IEEE TRANSACTIONS ON COMPONENTS, PACKAGING AND MANUFACTURING TECHNOLOGY, VOL. 7, NO. 3, MARCH 2017 371
Moisture Effects on NCF Adhesion and SolderJoint Reliability of Chip-on-Board Assembly
Using Cu Pillar/Sn–Ag MicrobumpYoungsoon Kim, Taeshik Yoon, Tae-Wan Kim, Taek-Soo Kim, and Kyung-Wook Paik
Abstract— Electronic devices as well as electronic packagingtechnology have required higher speed, I/O capability, and den-sity. To meet these requirements, flip-chip solder bumps intercon-nection combined with reflow assembly process has been a widelyused. In spite of the many advantages of flip-chip solder bumpjoints, there is a limitation for less than 100 µm fine pitch inter-connection due to the solder bump bridging during the assemblyprocess. Therefore, nonconductive films (NCFs) and Cu pillar/Sn–Ag microbump interconnection become a promising interconnec-tion solution for fine pitch assembly. However, NCFs technologyhas some problems for flip-chip assembly. Epoxy-based NCF caneasily absorb moisture, causing delamination reliability problemat moisture environment. In order to solve the interconnection,the NCF adhesion should be enhanced. Silane coupling agent(SCA) was added to NCFs to secure the microbump jointreliability for chip-on-board assembly by increasing the adhesionstrength in the pressure cooker test. After the humidity test, theNCFs’ modulus and Tg were increased by adding SCA content.Moreover, the measured adhesion strength and energy showedsimilar results after the humidity test that higher SCA contentshowed high adhesion strength and energy than lower SCAcontent and unmodified NCF at the interface between NCF andsolder resist of the printed circuit board substrate. The bumpjoint lifetime of 5wt% SCA NCF was longer than 1wt% SCAand unmodified NCF after the humidity test. In this paper, wereport results of our investigations on effects of employing SCAin NCF composition for improved moisture resistance.
Index Terms— Adhesion, moisture, nonconductive films(NCFs), solder joint reliability.
I. INTRODUCTION
To ensure the reliability of the flip-chip packaging, underfillsare necessary to protect solder bump joint from mechanicaland chemical attacks and to improve the reliability [1]–[3].Thus, the underfill material should have higher modulus,higher Tg, lower moisture absorption, and good adhesion.Due to the trends of integrated circuit devices such as highI/O capabilities, high speed, and high density, the bumppitch of the flip-chip packaging becomes narrower. The finepitch of flip-chip interconnection can cause severe reliability
Manuscript received August 24, 2015; revised August 31, 2016; acceptedNovember 6, 2016. Date of publication February 14, 2017; date of currentversion March 14, 2017. Recommended for publication by Associate EditorR. N. Das upon evaluation of reviewers’ comments.
Y. Kim, T.-W. Kim, and K.-W. Paik are with the Department of MaterialsScience and Engineering, Korea Advanced Institute of Science andTechnology, Daejeon 305-701, South Korea (e-mail: [email protected]).
T. Yoon and T.-S. Kim are with the Department of Mechanical Engineering,Korea Advanced Institute of Science and Technology, Daejeon 305-701, SouthKorea.
