Intl. Journal of Microcircuits and Electronic Packaging

Glob Top Plastic Ball Grid Array Package Encapsulation Process Development and Its Moisture Sensitivity Study L.Y. Yang, E. Padrinao, and Y.C. Mui Advanced Micro Devices 512 Chaichee lane, #03-06 Bedok Industrial Estate Singapore 469028 Phone: 65-240-9441 Fax: 65-448-2360-14603 e-mail: [email protected]

Abstract Liquid epoxy encapsulation is one of the important processes in the glob top PBGA assembly, which affect other subsequent process and package quality. The objective of the encapsulation is to protect the bonding wire and chip from harsh environment and make the device more reliable. In general, the basic concerns for encapsulation in glob top PBGA are encapsulation thickness, surface flatness, encapsulation area, voids in the encapsulation, and finished encapsulation surface. On the other hand, package reliability under accelerated condition should be addressed to assess the encapsulation quality as well as the assembly process reliability. In the encapsulation process development, design of experiment (DOE) is conducted in the first stage to help select the critical factors in the process, and then response surface methodology (RSM) is applied to optimize the process parameters. Additionally, statistic process control tools are used to monitor the process parameters and calculate the process capability. Basically, Plastic Ball Grid Arrays packages are known to have a higher moisture absorption rate when exposed to humid ambient conditions, so the moisture sensitivity becomes a concern when plastic packages are exposed to a humid environment for a prolonged time and then processed through reflow process. The objective of the moisture sensitivity study is to understand the effect of moisture ingress on the process and assembly reliability. The steam pressure pot test and Level-3 preconditioning test as well as temperature cycling test are used to evaluate the effects. The experimental result shows that the moisture contents inside the substrate will affect the die attach voids level and size. Moreover, the moisture level does play a part in the formation of delamination at the interface of die/substrate and encapsulation/substrate after acceleration test. However, there is no correlation between the die top delamination and the moisture contents absorbed by the package from the Level-3 preconditioning test. In addition, there was no delamination found at the die attach area or at the interface between the die and the substrate even when the package was moisture soaked fully saturated and went through temperature cycling. There is a trend to form more delamination in the interface of molding compound / solder mask, D/A after steam pressure test if more moisture is absorbed. Finally, it should be emphasized that plasma cleaning is found to have a significant effect on the package integrity for dry packed PBGA and moisture saturated units. Coupled with the stress test results, it implies that the most important thing in the PBGA assembly is to enhance the interface adhesion through plasma cleaning apart from the moisture preventing. The detailed experimental results and conclusions are explained in this paper.

Key words: Glob Top PBGA, Encapsulation, TAGUCHI, RSM, Moisture, Delamination, Moisture Desorption, Steam Pressure Pot, Temperature Cycling, Adhesion, and Plasma Cleaning.

1. Introduction The use of area array packages in products is on the rise in the semiconductor industry. Many assemblers have begun to develop the manufacturing processes to handle area arrays packages in-house. Under the concept of array pattern, Ball Grid Array Packages can be classified according to the type of material of the carrier substrate.

The International Journal of Microcircuits and Electronic Packaging, Volume 21, Number 3, Third Quarter 1998 (ISSN 1063-1674) 262

8 International Microelectronics And Packaging Society

Glob Top Plastic Ball Grid Array Package Encapsulation Process Development and Its Moisture Sensitivity Study

The Plastic Ball Grid Array (PBGA) has demonstrated a potential to become the leading BGA package type used in relatively high volume products. This fact can be attributed to reasons such as their ability to provide high lead count, high interconnection density, and high electrical performance. In general, the PBGA is based on a glass-filled substrate constructed of either BT or FR-4, which is usually 0.36 mm to 0.60 mm thick depending on the number of layers used. For the packages with less than 300 I/Os, the substrate has circuit trace with 150µm line width and spaces. Typically, 300µm vias are used to connect the wirebond pads to the ball pads on the opposite side of the substrate, commonly incorporated in two or four layers. The Integrated circuit (IC), or silicon die, is attached to the substrate using a die attach. The die attach is usually a silver filled epoxy. The IC is then wirebonded to the substrate using gold wires. To protect the gold wires and the IC, molding compound covers the assembly; this is one of the most sensitive steps in the constructions of PBGA. The next step is to attach the solder balls to the opposite side of the substrate. The solder balls are 0.76 mm in diameter and are made of eutectic solder. Finally, the PBGA are cleaned to remove the flux that was applied during the ball formation process and routed from the array1. In Glob Top PBGA assembly, instead of using molding transfer technology, liquid dispensing technology is used to encapsulate the IC and gold wirebonded on the top of the die. The schematic diagram is shown in Figure 1.

