JP4569058B2 - Evaluation method of anodic bonding substrate - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an anodic bonding substrate evaluation method for evaluating a bonding state of a bonding interface in an anodic bonding substrate in which a semiconductor substrate and a glass substrate are anodically bonded.
[0002]
[Prior art]
Several semiconductor pressure sensors, acceleration sensors, and structures using the same have been proposed as anodic bonding substrates in which a semiconductor substrate and a glass substrate are bonded by anodic bonding. Conventionally, for example, the following method is used to evaluate whether or not the state of the bonding interface is good by anodic bonding, which is a processing technique.
[0003]
Interference fringes (Newton rings) that appear when a silicon wafer and a glass substrate are anodically bonded from the back side of the glass substrate and there are unbonded parts (voids, gaps) at the bonding interface It is a method of visual observation or observation with an optical microscope.
[0004]
However, in the case where the back surface of the glass substrate is in a ground glass state and the observation light is scattered, as described in JP-A Nos. 2000-146733 and 2000-146734, the glass substrate is used. A method of improving by providing a mirror surface portion on the back surface is adopted.
[0005]
Japanese Patent Application Laid-Open No. 9-289238 also employs a method for calculating interference fringe information as a quantitative numerical value as in the above two publications. Further, in Japanese Patent Laid-Open No. 10-22354, a photodiode using a PN junction is formed on the silicon wafer side of the bonding interface, and light is emitted with and without an unbonded portion when irradiated with light. A method of detecting by the difference in the degree of flow of electromotive current is employed.
[0006]
[Problems to be solved by the invention]
However, the conventional evaluation method described above has the following problems. In the method of detecting interference fringes by irradiating from a certain light source, the state of interference fringes changes depending on the intensity of irradiated light, the incident angle of light, the observation angle for detecting interference fringes, and the wavelength, Detection is impossible if the gap at the joint is below the interference fringe detection limit, or if the joint is in close contact but is not anodic-bonded, that is, is not covalently bonded.
[0007]
In other words, at the atomic level where interference fringes are generated or a gap with a certain height is an essential condition for detection, and no interference fringes are generated at unjoined parts. In this case, it is not possible to detect such a gap that is not anodic-bonded, that is, has not been covalently bonded but is in close contact.
[0008]
Furthermore, in the above-mentioned JP-A-2000-146733, JP-A-2000-146734, and JP-A-10-22354, pretreatment and processing that require time and effort such as mirror finishing and formation of a photodiode are required. Moreover, the measuring apparatus provided with the optical system and the current measuring circuit used in JP-A-9-289238 and JP-A-10-22354 is expensive and cannot be manufactured at low cost.
[0009]
In addition, as a method capable of indirectly evaluating the state of the bonding interface, a method for measuring the bonding strength such as tensile strength is also known and practiced. However, this evaluation method depends on the size of the bonding area, and must be a sample shape with a certain bonding area, and it can detect local defects in the bonding interface area to be evaluated. Is unsuitable.
[0010]
In addition, even if there is a relatively wide defect area, if the strength of the base material of the anodic bonding substrate such as a silicon wafer or a glass substrate leads to destruction, that is, even if there is a defective portion in the bonding interface, The entire area is relatively large, and the bonding strength of the entire bonding interface may overcome the base metal fracture strength. Thus, it is difficult to accurately evaluate the bonding interface.
[0011]
Therefore, in view of the above problems, the present invention evaluates the bonding state of the bonding interface in the anodic bonding substrate in which the semiconductor substrate and the glass substrate are anodic bonded, and even the defect of the bonding interface to the extent that interference fringes do not occur. An object is to enable detection at a simple and low cost.
[0012]
[Means for Solving the Problems]
At the bonding interface of the anodic bonding substrate, bonding is performed by forming a covalent bond (Si-O or the like) between an atom (Si or the like) in the semiconductor substrate and an oxygen atom in the glass.
