JP4513513B2 - Manufacturing method of electronic parts - Google Patents

Description

The present invention relates to the production how electronic components and, more particularly, to joining and sealing techniques using molten solder.

  2. Description of the Related Art Conventionally, various electronic components have been provided in which a lid substrate is bonded to an element substrate on which elements are formed and a lid substrate is bonded, and a space is formed between the substrates. For example, a BAW device in which a piezoelectric thin film is sandwiched between a pair of electrodes is provided with a hollow portion around a portion where piezoelectric resonance occurs so that piezoelectric vibration is not hindered. Such electronic components are bonded in the state of a wafer and then divided into individual electronic components.

As a technique for performing such joining and sealing, a technique for thermocompression bonding using molten solder is known. For example, as shown in FIG. 15, a solder layer (3) is formed on one substrate (# 1), a metallization layer is provided on the other substrate (# 2), and reflow is performed with these two layers in contact with each other. The solder layer (3) is heated and melted, and a strong alloy compound is formed between the solder layer (3) and the metallization layer to bond the substrates (# 1, # 2) together. In the illustrated example, a spacer layer (9) that does not melt at the time of reflow is further provided in order to keep the distance between the substrates (# 1, # 2) above a certain level (see, for example, Patent Document 1).
JP 2000-141300 A

  When the wafer and the wafer are bonded by thermocompression bonding using the molten solder, the molten solder flows out between the wafers, and the solder protrudes. Furthermore, the protruding solder is agglomerated and greatly agglomerated in one place of the frame. This protrusion may cause deterioration of characteristics by short-circuiting between pads, for example, causing variation in capacitance between terminals without short-circuiting, resulting in characteristic variation, and blade wear or clogging in a subsequent dicing process. It brings about a problem that speeds up.

  As a countermeasure, if the pressure at the time of thermocompression bonding is reduced so as not to cause protrusion, an unbonded portion is generated due to warpage of the wafer and sealing becomes impossible. Further, when the pressure is reduced, the influence of the shape (flatness and parallelism) of the thermocompression bonding head and the stage is brought about, high processing accuracy is required, and the apparatus becomes expensive.

  Another countermeasure is to increase the clearance between terminals. However, although there is no short circuit, the number of wafers taken is reduced and productivity is lowered.

In view of the above circumstances, it is possible to suppress the protrusion of the molten layer, it is intended to provide a manufacturing how electronic components.

  In order to solve the above-described problems, the present invention provides a method for manufacturing an electronic component configured as follows.

The electronic component manufacturing method includes first to third steps. In the first step, the first substrate is made such that the weight ratio of Au to Sn (Au: Sn) in the first bonding layer and the second bonding layer is in the range of 84:16 to 62:38. The first bonding layer of Au is formed on the second substrate, and the second bonding layer of Sn is formed on the second substrate. In the second step, the first substrate and the second substrate are sandwiched so that the first bonding layer and the second bonding layer are in contact with each other, and are 200 ° C. or higher and lower than 280 ° C. Thermocompression bonding at temperature. In the third step, the first substrate and the second substrate are heat-treated at a pressure of 280 ° C. or higher and lower than 521 ° C. with a pressure lower than that of the second step.

  In the first step, the bonding layer may be formed in an appropriate shape according to the purpose of bonding between the substrates. For example, in the case of sealing, a bonding layer is formed in a frame shape. In the case of electrical connection, the bonding layer is formed in the shape of a pad or land.

In the second step, when the first bonding layer and the second bonding layer are brought into contact with each other and heat is applied while applying pressure, the melting point of the AuSn eutectic alloy is 280 ° C., and the melting point of Sn is 232 ° C. However, interdiffusion between Au and Sn occurs during the temperature rise process, and alloys such as AuSn ₄ , AuSn ₂ , and AuSn are formed. Since the weight ratio of Au to Sn (Au: Sn) is in the range of 84:16 to 62:38, when heating is continued, AuSn is generally formed. Since the melting point of the alloy is 309 ° C., it does not melt and does not protrude beyond 280 ° C. in the second step.

  In the third step, when the pressure is smaller than that in the second step and heated to 280 ° C. or higher, Au-20 wt% Sn is generated as a liquid phase from the interface between Au and AuSn, and this is the heat in the second step. It flows and fills in voids and gaps generated in the bonding layer during crimping. Since the pressure is reduced at the time of melting, no protrusion occurs. When the third step ends and cools, sealing is complete.

  According to the said method, the protrusion from the pattern of the AuSn alloy which arises at the time of thermocompression bonding can be suppressed.

Preferably, in the first step, the second to form a bonding layer having a smaller area than the first bonding layer, and between the second substrate and the second bonding layer, A third bonding layer having an area smaller than that of the first bonding layer is formed. In the second step, the third bonding layer does not melt.