attain its ideal temperature as it flows through the valve assembly in advance of reaching the dispensing tip. The heat allows the encapsulant to flow completely around the wirebonds, as well as eliminating trapped air bubbles under wirebonds and creating voids free parts. The second task involves the dispensing techniques, including the dispensing machine set-up and dispensing parameters selection. Basically, a syringe of epoxy is pressurized to force the fluid through a small needle, the pressure pulse is timed to dispense the proper amount of epoxy. The disadvantage of syringe dispensing is that the amount of epoxy applied will vary with the epoxy’s viscosity. As the temperature of environment changes, the quantity of epoxy dispensed also changes. Any material variables that cause the viscosity to change over the shelf life of the glue affect the dot sizes and dispensing volume. Therefore, there is a tight requirement for epoxy manufacturers, their manufacturing process should be highly repeatable and the viscosity should be a stable value with a very little variation. In this evaluation, a positive displacement pump is used to accurately control the amount of epoxy being dispensed. The dispensing repeatability is better than 1%. Similar to other processes, there are a plenty of dispensing factors affecting liquid dispensing process. For the Glob Top encapsulation process, volume and void control are two very important responses, since they affect the encapsulation appearance and quality. In the experience of the researchers, void levels can be controlled and eliminated by varying the dispensing patterns, thus in this study, the focus of process development is volume control. Based on the literature review and information from vendors, some important parameters identified to influence the volume are air pressure, dispensing valve, needle size, height of the needle above the die, needle dispensing speed, and fill routines. The objective of the process study is to obtain optimized parameters for Glob Top PBGA assembly. For Glob Top encapsulation process, the most widely accepted method of encapsulation is the two-step encapsulation, shown in Figure 2, which guarantees successful encapsulations.

Figure 1. Schematic diagram of Glob Top PBGA (courtesy: Asymtek). For the encapsulation process in Glob Top PBGA assembly, encapsulation materials and dispensing heating system play a very important and critical role. The epoxy materials should be with relatively low viscosity and comparable coefficient of thermal expansion (CTE) value, and temperature control plays a major part in the material viscosity variation. Most of the encapsulants used in advanced encapsulation applications are two-part epoxies, which are mixed and then frozen at -40 oC before delivery. Any variations of the temperature before use can result in a degradation of the product, premature curing and disastrous dispensing results such as rejected finished surface. The ideal solution is to slowly thaw the vacuum-sealed encapsulant in an upright position, then take it out from vacuum pack when it arrives at room temperature. A thermal Figure 2. Two step encapsulation (courtesy: Asymtek). heat source built into the dispensing system should be used to bring the material to a temperature between 80 oC to 100 oC during disFirst, the dispensing needle creates a dam around the die to be pensing to achieve lower viscosity to enhance the epoxy flow. Addiencapsulated. The dam holds all other material within the dispenstionally, temperature control will also allow the material to gently The International Journal of Microcircuits and Electronic Packaging, Volume 21, Number 3, Third Quarter 1998 (ISSN 1063-1674) 8 International Microelectronics And Packaging Society

263

Intl. Journal of Microcircuits and Electronic Packaging

Such a pressure induces a bending stress in the package. ing area for a neat, and accurate encapsulation, at the same time, its Tay et al.3 investigated the pressure exerting on the delaminated boundary covers a whole via, not half a via. Second, the needle dispenses a spiral fill routine to secure that the encapsulant pendie pad/encapsulant interface by employing the Finite Element simuetrates around and under all wirebonds. lation of heat and the moisture diffusion process that occur simultaThis paper will address design of experiment (DOE) and statistic neously. It established that stress developed in the encapsulant could process control (SPC)’s application in the process development to cause cracking to occur. The paper also concluded that the primary obtain a Robust Design\Process. The encapsulation materials used effect of the moisture in “popcorn” cracking is the reduction of the are Hysol CB-013 for filling and Hysol CB011-1R liquid epoxies adhesion of the die pad-encapsulated interface and not much on the for damming. The dispensing machine used is CAMALOT 5000 increasing of the stress due to the moisture vaporizing in the delamisystem. Due to the plastic materials in the package composition and nated interface. package structures that include too many interfaces inside the packKazuhiro et al.4 studied the effect of molding compound on packages, one of the characteristics that must be taken into account when age reliability especially cracking and delamination, one great probdealing with PBGAs is their moisture absorbing abilities. This fact lem in the reliability of plastic packages during reflow soldering. has required assemblers to take proper precautions to prevent delamiOne of the techniques used is to lower the Tg in order to decrease nation and/or cracking from becoming a serious problem. Reflow the absorbed moisture content of a compound. The technique of induced package cracking, more commonly known as “popcorn” lowering Tg of a molding compound is very effective for improvedelamination, is a major concern for manufacturers assembling ment of the package crack resistance, since the properties of low PBGAs. moisture absorption as well as high adhesion strength for a molding Popcorning is usually caused by the rapid vaporization of the compound can be realized. moisture absorbed inside the component into water vapor steam. Moore and Kelsall5 evaluated the impact of delamination on stressThe vaporization happens in the reflow process during the rapid induced and contamination-related failure in surface mount ICs. In increase in temperature. The rapid expansion of the vapor causes this paper, a temperature cycle evaluation is reported of a large die the package to crack. The failure regions are usually concentrated 132 pin plastic quad flat pack process. The effect of contamination under the die in the die attach area, either in the boundary between and moisture exposure on bond pad integrity in 68 pin leaded chip the die attach and the package substrate, or in the boundary between carriers damaged during solder reflow is also examined. The study the die attach and the back of the die. This failure can also go further results revealed that after 1000 temperature cycles, the corners of to the interface at the molding compound/BT substrate. It is also the die show significant amounts of delamination and it has been believed that the extent of the delamination is a function of the adheshown that this delamination will continue to spread from the shear sion strength between the die pad, the mold compound, the solder stress maximum at the die corners towards the die center and will produce wirebond degradation and damage to the device surface. mask, and the die attach. Normally, scanning acoustic microscopy Additionally, the delamination results after moisture / vapor phase (SAM) can be used for detecting internal defects from popcorn reflow showed that every unit from the high moisture group showed delamination. substantial delamination of mold compound from the die pad backThe phenomenon of moisture ingress and its detrimental effects side, the die surface, and the top-side of the lead-frame around the on the assembly characteristics of surface mount plastic IC devices die. The two lower moisture groups showed little delamination and has been an active research topic since it was first reported by no visual or acoustic evidence of package cracks. Fukuzawa et al.2. Moisture ingress in plastic molded ICs has become a generic reliability wide problem. Treated simply, the mechaShook6 presented effective characterization of moisture induced nism of moisture induced damage has been shown to be caused by damage of plastic surface mount devices has been achieved through the phase change and expansion if internally condensed moisture in the use of C-mode SAM. The results revealed that the sensitivity to the plastic during exposure to solder reflow temperatures. Rapid moisture ingress can vary significantly from device to device, the expansion of the confined vapor and the super-imposed reduction in degree of sensitivity is clearly not just a function of total pin count. strength of the molding compound as the temperature rises can ultiAlso, the potential susceptibility of a plastic surface mount device to mately result in package damage. A critical crossover point occurs moisture damage has been shown to depend greatly on the gradient when the strength of the plastic can no longer sustain the rising inof moisture content in the plastic. The level of ingress moisture at ternal stresses generated by the expanding vapor and the resultant the die paddle/plastic interface was shown to be the dominant factor effect. in inducing damage. In 1985, Fukuzawa et al.2 analyzed the mechanism of cracking in Therefore, it can be seen that delamination at interfaces between plastic IC packages during reflow soldering and established a model the different materials within the package is a major cause of moisfor crack occurrence. Fukuzawa suggested that during solder reflow ture ingress and subsequent premature package failure. Especially, process, owing to the differences between the coefficients of therwhen such packages are experiencing sudden temperature change. mal expansion (CTE) of the different materials constituting the IC For PBGA packages, this moisture-induced failure will be more stanpackage, high thermal stress is induced especially at the interface dard-out due to their materials used and structure applied. between the die pad and the encapsulant, causing delamination to In this paper, the impact of moisture absorption on the package occur. Moisture which has previously absorbed into the plastic enintegrity was studied in detail, in addition, the right conclusion to capsulant during the molding process, then begin to diffuse into the prevent the damage from moisture ingress were obtained. This study delaminated interface, creating a high vapor pressure in this area. was very beneficial for achieving a proper assembly process for the The International Journal of Microcircuits and Electronic Packaging, Volume 21, Number 3, Third Quarter 1998 (ISSN 1063-1674) 264

