Laser Soldering Process Optimization of MEMS Probe of Probe Card for Semiconductor Wafer Test

Myeongin Kim; Won Sik Hong; Mi-Song Kim

doi:10.5781/JWJ.2022.40.3.9

Abstract

In this study, the laser soldering process of probe-multilayer ceramic (MLC) substrates was optimized using Type 4 (T4) and Type 7 (T7) Sn-3.0Ag-0.5Cu (SAC305), T7 Sn-0.7Cu solders to improve the high-temperature durability of probe solder joints and satisfy the fine pitch requirements of the soldering process. Thermal cycle tests (TCTs) were performed to compare the heat resistance characteristics of the probe solder joint. The results showed that the bonding strength degradation rates of the SAC305 and Sn-0.7Cu solders after the TCT were within 40% compared to that of the as-soldered state. Because the probe surface and MLC were plated with Ni/Au, various intermetallic compounds (IMCs) were observed at the solder joint. The total IMC thickness of the T7 solder joint was thinner than that of the T4 solder because of the difference in redox reactions according to the particle size of the solder paste. The fracture surface of the probe-MLC solder joint after the shear strength test exhibited a mixed ductile and brittle fracture, and that of the Sn-0.7Cu solder joint exhibited a ductile fracture with numerous dimples.

Key words: Probe card, Laser soldering, Type 7 solder, MEMS probe, Thermal cycling test (TCT)

1. Introduction

A probe card is an essential component for a semiconductor wafer test and serves to check the electrical characteristics of a wafer chip before manufacturing individual packages. The defectiveness of an IC chip is determined by obtaining an electrical signal through contacts between a probe pin and a pad. A probe card is a high-value, custom-made component since it comes in different forms depending on the development model of an IC chip. The main components of the probe card are a printed circuit board (PCB) and a probe. A PCB serves as a support and simultaneously transmits electrical signals from the probe. Currently, the number of pads has been increasing, and the area has been decreasing as the number of I/O (Input/Output) of the semiconductor chip has been increasing. Therefore, a micro-electro-mechanical system (MEMS) type probe card is being used to improve the test efficiency and overcome the limitations of existing cantilever probe¹^-³⁾.

To reduce the thermal stress of a substrate and precisely control the position of multiple pins, the MEMS probe card uses a laser soldering process, which is one of the selective soldering processes. In the case of laser soldering, there is less thermal interference between surrounding components and a minimized possibility of short circuit and substrate damage because the laser is irradiated only to the solder joint. Currently, lead-free solder is used for probe laser soldering as lead (Pb) is regulated by the Restriction of the use of Hazardous Substances (RoHS). Typically, tin (Sn)-based solder with Sn-3.0Ag-0.5Cu (SAC305) composition is used³^,⁴⁾.

An improved property of a probe card for semiconductor wafer testing specialized for vehicles is demanded as automotive semiconductor needs have increased in recent years. Long-term reliability and high heat resistance of a solder joint between the probe and the substrate are required due to a high-temperature testing environment. Therefore, a solder application with a high melting point and suitable fatigue property is being studied to improve the properties⁵⁾.

The study used an SAC305 solder with a melting point of 217°C and Sn-0.7Cu solder with a melting point of 227°C to improve high heat resistance and optimize the high-speed laser soldering bonding process. Type 7 (T7) solder paste with an average powder particle size of 2-11㎛ was used to cope with a fine pitch, and the bonding characteristics were compared and analyzed with Type 4 (T4) solder paste. Thereafter, a thermal shock test was performed, and the mechanical strength before and after the test was measured to compare and analyze the bonding characteristics for each solder composition.

2. Test Method

2.1 Raw and Subsidiary Materials Analysis

Fig. 1 is an optical stereo-microscope image of a MEMS probe and an MLC substrate used for laser soldering. The length of the MEMS probe solder joint is 600㎛, and two holes are made to secure the molten solder’s fluidity. The probe used Ni as its base material, and gold (Au) of approximately 200 nm was electroplated on the surface. The MLC substrate’s dimension was 14.16 (L) × 12.25 (W) mm, and the solder joint pitch was 0.7 (L) × 0.1 (W) mm. Additionally, the Cu pad was surface treated with Ni/Au plating.

Fig. 1

Optical stereo-microscope images of (a) MEMS probe and (b) MLC substrate

Table 1 shows the three types of lead-free solder paste used for laser soldering. To prepare for a fine pitch, a conventional SAC305 solder paste was used, and two sizes of solder paste, T4 and T7, were used. Additionally, a T7 Sn-0.7Cu solder paste with a high melting point was applied before laser soldering to improve the high heat resistance characteristics.