In this case , in the second step, the first bonding layer having a large area and the second bonding layer having a small area are in contact with each other. Therefore, the portion where the second bonding layer is in contact with the first bonding surface and its The neighboring portion is melted to form a molten layer (for example, molten solder). The molten layer is pushed out from between the first bonding layer and the second bonding layer by the pressing force by sandwiching the first and second substrates, but wets along the first bonding layer having a large area. spread. Therefore, the first substrate and the second substrate can be joined without agglomerating the molten layer and without causing protrusion.

  Even if it melts in this way, the third bonding layer does not melt. Therefore, the distance between the first substrate and the second substrate can be kept at a certain level or more.

  Therefore, by appropriately selecting the shape and arrangement of the bonding layer, it is more convenient without performing complicated pressure control or improvement on the apparatus for achieving pressure uniformity during the thermocompression bonding in the second step. In this way, the overflow of the molten layer can be suppressed.

  Preferably, in the second step of each of the above methods, at least one of the first substrate and the second substrate is sandwiched through a flexible member.

  In this case, a substantially uniform pressure is applied to the entire surface of the substrate through the flexible member, and as a result, bonding can be made uniform in the surface. Because variations in pressure during thermocompression bonding can be suppressed, precision processing that increases the flatness of the members that sandwich the substrate and processing that suppresses in-plane variation in the substrate thickness are not required. The inside can be uniformly joined.

According to the manufacturing how the electronic component of the present invention, it is possible to suppress the protrusion of the molten layer.

  Examples of the present invention will be described below with reference to FIGS.

Example 1
A BAW (bulk elastic wave) device using a piezoelectric thin film vibrator (diaphragm type) will be described. However, the present invention can also be applied to elements that require hollow formation and that need to be sealed, such as SAW (surface acoustic wave) devices and MEMS devices having movable parts.

  As shown in FIG. 1, the BAW device 10 sequentially forms a lower electrode 14, a piezoelectric thin film 16, and an upper electrode 18 on a diaphragm of an insulating film 13 supported by a Si substrate 11 having a cavity 11x. This is an element that causes piezoelectric resonance between 14 and 18. Bumps 15 and 19 are connected to the electrodes 14 and 18, respectively. The resonance characteristics can be controlled by the film thickness and shape of the electrodes 14 and 18 and the thickness of the piezoelectric thin film 16. Since the BAW element 10 piezoelectrically vibrates the resonator element, a hollow portion is required so as not to inhibit the vibration. In addition, when the piezoelectric thin film 16 absorbs moisture due to humidity and its weight changes, the vibration characteristics change to cause a problem, so that hermetic sealing is required. Therefore, the lid glass substrate 20 is disposed on the Si substrate surface 11a side, and is joined and sealed by the sealing frames 17 and 28 so as to surround the resonator element along the outer edge.

  The lid glass substrate 20 has lands 26 and 27 formed on the surface facing the Si substrate 11 and external electrodes 22 and 23 exposed to the outside on the opposite surface. The lands 26, 27 and the external electrodes 22, 23 are connected to both ends of through holes 24, 25 formed in through holes that penetrate the lid glass substrate 20, respectively. The lands 26 and 27 are bonded to the bumps 15 and 19 of the Si substrate 11. Further, the back glass substrate 30 is also bonded to the Si substrate back surface 11b, and the diaphragm portion is sealed. By joining these three substrates 11, 20, and 30 in a collective substrate (wafer) state and dicing after the joining, a small chip size package can be obtained.

  In addition, since the piezoelectric thin film is not exposed on the back surface of the Si substrate, airtight sealing may not be necessary. In that case, a resin substrate is used instead of the back glass substrate, and the resin is bonded and sealed. Also good.

  Next, solder formation and a sealing process for joining and sealing the lid glass substrate 20 and the Si substrate surface 11a will be described with reference to FIGS. The bonding of the back glass substrate 30 and the Si substrate back surface 11b may be performed using the same process.

  (Step 1) As shown in FIG. 2, a bonding pattern 42 is formed on one side 41 of a Si wafer 40 on which a BAW resonator element (hereinafter referred to as “resonator element”), which is a functional element, is formed.

  As shown in FIG. 2A, the resonator element is formed on one surface 41 of the Si wafer 40 and the pads 15 and 19 and the sealing frame 17 for taking out the electrodes from the upper and lower electrodes of the resonator element are formed. Form. The sealing frame 17 is formed so as to surround the resonator element and the pads 15 and 19. In order to electrically ground the sealing frame 17, wiring is routed so that the ground terminal of the resonator element and the sealing frame 17 are electrically connected. The functional element is not limited to BAW as long as it is an element that requires a hollow part, such as an element having a movable part. Of course, the wafer material is not limited to a semiconductor Si wafer, and may be a piezoelectric wafer, an insulator wafer, or the like.

