BONDING STRUCTURE AND SEMICONDUCTOR DEVICE HAVING THE BONDING STRUCTURE

Abstract

A bonding structure to be formed in bonding of a Si semiconductor or the like and a substrate, as materials to be bonded, with transient liquid phase diffusion bonding. The bonding structure includes a bonding portion presenting a specific material texture. A metal composition of the bonding portion includes 78.0% by mass or more and 80.0% by mass or less of Ag, 20.0% by mass or more and 22.0% by mass or less of Sn, and an inevitable impurity element. A characteristic material texture configured from an island-like Ag phase including 95% by mass or more of Ag and a Ag.sub.3Sn phase including a Ag.sub.3Sn intermetallic compound and surrounding the island-like Ag phase is presented in observation of any cross section of the bonding portion.

Claims

1. A bonding structure comprising: a pair of materials to be bonded; and a bonding portion formed between the pair of materials to be bonded, wherein the bonding portion comprises, as constituent elements, 78.0% by mass or more and by mass or less of Ag, 20.0% by mass or more and 22.0% by mass or less of Sn, and an inevitable impurity element, and a material texture configured from an island-like Ag phase comprising 95% by mass or more of Ag and a Ag.sub.3Sn phase comprising a Ag.sub.3Sn intermetallic compound and surrounding the island-like Ag phase is observed in observation of any cross section of the bonding portion.

2. The bonding structure according to claim 1, wherein an area percentage of the island-like Ag phase in any cross section of the bonding portion is 18% or more and 35% or less.

3. The bonding structure according to claim 1, wherein a void ratio in any cross section of the bonding portion, in terms of area percentage, is 4% or less.

4. A semiconductor device comprising a semiconductor element and a substrate bonded to each other, wherein the semiconductor device comprises the bonding structure defined in claim 1, between the semiconductor element and the substrate.

5. A bonding material for transient liquid phase diffusion bonding, for use in bonding of a pair of materials to be bonded, with transient liquid phase diffusion bonding, wherein the bonding material comprises a sheet-shaped compression-molded article obtained by compression and rolling of a mixed powder of a Ag powder having an average particle size of 30 μm or more and 75 μm or less and a Sn powder having an average particle size of 1 μm or more and 20 μm or less, and the compression-molded article has a relative density of 95% or more based on a density of a bulky metal having the same composition as the mixing ratio in the mixed powder.

6. The bonding structure according to claim 2, wherein a void ratio in any cross section of the bonding portion, in terms of area percentage, is 4% or less.

7. A semiconductor device comprising a semiconductor element and a substrate bonded to each other, wherein the semiconductor device comprises the bonding structure defined in claim 2, between the semiconductor element and the substrate.

8. A semiconductor device comprising a semiconductor element and a substrate bonded to each other, wherein the semiconductor device comprises the bonding structure defined in claim 3, between the semiconductor element and the substrate.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0047] FIG. 1 A view schematically illustrating a material texture of the bonding portion in the bonding structure of the invention.

[0048] FIG. 2 A photograph representing a material texture of a bonding portion in a bonding structure with respect to each sample produced in the present embodiment.

[0049] FIG. 3 A mapping image with EDS, of a bonding portion in a bonding structure with respect to each sample produced in the embodiment.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0050] Hereinafter, an embodiment of the invention is described with reference to Examples described below. In the embodiment, a compression-molded article was produced from a Ag powder and a Sn powder, and was adopted as a bonding material for bonding a semiconductor chip and a substrate, and a bonding portion was subjected to material texture observation and evaluation of properties such as heat resistance.

[0051] First, a Ag powder (purity 99.9% by mass, average particle size 54.6 μm) produced by an atomization method, and a Sn powder (purity 99.9% by mass, average particle size 3.7 μm) produced by an atomization method were mixed in a shaker, and thus a mixture was obtained. The average particle sizes of the Ag powder and the Sn powder were determined in observation of each of the powders with SEM (magnifications: 200× (Ag powder) and 2000× (Sn powder), image resolution: 1024×768, acceleration voltage: 15 kV). The average area of each of the powders was calculated from the areas and the number of such powders with a function attached to a SEM apparatus, and the equivalent circle diameter was defined as the average particle size. In the embodiment, the above powder having an average particle size of 54.6 μm and the above powder having an average particle size of 3.7 μm were respectively mainly used as the Ag powder and the Sn powder, and in some comparative examples, a Ag powder having an average particle size of 90 μm and a Sn powder having an average particle size of 24 μm were used. These powders were also each produced by an atomization method.

