Characterization of the viscoelastic properties of an epoxy molding compound during cure

In the electronics industry epoxy molding compounds, underfills and adhesives are used for the packaging of electronic components. These materials are applied in liquid form, cured at elevated temperatures and then cooled down to room temperature. During these processing steps residual stresses are built up resulting from both cure and thermal shrinkage. These residual stresses add up to the stresses generated during thermal cycling and mechanical loading and may eventually lead to product failure.The viscoelastic properties of the encapsulation material depend highly on temperature and degree of cure. This paper investigates the increase of elastic modulus and the changes in the viscoelastic behavior of an epoxy molding compound, during the curing process. This is done using the shear setup of a Dynamic Mechanical Analyzer DMA-Q800. The cure dependent viscoelastic behavior is determined during heating scans of an intermittent cure experiment. In such an experiment the material is partially cured and then followed by a heating scan at 2 °C/min. During this heating scan continuous frequency sweeps are performed and the shear modulus is extracted. The Time–Temperature superposition principle is applied and the viscoelastic shear mastercurve is extracted. Analyzing the shear modulus, the cure dependent viscoelastic material behavior was modeled using the cure dependent glass transition temperature as a reference and a cure dependent rubbery modulus. It is shown that partial curing would increase the glass transition temperature and rubbery shear modulus. It also shifts the viscoelastic mastercurve to the higher time domain. Taking T_g as the reference temperature for different heating scans, the mastercurves collapse to one graph.In addition, using a Differential Scanning Calorimeter (DSC), the growth of the glass transition temperature, $T_g^{\mathit{DSC}}$, with respect to the conversion level is obtained. These values are coupled to the values of glass transition temperature in DMA apparatus, $T_g^{\mathit{DMA}}$, for calculating the conversion level at each step of curing process in shear mode test.     

Experimental identification of thermal induced warpage in polymer–metal composite films

Composite thick films consisting of multi-layered polymers and metals are widely used in integrated circuits(IC) and its packaging, and it arises intricate stress and warpage problems due to complex inner stress distribution and evolution. The wafer warpage origination and evolution of multi-layered polyimide (PI)/Cu composite film is measured in-situ by a Multi-beam Laser Optical Sensor (MOS) system. It's found that PI has an intricate influence on wafer warpage evolution and Cu plastic deformation due to viscoelasticity and glass-transition, and the influence differs in different structures and at different temperatures. Nonlinearity of the curvature–temperature curve of the composite occurs at much lower temperature than in single PI or Cu film, showing mutual effect of PI and Cu. Unlike bare or capping PI film that totally stress relaxed at high temperature, bottom PI coated by Cu film sustains a medium compressive stress, indicating that Cu coating film has restrained stress relaxation of PI. The warpage evolution during heating is different from that during cooling, perhaps due to different deformation mechanism.     

Study of thinned Si wafer warpage in 3D stacked wafers

3D (three-dimensional) wafer stacking technology has been developed extensively recently. One of the many technical challenges in 3D stacked wafers, and one of the most important, is wafer warpage. Wafer warpage is one of the root causes leading to process and product failures such as delamination, cracking, mechanical stresses, within wafer (WIW) uniformity and even electrical failure. In this study, the wafer warpage of thinned Si wafers in stacked wafers has been evaluated. Si wafer or glass was used as a thick substrate, and Cu or polyimide was used as the bonding material. The top Si wafer in the bonded stack was ground down to 20–100 μm, and wafer curvature was measured. Wafer curvature and how it relates to bonding material, substrate material of the stacked layers, and thickness of thinned Si wafer will be discussed.     