  Here, the dimensions of the sealing frame 17 are 0.03 mm in width, the outer dimension of the sealing frame 17 is 1.8 × 1.4 mm, and the corner portion has an inner R of 0.03 mm.

  A resist pattern of the sealing frame 17 and the pads 15 and 19 having the dimensions is formed by a photoresist method, and then, as shown in FIG. 2B, the first bonding pattern 42 is formed on one surface 41 of the Si wafer 40. Ti is deposited as the layer 43, Ni is deposited as the second layer 44, and Au is deposited as the third layer 45. Finally, the resist is removed with an organic solvent to form a pattern of the Ti / Ni / Au frame 17. A method other than such a vapor deposition lift-off method, for example, a metal mask vapor deposition method, an electrolytic plating method, an electroless plating method, or the like may be used.

  The thicknesses of the metal layers 43 to 45 were set such that the Ti of the first layer 43 was 0.05 μm, the Ni of the second layer 44 was 0.3 μm, and the Au of the third layer 45 was 2.3 μm. Ti here is a layer for ensuring adhesion, and may be about 0.005 μm to 0.5 μm. The material may also be NiCr, Cr, or the like. Ni is a barrier metal that suppresses diffusion of Ti into the solder, and may be Pt or the like. When the film thickness of Ti is large, Ti itself has a barrier property, so Ni may be omitted. Au forms an alloy by forming an alloy with Sn formed on the lid glass substrate 20 to be described later, and the alloy composition is intermediate between Au-29 wt% Sn (Au-20 wt% Sn eutectic point and δ phase). ) To set the film thickness.

  As the pads 15 and 19 that are the electrode extraction portions from the resonator element, the pads 15 and 19 having the same film thickness as the sealing frame 17 and a diameter of 150 μm were formed at the same time.

  (Step 2) As shown in FIG. 3, a bonding pattern 52 is formed on a bonding surface 51 of a heat-resistant glass lid glass wafer 50 to be the lid glass substrate 20.

  In the lid glass wafer 50, a through-hole is formed by sandblasting, a plating power supply film extending from the joint surface 51 to the external terminal portion through the side wall portion is formed, and then a metal film is formed up to the land, through-hole, and external electrode by electrolytic plating. Formed. The metal film was Ti / Cu / Ni / Au.

  As shown in FIG. 3A, on the bonding surface 51 side of the wafer 50, the sealing frame 28 corresponding to the Si wafer 40 formed in (Step 1) and pad bonding bumps (lands) 26, 27 are formed. A metal film was formed. The lands 26 and 27 were formed in a ring shape surrounding a through hole (diameter 100 mm). As shown in FIG. 3B, these metal films (bonding patterns) 52 are formed by sequentially depositing Ti of the first layer 53, Ni of the second layer 54, and Sn of the third layer 55 by the vapor deposition lift-off method. did.

  The thicknesses of the layers 53 to 55 were 0.05 μm for Ti of the first layer 53, 0.3 μm for Ni for the second layer 54, and 2.7 μm for Sn for the third layer 55. Here, Ti is a layer for ensuring adhesion, and may be about 0.005 to 0.5 μm, and the material may be NiCr, Cr, or the like. Ni is a barrier metal that suppresses the diffusion of Ti into the solder, and may be Pt or the like. When the film thickness of Ti is large, Ti itself has a barrier property, so it may not be Ni. Sn in the third layer 55 forms a bonding alloy by forming an alloy with the Au in the third layer 45 formed on the Si wafer 40, and the composition thereof is Au-29 wt% Sn (Au-20 wt% Sn eutectic). The film thickness was set so as to be between the point and the δ phase.

  Here, the lid glass wafer 50 is not limited to the heat-resistant glass as long as it has the same linear expansion coefficient as the Si wafer 40 to be bonded, and may be Si or the like.

  (Step 3) As shown in FIG. 4, the Si wafer 40 and the lid glass wafer 50 are aligned.

  Alignment is performed so that the bonding layers and bump portions (bumps and lands) of the sealing frames formed on each of them are bonded, and they are fixed with a holding jig (not shown). For alignment, an alignment mark (not shown) formed on the Si wafer 40 that can be seen through the lid glass wafer 50 and an alignment mark (not shown) formed on the lid glass wafer 50 are obtained by a method such as image recognition. After alignment, it was fixed with a holding jig.

  (Step 4) As shown in FIG. 5, the Si wafer 40 and the lid glass wafer 50 are heated while being pressurized.

  First, the Si wafer 40 and the lid glass wafer 50 fixed by the holding jig are loaded into the thermocompression bonding apparatus. As shown in FIG. 5A, the thermocompression bonding apparatus has a heater stage 2 and a heater head 6 and can be heated, and wafers 40 and 50 can be sandwiched between the stage 2 and the head 6 to apply pressure. The thermocompression bonding is possible in an inert atmosphere or vacuum.