[0052] In the embodiment, the mixing ratio between the Ag powder and the Sn powder was changed, and each mixture where the Sn content was 20% by mass, 25% by mass, 27% by mass, 30% by mass, 35% by mass, 40% by mass, 45% by mass, or 50% by mass was produced. Such each mixture of the Ag powder and the Sn powder was poured into a cylindrical mold, and compressed in a hydraulic pressing machine, and thus a flat disc-shaped molded article was obtained. The pressure in pressing was 5 ton. The compressed article completed was rolled on a desk roller, and thus a sheet-shaped compression-molded article having a thickness of 0.1 mm was obtained and adopted as a bonding material. The relative density of a Ag/Sn alloy (bulky metal) having the same composition as the mixing ratio in the bonding material produced in the embodiment was measured by the Archimedes method, and was confirmed to fall within the range from 95% to 100% even if the measurement error was counted.

[0053] Next, a plurality of bonding materials produced as above were used for bonding of a semiconductor chip and a circuit substrate, and thus a bonding structure was formed. A semiconductor chip was a Si chip, and bonded to a Cu substrate (KFC material) as a circuit substrate.

[0054] In the embodiment, a Si chip (2 mm×2 mm) having one surface metalized with Ti (0.01 μm)/Ni (0.3 μm)/Ag (0.2 μm) was prepared for production of a sample for a bonding strength evaluation test (die shear test) described below. Four of the Si chips were placed on a carbon jig so that surfaces metalized faced upward. Next, one of the bonding material (10 mm×10 mm) produced as above was placed on the four Si chips, and then a Cu substrate (11 mm×11 mm, thickness 0.2 mm) was placed on the bonding material. A carbon jig was placed thereon, and the Si chip/bonding material/Cu substrate was sandwiched and thereafter secured with a jig equipped with a spring (pressure applied: 2 MPa).

[0055] A Si chip (10 mm×10 mm) having the same configuration as described above was prepared for production of each sample for cross section observation and a heat cycle test of a bonding portion described below. One of the Si chip was placed in the same procedure as described above, one bonding material (10 mm×10 mm) was placed thereon, furthermore a Cu substrate (11 mm×11 mm, thickness 0.2 mm) was placed thereon, and the resultant was secured with a carbon jig (pressure applied: 2 MPa).

[0056] A sample of the Si chip/bonding material/Cu substrate secured as described above was introduced together with the jig into a low-oxygen oven (IPHH-202MS manufactured by Espec Corp.), and heated, and thus a bonding portion was formed. Such a heating treatment was performed with the start of temperature rise after the oven was purged with nitrogen for 30 minutes and the oxygen concentration reached 70 ppm or less. The heating was performed at a rate of temperature rise of 5° C./min under nitrogen flow until the temperature reached a bonding temperature set. Once the temperature in the oven reached the bonding temperature, the resultant was heated and retained at the bonding temperature under nitrogen flow for 30 minutes. After a lapse of 30 minutes, a blower of the oven was turned on and the temperature was dropped under nitrogen flow for 30 minutes (rate of temperature drop: about −5° C./min). Once the temperature in the oven reached 40° C., the sample was taken out and the jig was removed. In the embodiment, the bonding temperature was set to 250° C. with respect to each of the above eight bonding materials. Such bonding temperatures were set to 185° C., 225° C., 250° C., 275° C., and 315° C. with respect to some of the bonding materials (Sn content 27% by mass), and thus each bonding portion was formed.

Material Texture Observation and Composition Analysis of Bonding Portion

[0057] After the bonding step and observation of a material texture of a cross section of the bonding portion, the composition (composition analysis of an island-like Ag phase and a Ag.sub.3Sn phase, the area percentage of the island-like Ag phase, and the void ratio in the cross section of the bonding portion were measured.

[0058] In such observation and measurement, the sample (Si chip/bonding material/Cu substrate) subjected to bonding structure production described above was embedded into a resin, and around the center thereof was cut, and a cross section was polished. The cross section of the sample was observed with a scanning electron microscope (SEM: JSM-IT500HR manufactured by JEOL Ltd.), and a photograph of the cross section was taken. The magnification in SEM observation was adjusted to one, which enabled even upper and lower ends of the bonding portion to be observed, and observation in the embodiment was made with the acceleration voltage being set to 15 kV and the magnification being set to 1000× in SEM. In such observation and measurement, the entire bonding portion (but including no bonding interface) was an observation region.

[0059] EDS analysis (acceleration voltage 15 kV) was performed with an energy dispersive X-ray analyzer (EDS) attached to a SEM apparatus, for composition analysis of the bonding portion. First, each phase estimated to be the island-like Ag phase and the Ag.sub.3Sn phase with SEM observation was subjected to spot analysis at three points for a number of times of sampling (measurement time 30 seconds) of 10000 per point. As a result, it was confirmed that the composition of the island-like Ag phase was 100% by mass of Ag and the composition of the Ag.sub.3Sn phase was 73% by mass of Ag-27% by mass of Sn.

