JP2013098498A - Wafer thin-film processing control method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a wafer thin-film processing control method of carrying out thin-film processing based upon the remaining size from a via bottom part in a wafer to a wafer grinding surface.SOLUTION: After a via for through electrode is formed in the wafer, the wafer is held on a chuck table of a processing device with the grinding surface of the wafer up, and a grinding wheel is pressed against the grinding surface of the wafer to perform grinding processing. In this grinding processing, the overall film thickness of the wafer is measured by NCIG to acquire film thickness data, and the size from the via bottom part to the wafer grinding surface is measured to acquire remaining size data. In finish grinding processing, remaining size data on the remaining size from the via bottom part to the wafer grinding surface is derived by reference to film thickness data on the whole wafer to perform the grinding processing up to right before the bottom part of the via.

Description

  The present invention relates to a wafer thin film processing control method for controlling a thin film processing of a wafer to create a wafer with a through electrode.

In recent years, with the spread of computers and portable information terminal devices, the size and density of IC packages to be mounted have been increased, and in order to form a multilayered structure of chips, for example, a DRAM (Dynamic Random Access Memory), for example, through-holes An electrode (Through Silicon Via, hereinafter TSV) wafer has been proposed.
As a processing method for obtaining a TSV wafer, it is proposed that the TSV wafer is formed by using a via first process.
In this case, the Via First process is roughly as follows.
1. Forming vias for through electrodes (Si is etched by dry etching)
2. 2. Form side wall insulating film on via. 3. Fill the via with Poly-Si. Removal of Poly-Si on the wafer surface (CMP (Chemical Mechanical Polishing))

Through the above process, the wafer thinning process proceeds after the DRAM element is formed.
The wafer thinning process consists of a “grinding process” that mechanically grinds the wafer, and a “stress relief process” that removes stress layers (work-affected layers) that contain microcracks, etc. that occur on the ground surface of the wafer during grinding. It consists of. Since this work-affected layer causes warping of the wafer and generation of cracks, it is necessary to completely remove it by a stress relief process after the grinding process.

By the way, in the wafer thinning process, the via penetrates the wafer and becomes TSV by the “grinding process” and the “stress relief process”. During this thinning process, it is proposed to measure the shape of the via itself, such as the depth and diameter, before the TSV is formed.
For example, Patent Document 1 relates to a method for measuring the shape of a stepped structure such as a bottomed hole provided in a material capable of transmitting visible light, near-infrared light, or infrared light, for example, a stacked LSI (Large Scale). In order to ascertain in advance the shape of steps such as the depth of a through silicon via (TSV) used for stacking when creating an integrated circuit chip, before the through via is completed The shape of the non-through via is to be measured.

Thus, in the process for obtaining the TSV wafer, including the above-described inspection and measurement, in the “grinding process”, the thinning process is temporarily stopped just before the via penetrates the wafer and becomes the TSV. Therefore, it is required to perform so-called dimension stop processing.
To that end, for example, it has been proposed to control the thickness of only the wafer by, for example, mounting a non-contact thickness gauge (NCIG) on the grinder and controlling the thickness during grinding. Has been.
Further, it has been proposed that a wafer film thickness meter is similarly mounted on a polishing apparatus, and the polishing rate of the wafer being processed is fed back to the next wafer to manage the machining allowance.

JP 2009-158543 A

However, performing the thinning process based on the total thickness of the wafer as described above is a process that proceeds regardless of the depth of the via formed in the wafer, so depending on the depth of the via, There is a risk of vias penetrating.
Also, even within the same wafer, there are variations in via depth, and the grinding pressure on the wafer grinding surface directly above the via is higher than the grinding pressure on the wafer grinding surface without vias. Therefore, there arises a problem that grinding on the wafer grinding surface immediately above the via proceeds further and the amount of grinding increases. For this reason, in consideration of the via depth, it is necessary to perform grinding process control that reduces the amount of grinding and thins the wafer.
Therefore, reducing the amount of grinding and thinning the wafer increases the thickness of the wafer to be deleted in the stress relief process (polishing) and via exit process, and increases the processing time.
Moreover, the surface roughness (surface roughness and waviness) may be affected by the fluctuation of the grinding amount (removal allowance), and the difference between the maximum thickness value and the minimum value (TTV) on the entire wafer surface due to the increase in allowance allowance. May cause deterioration.