In a good bonded state (good bonded state), the bond density of the covalent bond is relatively large, and in a poor bonded state (bad bonded state), the bond density of the covalent bond is relatively small.
[0013]
According to the experiment by the present inventor, the difference in the bonding state due to the size of the covalent bond density is different in the etching state when the cross section of the bonding interface in the glass substrate is exposed to an etchable etchant. It was found to be expressed as
[0014]
That is, it was found that, in the poorly bonded state, regardless of the occurrence of interference fringes, the resistance of the bonding interface to the etching solution is lower than in the good bonding state, and etching of the bonding interface proceeds easily. The present invention has been created based on this finding.
[0015]
That is, the invention described in claim 1 is a method for evaluating the bonding state of the bonding interface (50) of the anodic bonding substrate (30) formed by anodic bonding of the semiconductor substrate (10) and the glass substrate (20). In the state where the exposed surface (31) in which the cross section of the bonding interface is exposed is formed in the anodic bonding substrate, the anodic bonding substrate is immersed in an etching solution capable of etching the glass substrate, and then immersed in the etching solution. Where the amount corresponding to the amount of the glass substrate etched in the state of the glass substrate alone is a, and the amount of the etchant soaked into the bonding interface from the exposed surface is b, the absolute value of the difference between the amount a and the amount b Based on the size of the interface penetration rate | b−a | / b, which is the ratio of | b−a | and the amount b, the bonding state of the bonding interface is evaluated. The amount a and the amount b are measured by observing the etching state of the cross section of the bonding interface from the side of the exposed surface by microscopic observation after immersion in the etching solution. It is characterized by that.
[0016]
According to the present invention, when the bonding interface of the anodic bonding substrate is in a defective bonding state, the interface penetration rate | ba− / | b is large, and when the bonding interface is in a good bonding state, the interface penetration rate | b−a | / b becomes smaller. Further, the size of the interface penetration rate | b−a | / b varies greatly depending on the presence or absence of bonding failure.
[0017]
Etching using an etchant that can etch a glass substrate is a processing method that can be easily performed in a normal semiconductor process, and compared with the above-described processing and processing such as conventional mirror processing and photodiode formation. It does not take time and effort. Moreover, the measurement of the etching amount and penetration amount by etching can be easily performed by observation with an optical or electron microscope.
[0018]
Therefore, according to the present invention, the anodic bonding substrate is processed by the etching solution, and the bonding state of the bonding interface is evaluated based on the size of the interface penetration rate | b−a | / b. Even a defect on the joint interface that does not lead to generation of interference fringes can be detected easily and at low cost.
[0019]
Here, as described above, if there is a bonding failure at the bonding interface, etching easily proceeds at the bonding interface, and the interface penetration rate | ba− || b increases. For example, the separation between the defective bonding state and the good bonding state can be specifically performed with the interface penetration rate | b−a | / b being approximately 0.3 as a boundary.
[0020]
In addition, a depletion layer in which cations in the glass are deficient exists in the vicinity of the bonding interface in the anodic bonded glass substrate. This depletion layer has a higher etching rate with the etchant than other portions of the glass substrate. That is, in the anodic bonded glass substrate, the amount b is larger than the amount a.
[0021]
Conversely, when the interface penetration rate is 0, the depletion layer is not formed, that is, anodic bonding is not performed. According to the experiment by the present inventors, in an anodic bonded glass substrate, the interface penetration rate is usually 0.1 or more.
[0022]
Therefore, more specifically, in the evaluation method of claim 1, when the interface penetration rate | b−a | / b is 0.1 or more and 0.3 or less, as in the invention of claim 2. In addition, it can be determined that the bonding state of the bonding interface (50) is good.
[0023]
Further, as in the invention described in claim 3, a hydrofluoric acid aqueous solution can be used as the etchant.
[0024]
In addition, the code | symbol in the bracket | parenthesis of each said means is an example which shows a corresponding relationship with the specific means as described in embodiment mentioned later.