With a thermocompression bonding apparatus, wafers 40 and 50 fixed on a stage 2 with a holding jig are loaded, a fluororesin sheet 4 having a thickness of 0.5 mm is placed on the wafers 40 and 50, and N ₂ is placed inside the apparatus. Then, the inert atmosphere was introduced, and then the head 6 was brought into contact with the load from above to apply a load. By this load, Au of the third layer 45 formed on the Si wafer 40 side and Sn of the third layer 55 formed on the lid glass wafer 50 side were brought into contact with each other. Note that due to the height variation of the Si wafer 40 and the lid glass wafer 50, Au of the third layer 45 formed on the Si wafer 40 side and Sn of the third layer 55 on the lid glass wafer 50 side are partially. There is a possibility that a non-contact portion may occur. The bonding pattern on the right side of FIG. 5 (a) shows such a state.

  Here, by using the fluororesin sheet 4 having a thickness of 0.5 mm, the pressure uniformity is improved over the entire surfaces of the wafers 40 and 50, and Au and the third layer 45 on the Si wafer 40 side over the entire surfaces of the wafers 40 and 50. The Sn of the third layer 55 on the lid glass wafer 50 side can be brought into contact. As another measure for improving the pressure uniformity, there is a method of increasing the flatness of the heater stage 2 and the heater head 6 and reducing the in-plane variation of the thickness of the wafers 40 and 50 themselves, but the cost becomes high. In addition, there are many manufacturing problems such as the heater stage 2 and the heater head 6 being distorted by heat or being scratched during work. In comparison, the fluororesin sheet is inexpensive and is an excellent method that can tolerate tolerances of the apparatus and the wafer itself.

  Here, the fluororesin sheet is used as a flexible member in consideration of heat resistance. However, other than the fluororesin sheet, a rubber-based material (for example, silicon rubber or Viton rubber), a metal or glass plate that softens at this thermocompression bonding temperature, or a composite material in which glass fibers are hardened with a resin material may be used. Of course, the thickness may be other than 0.5 mm. Further, the surface may be uneven to facilitate deformation. As the uneven shape, when a cone, a pyramid, a cylinder, a prism, a hemisphere, or the like is arranged on a sheet, the pressure uniformity is improved.

The atmosphere may be a gas atmosphere such as N ₂ or Ar, a vacuum atmosphere, or a reducing atmosphere as long as oxidation of the bonding metal can be prevented.

  Further, the load is set so that the wafers 40 and 50 having a diameter of 100 mm are pressurized with 2000 N. However, if the load allows the wafers 40 and 50 to be in close contact with each other without causing the wafers 40 and 50 to break, It can be large or small. In some cases, a larger load can eliminate voids generated during solid phase diffusion.

Subsequently, it heats to 240 degreeC with the temperature increase rate of 30 degrees C / min. In this temperature raising process, solid phase diffusion occurs between the Au contact portion of the third layer 45 formed on the Si wafer 40 side and the Sn contact portion of the third layer 55 formed on the lid glass wafer 50 side. AuSn ₄ , AuSn ₂ , and AuSn are formed and grown at the interface between Sn and Sn, and finally AuSn and AuSn ₂ are formed as an intermediate layer. As shown in the phase diagram of FIG. 7, the eutectic point of Au-90 wt% Sn is 219 ° C., but when it reaches 219 ° C. during the temperature rising process, AuSn ₂ and AuSn are formed by solid phase diffusion. Therefore, in this temperature rising process, Au and Sn are not melted, and the liquid phase of those alloys is not generated. Therefore, the solder does not melt and flow out at the time of thermocompression bonding, and the protruding solder does not aggregate.

  In this step, since it is solid phase diffusion, for example, as shown in FIG. 5B, the entire circumference of the sealing frame may not contact evenly and may not be solid phase diffused. Then, there may be a portion where the unbonded portion 57 remains and cannot be sealed.

Moreover, although it heated up to 240 degreeC, if the mechanism described here is possible, the temperature increase rate and ultimate temperature will not be limited to this. However, since the growth of the alloy phase by solid phase diffusion is slow when the ultimate temperature is low, generally 200 ° C. or higher is desirable. On the other hand, if it exceeds 280 ° C., it will be remelted, so it is necessary to make the temperature substantially below 275 ° C. In addition, if the rate of temperature rise is high, the temperature rises before solid phase diffusion sufficiently proceeds, a liquid phase is generated at 219 ° C. from the interface between Sn and AuSn ₄ , and the liquid phase component protrudes and aggregates due to pressure. End up. Therefore, it is necessary to optimize the heating rate.