Color versions of one or more of the figures in this paper are availableonline at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TCPMT.2016.2631540
problems due to the underfill voids and flux residue after thecapillary underfill process [4]. Therefore, B-stage preappliednonconductive films (NCFs) can be a promising technology tosolve the reliability problems of capillary underfill and makestable microsolder joints without underfill voids. Moreover,B-stage preapplied NCFs can reduce packaging process stepssuch as flux treatment and flux cleaning. To achieve stablesolder joints and void suppression at NCFs, NCFs should haveproper material properties to secure interconnection reliability.First, the resin material of the NCF should flow easily duringthe bonding process to avoid the resin trapping to occurbetween the metal bumps and substrate electrode. Second, theNCF resin material should have a flux function to removethe solder oxide at the bump in order to form stable solderjoints. Finally, there should be no voids when the resin iscompletely cured after final bonding at high temperature. Itwas reported that many NCFs used epoxy resin to secure thereliability of interconnection [5]–[8]. Epoxy resins are widelyused in electronic packaging due to its advantages such asexcellent mechanical and chemical properties. However, epoxyresin has a critical problem of moisture absorption. Epoxyresin can easily absorb moisture that can produce a severereliability issue of delamination since the absorbed moisture isexpanded when exposed to high temperature during packagingprocess [9]–[14]. The absorbed moisture within the epoxyresin can be classified into two states. The first state of theabsorbed moisture is the bound water where hydrogen bondingoccurs between the epoxy resin chains and the interfaces. Thesecond state is called the unbound water, which sits at voidsand free volume within an epoxy resin. Normally, more than90% of absorbed moisture in the epoxy resin is known as thebound water. Moreover, the bound water is mainly involved inhygroscopic swelling of epoxy resin, whereas unbound wateris not involved in the epoxy swelling [15]. As the moistureabsorption increases, the vapor pressure in epoxy resin is alsoincreased, while interfacial adhesion is decreased. Althoughthe vapor pressure saturates during the absorption process,the interfacial adhesion continually decreases. Thus, resindelamination can occur when the effects of interfacial adhesionare below the effects of the vapor pressure [16]. Preventing themoisture absorption is the main challenge at the solder–flip-chip interconnection, since this will improve adhesion, whichis the most important factor to prevent delamination at theinterface. Thus, preventing moisture absorption will make surethat NCFs have sufficient interfacial adhesion.
In this paper, the effects of silane coupling agent (SCA) onNCF adhesion between the Si chips and the printed circuit
372 IEEE TRANSACTIONS ON COMPONENTS, PACKAGING AND MANUFACTURING TECHNOLOGY, VOL. 7, NO. 3, MARCH 2017
Fig. 1. Molecular structure of glycidoxypropyltrimethoxysilane.
Fig. 2. (a) Image and (b) schematic of the die shear test.
board (PCB) substrates were evaluated. Here, NCFs wereapplied on the microbumps on a Si chip. The die shear andDCB test were performed to evaluate the NCF adhesion atthe interface between the NCFs and the solder-resist (SR)laminated PCB substrates depending on the SCA contents inthe NCFs. The pressure cooker test [121 °C, 2 atm, and 100%relative humidity (RH)] was performed up to 264 h to evaluatethe solder joint reliability depending on the SCA content inNCFs. The contact resistance was measured using the four-point probe system.
II. EXPERIMENTS
A. NCF Material Preparation and Mechanical Property
The NCFs used in this experiment are composed of threemain parts: 1) resin; 2) anhydride; and 3) additives. Epoxyresin was mainly used for the resin material, and multi-functional resin and phenoxy resin were mixed for enhancedfilm properties and had the characteristic of film formation,respectively. The anhydride material was used as a latent epoxycuring agent. In addition, some additives such as the silicafiller and the curing accelerator were used as a supporter.In this experiment, three different types of NCFs were used:1) the unmodified NCF; 2) the NCF with 1wt% SCA; and3) the NCF with 5wt% SCA, respectively. Fig. 1 shows themolecular structure of the glycidoxypropyltrimethoxysilane.The hydrolyzable group is made of hydrogen bonding withOH on the substrate surface, and the organofunctional groupis involved in cure of ether in NCFs.
The mechanical properties such as storage modulus and losstangent (tan δ) were measured using the dynamic mechanicalanalyzer (DMA) using a sinusoidal force of 100 ± 20 mN at0.02 Hz, and the temperature ranged from 30 °C to 180 °Cwith a heating rate of 5 °C/min. The moisture uptake ratiowas measured by weight changes after a humidity test using10 × 10-mm2 size and 500-μm-thickness NCFs.
B. Test Vehicle
1) Die Shear Test: For the top dummy test chip, a blankSi wafer was diced into 2.5 × 2.5 mm2 size. For the bottom
Fig. 3. (a) Schematic of the DCB specimen. (b) Photograph of the fabricatedDCB specimen.
Fig. 4. Image of a (a) Si test chip and a (b) PCB substrate for COB assembly.