[0025]
DETAILED DESCRIPTION OF THE INVENTION
(First embodiment)
DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments shown in the drawings will be described below. FIG. 1 is a schematic cross-sectional view showing a method of manufacturing an anodic bonding substrate according to the first embodiment of the present invention, and FIG. 2 is an explanatory view showing a method of anodic bonding.
[0026]
An anodic bonding substrate 30 shown in FIG. 1A is a substrate obtained by anodic bonding, for example, a silicon substrate (semiconductor substrate) 10 as a wafer having a diameter of 150 mm and a glass substrate 20. A broken line DL in FIG. 1A indicates a dicing line (scribe line) for dividing the anodic bonding substrate 30 into chips.
[0027]
Specifically, the anodic bonding is performed as shown in FIG. In a vacuum, the glass substrate 20 and the silicon substrate 10 are sequentially mounted on the cathode electrode 40 and the anode electrode 41 is pressed thereon to form a laminated state. A wafer electrode 11 (not shown in FIG. 1) made of Al (aluminum) or the like is formed on the silicon substrate 10, and the anode electrode 41 is in contact with the wafer electrode 11.
[0028]
The cathode electrode 40 is made of carbon or the like, and the anode electrode 41 is made of stainless steel (SUS) or the like. In addition, heaters 42 for heating the stacked bodies 10, 20, 40, 41 to a high temperature state are provided above and below the stacked bodies, for example.
[0029]
And while heating the said laminated body with the heater 42, a voltage is applied between the silicon substrate 10 and the glass substrate 20 by the cathode electrode 40 and the anode electrode 41, and the silicon substrate 10 and the glass substrate 20 are anodic-bonded. To do. For example, the heating temperature is about 360 ° C., the applied voltage is about 600 V, and the degree of vacuum is 1 × 10. ^-Five It can be about Torr.
[0030]
The anodic bonding substrate 30 thus formed is then divided into chips by performing dicing cut along the dicing line DL, as shown in FIG. 1B. Thereby, the anodic bonding substrate 30 formed by anodically bonding the silicon substrate 10a and the glass substrate 20 via the bonding interface 50 is formed as individual chips.
[0031]
Next, a method for evaluating the anodic bonding substrate according to this embodiment will be described. This evaluation can be performed using the anodic bonding substrate 30 divided by the dicing cut. By this dicing cut, as shown in FIG. 1B, the anodic bonding substrate 30 is in a state where an exposed surface 31 in which a cross section of the bonding interface 50 is exposed is formed.
[0032]
Here, the anodic bonding substrate (evaluation sample) 30 used for the evaluation may be a chip shape (usually rectangular shape) by the dicing cut, or desired along the dicing line DL without being divided into chips. It may be divided into strips of the size of.
[0033]
In other words, any state and shape may be used as long as the cross section of the bonding interface to be evaluated is exposed, but each chip shape is formed in a dicing cut process which is a final processing process of a general semiconductor substrate. It is very simple to use as an evaluation sample.
[0034]
In addition, this evaluation may be performed by checking the total number of each chip (evaluation sample), checking one to several chips for each lot (multiple wafers), or for each wafer. Alternatively, it may be performed by a method of determining the number of sampling inspections based on a general quality control method, such as checking one to several chips.
[0035]
Then, with the exposed surface 31 in which the cross section of the bonding interface 50 is exposed, the anodic bonding substrate 30 is immersed in an etchant that can etch the glass substrate 20. As this etchant, an etchant capable of glass etching in a semiconductor process such as a hydrofluoric acid aqueous solution can be used.
[0036]
FIG. 3 shows an etching state in the vicinity of the end face of the anodic bonding substrate 30 in a state where the anodic bonding substrate 30 on which the exposed surface 31 is formed is immersed in the etching solution 60. In the above hydrofluoric acid aqueous solution or the like as the etching solution 60, the silicon wafer is usually hardly etched, so that only the glass substrate 20 is etched.