Accordingly, a new technique has been proposed in which thinning is not performed based on the total thickness of the wafer, but is performed based on the remaining dimension from the bottom of the via inside the wafer to the wafer grinding surface.
According to this method, it is possible to perform a dimension stop process in which the grinding process is stopped immediately before the via penetrates to become TSV. In this sizing process, the remaining dimension from the bottom of the via to the wafer grinding surface is required to be as small as possible (preferably several μm). This is because the smaller the remaining dimension from the bottom of the via to the wafer grinding surface, the less the machining allowance in the stress relief process (polishing), leading to higher efficiency.
However, there is an upper limit of measurement performance of the non-contact thickness measuring instrument (about 10 μm at most), and a thickness smaller than 10 μm cannot be measured, and as it is, it cannot contribute to a desired dimensioning process.
In this new method as well, the grinding amount on the grinding surface of the wafer immediately above the via is further advanced and the grinding amount is increased, so that the accurate machining allowance is grasped so that the via does not penetrate. Therefore, it is necessary to measure at a plurality of locations.
Therefore, when performing TSV wafer processing using NCIG, a new method for performing thin film processing based on the remaining dimension from the bottom of the via inside the wafer to the ground wafer surface functions effectively. Accordingly, there is a need for a wafer thin film processing control method that enables high-efficiency in the stress relief process (polishing) and stable processing while suppressing fluctuations in machining allowance.

  In one aspect of the present invention, there is provided a wafer thin film processing control method, wherein a via for a through electrode is formed in advance on a via formation surface of a wafer, and then grinding is performed using a surface opposite to the via formation surface of the wafer as a grinding surface. The wafer thin film is characterized in that the remaining dimension data is obtained by measuring the remaining dimension from the bottom of the via inside the wafer to the ground surface of the wafer, and grinding is performed until the remaining dimension data reaches a predetermined value. A processing control method is provided.

  In another aspect of the present invention, there is provided a wafer thin film processing control method, wherein a via for a through electrode is previously formed on a via forming surface of a wafer, and then grinding is performed using a surface opposite to the via forming surface of the wafer as a ground surface. At the same time, the film thickness data is obtained by measuring the film thickness from the via formation surface to the grinding surface of the wafer, and the remaining dimension data is obtained by measuring the remaining dimension from the bottom of the via inside the wafer to the grinding surface of the wafer. Then, a rough grinding process for continuing grinding until the remaining dimension data reaches the first predetermined value, and after the remaining dimension data reaches the first predetermined value, from the via formation surface of the wafer to the ground surface. The film thickness data is obtained by measuring the film thickness of the wafer, and the derived remaining dimension data is obtained by deriving the remaining dimension from the via bottom in the wafer to the ground surface of the wafer corresponding to the film thickness data. Derived residual dimension data to provide a wafer thin film processing control method comprising: the finish grinding stage which is adapted to stop the grinding reaches the second predetermined value.

  In the wafer thin film processing control method described above, when the film thickness data and the remaining dimension data are acquired, the wafer can be acquired in a state where the wafer is rotated by a predetermined number.

  In addition, film thickness data or residual dimension data is acquired by a sensor head in a non-contact type film thickness measuring apparatus, and the film thickness data or residual dimension data can be acquired by changing the acquisition position of the sensor head with respect to the wafer. it can.

  Further, the film thickness data or the remaining dimension data is sampled at a predetermined time interval, and a mask process for acquiring only data within a predetermined peak level from the film thickness data or the remaining dimension data is performed for each sampling. it can.

  According to the present invention, when performing TSV wafer processing using NCIG, instead of based on the total thickness of the wafer, the remaining dimension from the bottom of the via inside the wafer corresponding to the total thickness of the wafer to the wafer grinding surface is obtained. By adopting a technique for controlling the machining, it is possible to provide a machining control technique that can stabilize the machining while improving the efficiency in the next process and suppressing fluctuations in the machining allowance.