[0060] Next, mapping with EDS was performed and the area percentage of the island-like Ag phase was measured. An observation region (magnification: 10000×) of the cross section of the bonding portion was first subjected to element mapping. In the embodiment, the mapping resolution was 256×192 pixels, and the analysis elements were Ag, Sn, Cu, Si, and Ni. In such mapping analysis with EDS, sampling (counting) at each pixel was preferably sufficiently performed for precise analysis of the composition at each pixel as much as possible. A suitable index exemplified was to perform sampling 2000 times or more per pixel, and in the embodiment, such each element for analysis was subjected to measurement for a number of times of sampling of 2500 per pixel.

[0061] It was confirmed from the results of mapping analysis of each of the samples that a region having a Ag concentration of about 100% by mass and a region having a Ag concentration of about 73% by mass were present in the cross section of the bonding portion. The resulting mapping image was binarized with a predetermined Ag concentration as a threshold, and a mapping image (Ag mapping image) for distinction between the Ag phase and the Ag.sub.3Sn phase was created. The Ag concentration serving as a threshold was a Ag concentration of 85% by mass set in the embodiment, and a region having a concentration equal to or more than the threshold was defined as the Ag phase and a region having a concentration equal to or less than the threshold was defined as the Ag.sub.3Sn phase for creation of a Ag mapping image. Such a Ag mapping image created was used and image analysis software (trade name: MIPAR) was applied for calculation of the area percentage of the Ag phase. FIG. 3 illustrates a mapping image of each of the samples, prepared based on the Ag concentration. A light color portion corresponds to the island-like Ag phase and a dark color portion corresponds to the Ag.sub.3Sn phase in such each image.

[0062] The void ratio was determined with measurement of the area percentage in the SEM image taken in the SEM observation, by means of the same image analysis software as described above. In the image analysis, a portion clearly determined to correspond to a void in the SEM image (a portion of a color close to black in the image) was adopted as reference, and the area percentage of a portion in the same color tone as such a void portion was measured.

Evaluation (Die Shear Test) of Bonding Strength

[0063] Next, the bonding strength of the bonding portion of each of the samples was measured. A bonding strength evaluation test (die shear test) was performed with a test at room temperature and a test in the state of heating at 260° C., of each of the samples produced as described above. In bonding strength measurement, such each sample was installed in a bond tester (MFM1200L manufactured by TRY Precision Technology Co., Ltd.) as a measurement apparatus so that the Cu substrate faced downward, a tool (click) of the tester was hung to an end of the chip, and the die shear strength was measured under application of a shear load of 200 kg. A case where the shear strength was 20 MPa or more was rated as “pass (o)”, and a case of pass in both the tests at room temperature and a high temperature was rated as “suitable bonding portion”.

Evaluation of Heat Cycle Durability

[0064] Furthermore, a heat cycle test of the bonding portion of each of the samples was performed for confirmation of durability in heating/cooling cycles. In the heat cycle test, a heat cycle was defined as one cycle of retention at −50° C. for 30 minutes/retention at 175° C. for 30 minutes and each of the samples was subjected to 250 cycles. The bonding portion after 250 cycles was confirmed with respect to peeling and the presence of cracks.

[0065] Herein, no heat cycle test was performed in any sample having a bonding strength of the lower limit value or less (5 MPa or less) at room temperature or a high temperature in the die shear test.

[0066] The above test results are shown in Table 1. Table 1 also shows the composition of the bonding portion and the area percentage of the island-like Ag phase, of each of the samples, and furthermore the measurement results of the void ratio. Each SEM photograph of bonding portions (bonding temperature 250° C.) of Nos. 1, 2, 5, 10, 11, 12, 13, and 14 in Table 1, as an observation image example of the material texture of the bonding portion of each of the samples, is illustrated in FIG. 2, and each mapping image thereof is illustrated in FIG. 3.