Fig. 5. Top and cross-sectional images of the Cu pillar/Sn–Ag microbumpon a Si test chip.
substrate, copper clad lamination (CCL) with a thickness of60 μm was used as the core layer. The top and bottom ofthe CCL was first electroplated with a 15-μm Cu layer, and a20-μm-thick solder resist was coated on top of the Cu layer.Fig. 2 shows the top view and the schematic of the die sheartest specimen.
2) Double Cantilever Beam Test: PCBs of size10 × 40 mm2 that consisted of CCL, electroplated Cu,and solder resist were used as beam materials in the DCBspecimen. As seen in Fig. 3, 25-μm-thick NCFs werelaminated between two symmetric PCB beams and werecured at 210 °C under the 50-mbar pressure. NCFs werepartially applied on PCB beams to make a precrack region of10 mm, which ensures the stable crack initiation during thefracture test. After the bonding of PCB beams with NCF, thealuminum loading tabs were attached on both edges of PCBbeams to apply tensile forces on the DCB specimen.
3) Reliability Evaluation: For the top chip, 10 × 10-mm2-size Si chips with a bump pitch of 130 μm were used.The microbump structure on a Si chip was Cu pillar/Sn–Ag where the Cu pillar and Sn–Ag heights were 50 and20 μm, respectively. When observing the top view of themicrobump, the diameter was 70 μm. For the bottom PCB,20 × 20-mm2-size PCB was used. For the PCB substrate, thecore thickness of PCB was 150 μm, and a 15-μm thickness
KIM et al.: MOISTURE EFFECTS ON NCF ADHESION AND SOLDER JOINT RELIABILITY OF COB ASSEMBLY 373
Fig. 6. Top and cross-sectional images of coined solder bumps on a PCBsubstrate after the coining process.
Fig. 7. Photograph of the DCB fracture test.
of Cu layer was electroplated followed by a 20-μm solderresist covering. A 75-μm-diameter size (top view) coined SACsolder bump with a height of 35 μm was formed on Cu bumppads with electroless nickel electroless palladium immersiongold (ENEPIG) surface finish (Ni: 7 μm, Pd: 0.06 μm, andAu: 0.1 μm). The coined SAC solder bump was 15 μm higherthan the solder-resist surface. Fig. 4 shows the top images of aSi chip and a PCB substrate used for the chip-on-board (COB)assembly. Figs. 5 and 6 show the top and cross-sectionalimages of the Cu pillar/Sn–Ag microbump on a Si test chipand coined solder bumps on a PCB substrate after the coiningprocess, respectively.
C. Evaluation of Adhesion Strength of NCF
The adhesion strength of the NCFs with various SCAcontent (0wt%, 1wt%, and 5wt%) were evaluated using the dieshear test before and after the pressure cooker test at 121 °C,2 atm, and 100% RH.
For the die sheer test, COB samples were prepared asfollows. First, a silicon wafer was diced into 2.5 × 2.5 mm2
size and the PCB substrate was diced into 10 × 10 mm2 size.Then, the 20-μm-thickness NCF was laminated on the dicedsilicon wafer surface using a roll laminator. After the NCFwas laminated, thermal compression bonding was performedat 250 °C and 1 Mpa for 60 s to attach the silicon chip on aPCB substrate. Finally, vacuum lamination was performed at210 °C for 10 min to fully cure the NCF resin. The die sheertest was performed at room temperature, and the shear speedand shear height were 700 μm/s and 100 μm, respectively.
D. Evaluation of Adhesion Energy of NCF/Solder-ResistInterface Using DCB Test
The adhesion energy between the NCF and SR was mea-sured using the DCB fracture mechanics test [17]. The con-tents of SCA were 0wt%, 1wt%, and 5wt%. The as-bonded
Fig. 8. Modulus and tan δ of NCFs before and after the humidity testdepending on various amounts of SCA. (a) Unmodified NCF. (b) 1wt% SCAadded NCF. (c) 5wt% SCA added NCF.
and pressure-cooker-tested (48 and 120 h) samples were testedto characterize the humidity reliability. The PCB was used asa beam material, and the flexural modulus of a single PCBbeam was measured by a three-point bending test. The DCBspecimens were tested by a high-precision micromechanicaltester, which controls the displacement rate and monitorsthe applied load as shown in Fig. 7. The DCB specimenswere loaded and unloaded under a constant displacement
374 IEEE TRANSACTIONS ON COMPONENTS, PACKAGING AND MANUFACTURING TECHNOLOGY, VOL. 7, NO. 3, MARCH 2017
TABLE I
SUMMARY OF THE DMA RESULTS OF NCFs
Fig. 9. Moisture absorption ratios of various NCFs.
