the process of transferring the dies 1522 onto destination substrate 1048. As shown in the total correction equation above, ErrorAlignmentMarks2 and ErrorDiePlacement are fixed values supplied as offsets to a control, and ErrorBHplacement is a function of time and can act as the feedback signal in a feedback control loop. During the process of transferring the plurality of dies 1522 from the bonding heads 124 onto the destination substrate 1048, further movement and potential misalignment may occur. Thus, bonding apparatus 100 is capable of continuously determining the bonding head alignment error so that further adjustments and corrections can be performed in real-time to minimize alignment errors.
The alignment can be performed under the control of the controller 160 or a local controller. The alignment can be performed more than once per transfer operation and may be performed on average about one time per transfer cycle.
Then, the set of dies can be bonded to the corresponding destination bonding sites of the destination substrate 1048 at step S512 in
Pressure is exerted to bond the dies 1522 to the destination bonding sites of the destination substrate 1048 to each other. In an embodiment, the bonds can be oxide-to-oxide bonds. The pressure during bonding can be in a range 5 kPa to 200 kPa up to 2 MPa. The bonding can be performed at room temperature (for example, at a temperature in a range from 20° C. to 25° C.) or higher. Bonding is performed at a temperature less than a subsequent anneal to expand conductive metal within the dies and at the destination bonding sites. The temperature and pressure may be limited depending on films present during bonding or components within the apparatus 100. For example, the temperature may be no higher than approximately 200° C. After reading this specification, skilled artisans will be able to determine the pressure and temperature used for bonding. The controller 160 or a local controller can transmit an instruction to allow sufficient pressure to be generated for the bonding operation. At this point in the method, one transfer cycle has been completed for the set of dies that include the dies 1522.
A determination is made whether more dies are to be transferred from the source substrate to the destination substrate at decision box S513 in
In a hybrid bonding scenario, there are three steps in a bonding operation, a first anneal to cause the metal within the dies and at the destination bonding sites to expand and contact each other, and a second anneal to cause metal atoms to cross the metal-metal interface and reduce contact resistance. The method previously described with respect to the flow chart in
Embodiments described herein allow for less force to be used when picking up dies during a die transfer operation. Embodiments can allow for less contact area between an adhesive material of a die substrate and a contacting side of a die. A gas flow rate for a Bernoulli chuck or a vacuum to be drawn when removing the die from the die substrate can be less as compared to a configuration where a convention planar substrate chuck and conventional die substrate are used. Thus, equipment in support of a Bernoulli chuck or a vacuum system used with the apparatus 100 may be less aggressively designed because it will not require a relatively larger gas flow or a relative stronger vacuum as compared to another apparatus that does not have the features or use the techniques as described herein.
Embodiment(s) of the present disclosure can also be realized by a computer of a system or apparatus that reads out and executes computer executable instructions (e.g., one or more programs) recorded on a storage medium (which may also be referred to more fully as a ‘non-transitory computer-readable storage medium’) to perform the functions of one or more of the above-described embodiment(s) and/or that includes one or more circuits (e.g., application specific integrated circuit (ASIC)) for performing the functions of one or more of the above-described embodiment(s), and by a method performed by the computer of the system or apparatus by, for example, reading out and executing the computer executable instructions from the storage medium to perform the functions of one or more of the above-described embodiment(s) and/or controlling the one or more circuits to perform the functions of one or more of the above-described embodiment(s). The computer may comprise one or more processors (e.g., central processing unit (CPU), micro processing unit (MPU)) and may include a network of separate computers or separate processors to read out and execute the computer executable instructions. The computer executable instructions may be provided to the computer, for example, from a network or the storage medium. The storage medium may include, for example, one or more of a hard disk, a random-access memory (RAM), a read only memory (ROM), a storage of distributed computing systems, an optical disk (such as a compact disc (CD), digital versatile disc (DVD), or Blu-ray Disc™ (BD)), a flash memory device, a memory card, and the like.
While the present disclosure has been described with reference to exemplary embodiments, it is to be understood that the disclosure is not limited to the disclosed exemplary embodiments. The scope of the following claims is to be accorded the broadest interpretation so as to encompass all such modifications and equivalent structures and functions.