8 International Microelectronics And Packaging Society

Glob Top Plastic Ball Grid Array Package Encapsulation Process Development and Its Moisture Sensitivity Study

Glob Top Plastic Ball Grid Array Packages.

2. Process Development and Experiment Implementation For Moisture Sensitivity Study 2.1. Encapsulation Process Development The intention of the encapsulation process study is to understand the process characteristics, define critical process parameters and optimize these parameters. Since the dispensing pattern has been defined prior to this study, factors that affect dispensing volume are the focus of this study. In this study, TAGUCHI method is used to evaluate the effect of different factors on encapsulation response and select the critical parameters, and then response surface methodology (RSM) techniques are applied to optimize those parameters.

2.2. Critical Process Parameter Evaluation Using TAGUCHI Method TAGUCHI method is developed based on a number of existing statistical methods, it basically constructs a special set of orthogonal arrays (OAs) to layout this experiment. By combining the orthogonal Latin squares in a unique manner, it can prepare a new set of standard OAs to be used for a number of experimental situations. In this paper, TAGUCHI method is applied in two processes, called damming process and filling process (explained in the introduction part). a. Damming Process The requirement of dam in the Glob Top is to constrain the potting material and then achieve a fixed encapsulation area and flat encapsulation area. Dispensing needle height from the substrate, dispensing head transverse speed, dispensing needle size, air pressure, dispensing work holder temperature, environmental factors, and other machine settings all play parts in the process. However, the task of controlling all possible input factors of a process is impractical concerning time and cost. Therefore, only some critical parameters are selected for these parameters optimization. In the experiment, four factors are selected and L8 orthogonal array is used to conduct the experiment. The experiment response is dispensing dam height and dispensing volume. Two replicates are used for this experiment and the sample size is 6 units. A total of 12 units were built for each run. The dam height is measured on the Litematic stylus measurement system. The dispensing weight was measured on 4 digit weighting machine. The results were processed following the standard TAGUCHI ANOVA style and showed the significant factors that affect the dam height are the factors A, B and C (Table 1) at a = 0.05 level of significance.

b. Potting Process Potting process, known also as filling, is the more critical process in Glob Top PBGA encapsulation process. Since it affects the encapsulation surface, the height of the encapsulation materials, and the flatness of the surface. It involves more process parameters and has a bigger effect on the reliability of units as well such as voids. The experimental parameters and levels are shown in Table 1. The experimental responses are dispensing volume. Experimental replicate is 2 and 6 units for each run. The results show that needle transverse speed, needle heating temperature, and air pressure all have significant influence on the filling process at a = 0.05 level of significance. Table 1. Experimental setting conditions. Factors A: Air pressure, psi B: needle transverse speed, mm/s C: Work holder temp, oC D: Needle height, mm E: Needle heating temp, oC F: Dam height, mm