Fig. 10. Adhesion strength depending on the added amount of SCA.
rate of 50 μm/s, and then a load–displacement curve can beobtained. By considering the dimensions and flexural modulusof the PCB beam, the adhesion energy can be calculated asa function of crack length and critical load, based on thefracture mechanics model [18], [19]. The multiple values ofthe adhesion energy can be probed in a specimen by repeatingloading, crack growth, and unloading cycles.
E. Moisture Absorption
The moisture uptake rate is the key parameter when con-sidering NCFs’ adhesion strength since the adhesion strengthof the NCFs decreased when vapor pressure increased. Threetypes of NCFs with various SCA content (0wt%, 1wt%, and
Fig. 11. Adhesion strength of various NCFs on PCB substrates after a PCTtest.
5wt%) were cured at 210 °C for 10 min using a vacuumlaminator and baked at 150 °C for 24 h for full cure beforemeasuring the moisture weight gain. The moisture weightgain of the NCFs was calculated using the electronic balanceequipment. After all the weights of NCFs before the PCT weremeasured, they were placed in a PCT chamber to perform thePCT test at 121 °C, 100% RH, and 2 atm. The weights ofthe NCFs were measured after 0, 2, 4, 8, 12, 24, 48, 72, and120 h to observe the amount of moisture absorbed dependingon the SCA content.
F. Reliability Evaluation
Three NCFs with various SCA content (0wt%, 1wt%, and5wt%) were used to investigate the NCF adhesion strengtheffects on the microbump joint reliability of the COB assem-bly. A Si chip having dimension 10 × 10 mm2 size was bondedon 20 × 20-mm2 PCB using thermal compression bondingafter laminating the B-stage preapplied NCFs on a Si chipsurface. The thermal compression bonding was performed at250 °C for 30 s with a heating rate of 6 °C/s. After the thermalcompression bonding, the PCT test was performed at 121 °C,100% RH, and 2 atm up to 216 h, and the contact resistancesof the microbump joints were measured using a four-pointprobe to verify microbump reliability.
KIM et al.: MOISTURE EFFECTS ON NCF ADHESION AND SOLDER JOINT RELIABILITY OF COB ASSEMBLY 375
Fig. 12. Type of fracture mode after a die shear test of a Si chip bonded on a solder-resist applied PCB using NCFs. (a) Si, NCF, and SR. (b) Si and NCF.(c) NCF.
III. RESULTS AND DISCUSSION
A. Evaluation of Mechanical Properties of NCFs
The mechanical properties of the cured NCFs were deter-mined by measuring the changes of modulus and loss tangent(tan δ) of the three NCFs depending on the SCA content beforeand after 72 h of the PCT reliability test at 121 °C, 100% RH,and 2 atm.
Fig. 8 shows the modulus and tan δ results of three NCFs.The modulus of all NCFs decreased compared to as-curedNCFs. Especially at high temperature, the degradation ofthe mechanical properties of NCFs was presumably due tomoisture absorption, which acts as a plasticizer [20]. Themodulus slightly increased by adding SCA into NCFs. Theglass temperature (Tg) was defined at the maximum peak oftan δ. Tg also increased by adding SCA. This is presumablybecause the organofunctional groups of SCA participated inthe NCF resin curing process. As a result, the cross-linkingdensity increased by the added SCA. The modulus and Tg
data results were measured by the DMA and summarized inthe Table I.