It is a principal system block diagram at the time of performing the process of measuring the film thickness of a wafer while grinding a wafer in the case of the wafer thinning process for obtaining the TSV wafer concerning this invention. It is a system configuration | structure figure which shows an example of the non-contact-type film thickness measuring apparatus mounted in the apparatus shown in FIG. 1, and measures the film thickness of a wafer. It is the flowchart which showed the process of measuring the film thickness of the wafer performed with a rough grinding process in the process of wafer thinning. It is a typical explanatory view when reading the film thickness of the wafer and the remaining dimension of the via formation position by the sensor head by the non-contact type film thickness measuring device during rough grinding. With the sensor head of the non-contact type film thickness measuring apparatus shown in FIG. 4, the film thickness data of the wafer obtained by reading the film thickness of the wafer and the remaining dimension of the via formation position and the via bottom inside the wafer to the wafer grinding surface It is the graph which showed the relationship with residual dimension data. When the chuck table on which the wafer is placed is rotated at a predetermined number of rotations, the wafer thickness data obtained by sampling every predetermined time and the remaining dimension data from the via bottom inside the wafer to the wafer grinding surface are shown. It is a graph. In the wafer thinning process, the wafer thickness measurement is performed along with finish grinding, and the grinding process is performed based on the remaining dimension data from the bottom of the wafer via to the wafer grinding surface obtained from the film thickness data. It is the flowchart which showed.

FIG. 1 shows an example of a main part of a processing apparatus 1 used for wafer grinding. This processing apparatus 1 is used in a grinding process for mechanically grinding a wafer in executing a wafer thinning process in a TSV wafer process for obtaining a TSV wafer. The processing apparatus 1 includes a grinding / polishing apparatus 1 (hereinafter referred to as a PG apparatus 1) having a function of performing a polishing process as well as a double-sided cleaning of a wafer, starting with a grinding process required in the TSV wafer process. Can be used.
Use of the PG device 1 is preferable in that it is not necessary to transfer the wafer between the steps, and damage during the transfer to the wafer can be eliminated.
In addition, the grinding process here employs a processing method for executing rough grinding and finish grinding as grinding.
The PG device 1 includes a chuck table 3 that holds a wafer 2 on which a via for a through electrode (Tsv) is formed in advance. The chuck table 3 has a disk shape, and is configured to be rotationally driven by a motor (not shown) on the lower surface thereof. The chuck table 3 is held in the suction state via the suction plate 4.
In this PG device 1, a grinding wheel 5 for grinding the wafer 2 on the chuck table 3 is disposed so that the grindstone portion 5 g can be pushed toward the grinding surface of the wafer 2.

  For example, as described above, the wafer 2 has a through electrode via 6 formed in advance through a via first process, and a support base material 7 is bonded to the surface (device surface) side on which the via 6 is formed. It is configured. In FIG. 1, in order to clearly show the formation state of the vias 6, the vias 6 having the same diameter at regular intervals are shown in a schematic cross-sectional view.

The PG device 1 as described above includes a non-contact type film thickness measuring device 10 (hereinafter referred to as “film thickness measuring device 10”) for measuring the film thickness or the like (described later) of the wafer 2 on the chuck table 3 in the grinding stage. A sensor head 10h of NCIG 10) is arranged in close proximity. In this case, it is desirable that the sensor head 10 h can move the reading position on the wafer 2.
This is because the via 6 formed in the wafer 2 is unevenly distributed depending on the position of the wafer 2 due to the circuit design of the DRAM to be commercialized, which hinders accurate measurement of the film thickness and the like.
In addition, since grinding on the grinding surface of the wafer 2 immediately above the via 6 progresses during grinding and the amount of grinding increases, an accurate machining allowance can be made to enable grinding processing control considering the depth of the via 6. This is because it is necessary to measure at a plurality of locations in order to grasp.
For example, as shown in FIG. 2, the NCIG 10 includes a controller 10c connected to the PG device 1 via an RS232C cable Crs, a spectrometer 10s connected to the controller 10c via a dedicated cable C, a spectrometer 10s and a fiber. A sensor head 10h connected via a cable Cf.