[0037]
After anodic bonding substrate 30 is immersed in etching solution 60 to some extent and anodic bonding substrate 30 is subjected to pure water cleaning and drying treatment, the cross section of bonding interface 50 is observed by observation with an electron microscope such as an optical microscope or a relatively simple SEM. The etching state is observed from the side of the exposed surface 31.
[0038]
Then, as shown in FIG. 3, due to the difference in the etching rate of each part, the amount (glass etching amount) corresponding to the amount of the glass substrate 20 etched in the state of the glass substrate 20 alone is a, and from the exposed surface 31 to the bonding interface When the amount of the etchant soaked in 50 (interfacial etching amount) is b, the etched state is observed so that a <b.
[0039]
Here, FIG. 4 is a diagram schematically showing a bonding state at the bonding interface 50. In anodic bonding, an oxygen-containing component (Na ₂ O)) moves in the opposite direction to the contact interface with silicon, whereby a covalent bond (Si—O) between oxygen ions and silicon that has been negatively charged is formed at the contact interface. .
[0040]
The bonding of the bonding interface 50 is achieved by the covalent bonding of Si-O. However, when a site that does not reach the covalent bonding of Si-O is generated as shown by the x mark in FIG. The part becomes poorly bonded.
[0041]
That is, in a good bonding state (good bonding state), the bond density of the Si—O covalent bond is relatively large, and does not occur as the above-described interference fringes, or does not occur as interference fringes such as voids. In the bonded state (bad bonded state), the bond density of the covalent bond is relatively small.
[0042]
For example, when using hydrofluoric acid as the etching solution 60, hydrofluoric acid is generally SiO 2 ₂ + 6HF → H ₂ SiF ₆ + H ₂ As represented by the formula of O, SiO ₂ That is, the glass substrate 20 is melted.
[0043]
Therefore, the difference in the bonding state depending on the bond density of the covalent bond is the penetration amount | b−a | (that is, the absolute value of the difference between the interface etching amount b and the glass etching amount a in the etching described above. It is expressed as a difference of b-a) or (ab).
[0044]
That is, when the bonding density of Si—O is relatively high and the bonding interface 50 is in a good bonding state, the bonding density of Si—O is relatively small and the bonding interface 50 is in a defective bonding state. Compared with the case of the above, etching to the bonding interface 50 by the etching solution 60 does not proceed, and the penetration amount | b−a | is small.
[0045]
In this embodiment, the bonding state of the bonding interface 50 is evaluated based on the size of the interface penetration rate | b−a | / b, which is the ratio of the penetration amount | b−a | and the interface etching amount b. .
[0046]
Since this evaluation involves an etching process, it is inevitable that the etching amounts | b−a |, a, and b slightly vary due to variations in the etching rate due to variations in the concentration of the etchant 60 and the like. On the other hand, if the evaluation is based on the interface penetration rate, it is possible to exclude variations in the etching rate for each evaluation lot as compared with the absolute penetration amount | ba− | Can be realized.
[0047]
When the bonding interface of the anodic bonding substrate is in a defective bonding state, the interface penetration rate | b−a | / b is large. When the bonding interface is in a good bonding state, the interface penetration rate | b−a | / B becomes smaller. Further, the size of the interface penetration rate | b−a | / b varies greatly depending on the presence or absence of bonding failure.
[0048]
Etching using an etchant 60 that can etch the glass substrate 20 is a processing method that can be easily performed in a normal semiconductor process, and is applicable to the above-described conventional mirror processing, photodiode formation, and the like. Compared to the time required. Moreover, the measurement of the etching amount and penetration amount by etching can be easily performed by observation with an optical or electron microscope.
[0049]
Therefore, according to the evaluation method of the present embodiment, the anodic bonding substrate 30 is processed with the etching solution 60 and the bonding interface 50 is bonded based on the size of the interface penetration rate | b−a | / b. By evaluating the state, even a defect of the bonding interface 50 that does not lead to generation of interference fringes can be detected easily and at low cost.