Electrical conduction in polyimide between 20 and 350° C

Although conduction in polyimides at elevated temperatures has been widely reported, measurements at ordinary device temperatures have been less well documented. Quantitatively reproducible low field conduction measurements on two device-grade polyimides (PMDA-ODA, BTDA-ODA/MPDA) in the temperature range of 20–350° C and under dry conditions are reported. Aluminum—polyimide—aluminum capacitors are prepared by spin coating an aluminized silicon wafer with between two and four coats of polyimide (prebake at 135° C for 10 min between coats). Samples are cured in dry nitrogen at 400° C for 45 min. Final thickness ranged between 3.3 and 6.6 μm. To permit rapid equilibration of moisture between the film and ambient, the upper electrode is patterned into multiple 25 μm stripes with 5 μ spaces for a total area of 5.1 cm². After a bake-out at 120° C under dry air and subsequent equilibration in a dry ambient at the test temperature, a voltage step is applied to the sample and the current versus time is recorded for 16,000 sec (the charging current). The sample is then shorted, and the discharging current is recorded. Below 100° C, both charging and discharging currents are dominated by a reversible polarization that follows a power law (approximately t^−0.8). Isochronal plots of the polarization current reveal a linear dependence on the applied voltage for fields in the range 10⁴–10⁵ V/cm. The polarization current is nearly independent of temperature and is well modeled by the Lewis molecular dipole theory of polarization. Above 150° C, the current is increasingly dominated by a relatively constant transport current, defined as the difference between charging and discharging currents. This current is ohmic over the field range examined, and shows a complex, activated temperature dependence. For PMDA-ODA the transport current has an activation energy (E _a ) of 0.5 eV below 175° C and 1.5 eV above that temperature. For BTDA-ODA/MPDA the E_a is 0.6 up to 250° C and 2.1 eV above. This corresponds to a resistivity of 9 × 10¹⁸ Ω-@#@ cm at 23° C and 3.5 × 10¹⁴ Ω-cm at 200° C for PMDA-ODA and 5 × 10¹⁹ Ω-cm at 23° C and 5.6 × 10¹³ Ω-cm at 300° C for BTDA-ODA/MPDA. This work demonstrates that the low temperature behavior of polyimide cannot be extrapolated from high temperature measurements. Work sponsored in part by E. I. DuPont de Nemours & Co., Inc.     

Ohmic contacts to InP-based materials induced by means of rapid thermal low pressure (metallorganic) chemical vapor deposition technique

A rapid-thermal-low-pressure-metallorganic-chemical-vapor-deposition (RT-LPMOCVD) technique was executed in order to deposit non-semiconductor thin layer materials, necessary for producing metal contact to InP-based microelectronic devices. Silicon dioxide (SiO₂) films were deposited onto InP substrates in rapid thermal cycles, using O₂ and 2% diluted SiH₄ in Ar, with very fast growth kinetics and low activation energy. The SiO₂ film exhibited excellent properties, such as refractivity index, density, internal stress, and wetp-etch rates. The SiO₂ films were dry etched in a given pattern to allow for the formation of a small metal contact to the InP-based material, onto which the SiO₂ layer was deposited. Subsequently, titanium-nitride (TiN _x ) thin films were deposited onto the InP substrate through rapid thermal deposition cycles, using a tetrakis (dimethylamido) titanium (DMATi) metallorganic liquid source as the precursor for the process, with fast kinetics. The deposited TiN _x films had a stoichiometric structure and contained nitrogen and titanium in a ratio close to unity, but incorporated a large amount of carbon and oxygen. The film properties, such as resistivity (40–80 μΩ·mm) and stress (compressive; ?0.5 to ?2.0×10⁹ dyne·cm^?2), were studied in addition to an intensive investigation of its microstructure and morphology, and their performance as an ohmic contacts while deposited ontop?In_0.53Ga_0.47As material (Zn doped 1.2×10¹⁸ cm^?3).     

Improving the Magnetic Properties of Li-Zn-Ti Ferrite by Doping with H3BO3-Bi2O3-SiO2-ZnO Glass for LTCC Technology

Li-Zn-Ti ferrite doped with 0.5 wt.% to 16 wt.% H₃BO₃-Bi₂O₃-SiO₂-ZnO (BBSZ) glass was synthesized using a low-temperature ceramic sintering process. Selected parameters of saturation induction (B _S), coercivity (H _C), Curie temperature (T _C), and complex permeability spectra were measured as functions of doping content, and their relationships with ferrite density and microstructure are discussed. It was found that Li-Zn-Ti ferrite can be fired at low temperature (900°C) with BBSZ glass content varying from 0.5 wt.% to 2 wt.%. The real permeability increased from 80 to 190 in the frequency range from 1 MHz to 3 MHz, the saturation induction B _S increased from 105 mT to 150 mT at 1 kHz, whereas the coercivity H _C decreased from 165 A/m to 65 A/m at 1 kHz and the Curie temperature T _C slightly declined from 155°C to 143°C. These results confirm that this new ferrite material could be used in low-temperature cofired ceramic (LTCC) devices.     

Electrical Characteristics of Al/Poly(methyl methacrylate)/p-Si Schottky Device

Al/Poly(methyl methacrylate)(PMMA)/p-Si organic Schottky devices were fabricated on a p-Si semiconductor wafer by spin coating of PMMA solution. The capacitance–voltage (C–V) and conductance–voltage (G–V) characteristics of Al/PMMA/p-Si structures have been investigated in the frequency range of 1 kHz–10 MHz at room temperature. The diode parameters such as ideality factor, series resistance and barrier height were calculated from the forward bias current–voltage (I–V) characteristics. In order to explain the electrical characteristics of metal–polymer–semiconductor (MPS) with a PMMA interface, the investigation of interface states density and series resistance from C–V and G–V characteristics in the MPS structures with thin interfacial insulator layer have been reported. The measurements of capacitance (C) and conductance (G) were found to be strongly dependent on bias voltage and frequency for Al/PMMA/p-Si structures. The values of interface state density (D _it) were calculated. These values of D _it and series resistance (R _s) were responsible for the non-ideal behavior of I–V and C–V characteristics.     