When the function of NCIG10 is demonstrated roughly, in NCIG10, infrared light is used and the measuring method by a spectral interferometry is performed. Infrared light irradiated from the sensor head 10h onto the wafer 2 to be measured is reflected from the ground surface and the device surface of the wafer 2 to become reflected light, and the reflected light interferes to become an interference wave. Is received by the sensor head 10h. The interference wave received by the sensor head 10h is sent to the spectroscope 10s via the fiber cable Cf, and the spectroscope 10s samples the interference wave every predetermined time and digitizes it, and the numerical data is converted via the dedicated cable C. It is sent to the controller 10c. The controller 10c processes numerical data from the spectroscope 10s, and the PG device 1 displays it as readable data or prints it out.
The PG device 1 stores software for executing the TSV wafer process along with the measurement operation of the NCIG 10, and the controller 10c in the NCIG 10 executes measurement methods such as reading of film thickness data and digitization processing. Software for storing.

Next, the wafer thinning process performed together with the measurement technique performed using the NCIG 10 as described above will be described with a specific example.
First, the support base material 7 is bonded to the surface (device surface) side where the vias 6 are formed on the wafer 2 on which the vias 6 for TSV have been previously formed through the Via First process, and the chuck table 3 of the PG device 1. Is held in the suction state via the suction plate 4 (see FIG. 1).
Then, an operation command is issued to the PG device 1, and the polishing process is executed starting with the grinding process.

Of the grinding process, the rough grinding step will be described based on the flowchart shown in FIG.
At the start of rough grinding (STEP 1), the chuck table 3 is rotationally driven by a predetermined number (STEP 2), and the grinding wheel 5 is pushed toward the grinding surface of the wafer 2 via the grindstone 5g (STEP 3). As a result, a considerable amount of grinding is performed with the surface opposite to the device surface on which the via 6 of the wafer 2 is formed as a grinding surface. At the same time as the rough grinding starts, the NCIG 10 is activated (STEP 4). Infrared light irradiated to the wafer 2 from the sensor head 10 h disposed close to the upper side of the wafer 2 is reflected from the ground surface and the device surface of the wafer 2 to become reflected light. These reflected lights interfere to form an interference wave and are received by the sensor head 10h. The interference wave received by the sensor head 10h is sent to the spectroscope 10s via the fiber cable Cf, and the spectroscope 10s samples the interference wave every predetermined time and digitizes it, and the numerical data is converted via the dedicated cable C. It is sent to the controller 10c. Then, numerical data from the spectroscope 10s is processed in the controller 10c (STEP 5), and displayed as readable data in the PG device 1 or printed out to easily grasp the progress of the grinding process. Can do.
The above rough grinding step is performed by moving the sensor head 10h of the NCIG 10 above the wafer 2 in parallel to the grinding surface of the wafer 2 and changing the measurement position with respect to the wafer 2.

  As shown in FIG. 4, the infrared light from the sensor head 10h receives the interference wave between the ground surface of the wafer 2 and the reflected light from the device surface, and at the same time, the ground surface of the wafer 2 and the via 6 in the wafer 2. The interference wave with the reflected light from the bottom of the light is received. These received light data are sent to the spectroscope 10s via the fiber cable Cf, and the interference wave is sampled and digitized every predetermined time in the spectroscope 10s, and the numeric data is sent to the controller 10c via the dedicated cable C. The controller 10c derives the film thickness T1 between the ground surface of the wafer 2 and the device surface based on the numerical data from the spectroscope 10s. At the same time, the controller 10c derives the remaining dimension T2 between the ground surface of the wafer 2 and the bottom of the via 6 in the wafer 2 (STEP 5), and can display it as readable data in the PG device 1 ( (See FIG. 5).