[0050]
A specific example of this evaluation method is shown. In this example, the anodic bonding substrate 30 used for the evaluation was divided into strips having a desired dimension along the dicing line DL without being divided into chips.
[0051]
As the etching solution 60, a general hydrofluoric acid aqueous solution for semiconductor (49 wt% hydrofluoric acid) and pure water mixed at a ratio of 1: 1 was used, and the anodic bonding substrate 30 was immersed in the etching solution 60. In the case of this hydrofluoric acid aqueous solution 60, the immersion time is 10 minutes, and the etching rate is about 10 μm / minute for the glass substrate 20 used this time.
[0052]
After the anodic bonding substrate 30 was immersed in a hydrofluoric acid aqueous solution, the anodic bonding substrate 30 was washed with pure water and dried, and the cross-section of the bonding interface 50 was observed with an optical microscope. In the hydrofluoric acid aqueous solution 60, the silicon substrate (silicon wafer) 10 was hardly etched, and only the glass substrate 20 was etched.
[0053]
As a result of the observation, as the anodic bonding substrate 30 after etching, etching of the bonding interface 50 progresses by hydrofluoric acid etching, and is relatively etched into the bonding interface 50 (hereinafter referred to as sample A). The sample was divided into two (hereinafter referred to as “sample B”) which were etched without relatively penetrating the bonding interface 50.
[0054]
FIG. 5 shows the results of obtaining the interface penetration rate (b−a) / b for sample A (n = 8) and sample B (n = 20) using (b−a) as the penetration amount. FIG. In this example, all the evaluated anodic bonding substrates 30 used were those in which no interference fringes could be detected by optical microscope observation.
[0055]
From FIG. 5, the sample A and the sample B are greatly different in the size of the interface penetration rate (b−a) / b, and the sample B having a low interface penetration rate (b−a) / b is in a good bonded state. The sample A having a large interface penetration rate (ba) / b is considered to be in a poorly bonded state.
[0056]
In order to confirm this, it verified by another evaluation method. As a result of trying to peel off the silicon substrate 10 and the glass substrate 20 in each sample by tweezers or the like, it was the sample A that was relatively easily peeled off, and the sample B was not peeled off at all.
[0057]
In addition, when a tensile test is performed as a destructive test and the bonding interface 50 is observed, the base material that the glass substrate 20 or the silicon substrate 10 is broken in the sample B that is likely to have a good state of the bonding interface 50. In the sample A in which the state of the bonding interface 50 may not be favorable, the portion where the bonding interface 50 was peeled off was observed, although it was fracture.
[0058]
Thus, experimentally, in the samples A and B divided by the interface penetration rate (b−a) / b shown in FIG. 5, the sample A is in a poorly joined state, and the sample B is in a well joined state. It was confirmed that there was. And it turns out that the interface penetration rate by this embodiment can take correspondence with another general evaluation method (tensile test etc.) clearly.
[0059]
Here, in FIG. 5, the interface penetration rate in the sample B realizing the good bonded state is 0.13 to 0.29 even when the average value of 4σ is taken. From this, it can be said that if the penetration rate of the interface is 0.1 or more and 0.3 or less, it can be judged as a non-defective product.
[0060]
Incidentally, when the interface penetration rate | b−a | / b is too small, or extremely zero (that is, when b = a), anodic bonding is not realized. This is due to the following reason.
[0061]
As shown in FIG. 3 above, in the glass substrate 20 that has been anodically bonded, the cations (Na ⁺ Etc.) is depleted. As described with reference to FIG. 4, the depletion layer 51 is configured such that the cation in the glass substrate 20 is opposite to the bonding interface 50 in order to realize the covalent bond of Si—O at the bonding interface 50. Formed from moving to.
[0062]
The depletion layer 51 has a higher etching rate with the etchant 60 than other portions of the glass substrate 20. Therefore, even if the bonding state of the bonding interface 50 is perfect, as shown in FIG. 3, the vicinity of the bonding interface 50 including the depletion layer 51 in the glass substrate 20 is greatly etched compared to other portions. Is done.