  (Step 5) As shown in FIG. 6, the pressure is set to zero, the temperature is set to the melting point of Au-20 wt% Sn or more, and then cooled.

  The pressure (load) applied to the wafers 40 and 50 is reduced or zero, and then the temperature exceeds the melting point (280 ° C.) of Au-20 wt% Sn. Here, as shown in FIG. 6 (a), the heater head 6 is raised so that the load applied to the wafers 40 and 50 becomes zero, and the variation in the surface temperature of the heater stage 2 is ± 10 ° C. Heated to 300 ° C. After reaching 300 ° C., the heater 2 is cooled.

Here, by heating to 300 ° C., AuSn or AuSn ₂ formed in the previous step reacts with Au remaining on the pattern to form a liquid phase part. The liquid phase flows to the portion 57 that was not joined at the time of the previous thermocompression bonding, and the unjoined portion 57 is joined as shown in FIG. As a result, the entire circumference of the sealing frame is joined and sealed. In addition, the load is zero, and the molten component does not flow out of the sealing frame. In addition, since there is no crushing due to crimping, the thickness of the solder joint can be made substantially equal to the thickness at the time of film formation, and the vibration problem of the resonator element can be secured. Similarly, the bump portion is also bonded to ensure conduction, and the periphery of the through hole is bonded, so that the element can be sealed from the through hole. In addition, even if it does not make a load zero, by making smaller than (process 4), the protrusion of the fusion | melting component from the sealing frame and the collapse of a sealing frame can be suppressed.

  The component of the liquid phase generated here is the composition ratio of the deposited Au and Sn, and is about Au-29 wt% Sn (intermediate between Au-20 wt% Sn eutectic point and δ phase). In addition, Ni formed as a barrier in this process is also partially eroded in the AuSn liquid phase, and although the bonding alloy is actually an AuSn alloy containing a slight amount of Ni, sufficient bonding strength and reliability are ensured. Yes, there are no defects.

  Through the above (Step 1) to (Step 5), the Si wafer surface 11a on which the BAW element was formed and the lid glass wafer 50 were bonded and sealed in the wafer state. In the same manner, the Si wafer back surface 11b was bonded to the back glass 30 and sealed, and then diced to form a BAW device. Next, in order to perform sealing and electrical property inspection, the BAW device is depressurized in the chamber, immersed in Fluorinert (an inert liquid having a small surface tension) in a depressurized state, and the pressure of the Fluorinert liquid is immersed in the immersed state. Raised. As a result, when the sealing is incomplete, the fluorinate penetrates into the BAW device and adheres to the resonator element, resulting in a deteriorated electrical characteristic. By measuring the electrical characteristics of such BAW devices, the characteristics and sealing were inspected. In this way, a small BAW device with high reliability can be obtained.

  The operations and effects of Example 1 are as follows.

  First, the occurrence of protrusion can be suppressed. That is, when AuSn solder is melted, no pressure is applied and the solder does not protrude. For this reason, no defects such as a short circuit defect or an adverse effect on the subsequent dicing process occur, and the defects are reduced.

  Second, uniform pressurization within the wafer surface can be achieved, and sealing failure is unlikely to occur. In other words, a uniform material can be applied in the wafer surface with a flexible material, and the surface can be substantially uniformly solid-phase diffused at a low temperature and at a low pressure without increasing the apparatus price or wafer price. Therefore, since there are few uncompressed parts, the sealing failure after performing the subsequent fusion bonding can be reduced.

  Third, the size can be reduced, and the productivity is improved. That is, since the occurrence of protrusion can be suppressed, the distance between the frames and the distance between the frames and the bumps and the wiring pattern can be reduced, which contributes to the miniaturization of the element. Furthermore, by reducing the size, the number of wafers taken can be increased, and productivity is improved.

  Fourth, a vibration space can be secured. That is, the bonding solder thickness (height) is substantially equal to the film formation value, and a sufficient vibration space can be secured.

(Example 2)
A BAW filter having the same configuration as that of the first embodiment but different in the configuration of the solder joint and the thermocompression bonding method will be described with reference to FIGS. Below, it demonstrates centering around difference with Example 1 along process flow.

  (Step 1) As shown in FIG. 8, a bonding pattern 62 including a frame electrode 17 is formed on one surface 41 of a Si wafer 40 on which functional elements are formed.

  As shown in FIG. 8A, as in the first embodiment, a resonator element is formed on one surface 41 of the Si wafer 40, and electrodes are taken out from the upper and lower electrodes of the resonator element by a vapor deposition lift-off method. Pads 15 and 19 and a sealing frame 17 are formed. The sealing frame 17 is formed so as to surround the resonator element and the pads 15 and 19. In order to electrically ground the sealing frame 17, wiring is routed so that the ground terminal of the resonator element and the sealing frame 17 are electrically connected.