Determination of residual strains of the EMC in PBGA during manufacturing and IR solder reflow processes

Even though recently published results indicated that residual strains of the epoxy molding compound (EMC) play a key role on the warpage values and shapes of the plastic ball grid array (PBGA) packages, it is still unknown about how these residual strains build up and change during the manufacturing and infrared (IR) solder reflow processes. The purpose of this study is to quantify the residual strains of the EMC in the PBGA packages during the aforementioned processes using a combination of experimental, theoretical and numerical approaches. In the experiments, a full-field shadow moiré is used for measuring their real-time out-of-plane deformation (warpage), during heating and cooling conditions, of two types of the PBGA specimens (without a silicon chip inside) with the same EMC but different substrates (with glass transition temperature T_g = 172 and 202 °C). Furthermore, Timoshenko’s bi-material theory associated with the measured and temperature-dependent elastic moduli and coefficients of thermal expansion for the EMC and substrates is applied for extracting residual strains of the EMC from shadow moiré results. In the analysis, the finite element method cooperating with those determined residual strains is employed to numerically simulate the thermal-induced deformations of the PBGA specimens, in order to verify mechanics. The full-field warpage of the specimens from shadow moiré is documented before and after post-mold curing, solder reflow and during the temperature cycling (from room temperature to 260 °C). The residual strains of the EMC for the specimens with low-T_g and high-T_g substrates after post-mold curing are found to be 0.059% and 0.134%, respectively, which double those before post-mold curing, and further down to 0.035% and 0.08% after the first thermal cycling. After the first cycling, the residual strains remain almost constant during heating and cooling processes. This phenomenon is also observed at lead-free solder reflow processes. Therefore, the residual strains of the EMC induced by the chemical shrinkage of the EMC curing and possibly mold flow pressure are different between the specimens with low-T_g and high-T_g substrates, and these residual strains could change during post-mold curing and the first solder reflow processes.     

Effects of Partial Replacement of Iron with Tungsten on Microstructure,Electrical, Magnetic and Humidity Properties of Copper-Zinc Ferrite Material

This paper presents the analysis of the influence of partial replacement of iron with tungsten on the properties of copper-zinc spinel ferrite material. The samples of Cu_0.5Zn_0.5W _x Fe_2?x O₄ spinel powder ferrites were prepared by using a sol–gel self-combustion technology. The ferrite samples were treated for 30 min at 1000°C. The x-ray diffraction was used in order to establish the differences between the phase compositions of ferrites with different tungsten content. Scanning electron microscopy was used to highlight the influence of the tungsten content on the crystallites. All the samples of Cu_0.5Zn_0.5W _x Fe_2?x O₄ were subject to investigation of the influence of the substitution of tungsten upon their electrical and magnetic properties. The measurements of the electrical properties were performed in different humidity conditions, in order to highlight the effect of moisture conditions on the electrical properties of the material and to analyze the applicability of Cu_0.5Zn_0.5W _x Fe_2?x O₄ ferrite for resistive or capacitive humidity sensors.     

Composition and Band Structure of the Native Oxide Nanolayer on the Ion Beam Treated Surface of the GaAs Wafer

Detailed information on GaAs oxide properties is important for solving the problem of passivating and dielectric layers in the GaAs-based electronics. The elemental and chemical compositions of the native oxide layer grown on the atomically clean surface of an n-GaAs (100) wafer etched by Ar⁺ ions have been studied by synchrotron-based photoelectron spectroscopy. It has been revealed that the oxide layer is essentially enriched in the Ga₂O₃ phase which is known to be a quite good dielectric as compared to As₂O₃. The gallium to arsenic ratio reaches the value as high as [Ga]/[As] = 1.5 in the course of oxidation. The Ga-enrichment occurs supposedly due to diffusion away of As released in preferential oxidation of Ga atoms. A band diagram was constructed for the native oxide nanolayer on the n-GaAs wafer. It has been shown that this natural nanostructure has features of a p–n heterojunction.     