In the rough grinding step as described above, the film thickness T1 between the grinding surface of the wafer 2 and the device surface is, for example, 77 μm, and the remaining dimension T2 between the grinding surface of the wafer 2 and the bottom of the via 6 in the wafer 2 is the first. It continues until it reaches the lower limit value of the reading thickness by NCIG10 as a predetermined value (STEP 6). This is because the lower limit value of the reading thickness by the NCIG 10 is about 10 μm, and a thickness below that cannot be read directly.
In the rough grinding process, as shown in FIG. 5, as the grinding process proceeds, a data group α indicating the film thickness T1 between the grinding surface of the wafer 2 and the device surface, the grinding surface of the wafer 2 and the inside of the wafer 2 It can be seen that the data group β indicating the remaining dimension T2 between the bottoms of the vias 6 changes in parallel with each other. In FIG. 5, a data group β indicating the remaining dimension T2 between the ground surface of the wafer 2 and the bottom of the via 6 in the wafer 2 is a data group α indicating the film thickness T1 between the ground surface of the wafer 2 and the device surface. Compared with, the number of data is clearly small. This is because the film thickness T1 and the remaining dimension T2 are measured simultaneously by the NCIG 10 here, and when measured simultaneously, the higher level is taken in as measurement data.
Therefore, in order to increase the data group β indicating the remaining dimension T2 between the ground surface of the wafer 2 and the bottom of the via 6 in the wafer 2, for example, a data masking process in which data exceeding a predetermined level is not taken in is considered. It is done.

Here, the film thickness T1 becomes 77 μm, for example, and the remaining dimension T2 between the ground surface of the wafer 2 and the bottom of the via 6 in the wafer 2 reaches 10 μm. This can be verified by the relationship with the thickness (μm) (see FIG. 6).
FIG. 6 shows a case where the chuck table 3 has a predetermined rotation speed (for example, 60 rpm) and the range of the remaining dimension data extracted by the mask process is 0 to 60 μm.
The remaining dimension data displayed as a plurality of layers indicates that there is a variation in the remaining dimension of the via 6 in the wafer 2.

Next, the finish grinding step in the grinding process will be described based on the flowchart shown in FIG.
When the remaining dimension T2 between the grinding surface of the wafer 2 and the bottom of the via 6 in the wafer 2 reaches 10 [mu] m, a finish grinding step is started (STEP 7). The chuck table 3 is driven to rotate (STEP 8), the grinding wheel 5 is pushed in (STEP 9), and the NCIG 10 is started (STEP 10). Here, while moving the sensor head 10h of the NCIG 10 above the wafer 2 in parallel with the grinding surface of the wafer 2 (STEP 11), infrared light is irradiated from the sensor head 10h to the grinding surface of the wafer 2 as follows. Measurement is performed.
On the PG device 1 side, the film thickness T1 _{(T2 = 10 μm)} (here 77 _μm) at the start of finish grinding is stored, and then T1 when finish grinding is in progress (STEP 12) to finish the remaining dimension T2. It is calculated and stored from the film thickness T1 at the start of grinding _{(T2 = 10 μm)} and T1 at the time of finishing grinding. That is, T2 ← T2 (10 μm) − | T1−T1 _{(T2 = 10 μm)} | is calculated (STEP 13) and stored.
While the finish grinding process proceeds, the value of T1 is updated every sampling time, and it is determined by the above calculation whether the target value of T2, that is, the second predetermined value, has reached, for example, 1 μm (STEP 14). . That is, the PG device 1 derives T2 ← T2 (10 μm) − | T1 (68 μm) −T1 (77 μm) _{(T2 = 10 μm)} | = 10− | 68−77 | = 1 μm by the above-described calculation. When T1 is thinned to 68 μm, it can be understood that the remaining dimension T2 between the ground surface of the wafer 2 and the bottom of the via 6 in the wafer 2 has reached the target value of 1 μm. When the remaining dimension T2 reaches the target value of 1 μm, it can be seen that, in the grinding process, the dimension stop processing for temporarily stopping the thinning process immediately before the via penetrates to TSV has been achieved. The process can be terminated.
In the above finish grinding stage, the remaining dimension T2 between the grinding surface of the wafer 2 and the bottom of the via 6 in the wafer 2 is targeted while moving the sensor head 10h above the wafer 2 in parallel with the grinding surface of the wafer 2. Since it is grasped that the value has reached 1 μm, it is possible to avoid the penetration of the via due to the grinding unevenness in which the grinding on the grinding surface of the wafer immediately above the via proceeds further and the grinding amount increases.