Levels L1 10 8 40 1.5 No 1

L2 40 15 80 2.3 45 1.2

2.3. Optimization Studies Once the critical parameters are identified from TAGUCHI implementation, the optimization process using response surface methodology will be conducted for both damming and filling processes. In the response surface study, a series of experiments are conducted on the process and data collected are then fitted to a mathematical model that described the relationship between the input variables and the response. The mathematical model may then be used to determine the optimum levels of each input variable that provide either a maximum or minimum value of the response. For damming process, based on results from screening experiments, three parameters were identified and studied, three levels for each factor. Table 2 shows the experimental planning. Figures 3, 4 and 5 show the estimated dispensing volume response function from respective input factors. In Figure 3, the effect of dispensing speed and needle height from the substrate on dispensing volume is demonstrated. It is clearly shown that the dispensing volume will be significantly decreased upon an increase of the needle moving speed, and it seems needle height settings have a little effect on the amount of volume dispensed. The effect of air pressure on the amount of dispensing volume is clearly shown in Figure 4. It is clear that with the increase of the air pressure, the volume increased dramatically. When the dispensing air pressure reaches around 40 psi, the response is very stable. It can be concluded that dispensing air pressure and needle moving speed influence the amount of volume dispensed dramatically compared to the effect of needle height setting. Figure 5 shows the effect of pressure and speed. Furthermore, if one select fast dispensing speed, then the dispensed volume is relatively stable within the air pressure range.

The International Journal of Microcircuits and Electronic Packaging, Volume 21, Number 3, Third Quarter 1998 (ISSN 1063-1674) 8 International Microelectronics And Packaging Society

265

Intl. Journal of Microcircuits and Electronic Packaging

Table 2. RSM experimental setting conditions. Factors A: Speed (mm/s) B: Air pressure (psi) C: Needle height (mm)

Level 1 10 10 0.8

Level 2 15 25 1

Estimated Response Function

0.54

Volume

0.49 0.44 0.39

(X 0.01)

0.34

Level 3 20 40 1.2

The normality of the data was also plotted and is shown in Figure 6 (while the normal probablity plot for residuals is demonstrated later in Figure 10). The straight line indicated that the data conformed to a normal distribution. Similarly, RSM was conducted on filling process as well. The experimental factors are shown in Table 3. A total of 54 units were built. The substrates are baked for 4 hours at 125 oC, the work holder temperature is set at 90 oC and the syringe is heated at 45 oC. Figures 7, 8 and 9 show the estimated response function when the encapsulated volume is monitored. In Figure 7, with the increase of needle moving speed, the encapsulation volume decreases dramatically and the same trend can be found in Figure 8, compared to the effect of the speed, the effect of the pressure is quite flat. Additionally, the air pressure plays a more critical role to the volume response compared to the heating temperature.

146 126 106

0.29 0.24

86 7

9

11 13 15 17 19

Normal Probability Plot for Residuals

Height

66

Speed

99.9

Figure 3. 3-D surface plot (Response: Volume).

Volume

Estimated Response Function

cumulative percent

99 95 80 50 20 5

0.36

1

0.33

0.1

0.3

-23 (X 0.01)

0.27

0.21

17

37

residual var_1

146 126 106

0.24

-3

57 (X 1E-3)

86 0

10

20

Height

66 30

40

Pressure

Figure 6. Normal probability plot for residuals. Table 3. RSM experimental setting conditions.

Figure 4. 3-D surface plot (Response: Volume).

Experimental factors A: Air Pressure, psi B: Needle Transverse Speed, mm/s C: Work holder Temp, oC D: Needle Height, mm

Estimated Response Function

L1 10 8 40 1

Levels L2 40 20 80 1.2

0.59

Volume

0.49 0.39 40

0.29 30 20

0.19

10 7

9

11 13 15 17 19

0

Pressure

Speed

Figure 5. 3-D surface plot (Response: Volume). The International Journal of Microcircuits and Electronic Packaging, Volume 21, Number 3, Third Quarter 1998 (ISSN 1063-1674) 266

8 International Microelectronics And Packaging Society

Glob Top Plastic Ball Grid Array Package Encapsulation Process Development and Its Moisture Sensitivity Study

Estimated Response Function

0.73

Volume

0.63 0.53 16

0.43 14 12

0.33

10 20

24

28

32

Speed

8 36

40

Pressure

Figure 9. 3-D surface plot.

Figure 7. 3-D surface plot.

Estimated Response Function

0.73

Volume

0.63 0.53 16

0.43 14 12

0.33

10 30

33

36

39

8 42

Speed

45

Temp

Figure 8. 3-D surface plot. The confirmation runs are conducted for each study and the optimized process parameters are obtained based on the desired volume and experimental results. Table 4 shows the final optimized encapsulation process parameters. The confirmation run data show also that the available process is under process control regarding to the control chart and CPK. Table 4. Optimized parameters for dam and fill process. Dam process

Fill process

Needle size: 14 Air pressure: 40 psi Speed: 30 mm/s Lift gap : 7 mil Lead screw: 200 mil Needle height: 1.2 mm fiducial point: 2 point on substrate Needle size: 14 Air pressure: 40 psi Speed: 22 mm/s Needle heating temp: 45 deg C Lead screw: 200 mil Needle height 2.5 mm Lift gap: 14 mil

Figure 10. Normal probability plot for residuals.

3. Package Moisture Sensitivity Study In order to understand the moisture sensitivity characteristics of concurrently assembled in-house 292 perimeter Plastic Ball Grid Array Packages, moisture characteristics as well as the effect of moisture on package integrity are studied. C-Mode scanning acoustic microscopy (CSAM) was used to fully characterize the package integrity of plastic BGAs.