  Here, the sealing frame 17 has a width of 0.05 mm, an outer dimension of the sealing frame 17 of 1.1 × 0.9 mm, and the corner portion has an inner R of 0.03 mm. A resist pattern of this size is formed by a photoresist method, and then, as shown in FIG. 8B, Ti of the first layer 63 of the bonding pattern 62, Cu of the second layer 64, and third layer 65 of the third layer 65 Au was sequentially deposited, and finally the resist was removed with an organic solvent to form a pattern of the Ti / Cu / Au frame 17 on the Si wafer 40.

The thickness of each of the layers 63 to 65 was 0.05 μm for Ti of the first layer 63, 5.0 μm for Cu for the second layer 64, and 0.1 μm for Au for the third layer 65. Ti here is a layer for ensuring adhesion, and may be about 0.005 μm to 0.5 μm, and the material may be NiCr, Cr or the like. Cu mutually diffuses with Sn of the third layer 75 formed on the lid glass wafer 50 described later to form a bonding alloy. The film thickness was set so that the composition of the bonding alloy was Cu richer than the composition of Cu ₃ Sn. Au in the third layer 65 is an antioxidant film on the Cu surface of the second layer 64. If Au is thin, it is interdiffused with Cu before bonding, and Cu forms an oxide on the surface, so an appropriate thickness is required. The thickness is preferably 0.05 μm or more. On the other hand, if the thickness is large, it becomes expensive. Taking these into account, the film thickness was set as described above. Bumps 15 and 19 having the same film thickness as the sealing frame 17 were formed at the same time on the electrode extraction portion from the resonator element. Here, unlike Example 1, the diameter was 100 μm.

  Further, since the Cu wiring can be shared with that used for the electrode lead-out wiring from the resonator element, it may be formed simultaneously with the lead-out wiring.

  (Step 2) As shown in FIG. 9, a bonding pattern 72 including the sealing frame 28 is formed on the bonding surface 51 of the heat-resistant glass lid glass wafer 50.

  In order to form lands, through-holes, and external electrodes on the lid glass wafer 50, through-holes are formed by sandblasting, and after forming a power supply film for plating from the joint surface 71 to the external terminal portion through the side wall portion, electrolytic plating is performed. Then, a metal film was formed up to the bump bonding surface, side wall, and external terminal portion. The metal film was Ti / Cu / Ni / Au.

  As shown in FIG. 9A, the sealing frame formed on the Si wafer 40 on the bonding surface 51 side of the lid glass wafer 50 corresponds to the Si wafer 40 formed in (Step 1). The metal film of the sealing frame 28 and the lands 26 and 27 was formed so that the line width was smaller than the width of 17 and the diameter of the bumps 15 and 19. Here, the width of the sealing frame 28 on the side of the lid glass wafer 50 is set to 25 μm, and the bump portions (lands) 26 and 27 are formed in a ring shape having a width of 20 μm surrounding the through hole. As shown in FIG. 9B, these metal films (bonding patterns) 72 are formed by sequentially depositing Ti of the first layer 73, Cu of the second layer 74, and Sn of the third layer 75 by the vapor deposition lift-off method. Filmed.

The thicknesses of the layers 73 to 75 were set such that the Ti of the first layer 73 was 0.05 μm, the Cu of the second layer 74 was 7.5 μm, and the Sn of the third layer 75 was 5.0 μm. Ti here is a layer for ensuring adhesion, and may be about 0.005 μm to 0.5 μm, and the material may be NiCr, Cr or the like. Sn melts at the time of bonding and interdiffuses with Cu to form a bonding metal. Regarding the film thickness, in-plane variations in the wafer thickness of Si and heat-resistant glass in wafer level bonding, and the flatness of the bonding head and stage In consideration of the above, the thickness that can be surely contacted and joined by the molten Sn is set to 5 μm. Cu mutually diffuses with Sn of the third layer 75 formed on the lid glass wafer 50 described later to form a bonding alloy. The composition of the bonding alloy was made to be richer than the composition of Cu ₃ Sn, and the film thickness of Cu was set.

The film thickness and line width of the metal layer should be set to the following dimensional relationship. As shown in FIG. 14, the width of one of the sealing frame and W _1, and the thickness of the metal layer which does not melt by the temperature at the time of joining of the sealing frame and H _1, the metal layer melts by a temperature at the time of joining the thickness and _{H 2.} Further, when the width of the other sealing frame is W ₂ and the thickness of the metal layer that melts during bonding is H ₃ among the metal layers of the sealing frame, the following formulas (1) and (2) It is preferable to form the metal layer so as to satisfy the relationship.
W ₁ · H ₂ + W ₂ · H ₃ ≦ (W ₂ −W ₁ ) · H ₁ (1)
W ₂ > W ₁ (2)
When there is no metal layer that melts at the time of joining on the other sealing frame side, calculation is performed assuming that H ₃ = 0, and the metal layer may be formed.