B. Moisture Uptake Ratio
The moisture uptake ratio of epoxy resin is a critical factorbecause the moisture can be easily absorbed in NCFs and dif-fuse at the interface between the NCFs and the PCB substrates.Absorbed moisture can increase the vapor pressure at the inter-face between the NCFs and the PCB substrates resulting inweakening of interfacial adhesion. Thus, the moisture uptakeratio of NCFs and the diffusivity at the interface between theNCFs and the PCB substrates are important material properties[12], [21], [22]. The moisture uptake ratio is defined as thedifference in weight before and after the humidity test. AllNCFs were exposed at 121 °C, 100% RH, and 2 atm in thePCT chamber for 120 h. The moisture absorption of NCFsdecreased by the added SCA, as shown in Fig. 9. Althoughthe moisture uptake ratio of the three types of NCFs wasdifferent, all NCFs reached 90% absorption ratio of moistureat an early stage. This corresponds to the general moistureabsorption mechanism [23]. The moisture uptake ratio wasdifferent depending on the amount of SCA added in NCFsbecause the organofunctional group of SCA was involved inthe epoxy resin curing system. The moisture uptake ratio was4.83% for the unmodified NCF, 4.15% for the NCF with 1wt%SCA, and 3.53% for the NCF with 5wt% SCA, respectively.By increasing the SCA content, the cross-linking density ofthe NCFs increases because the moisture absorption site suchas nanovoid/microvoid or free volume decreases. Eventually,
Fig. 13. Failure modes of a die shear test after a humidity test. (a) Fracturemode of the unmodified NCF. (b) Fracture mode of the 1wt% SCA addedNCF. (c) Fracture mode of the 5wt% SCA added NCF.
the mechanical property of NCFs can be improved by addingSCA.
C. Evaluation of the Adhesion Strength of NCFs
Fig. 10 shows the adhesion strength results of the NCFsmeasured by a die shear test. As the SCA contents in NCFs
376 IEEE TRANSACTIONS ON COMPONENTS, PACKAGING AND MANUFACTURING TECHNOLOGY, VOL. 7, NO. 3, MARCH 2017
Fig. 14. DCB specimens before and after a humidity test. (a) After as-bonded.(b) After 48 h of humidity test. (c) After 120 h of humidity test.
Fig. 15. Measured adhesion energy of NCFs on the SR during the PCT testusing the DCB test.
increase, the adhesion strength increases slightly. This ispresumably because the hydrolyzable group in the SCA reactswith the OH sites on the organic substrate surface.
Also when observing the die shear test results of NCFs inthe humidity test at 121 °C, 100% RH, and 2 atm, all the threeNCFs showed a decrease in adhesion strength value compared
to the adhesion strength before the humidity test, as shownin Fig. 11. However, the loss of adhesion strength was lesssignificant for 1wt% SCA and 5wt% SCA NCFs compared tothe same for the unmodified NCF.
The failure sites after the die shear test were classified intofive categories depending on the remaining component on thefailure surface, as shown in Fig. 12. The initial fracture surfacemode after the die shear test showed the remaining three layersof Si chip, NCF, and solder resist, as shown in Fig. 13(a).The fracture mode of the SCA-added NCFs shows that theNCF portion decreases at the failure site, as the SCA contentincreases, because the adhesion strength between the NCFsand the solder resists increased.
The adhesion strength after die shear test showed a dramaticchange after a humidity test. In the case of unmodified NCF,the adhesion strength between the SR and the NCF disappearsafter 12 h of PCT, as shown in Fig. 13(a). Here, the fracturesite was the interface between the NCF and the SR. However,as the added SCA in NCFs increased, the adhesion betweenthe SR and the NCF was somewhat maintained even after thehumidity test, as shown in Fig. 13(b) and (c). The fracturesurface of the NCF with 1wt% SCA was the mixed mode ofFig. 12(a) and (d) up to 24 h of humidity test. After 24 h ofhumidity test, the fracture surface was the mode of Fig. 12(d).On the other hand, the NCF with 5wt% SCA showed themodes of Fig. 12(a) and (d) for the fracture surface up to 120 hof humidity test. The remaining Si chip at the fracture surfaceimplies that the adhesion strength was significantly enhancedat the interface between the NCFs and the SRs by addingSCA. The decrease in adhesion strength at the interface waspresumably due to the moisture absorption effect in the NCFs.The absorbed moisture was squeezed between the OH sitesat interface between the NCF and the SR during a humiditytest. Therefore, the absorbed moisture prevents the chemicalbonding between the NCFs and SRs through the hydrolysisreaction. By adding epoxy SCA into NCFs, this allows thehydrozable group in the SCA to form a hydrogen bondingwith the substrate surface. Eventually, this strengthens thechemical bonding at the interface between NCFs and SR aswell as reduces the OH sites on the solder-resist surface thatcan absorb moisture during the humidity test.