3.1. Moisture Absorption and Desorption Characterization

The intention of this evaluation is to understand the moisture absorption and desorption characteristics. The sample pictures are shown in Figure 11 and the evaluation flow1998 chart(ISSN is shown in Figure The International Journal of Microcircuits and Electronic Packaging, Volume 21, Number 3, Third Quarter 1063-1674) 8 International Microelectronics And Packaging Society

267

Intl. Journal of Microcircuits and Electronic Packaging

12. Initially, the bare substrates are baked for 24 hours at 125 oC to drive out all the moisture to ensure the initial dry condition, then 42 of these substrates were die attached. 21 of 42 die-attached units were encapsulated using liquid dispensing epoxy. All prepared samples were transferred to Level-3 moisture environment (30oC/ 60% RH) and soaked for the required hours. The moisture weight gains are recorded after different moisture soak hours.

Bare substrates (45)

Substrates bake & weight (45)

Die attachment (30) Bare substrates (15) Cure & weight (30)

D/A substrates (15)

Encapsulation (15) Cure & weight (15)

Moisture soak (45)

Weight (45)

(a) Bare substrate

(b) Substrate with die

(c) Substrate with die & glob top

Encap parts (15)

Figure 13. Moisture desorption evaluation flow chart.

Bare substrate preparation ( 63 )

3.2. Effect of Moisture Content on Die Attached Quality

Bake for 24 hrs & weight (63)

Substrates for die attach (42) Cure & weight (42)

Die attached substrate (21)

D/A substrates (15)

Bake for 2 , 4, 6, 8, 10 hrs & weight individually (45)

Figure 11. Pictures of test samples.

Bare substrates (21)

Bare substrates (15)

Encapsulation (21)

Cure and weight (21) Moisture soak for 0, ,4, 8, 16, 32, 64, 128, 192 hrs & weight (63)

Figure 12. Moisture absorption evaluation flow chart. In the moisture desorption test, dry experimental samples were soaked in the JEDEC Level-3 condition for longer time to make sure all samples are moisture saturated. The moisture-saturated units are then sent to bake oven to drive out the moisture content. The moisture content was recorded at different time frame and Figure 13 shows the evaluation flow chart.

The procedure used to evaluate the effect of moisture on die attached quality is summarized in Figure 14. In this evaluation, similarly, the BT substrates were baked for 24 hours at 125 oC to drive out all moisture and then the dried substrates are transferred to a moisture chamber and soaked for the required hours. The soak condition is 30oC/60%RH. After soaking, the silicon chip was attached on the moisture soaked substrate. The die attach material used in this case was Hitachi EN4702 and the bonding line thickness (BLT) was set around 1.5 mil, in addition, the curing condition is at 125 oC for two and half hours. All die-attached samples were then sent to accelerated stress test chamber, such as thermal shock test and temperature test. X -Ray technique was used to detect the voids levels under the die in this evaluation. Die grounding technique was applied to observe the presence of the void levels as well. Substrate preparation (30) Bake for 24 hrs & weight (30) Moisture soak for 0, 4, 8, 16,?hrs & weight (30) Die attachment for all units (30)

Cure & weight, Voids check, (30)

Die shear test (10)

Stress test(TS, PRECON+ T/C) (20) SAM & Die shear, Cross sectioon

Figure 14. Ingress moisture effect on D/A reliability. evaluation flow chart.3, Third Quarter 1998 (ISSN 1063-1674) The International Journal of Microcircuits and Electronic Packaging, Volume 21, Number 268

8 International Microelectronics And Packaging Society

Glob Top Plastic Ball Grid Array Package Encapsulation Process Development and Its Moisture Sensitivity Study

3.3. Effect of Moisture Content on Encapsulated PBGA Units Due to the special structure of PBGA package, it is doughtful that this type package is very moisture sensitive, especially if the interface strength is not strong enough. In this part, the effect of moisture ingress on the Glob Top PBGA packages integrity is studied. Figure 15 shows the evaluation flow chart. The encapsulation materials are Hysol CB011-1R for dam and Hysol CB013 for fill, respectively. The encapsulation area is 22 mm x 22 mm. The thickness of the encapsulation materials is 1.2 mm. Evaluation samples are sent for JEDEC Level-3 preconditioning test, temperature cycling test (-65oC/145oC) and thermal shock test (-55oC/125oC). The package integrity of units after the stress test is then inspected using C-SAM technique.

than the moisture diffusion rate. The rate for moisture diffusion depends on sample dimensions and environmental humidity conditions as well as material properties. In this case, moisture diffusion problem is taken as one-dimensional problem since the sample has a very bigger area/thickness ratio. Figure 16 shows the samples moisture absorption curves. Xaxis is the moisture soak time and Y-axis is the moisture weight gain percentage. Three curves are plotted for bare substrates, substrates with die attached, and fully encapsulated units. It is clearly shown that the bare substrates have the biggest moisture weight gain and achieved the saturated state in the fast rate. The substrates with die attached have less moisture weight gain, and the fully encapsulated units have the smallest moisture weight gain and took longer time to get saturated as well. One of the possible reasons is the difference of the exposed substrate area, since most of the moisture is penetrated to the samples through the substrate surface (solder mask surface as well) instead of the side walls. The results show that the more exposed substrate area, the higher moisture absorption capability. On the other hand, the results also demonstrate that dieattached substrates and encapsulated substrates have less potential to absorb moisture, that implicit that die-attach materials and encapsulants have higher moisture resistance compared to solder mask and BT substrate. It can be concluded that fully encapsulated units have less potential to absorb more moisture, it is more imperative to prevent the incoming bare substrates from high humidity environment.