Table 1

Composition, type, and powder size of solder pastes

Solder alloy composition (wt%)	Powder type	Powder size (μm)	Melting point (°C)	Manufacturer
Sn-3.0Ag-0.5Cu (SAC305)	T4	20-38	217	Ecojoin Co., Ltd (S. Korea)
Sn-3.0Ag-0.5Cu (SAC305)	T7	2-11	217	MK Electron Co., Ltd (S. Korea)
Sn-0.7Cu	T7	2-11	227	SNF Co., Ltd. (S. Korea)

2.2 Laser Soldering Joint

Fig. 2 is an image showing the laser soldering process. A probe pin cartridge and a solder pot were prepared and mounted before laser soldering. The laser soldering process was proceeded in the order of MEMS probe pick-up, solder paste dipping, probe positioning, and laser soldering with the probe pin gripper moving from left to right. The laser soldering process was performed in the order of MEMS probe pick-up, solder paste dipping, probe positioning, and laser soldering while the probe pin gripper was moving from left to right. The diode laser source used in the process had a wavelength of 808 nm and a maximum output value of 55 W. The laser soldering process was optimized by setting the laser irradiation time and current value as variables.

Fig. 2

Photograph of the laser soldering process

2.3 Solder Joint Mechanical Strength Measurement

A shear strength test was performed to measure the mechanical strength of the laser soldering joint. Fig. 3(a) shows the shear strength test being conducted, and Fig. 3(b) and (c) show images of comparing coupons before and after the test, respectively. The shear strength test was performed using Dage 4000 (Nordson Co., Ltd., USA) equipment with 50 ㎛/sec of shear rate and 500 ㎛ of shear test jig height.

Fig. 3

Photographs of (a) shear strength test of the solder joint and (b) probe-MLC bonded coupon comparison between before and after test

2.4 Thermal Shock Test

A thermal cycling test (TCT) was performed to evaluate the high heat resistance of the solder joint. 500 cycles were performed at -25 - 115°C for 10 minutes at each temperature⁶^-⁸⁾ for TCT using NT-1200W (ETAC Engineering Co., Ltd., Japan) equipment.

3. Experimental Results

3.1 Laser Soldering Process Optimization

Soldering was performed while changing the laser irradiation time and current value considering the different solder melting points and sizes of the three types of solder pastes used in the test.

Fig. 4 shows images of the two types of soldering defects mainly observed during the laser soldering process optimization process. Fig. 4(a) demonstrates that the solder was incompletely melted because the laser irradiation amount did not reach the solder melting temperature. Therefore, the test was modified to increase the laser profile’s maximum current value. Fig. 4(b) shows that the solder was scattered around the joint. The insufficient preheating time during the laser soldering process was speculated to be the reason. Therefore, the bonding process was processed with an increased laser irradiation time. A laser soldering profile, as shown in Fig. 5, was obtained through the above laser soldering bonding process optimization process. Varying laser profiles were applied for each solder composition. The laser soldering bonding process was performed in 2.5 seconds for the SAC305 solder paste and 3 seconds for the Sn-0.7Cu solder paste.

Fig. 4

Optical stereo-microscope images of soldering process defects: (a) de-wetting and (b) solder scattering

Fig. 5

Laser soldering profiles for (a) Sn-3.0Ag-0.5Cu and (b) Sn-0.7Cu solders

3.2 Mechanical Strength of the Joint

A shear strength test was performed on the MEMS probe joint, as shown in Fig. 6, to compare the joint solder joint strength before and after the TCT. Bonding strengths of the probe joints before the TCT joined using T4 SAC305, T7 SAC305, and T7 Sn-0.7Cu solder paste were 34.9, 36.2, and 38.8 gf, and after the TCT were 25.2, 26.2, and 23.3 gf, respectively. T4 and T7 SAC305’s shear strength values before and after the TCT were similar, and the reduction rates were also similar, showing 27.6% and 27.7%, respectively.

Fig. 6

Shear strength comparison of the laser soldered probe-MLC joints before and after TCT

SAC305 composition showed almost no difference in bonding strength depending on the powder particle size, but the solder joint demonstrated a better bonding when bonding under the same temperature profile setting. The difference is speculated to be due to the powder particle size. The results of previous studies on laser bonding display that insufficiency in laser beam melting increases as the powder particle size increases under the same test setting⁹⁾, and a more significant amount of energy is required as the average particle size increases¹⁰⁾. Therefore, the fluidity of the joint increased as the T7 solder paste melted easier than the T4 under the identical Fig. 5(a) laser profile condition, indicating that the fillet formation of the solder joint was better formed. T7 Sn-0.7Cu showed a relatively high initial strength value, but the strength value after the TCT decreased to 23.3 gf, showing a decrease rate of approximately 39%. It is common for Sn-0.7Cu solder to offer a relatively low degradation rate as Ag3Sn is not formed¹¹⁾.

However, the test measured a more significant deterioration rate for Sn0.7-Cu solder. The laser soldering profile optimization for each solder composition was speculated to be insufficient. Overall, a significant result was collected, considering that the intensity value’s reduction rate after the TCT was 40%.