D. Evaluation of the Adhesion Energy of NCFs
The adhesion energy between NCFs and SRs was measuredusing the DCB test while the adhesion strength was measuredusing the die shear test. The adhesion energy represents therequired energy to fracture the interface; therefore, the char-acterization of the adhesion energy is also critical, particularlyat the impact loading condition.
Fig. 14 shows the interface of DCB specimens after they areas-bonded in the DCB test and after 48 and 120 h of humiditytest. As shown in Fig. 15, the SCA successfully enhanced theNCF adhesion energy, and the energy was decreased as thePCT time increased. The trends are similar to the adhesionstrength data using the die shear test. However, the adhesionenergy values of NCF with 1wt% SCA were slightly higherthan those of NCF with 5wt% SCA, presumably due to the
KIM et al.: MOISTURE EFFECTS ON NCF ADHESION AND SOLDER JOINT RELIABILITY OF COB ASSEMBLY 377
Fig. 16. Microbump joint images of (a) unmodified NCF, (b) 1wt% SCA added NCF, and (c) 5wt% SCA added NCF after bonding.
Fig. 17. Contact resistances of microbump joints of (a) unmodified, (b) 1wt%SCA, and (c) 5wt% SCA added NCFs after PCT.
high storage modulus of the NCF with 5wt% SCA as describedin Table I. In general, polymers become brittle at highermodulus. Therefore, too much SCA reduced the ductility ofNCFs resulting in the reduced fracture absorbing ability. Itis good that silane-contained NCFs have a sufficiently highadhesion energy compared to the unmodified NCF.
E. Verification of Silane Coupling Agent on the MicrobumpJoint Reliability of COB Assembly
Fig. 16 shows the microbump joint images bonded by athermocompression bonding using preapplied NCFs. Threedifferent SCA contents were added in the NCFs for compari-son. All microbump joints showed well-defined solder joints.The contact resistance results of microbump joints after thePCT using various silane contents (0wt%, 1wt%, and 5wt%)are shown in Fig. 17. For the unmodified NCF, the contactresistance increased by 20% after 12 h. Some joint regionsshowed open failure after 120 h and complete open failureafter 264 h during a PCT test. For the NCF with 1wt% SCA,the contact resistance increased by 20% after 48 h, and somejoint regions show a complete open failure after 264 h during aPCT test. However, for the NCF with 5wt% SCA, the contactresistances increased less than 20% after 72 h, and no bumpjoints were failed even after 264 h of the PCT test, showing
excellent bump joint stability. The contact resistance resultsof three different NCFs obviously showed a significant SCAeffect on the bump joint reliability after the PCT test.
Fig. 18 shows the microbump joint failure mode of theCOB assemblies after the PCT test. Three different typesof failure modes were observed: 1) microsolder bump crack;2) delamination at the NCFs and the SRs; and 3) opened bumpjoints. At the early stage of PCT test, microsolder crack occursas shown in Fig. 18(a). As the adhesion strength between theNCF and SR decreased further due to the moisture absorptionat NCFs, delamination occurs at the NCF and SR interface asshown in Fig. 18(b). Microsolder bump joints were completelyfailed as shown in Fig. 18(c). In the case of the unmodifiedNCF and the NCF with 1wt% SCA, completely failed solderjoints were observed after 264 h of PCT test. On the otherhand, the NCF with 5wt% SCA showed no bump joint failureeven after 264 h of PCT, which demonstrates the excellentNCF adhesion property causing the stable solder bump jointreliability.