[0063]
Further, when the evaluation method of the present embodiment is compared with the tensile test, the following can be said.
[0064]
A method of performing a destructive test and observing the interface state like a tensile test cannot be detected when the base material is first destroyed. In addition, since it is peeled off in a comprehensive manner with a certain bonding area, a defective portion cannot always be detected when a bonding interface that is not locally good exists.
[0065]
However, in the evaluation method of the present embodiment, evaluation can be performed without depending on the size of the bonding area of the evaluation sample. Therefore, by using a sample with a small bonding area, a locally poor bonding interface is also compared. Can be detected easily.
[0066]
In this embodiment, since the non-destructive inspection is performed, the anodic bonded substrate 30 determined as a non-defective product (good bonded state) and the anodic bonded substrate 30 manufactured from the same wafer or lot can be shipped.
[0067]
Although the glass substrate 20 is slightly etched in the etching process of this evaluation method, the anodic bonding substrate (chip) 30 that has been evaluated and determined to be non-defective is shipped after being compared with the final product size specification. do it. In the hydrofluoric acid-based etching solution 60 of the above example, the immersion time is 10 minutes, the etching rate of the glass substrate is 10 μm / minute, and the product has a glass etching amount b of about 100 μm (see FIG. 3). There is no.
[0068]
(Second Embodiment)
By the way, in an actual anodic bonding substrate, a protective film, wiring, and the like are formed on a silicon substrate. In the evaluation method of the first embodiment, these protective film, wiring, and the like are formed by an etching solution such as a hydrofluoric acid aqueous solution. There is a risk of being etched.
[0069]
In the second embodiment, in the evaluation method described above, a step of protecting the protective film, wiring, etc. on the element formed on the silicon substrate 10 from the etching solution 60 is added. This embodiment is suitable for use in a semiconductor film having a protective film or wiring formed on the silicon substrate 10 side, such as a semiconductor pressure sensor, an infrared sensor, a flow sensor, a gas sensor, or a humidity sensor.
[0070]
FIG. 6 is a schematic cross-sectional view showing the method for manufacturing the anodic bonding substrate according to the second embodiment. FIG. 6 shows an anodic bonding substrate 30 in which a diaphragm 12 as a thin portion is formed on a silicon substrate 10 as a configuration provided in each sensor described above.
[0071]
As shown in FIG. 6A, an element (not shown), an interlayer film 13, a wiring 14, and a protective film 15 are formed on the surface side of a silicon substrate 10 as a wafer by a normal semiconductor process. Diaphragm 12 is formed by forming a recess on the back side of 10 by anisotropic etching or the like. In the same manner as in the first embodiment, the glass substrate 20 is anodically bonded to the back side of the silicon substrate 10.
[0072]
Next, as shown in FIG. 6B, the outermost surface of the silicon substrate 10 is covered with a surface protective material 16 having resistance to the etching solution before dicing cutting after anodic bonding. The surface protective material 16 can be formed using, for example, a material for a hydrofluoric acid aqueous solution such as a resist, and can be formed by spin coating or the like in the same manner as a general resist coating process.
[0073]
Thereafter, as shown in FIG. 6C, dicing cut is performed along the dicing line DL, and the anodic bonding substrate 30 is divided into a plurality of chips. Then, as shown in FIG. 6 (d), the bonding interface 50 is evaluated in the same manner as in the first embodiment by performing a total number check or a sampling check on the chips that are cut and formed with the exposed surface 31.
[0074]
At this time, the protective film 15 and the interlayer film 13 and the like are slightly etched from the chip cross section (exposed surface 31) not protected by the surface protective material 16, but the etching solution concentration or the like so as not to affect the product. Adjust the etching time. The same applies to the glass substrate 20. However, since not only the cross section but also the back surface (surface opposite to the bonding interface 50) is etched in the glass substrate 20, this is also an etching condition that does not affect the product.