  The lid glass wafer 50 is not limited to heat-resistant glass as long as it is a material having the same linear expansion coefficient as the Si wafer 40 to be joined, and may be Si or the like.

  (Step 3) As shown in FIG. 10, the Si wafer 40 and the lid glass wafer 50 are aligned.

  In the same manner as in the first embodiment, the sealing frames and the bumps formed on the respective parts are aligned so as to be joined and fixed with a holding jig.

  (Step 4) As shown in FIG. 11, the wafers 40 and 50 are heated while being pressurized.

  First, as shown in FIG. 11A, as in the first embodiment, the wafers 40 and 50 fixed by the holding jig are loaded into the thermocompression bonding apparatus.

With the thermocompression bonding apparatus, the wafers 40 and 50 fixed on the holding jig were loaded on the stage 2 to create an inert atmosphere, and then the head 6 was brought into contact with the load from above to apply a load. The load was applied so that the load / bonding area = 7.5 MPa. The atmosphere may be a gas atmosphere such as N ₂ or Ar or a vacuum atmosphere as long as the bonding metal can be prevented from being oxidized. Moreover, the load of 7.5 MPa may be larger or smaller than this.

  Subsequently, it heats to 280 degreeC. In this temperature raising process, Sn of the third layer 75 formed on the lid glass wafer 50 is melted and wetted and bonded to the bonding pattern 62 of Cu or Au formed on the Si wafer 40. At this time, since the load is applied to the wafers 40 and 50 even when Sn of the third layer on the lid glass wafer 50 side melts, the melted Sn flows and the second layer formed on the lid glass wafer 50 side. It does not stay on Cu. Since the Cu width of the second layer of the second layer 64 formed on the electrode on the Si wafer 40 side is wide, Sn flows along it. Therefore, Sn hardly protrudes outside the bonding pattern 72 on the Si wafer 40 side. In addition, as described in Example 1 during pressurization, when a material having flexibility such as a fluororesin sheet or a rubber sheet is inserted and thermocompression bonded, the pressure uniformity is increased and the sealing good product rate can be further improved. .

  Thus, Sn is wet and spread along the sealing frame on the Si wafer 40 side, and the entire periphery of the sealing frame is joined and then cooled, whereby the sealing is completed. At the same time, the bump part is also bonded.

  According to this method, since the second layer of Cu on the lid glass wafer 50 side is not melted, the bonding height between the wafers 40 and 50 can be made a certain level or more, and the vibration space of the resonator element can be secured.

Next, a pressure of 7.5 MPa was applied at 280 ° C., and thermal annealing was performed for 30 minutes. At this time, as shown in enlarged view in FIG. 12 (a), initially, _Cu 3 near Cu layer 64, 74 _S n, alloy portion of the Cu and Sn, including _Cu 6 Sn ₅ 81 and 83 Is formed, and the Sn portion 82 is interposed between the Cu layers 64 and 74. Finally, as shown in FIG. 12B, Cu ₃ Sn is inserted between the Cu layers 64 and 74. An alloy portion 84 of Cu and Sn as main components is continuously formed, and the Sn portion 85 is almost eliminated. Thereby, it will be in the joining state which is not remelted at the time of mounting to a motherboard (about 240 degreeC). That is, as shown in the state diagram of FIG. 13, Cu ₆ Sn ₅ (η) and Cu ₃ Sn (ε) were formed, the melting point of the bonding alloy was increased, and a highly heat-resistant bonding sealing alloy was formed. The temperature and time are not limited as long as the high melting point alloy is formed. As for pressurization, Kirkendall voids may occur due to the difference in diffusion rate between Cu and Sn, which may result in poor sealing. In order to suppress this, pressurization of several MPa or more is required.

  Here, although Cu and Sn are used, any metal other than Cu may be used as long as it does not melt into the solder at the bonding temperature, and any metal other than Sn that melts at the bonding temperature may be used. For example, Ni is deposited on Cu and made into a high melting point alloy of NiSn, or AuSn is deposited on Cu and AuSn eutectic bonding is used, Cu or Au using In solder is used. For example, a high melting point alloy may be obtained by reacting with Ag.

  The Si wafer surface 11a on which the resonator element was formed by the above (Step 1) to (Step 4) and the lid glass wafer 50 were bonded and sealed in a wafer state. In the same manner, the Si wafer back surface 11b was bonded to the back glass plate 30 and sealed, and then diced to form a BAW device. Next, in order to perform sealing and electrical property inspection, the BAW device is depressurized in the chamber, immersed in Fluorinert (an inert liquid having a small surface tension) in a depressurized state, and the pressure of the Fluorinert liquid is immersed in the immersed state. Raised. As a result, when the sealing is incomplete, the fluorinate penetrates into the BAW device and adheres to the resonator element, resulting in a deteriorated electrical characteristic. By measuring the electrical characteristics of such BAW devices, the characteristics and sealing were inspected. In this way, a small BAW device with high reliability can be obtained.