IV. CONCLUSION
In this paper, the mechanical properties, moisture absorp-tion, and adhesion strength of NCFs modified with variousamounts of SCA were investigated. In addition, we evaluatedthe lifetimes of the modified NCFs upon exposure to temper-ature and humidity.
The added SCA significantly affects the NCF mechanicalproperties such as the adhesion strength of the NCFs and thesolder resist interface of PCB substrates.
When observing the NCF properties, higher SCA-addedNCFs showed higher modulus and Tg as well as less moistureabsorption rate during high temperature and high humidityaging presumably due to the higher cross-linking density ofSCA-added NCFs.
The adhesion strength was significantly improved by addingSCA to the NCFs. SCA allows the OH bond to chemicallybond with the solder resist, resulting in an increase of theadhesion strength between NCF and SR. Furthermore, SCAprobably reduced the moisture absorption sites on the solderresist surface. The SCA effectively improved the adhesionenergy of NCFs. The NCF with 1wt% SCA had higheradhesion energy than the NCT with 5wt% SCA presumablybecause of the increased storage modulus of the NCF with5wt% SCA.
378 IEEE TRANSACTIONS ON COMPONENTS, PACKAGING AND MANUFACTURING TECHNOLOGY, VOL. 7, NO. 3, MARCH 2017
Fig. 18. Failure modes of the microbump joint. (a) Microsolder bump crack. (b) Delamination at NCFs and SR. (c) Opened bump joint failure.
The microbump joint reliability evaluation of the unmodi-fied NCF and the NCF modified with two different loadings ofSCA shows that NCFs with a higher silane content offer longerlifetime upon exposure to humidity. This confirms excellentimprovement in the adhesion property of the NCFs upon themodification of SCA.
REFERENCES
[1] Z. Zhang and C. P. Wong, “Recent advances in flip-chip underfill:Materialsprocess, and reliability,” IEEE Trans. Adv. Packag., vol. 27,no. 3, pp. 515–524, Mar. 2004.
[2] B. Han and Y. Guo, “Thermal deformation analysis of various electronicpackaging products by Moiré and microscopic Moiré interferometry,”J. Electron. Packag., vol. 117, no. 3, pp. 185–191, 1995.
[3] D. Suryanrayana, R. Hsiao, T. P. Gall, and J. M. McCreary, “Enhance-ment of flip-chip fatigue life by encapsulation,” IEEE Trans. Compon.Hybrids Manuf. Technol., vol. 14, no. 1, pp. 218–223, Mar. 1991.
[4] Y. Y. Ong et al., “Assembly and reliability of micro-bumped chipswith through-silicon vias (TSV) interposer,” in Proc. EPTC, Dec. 2009,pp. 452–458.
[5] Y. C. Chan, S. C. Tan, S. M. Nelson Lui, and C. W. Tan, “Elec-trical characterization of NCP- and NCF-bonded fine-pitch flip-chip-on-flexible packages,” IEEE Trans. Adv. Packag., vol. 29, no. 4,pp. 735–740, Nov. 2006.
[6] Y. Choi, J. Shin, I. Kim, Y.-S. Kim, and K. W. Paik, “Fine pitch chip onboard (CoB) bonding using B-stage non-conductive film (NCF) for 3DTSV vertical interconnection,” in Proc. IEEE Int. Symp. Adv. Packag.Mater., Mar. 2013, pp. 186–191.
[7] Y.-M. Lin, S.-T. Lu, and T.-H. Chen, “Reliability of 20μm pitch NCFtype COF package with compliant-bump,” in Proc. IMPACT, Oct. 2009,pp. 568–571.
[8] S. Kawamoto, M. Yoshida, S. Teraki, and H. Iida, “Effect of NCF designfor the assembly of flip chip and reliability,” in Proc. ECTC, Jun. 2012,pp. 399–405.
[9] J. C. W. van Vroonhoven, “Effects of adhesion and delamination onstress singularities in plastic-packaged integrated circuits,” J. Electron.Packag., vol. 115, no. 1, pp. 28–33, Mar. 1993.