[0075]
Specifically, the conditions of the hydrofluoric acid aqueous solution 60 shown in the specific example of the first embodiment can be adopted as such etching conditions. Or you may adjust with the scribe width which is outside an element formation area. Further, the rear surface of the glass substrate 20 may be protected by the surface protective material 16 in the same manner as the front surface side of the silicon substrate 10.
[0076]
Thereafter, for the anodic bonding substrate 30 subjected to the etching process, the interface penetration rate | ba− / | b is obtained as in the first embodiment, and the bonding state of the bonding interface 50 is evaluated based on the size. To do. For example, an interface penetration rate of 0.1 or more and 0.3 or less is determined as a good product (good bonded state).
[0077]
Thereafter, as shown in FIG. 6 (e), the surface protective material 16 is removed only from the chip (anodic bonding substrate 30) determined to be a non-defective product and the anodic bonding substrate 30 manufactured from the same wafer or lot and shipped. To do. The removal of the surface protective material 16 can be performed using a stripping solution similar to a general resist removing step. If it manufactures in such a process, it becomes possible to ship a product with high reliability of anodic bonding.
[0078]
Thus, also in the second embodiment, the same effect as in the first embodiment can be obtained. In particular, this embodiment is suitable for use in a case where a protective film, wiring, or the like is formed on the silicon substrate 10 side. In addition, it is preferable to use the evaluation method of the said 1st Embodiment which does not use the surface protection material 17 as an anodic bonding board | substrate only for an evaluation sample in the manufacturing process of these products.
[Brief description of the drawings]
FIG. 1 is a schematic cross-sectional view showing a method for producing an anodic bonding substrate according to a first embodiment of the present invention.
FIG. 2 is an explanatory view showing a method of anodic bonding.
FIG. 3 is a diagram showing an etching state of the anodic bonding substrate in the anodic bonding substrate evaluation method according to the first embodiment.
FIG. 4 is a diagram schematically showing a bonding state of a bonding interface by anodic bonding.
FIG. 5 is a diagram showing a result of obtaining an interface penetration rate by the anodic bonding substrate evaluation method according to the first embodiment.
FIG. 6 is a schematic cross-sectional view illustrating a method for manufacturing an anodic bonding substrate according to a second embodiment of the present invention.
[Explanation of symbols]
10 ... Silicon substrate (semiconductor substrate), 20 ... Glass substrate,
DESCRIPTION OF SYMBOLS 30 ... Anode-bonding board | substrate, 31 ... Exposed surface, 50 ... Bonding interface, 60 ... Etching liquid.

Claims (3)

A method for evaluating a bonding state of a bonding interface (50) of an anodic bonding substrate (30) formed by anodically bonding a semiconductor substrate (10) and a glass substrate (20),
After immersing the anodic bonding substrate in an etching solution (60) capable of etching the glass substrate with an exposed surface (31) in which a cross section of the bonding interface is exposed in the anodic bonding substrate,
During immersion in this etching solution, the amount corresponding to the amount of etching of the glass substrate in the state of the glass substrate alone was a, and the amount of the etching solution soaked into the bonding interface from the exposed surface was b. Then, based on the magnitude of the interface penetration rate | b−a | / b, which is the ratio of the absolute value | b−a | of the difference between the amount a and the amount b to the amount b, It evaluates the bonding state ,
The amount a and the amount b are measured by observing the etching state of the cross section of the bonding interface from the side of the exposed surface by microscopic observation after immersion in the etching solution. Evaluation method of substrate. 2. The bonding state of the bonding interface (50) is determined to be good when the interface penetration rate | b−a | / b is 0.1 or more and 0.3 or less. The evaluation method of the anodic bonding board | substrate as described in 2. 3. The method for evaluating an anodic bonding substrate according to claim 1, wherein a hydrofluoric acid aqueous solution is used as the etching solution.

Images