3.3 Analysis of Joint Cross-Section and Fracture Surface

The SEM analysis was performed on a cross-section for the joint interface analysis of a solder joint. Fig. 7 shows full and enlarged SEM images of the laser soldered joints using T4 SAC305 (Fig. 7(a, b)), T7 SAC305 (Fig. 7(c, d)), and T7 Sn-0.7Cu solder (Fig. 7(e, f)). Full cross-section images display that the solder joints were formed in all areas. Additionally, according to Fig. 7(b, d, f), various intermetallic compounds (IMC) were formed in the joints joined using three types of solder pastes. It demonstrates that when T4 size solder is used, the amount of IMC is relatively large and thick regardless of the solder composition. The solder flux reduction effect explains the above phenomenon. The solder flux reduction reaction can be expressed as Equation 1 below.

(1)

2(R−COOH)+SnO→(R−COO)2Sn+H2O2(R−COOH)+SnO2→(R−COO)2SnO+H2O4(R−COOH)+SnO2→(R−COO)4SnO+2H2O

Fig. 7

Cross-sectional SEM micrographs of laser solder joints, (a, b) T4 Sn-3Ag-0.5Cu, (c, d) T7 Sn-3Ag-0.5Cu, and (e, f) T7 Sn-0.7Cu

According to the results of research by Nobari¹²⁾, the surface oxidation of the solder is formed in the same thickness regardless of the particle size, and the measured and calculated values are consistent with the SnO value of 4.5 nm and SnO2 value of 2 nm. Since T4 and T7 size solder particles show approximately three times or more diameter differences from each other, a proportional difference in the total surface area occurs assuming the exact weight of solder particle amount. Therefore, the smaller the solder particle size, the less flux amount compared to the solder oxide layer when the tin oxide is reduced, as shown in the above equation, which causes the total IMC amount to decrease. Additionally, a difference in the IMC amount formed in the joint may occur as a significant alteration in a joint’s maximum heating temperature level can be caused by a minor change during the laser soldering process. Fig. 8 resulted from the EDS component analysis of the probe/solder and solder/MLC interface. Fig. 8(a, b) is the solder joint that is laser soldered using T7 SAC305 solder, and Fig. 8(c, d) is the same using T7 Sn-0.7Cu solder. Since both the probe and MLC are Ni/Au plated, Ni₃Sn and AuSn2 are formed following the interface regardless of the solder composition. Additionally, (Cu, Ni, Au)₆Sn₅ and Cu₆Sn₅ IMC were formed inside the solder, while Ag₃Sn was formed when T7 SAC305 was used¹³^-¹⁵⁾.

Fig. 8

Magnified SEM images and EDS data of laser soldered joints, (a, b) T7 SAC305 and (c, d) T7 Sn-0.7Cu

Fig. 9 shows the fracture surfaces’ SEM analysis after the solder joint shear strength test. The joint’s fracture surface showed a mixed fracture pattern of ductile fracture and brittle fracture. Fig. 9(a) demonstrates that the fracture surface of T4 SAC305 exhibits relatively brittle fracture characteristics because T4 solder formed a more considerable IMC amount than T7 solder regardless of the solder composition. T7 solder yielded a different fracture surface pattern depending on the composition. Ag₃Sn IMC and Sn-Cu IMC were formed inside the SAC305 solder, but only Sn-Cu IMC was observed inside the Sn-0.7Cu solder due to the absence of Ag. Therefore, it was discovered that multiple dimples were formed due to the increased ductility of the Sn-0.7Cu solder joint.

Fig. 9

SEM images of fracture surfaces of laser solder joints after shear strength test, (a) T4 Sn-3Ag-0.5Cu, (b) T7 Sn-3Ag-0.5Cu, and (c) T7 Sn-0.7Cu

4. Conclusion

The study optimized the high-speed laser soldering bonding process using SAC305 solder with a melting point of 220°C, a typical mesophilic lead-free solder, and Sn-0.7Cu with a melting point of 227°C to improve high heat resistance. The bonding properties were evaluated through a thermal shock test and shear strength measurement. The results are as follows.

1) Comparing the initial bond strength values with T4 SAC305, the T7 SAC305 and T7 Sn-0.7Cu solder joints showed 34.9, 36.2, and 38.8 gf strength values, respectively, while Sn-0.7Cu displayed a relatively high shear strength value. The post-shock test shear strength values decreased to 25.2, 26.2, and 23.3 gf, respectively, and the bond strength deterioration rate was within 40% of the initial value.
2) Laser soldering joint interface analysis revealed the formation of various IMC. T4 solder with a large solder paste powder particle size formed a more considerable IMC amount compared to the T7 size solder. Ni₃Sn and AuSn₂ were formed near the interface, and (Cu, Ni, Au)₆Sn₅, Cu₆Sn₅, and Ag₃Sn were formed inside as the probe and MLC were surface-treated with Ni/Au.
3) The fracture surfaces of SAC305 and Sn-0.7Cu solders showed a mixture of ductile and brittle fracture patterns. A smaller amount of IMC formed in the joint created more ductile fracture characteristics with dimples. Therefore, using a solder paste with a larger powder particle and SAC305 solder caused an increased exhibition of brittle fracture properties. Contrarily, the T7 Sn-0.7Cu solder joint displayed ductile fracture characteristics.