[10] M. H. Shirangi, J. Auersperg, M. Koyuncu, H. Walter, W. H. Muller,and B. Michel, “Characterization of dual-stage moisture diffusion,residual moisture content and hygroscopic swelling of epoxy moldingcompounds,” in Proc. 9th EuroSime, Freiburg, Germany, Apr. 2008,pp. 455–462.
[11] M. H. Shirangi, X. J. Fan, and B. Michel, “Mechanism of moisturediffusion, hygroscopic swelling and adhesion degradation in epoxymolding compounds,” in Proc. 41st Int. Symp. Microelectron. (IMAPS),Providence, RI, USA, 2008, pp. 1082–1089.
[12] X. J. Fan, J. Zhou, and A. Chandra, “Package structural integrityanalysis considering moisture,” in Proc. Electron. Compon. Technol.Conf. (ECTC), Orlando, FL, USA, 2008, pp. 1054–1066.
[13] M. G. Pecht, L. T. Nguyen, and E. B. Hakim, Plastic-EncapsulatedMicroelectronics. New York, NY, USA: Wiley, 1995.
[14] H. Ardebili, C. Hillman, M. A. E. Natishan, P. McCluskey, M. G. Pecht,and D. Peterson, “A comparison of the theory of moisture diffusionin plastic encapsulated microelectronics with moisture sensor chip andweight-gain measurements,” IEEE Trans. Compon. Packag. Technol.,vol. 25, no. 1, pp. 132–139, Mar. 2002.
[15] R. H. Hodge, G. H. Edward, and G. P. Simon, “Water absorption andstates of water in semicrystalline poly (vinyl alcohol) films,” Polymer,vol. 37, no. 8, pp. 1371–1376, 1996.
[16] X. Fan, G. Q. Zhang, D. Willem van Driel, and J. L. Ernst, “Interfacialdelamination mechanisms during soldering reflow with moisture pre-conditioning,” IEEE Trans. Compon. Packag. Technol., vol. 31, no. 2,pp. 252–259, Feb. 2008.
[17] T. Yoon, W. C. Shin, T. Y. Kim, J. H. Mun, T.-S. Kim, and B. J. Cho,“Direct measurement of adhesion energy of monolayer grapheneAs-grown on copper and its application to renewable transfer process,”Nano Lett., vol. 12, no. 3, pp. 1448–1452, 2012.
[18] M. F. Kanninen, “An augmented double cantilever beam model forstudying crack propagation and arrest,” Int. J. Fracture, vol. 9, no. 1,pp. 83–92, Mar. 1973.
[19] R. J. Hohlfelder, D. A. Maidenberg, and R. H. Dauskardt, “Adhesionof benzocyclobutene-passivated silicon in epoxy layered structures,”J. Mater. Res., vol. 16, no. 1, pp. 243–255, 2001.
[20] R. A. Jurf and J. R. Vinson, “Effect of moisture on the static andviscoelastic shear properties of epoxy adhesives,” J. Mater. Sci., vol. 20,no. 8, pp. 2979–2989, 1985.
[21] L. K. Teh et al., “Moisture-induced failures of adhesive flip chipinterconnects,” IEEE Trans. Compon. Packag. Technol., vol. 28, no. 3,pp. 506–516, Sep. 2005.
[22] P. Alpern and K. C. Lee, “Moisture-induced delamination in plasticencapsulated microelectronic devices: A physics of failure approach,”IEEE Trans. Device Mater. Rel., vol. 8, no. 3, pp. 478–483, Sep. 2008.
[23] R. Aschenbrenner, J. Gwiasda, J. Eldring, E. Zakel, and H. Reichl, “Flipchip attachment using non-conductive adhesives and gold ball bumps,”Int. J. Microcircuits Electron. Packag., vol. 18, no. 2, pp. 154–161,1995.
Youngsoon Kim, photograph and biography not available at the time ofpublication.
Taeshik Yoon, photograph and biography not available at the time ofpublication.
Tae-Wan Kim, photograph and biography not available at the time ofpublication.
Taek-Soo Kim, photograph and biography not available at the time ofpublication.
Kyung-Wook Paik, photograph and biography not available at the time ofpublication.