Weight gain vs soak time 0.4 A: Bare substrate B: substarte w D/A C: Encapsulated device

0.35

Figure 15. Ingress moisture effect on package integrity. Generally, based on moisture diffusion theory and PBGA package structure, it was assumed that, in Plastic Ball Grid Array packages, most moisture would be absorbed by the substrate that is composed of BT materials and solder mask. At the same time, the dieattach materials and encapsulant will also absorb moisture, although the liquid encapsulant is claimed to have a very good moisture absorption resistance. The moisture accumulated at the interfaces has a very significant effect on the package integrity.

Weight gain, %

0.3 0.25 0.2 0.15 0.1 0.05 0 0

50

100

150

200

Soak time, hrs

Figure 16. Sample moisture absorption curves (JEDEC Level 3 condition).

4.2. Moisture Desorption Evaluation Moisture desorption refers to the capability of the package to drive out the moisture through baking. The baking condition used in 4. Results and Analysis For Moisture this evaluation is 125 oC at which less damage will be induced to Sensitivity Study solder mask. Figure 17 shows the moisture desorption curves for bare substrate, substrate with die attached, and fully encapsulated packages. In contrast to conventional plastic packages, the moisture 4.1. Moisture Absorption Evaluation desorption rate for PBGA substrates is very fast, and the saturated moisture will be totally driven out by only 1-2 hours baking, a true Moisture absorption or diffusion is actually similar to heat transresult for all samples. This information will be very useful eventufer phenomenon. The difference is the thermal diffusion rate is higher ally for the assembly process flow evaluation. The International Journal of Microcircuits and Electronic Packaging, Volume 21, Number 3, Third Quarter 1998 (ISSN 1063-1674) 8 International Microelectronics And Packaging Society

269

Intl. Journal of Microcircuits and Electronic Packaging

Moisture level Vs. Die pull strength 100

64

100

60 40 20

80

60 58

60

56 40

54 52

20

50

Baking time, hrs

9

8

7

6

5

4

0 3

48 2

47

28

-20

22

7

4

3

2

1

0

0

1

Moisture %

Die pull strength, N

62

D/A units Encap units Substrate

80

Moisture concentration, %

Effective Baking Evaluation

Sam ple No. Pull strength WT%

Figure 17. Moisture desorption characteristics.

4.3. Moisture Effect on Die Attach Process

Figure 18. Correlation between the die shear and moisture levels.

It is generally assumed that moisture absorption before die atVoid tach process will induce more die attach defects. These defects such as die attach voids will cause subsequent package failures when the Void package goes through some harsh tests. However, this evaluation results show that the die attach integrity is really depended on other typical factors, such as die attach materials used, die-attach process parameters settings, and curing conditions, instead of moisture con(a) Die attach voids with little moisture (b) Die attach voids with more moisture tents in the substrate, although the moisture will relatively increase the die attach voids size and levels. Figure 19. Comparison of voids levels and size for different In this evaluation, die attach voids and die bonding adhesion are moisture weight gain. the responses. The die attach material used is Hitachi EN4702 and bonding machine is K&S 6900 die bonder. The optimized BLT for Die attach voids will affect the package integrity if the package die-attach process was 1.5 mil. Based on material vendor informagoes through stress test, especially for larger size voids. The voids o tion, the curing condition was cure for 2.5 hours at 125 C. In order in the die attach area will help absorb moisture and accumulate the to take a close look at the effect of moisture content on die attach contamination as well. During the stressing tests, the accumulated process, the bare substrates were soaked for 0, 4, 8, 16, and 128 moisture will evaporate and the vaporized pressure will cause the o hours at 30 C/60 % RH condition. The moisture soaked substrates die attach delamination or cracks, if the die attach adhesion and /or were then die bonded and cured. The experimental results showed adhesion between the solder mask and copper clad are not strong that the higher moisture weight gain, the more possibility to form enough. the voids for current setting. Additionally, with the increase of moisIn the following, thermal shock test and preconditioning test were ture content in the substrate, the more and bigger voids formed in conducted to evaluate these effects using the substrate with die atthe die attach area for the evaluated die attach materials. tached only. In order to examine package integrity, die shear and die pull test as well as accelerated stress testing are conducted to evaluate the die a. Thermal Shock Test for Die Attached Units attach strength. The die shear machine used is Dage 2400PC and the The main purpose for this test is to evaluate the effect of die bond die pull test is conducted by INSTRON pull machine. The results strength for different level moisture soaked units before die attach. were very unexpected, all of the die shear strengths are larger than Surprisingly, after 100 cycles and 200 cycles, there was no die at80 kg, whether the units are moisture saturated or not. The pull test tach failures found through the SAM pictures. One possible reason results, shown in Figure 18, similarly implied that there is no relais that the vapor pressure could find its way out of the die attach area tionship between the die shear strength and voids levels or moisture due to the short of barrier materials such as encapsulation materials. contents based on the evaluation for current die attach materials and On the other hand, it demonstrates that die attach strength is strong machine setting. A possible justification can be attributed to the fairly enough to overcome applied stress. higher die bond strength is enough to resist the shear strength. However, the void levels do increase with the increase of moisture weight b. Preconditioning Test Plus T/C Test gain, and moreover, the small voids join together and form bigger Similar phenomenon is observed in preconditioning and temirregular voids for higher moisture content substrates. Figure 19 perature cycling test for the samples. Level-3 preconditioning test shows the comparison of the voids levels and size. The International Journal of Microcircuits and Electronic Packaging, Volume 21, Number 3, Third Quarter 1998 (ISSN 1063-1674) 270