  The operation and effect of the second embodiment is as follows.

  First, the occurrence of protrusion can be suppressed. That is, even when the solder is melted, the wetting is controlled by the electrode on the Si side, so that the solder does not protrude. For this reason, there are no problems such as a short circuit defect or an adverse effect on the subsequent dicing process, and the defect is reduced.

  Second, by providing a molten Sn thickness that can absorb variations, sealing failure is unlikely to occur. That is, there is a thickness of molten solder that can absorb variations in the wafer surface and variations in the flatness of the head and stage of the apparatus, and sealing defects can be reduced.

  Third, the size can be reduced, and the productivity is improved. That is, since the occurrence of protrusion can be suppressed, the distance between the frames and the distance between the frames and the bumps and the wiring pattern can be reduced, which contributes to the miniaturization of the element. Furthermore, by reducing the size, the number of wafers taken can be increased, and productivity is improved. However, compared with the first embodiment, the effect is small.

  Fourth, the height can be controlled by the unmelted layer of Cu, and a vibration space can be secured. This is because the Cu post does not melt during thermocompression bonding.

  Fifth, material costs are lower than in the first embodiment. This is because the amount of Au used is small.

  In addition, the manufacturing method of the electronic component of this invention and an electronic component using the same are not limited to above-mentioned embodiment, It can implement by adding a various change.

  The present invention relates to an element that requires the formation of a hollow part in order to maintain a vibration space and that needs to be sealed to protect the vibration part from moisture, dust, etc., particularly an elastic surface in the communication field Although it is suitable for a piezoelectric filter (BAW filter) using a wave filter (SAW filter) or a piezoelectric thin film resonator, the present invention is not limited to this, and other devices as well as a micro machine having a movable part (Micro Electro Mechanical). (System: MEMS) can be widely applied.

It is a block diagram of a BAW device. Example 1 It is explanatory drawing of the manufacturing process of a BAW device. Example 1 It is explanatory drawing of the manufacturing process of a BAW device. Example 1 It is explanatory drawing of the manufacturing process of a BAW device. Example 1 It is explanatory drawing of the manufacturing process of a BAW device. Example 1 It is explanatory drawing of the manufacturing process of a BAW device. Example 1 It is a phase diagram of an AuSn alloy. Example 1 It is explanatory drawing of the manufacturing process of a BAW device. (Example 2) It is explanatory drawing of the manufacturing process of a BAW device. (Example 2) It is explanatory drawing of the manufacturing process of a BAW device. (Example 2) It is explanatory drawing of the manufacturing process of a BAW device. (Example 2) It is explanatory drawing of the manufacturing process of a BAW device. (Example 2) It is a phase diagram of a CuSn alloy. (Example 2) It is explanatory drawing of the manufacturing process of a BAW device. (Example 2) It is a block diagram of a BAW device. (Conventional example)

Explanation of symbols

10 BAW devices (electronic components)
40 Si wafer (first substrate)
42 bonding pattern 43 first layer 44 second layer 45 third layer (first bonding layer)
50 Lid glass wafer (second substrate)
52 bonding pattern 53 first layer 54 second layer 55 third layer (second bonding layer)
62 bonding pattern 63 first layer 64 second layer 65 third layer (first bonding layer)
72 bonding pattern 73 first layer 74 second layer (third bonding layer)
75 Third layer (second bonding layer)

Claims (3)

The first substrate is made of Au on the first substrate so that the weight ratio of Au to Sn (Au: Sn) in the first bonding layer and the second bonding layer is in the range of 84:16 to 62:38. Forming a bonding layer and forming the second bonding layer of Sn on a second substrate;
The second substrate is sandwiched between the first substrate and the second substrate so that the first bonding layer and the second bonding layer are in contact with each other, and is thermocompression bonded at a temperature of 200 ° C. or higher and lower than 280 ° C. And the steps
An electronic component comprising: a third step of heat-treating the first substrate and the second substrate at a pressure of 280 ° C. or higher and lower than 521 ° C. with a pressure lower than that of the second step. Production method. In the first step, the second bonding layer having an area smaller than that of the first bonding layer is formed, and the first bonding layer is formed between the second bonding layer and the second substrate. Forming a third bonding layer having an area smaller than the bonding layer of
The method for manufacturing an electronic component according to claim 1, wherein in the second step, the third bonding layer does not melt.   3. The electronic component according to claim 1, wherein in the second step, at least one of the first substrate and the second substrate is sandwiched through a flexible member. Production method.

Images