TABLE-US-00001 TABLE 1 Bonding portion configuration Bonding strength evaluation Composition (% by mass) Island-like Pass/ of bonding material Ag phase Room Heating at Fail Ag powder Sn powder Bonding Composition Ag temperature 260° C. in Particle Particle temper- (% by mass) content Area Shear Shear heat size Mixing size Mixing ature Ag Sn (% by per- Void strength Pass/ strength Pass/ cycle No. (μm) ratio (μm) ratio (° C.) content content mass) centage ratio [MPa] Fail [MPa] Fail test  1 54.6 80% 3.7 20% 250 81.2% 18.8% 100% 35.4% 4% 17 Fail  8 Fail Fail Com- parative Example  2 75% 25% 80.0% 20.0% 100% 33.6% 3% 25 Pass 45 Pass Pass Example  3 73% 27% 185 68.3% 31.7% 100% 36.0% 1% 5 or less Fail 5 or less Fail — Com- parative Example  4 225 78.1% 21.9% 100% 20.3% 4% 52 Pass 35 Pass Pass Example  5 250 78.9% 21.1% 100% 29.6% 2% 76 Pass 52 Pass Pass Example  6 275 79.8% 20.2% 100% 24.2% 4% 50 Pass 36 Pass Pass Example  7 315 81.2% 18.8% 100% 41.2% 3% 38 Pass 19 Fail Fail Com- parative Example  8 90.0 250 84.1% 15.9% 100% 40.3% 3% 32 Pass 11 Fail Fail Com- parative Example  9 54.6 24.0 85.0% 15.0% 100% 39.2% 3% 31 Pass 12 Fail Fail Com- parative Example 10 70% 3.7 30% 78.2% 21.8% 100% 25.8% 2% 55 Pass 31 Pass Pass Example 11 65% 35% 79.6% 20.4% 100% 30.3% 3% 45 Pass 35 Pass Pass Example 12 60% 40% 79.5% 20.5% 100% 26.1% 4% 38 Pass 21 Pass Pass Example 13 55% 45% 77.8% 22.2% 100% 17.0% 7% 34 Pass 5 or less Fail — Com- parative Example 14 50% 50% 77.2% 22.8% 100% 5.2% 4% 22 Pass 12 Fail Fail Com- parative Example

[0067] It was found with reference to Table 1, and FIG. 2 and FIG. 3 that the bonding portion having the material texture configured from the Ag phase and the Ag.sub.3Sn phase was formed with transient liquid phase diffusion bonding with the compression-molded article of the Ag powder and the Sn powder, produced in the embodiment, as a bonding material. However, the sample lack in bonding strength or inferior in heat cycle durability was formed depending on the composition of the bonding portion. Specifically, the bonding portion, when had a Ag content of less than 78.0% by mass (No. 3, 13, and 14), was inferior in bonding strength at a high temperature. The bonding portion, when had a Ag content of more than 80.0% by mass (No. 1, Nos. 7 to 9), failed in terms of bonding strength at a high temperature, although was likely higher in bonding strength than a sample having a Ag content of less than 78.0% by mass. Such a bonding portion, when did not fall within such a suitable composition range, was peeled/cracked also in the heat cycle test. Accordingly, it is considered that the bonding portion in the bonding structure of the invention, with transient liquid phase diffusion bonding, ensures bonding strength and durability due to optimization of not only the material texture of the Ag phase and the Ag.sub.3Sn phase, but also the composition range.

[0068] It is preferable from the viewpoint of optimization of the material texture to allow the area percentage of the island-like Ag phase to be suitable. The area percentage of the island-like Ag phase is not highly varied with respect to any bonding portion (No. 2, No. 4 to No. 6, and Nos. 10 to 12) having a suitable composition. Such any bonding portion has suitable bonding strength and heat cycle durability. On the other hand, when the composition of the bonding portion does not fall within the scope of the invention, the area percentage of the island-like Ag phase is likely to be less than 18% or more than 35%. In particular, a bonding portion of No. 13, although had a composition of the bonding portion, extremely close to the present range, was slightly lack in bonding strength. Such lack was considered to be due to lack of the island-like Ag phase.

[0069] While the bonding portion was formed with the compression-molded article of the Ag powder and the Sn powder, as a bonding material, in the embodiment, the configuration and bonding conditions (bonding temperature) of the bonding material also had an influence on the composition of the bonding portion. A bonding material with a coarse Ag powder or Sn powder was applied in each bonding portion in Nos. 8 and 9, and such each bonding portion had a high Ag content and was also high in area percentage of the island-like Ag phase, as compared with the bonding portion with application of the bonding material having the same mixing ratio. No. 3 (bonding temperature 185° C.) and No. 7 (bonding temperature 315° C.), in which the bonding temperature was out of a condition of 200° C. or more and 300° C. or less, respectively had a low Ag content and a high Ag content. It is presumed that the bonding temperature has an influence on diffusion of Sn molten, in bonding. Herein, the bonding portions formed in the embodiment, except for that of No. 13, each had a low void ratio of 4% or less. This is considered to be due to the effect with transient liquid phase diffusion bonding of a solid bonding material (the compression-molded article of the Ag powder and the Sn powder) unlike a conventional paste.

INDUSTRIAL APPLICABILITY

[0070] The bonding structure of the invention includes a bonding portion having a specific material texture configured from a Ag phase and a Ag.sub.3Sn phase. The bonding portion is favorable in bonding strength and excellent in heat resistance/durability due to heat resistance of the Ag phase and the Ag.sub.3Sn phase and softness of the Ag phase. The invention can be suitably applied in bonding or the like of a device element in a semiconductor device such as a power device such as a hybrid car or an EV.