8 International Microelectronics And Packaging Society

Glob Top Plastic Ball Grid Array Package Encapsulation Process Development and Its Moisture Sensitivity Study

4.4. Moisture Effect on Fully Encapsulated Units Above, the researchers have evaluated the moisture absorption, moisture desorption as well as the effect of moisture on die attach process, in this section, the effect of moisture on the whole package integrity will be studied. For Plastic Ball Grid Array Packages, the encapsulation materials and some interfaces are potentially easy for moisture ingress and accumulation as well. The moisture concentration in the interface will cause not only a decrease of the adhesion but also popcorning if the package goes through the stress test. In the evaluation process, the samples are soaked in 30oC/60% RH environment for different time right after die-attach process (same materials and process parameters), therefore, there will be different level moisture contents in the substrate, then liquid epoxy was dispensed to cover the die and substrate. The encapsulation area is 22mm by 22mm, and then went through curing. The curing condition is 110 oC for 1 hour and 165 oC for 2 hours. When the units are fully cured, some destructive adhesion tests and stress tests are conducted; the detail results were discussed in the following part A, B, and C. a. Pull Test INSTRON pull test was conducted to evaluate the adhesion of the encapsulant/solder mask interface. Figure 20 shows the relationship of moisture levels and encapsulation pull test strength. As shown in Figure 21, the failure mode of the pull test is between the solder mask and BT core for all tested samples, which did not go through the stress test such as SPP test as well as temperature cycling test.

100 90 80 70 60 50 40 30 20 10 0

300 250 200 150 100 50

10

9

8

7

6

5

4

3

2

1

0

Moisture wt%

Moisture levels Vs. Encap pull strength

Encap pull strength, N

and standard reflow temperature is applied and the results showed good integrity for the test samples. There is no delamination and other failures found even after 500 temperature cycle. It can be concluded that the adhesion of die to substrate is strong enough to overcome the stresses induced by the accelerated test under the test section although the samples are with different level and size of die attach voids. This conclusion actually implied that it is not necessary to bake the substrate if die attach integrity is the concern only in this current Glob Top PBGA package, although the moisture absorbed in the substrate does increase the die attach voids numbers and size. However, from the whole package point of view, it might be a different alterative. In the following test, the effect of moisture on encapsulation adhesion will be evaluated and it is disclosed that baking is a necessary step.

Sam ple No Pull strength WT%

Figure 20. Pull strength vs. moisture levels. fibre exposed, copper pull off .

Figure 21. Failure mode for the pull test. The results suggested that the adhesion between the solder mask and the encapsulant was strong enough to withstand the pull strength, even through the moisture content is very high in the substrate before encapsulation. It is also contributed that Hysol CB013 material can achieve a good adhesion with Taiyo solder mask. Additionally, there is no plasma cleaning conducted prior to encapsulation. However, when the units went through the SPP test, the delamination observed at the interface of the solder mask and copper trace and the pull strength was a lot smaller too. It can be partially contributed to the decrease of adhesion at the interface between the solder mask and the copper trace due to the high temperature and high humidity environment. However, there is no failure mode observed at the interface of encapsulation and solder mask for SPP tested units, even after preconditioning and temperature cycling test. b. SPP Stress Test SPP test is very useful to evaluate the delamination of substrate and encapsulation materials. Usually, SPP test condition is 121oC temperature, 15 psi pressure, and 100% humidity. During SPP test, the ingress moisture will try to evaporate due to the higher tempera-

The International Journal of Microcircuits and Electronic Packaging, Volume 21, Number 3, Third Quarter 1998 (ISSN 1063-1674) 8 International Microelectronics And Packaging Society

271

Intl. Journal of Microcircuits and Electronic Packaging

ture and the vaporization pressure will cause the interface delamination if the interface adhesion strength is not strong enough. Figure 22 shows the SAM pictures after SPP test. From the top-scan SAM pictures, there is no die top delamination observed at the interface of die/encapsulants, however, delamination is observed in the interface of solder mask and encapsulation or BT core/solder mask as shown in Figure 23.

ing test. Moreover, the die attach integrity was evaluated to be very well. Delamination

Delamination

Moisture %: 95%-100%

(a) top scan

Delamination

Delamination 1

2

Moisture %: 40-50%

Figure 22. SAM pictures for 168 hrs SPP test results.

Moisture: 60% Delamination (die attach area)

Moisture: 95%

Figure 23. SAM pictures shown die attach delamination for different moisture levels. Figure 23 also shows that there are some delamination in the die attach area. Most likely, the delamination started at the edge of the die and propagated into the middle. Again, it seemed that with the increase of the moisture levels, a more delamination was observed under the die in the die attach area after the SPP test. The reason may be explained as follows, during moisture soak, moisture will also accumulate in the voids under the die attach and the moisture will try to evaporate and expand and then induce higher residual stress left after cure. It becomes easier to fail if these units were undergoing the SPP test.