After the grinding, the polishing process is performed in the PG device 1 while the wafer 2 is held on the chuck table 3 to remove the work-affected layer and the like. As a result, breakage such as inadvertent cracking of the wafer 2 can be prevented and a TSV wafer can be obtained.
The polished wafer 2 is removed from the chuck table 3 and transferred to the next process such as a wafer processing process, where coating and dicing are performed.

As described above, the lower limit value of the reading thickness by the NCIG 10 is about 10 μm, and even if the thickness below that value cannot be read directly, the wafer 2 can be calculated by simple calculation based on the film thickness data measured at the same time. The remaining dimension T2 between the ground surface and the bottom of the via 6 in the wafer 2 can be obtained. Based on the value of T2, the thinning process is temporarily stopped immediately before the via penetrates to the TSV. Stop processing can be achieved.
Thus, according to the measurement method as described above, when performing TSV wafer processing using NCIG, there is a new method for performing thinning processing based on the remaining dimensions from the bottom of the via inside the wafer to the ground surface of the wafer. It is possible to provide a wafer thin film processing control method that functions effectively and thereby enables stabilization of processing by suppressing high efficiency in the next process and fluctuation in machining allowance.

DESCRIPTION OF SYMBOLS 1 PG apparatus 2 Wafer 3 Chuck table 4 Suction plate 5 Grinding wheel 5g Wheel part 6 Via 7 Support base material 10 NCIG
10h Sensor head 10c Controller 10s Spectrometer Crs RS232C cable C Dedicated cable Cf Fiber cable

Claims (5)

A wafer thin film processing control method comprising:
After forming the via for the through electrode on the via forming surface of the wafer in advance, the surface opposite to the via forming surface of the wafer is ground as a grinding surface.
Wafer thin film processing characterized in that residual dimension data is obtained by measuring a residual dimension from a via bottom inside the wafer to a ground surface of the wafer, and grinding is performed until the residual dimension data reaches a predetermined value Control method. A wafer thin film processing control method comprising:
After forming the via for the through electrode on the via forming surface of the wafer in advance, the surface opposite to the via forming surface of the wafer is ground as a grinding surface.
The film thickness data is obtained by measuring the film thickness from the via formation surface of the wafer to the grinding surface, and the residual dimension data is measured by measuring the residual dimension from the via bottom inside the wafer to the grinding surface of the wafer. A rough grinding step of obtaining and continuing grinding until the remaining dimension data reaches a first predetermined value;
After the remaining dimension data reaches the first predetermined value, the film thickness data is obtained by measuring the film thickness from the via formation surface of the wafer to the grinding surface, and corresponds to the film thickness data. The derived remaining dimension data is obtained by deriving the remaining dimension from the bottom of the via inside the wafer to the ground surface of the wafer, and when the derived remaining dimension data reaches the second predetermined value, the grinding process is performed. A finish grinding stage that stops the operation,
A method for controlling the processing of a wafer thin film.   3. The wafer thin film processing control method according to claim 1, wherein the film thickness data or the remaining dimension data is acquired in a state in which the wafer is rotated a predetermined number of times.   The film thickness data or residual dimension data is acquired by a sensor head in a non-contact type film thickness measuring apparatus, and the film thickness data or residual dimension data is acquired by changing the acquisition position of the sensor head with respect to the wafer. The wafer thin film processing control method according to any one of claims 1 to 3, wherein:   The film thickness data or the remaining dimension data is sampled at a predetermined time interval, and a mask process for obtaining only data within a predetermined peak level of the film thickness data or the remaining dimension data is performed for each sampling. The wafer thin film processing control method according to any one of claims 1 to 4.

Images