(b) Through scan

Figure 24. SAM pictures after level 3 preconditioning test (a: top scan b: bottom scan). It should be addressed that there is no die top delamination observed any more after plasma cleaning, which was conducted just before encapsulation, even if there were moisture absorbed in the interface before plasma cleaning. This fact indicated that plasma cleaning could effectively reduce the potential for the die top delamination in the liquid dispensing whether the substrates were baked or not. Consequently, this conclusion means interface adhesion is a more critical control factor than moisture. After preconditioning test, the subsequent temperature cycling test was conducted. The C-SAMs were evaluated for those units after 200 cycles and 500 cycles. From the data, the researchers obtained after temperature cycling tests, there was no delamination in the interface of die and substrate. Also, it seems there is no correlationship between the moisture levels and package integrity failures during the temperature cycling test, however, plasma cleaning does help prevent the die top from delamination even after moisture soak.

5. Conclusions 5.1. Encapsulation Process Development

Encapsulation process is a very critical process for the Glob Top PBGA assembly. First, dispensing equipment should be evaluated followed by the encapsulation materials. Second, several DOE studies have been conducted and the critical parameters are identified and the optimized parameters are selected. Some important conclusions are as follows, 1. For damming process, the speed and air pressure play very important role. Their value will affect the dispensing volume and c. Preconditioning Test + Temperature Cycling Test achieved dam height. Another very important test for the evaluation is Level-3 precon2. For filling process, similarily, the speed and air pressure as ditioning test plus temperature cycling test. The temperature cywell as needle heating temperature also influence the dispensing cling condition is –45 oC/ 165 oC, dwell time is 15 minutes at each volume result significantly. Additionally, the dispensing pattern, tostage. The SAM pictures, shown in Figure 24, obtained after pregether with speed and needle heating temperature also affect the conditioning test, illustrated that there is heavy die top delamination dispensing quality such as voids formed in the encapsulation proin the packages. However, there was no delamination at the intercess. face of the encapsulation and the substrate at all after preconditionThe International Journal of Microcircuits and Electronic Packaging, Volume 21, Number 3, Third Quarter 1998 (ISSN 1063-1674) 272

8 International Microelectronics And Packaging Society

Glob Top Plastic Ball Grid Array Package Encapsulation Process Development and Its Moisture Sensitivity Study

3. Substrate heating is required.

5.2. Moisture Sensitivity Study In this paper, the moisture absorption, the desorption as well as the effect of moisture on die attach/encapsulation integrity were evaluated and a wealth of very important information were obtained. First, for the BT bare substrate, they are vary liable to absorb moisture in the high humidity environment and easy to drive them off too. In addition, most of the moisture absorbed in the PBGA packages were by BT and solder materials, the die attach materials as well as encapsulation materials, as the supplier claimed, do not absorb much moisture. As the effect of moisture on the die attach and encapsulation integrity, it can be concluded as follows: 1) Substrate bake before die-attach can reduce the voids number and voids size. 2) The higher moisture levels in the substrate before encapsulation can cause more delamination in the interface between the encapsulant/substrate. If the wedge bonding is weak, there will be some failures. 3) The higher levels moisture absorbed in the die attach materials, there will be a potential to cause more die attach delamination after SPP test. 4) There seems no relation between the die top delamination after preconditioning and moisture levels. In addition, plasma cleaning which enhance adhesion strength is an effective way to prevent die top delamination when the package goes through preconditioning test. 5) Using current materials and structure, the die shear strength and encapsulation strength are very strong. However, more powerful encapsulation materials are needed to prevent the delamination at the interface of encapsulant / substrate.

5. T. Moore and S. Kelshall, “The Impact of Delamination on Stress-Induced and Contamination-Related Failures in Surface Mount ICs,” Proceedings of the 30th International Reliability Physics Symposium, pp. 169-176, 1992. 6. Richard L. Shook, “Moisture Sensitivity Characterization of Plastic Surface Mount Devices Using Scanning Acoustic Microscopy”, IEEE/IRPS, pp. 157-168, 1992.

About the author L.Y. Yang received a B.E. Degree and M.E. Degree in Power and Energy Engineering from Xi’an Jiaotong University in 1991, and 1994, respectively. In 1996, he obtained another M.E. Degree in Thermal Engineering from Nanyang Technological University. Currently, he is working as an engineer at Advance Micro Devices Inc. His research areas cover thermal management for IC packages and process development for new advanced packages.

References 1. Jennie. S Hwang, “Ball Grid Array & Fine Pitch Peripheral Interconnections: A Handbook of Technology & Applications for Microelectronics/Electronics Manufacturing”, Electrochemical Publications LTD, 1995. 2. Fukuzawa, S. Ishiguro, and S. Nambu, “Moisture Resistance Degradation of Plastic LSIs by Reflow Soldering”, Proceedings of the 23rd International Reliability Physics Symposium, pp.192-197, 1985. 3. Andrew A. O. Tay and Tingyu Lin, “Moisture Diffusion and Heat Transfer in Plastic IC Packages,” IEEE Transactions on Components, Packaging, and Manufacturing Technology, CPMT-Part A, Vol. 19, No. 2, June 1996. 4. Kazuhiro Tada, Masakazu Murayama, Hirofumi Fujioka, and Hirozoh Kanegae, “Properties of Molding Compounds to Improve Package Reliability of SMDs”, IMC 1994 Proceedings, Omiya, Japan, pp. 162-166, April 20-22, 1994. The International Journal of Microcircuits and Electronic Packaging, Volume 21, Number 3, Third Quarter 1998 (ISSN 1063-1674) 8 International Microelectronics And Packaging Society

273