Ion-beam synthesis of InSb nanocrystals in the buried SiO2 layer of a silicon-on-insulator structure

The ion-beam synthesis of InSb nanocrystals in the buried SiO₂ layer of a silicon-on-insulator structure is investigated. The distributions of In and Sb atoms after annealing at a temperature of T _a = 500–1100°C are studied. It is established that the redistribution of implanted atoms is unsteadily dependent on the annealing temperature. The formation of InSb nanocrystals occurs at Ta ≥ 800°C near the Si/SiO₂ interface and at a depth corresponding to the mean paths R _p . Analysis of the profiles of implanted atoms and of the structure and depth distribution of nanocrystals formed allows an inference regarding the two-stage character of formation of the InSb phase. In the initial stage, antimony precipitates are formed; further the precipitates serve as nuclei for indium and antimony to flow to them.     

Effects of Thermal Cycling on Material Properties of Nonconductive Pastes (NCPs) and the Relationship Between Material Properties and Warpage Behavior During Thermal Cycling

In this paper, the effects of thermal cycling on material properties such as coefficient of thermal expansion (CTE), modulus, and glass transition temperature $({rm T}_{rm g})$ of nonconductive pastes (NCPs) for flip chip applications were investigated. Using a thermomechanical analyzer, the dimensional changes of NCPs and an underfill material were measured. The dimensional changes of all materials during the first cycle rapidly increased near ${rm T}_{rm g}$. However, the rapid increase of dimensional change near ${rm T}_{rm g}$ was not observed during the second and third cycles. Furthermore, using a dynamic mechanical analyzer, the modulus and ${rm T}_{rm g}$ were measured. The modulus in the first cycle was smaller than that in the second cycle for all materials. After the first cycle, the modulus curves followed the second cycle curve. Next, the warpage behavior of the flip chip assemblies was observed using the Twyman–Green interferometry method to investigate how material property changes affect warpage behavior during thermal cycling (T/C) and it was found that the warpage of the flip chip assembly decreased after the first cycle. However, after the first cycle, the amount of warpage was constant for the following five cycles. As a result, it was verified that the material properties of NCPs and the underfill material change after the first thermal cycle, and the material property changes are closely related to the warpage hysteresis behavior during T/C. Finally, warpage hysteresis was understood as shear strain.      

Properties of low-temperature (80-300° C) pyrolytic SiO2 on Si and InP

We report on a new low-temperature pyrolytic deposition technology for silicon dioxide. We present data characterizing the electrical and optical properties of this dielectric deposited on Si and InP substrates. The effects of thermal processing are also reported. Deposition of high-quality SiO₂ is achieved by reacting SiH₄ and O₂ at pressures of 2–12 Torr. Reactions occur by pyrolysis only, promoting stoichiometric SiO₂ deposition and good interfacial properties. No plasma- or photo-enhancement is required. Deposition is achieved at temperatures as low as 80° C, the lowest temperature ever reported for pyrolytic SiO₂ deposition. Rates as high as 65 Å/min at 100° C and 100-150 Å/min at 150-300° C are attained. The leakage current densities measured for both Si and InP MIS capacitors (e.g. 10^-9 Å/cm² for 150° C SiO₂) are two to six orders of magnitude lower than values reported for plasma- and photo-enhanced SiO₂ deposited at equivalent temperatures. The high-temperature integrity of this dielectric also makes it a promising annealing cap for group III-V compound semiconductors. Our annealing studies show that SiO₂-capped indium phosphide surfaces remain specular up to 850° C.     

Structural Characterization and Thermoelectric Properties of Hot-Pressed CoSi Nanocomposites

Fabrication of nanocomposites by introduction of SiO₂ metal oxide nanoparticles into a cobalt silicide thermoelectric matrix is studied. The CoSi matrix material was prepared through solid-state synthesis, and the nano-SiO₂ metal oxide was introduced by mechanical grinding. The mixed powders were hot pressed to fabricate nanocomposites. The structural and morphological modifications were studied by powder x-ray diffraction analysis and scanning electron microscopy. The thermoelectric properties of the materials were followed through the Hall effect, Seebeck coefficient, and electrical and thermal conductivities in the temperature range from 300 K to 1000 K.     

Formation of the Thermoelectric Candidate Chromium Silicide by Use of a Pack-Cementation Process

Transition-metal silicides are reported to be good candidates for thermoelectric applications because of their thermal and structural stability, high electrical conductivity, and generation of thermoelectric power at elevated temperatures. Chromium disilicide (CrSi₂) is a narrow-gap semiconductor and a potential p-type thermoelectric material up to 973 K with a band gap of 0.30 eV. In this work, CrSi₂ was formed from Si wafers by use of a two-step, pack-cementation, chemical diffusion method. Several deposition conditions were used to investigate the effect of temperature and donor concentration on the structure of the final products. Scanning electron microscopy and x-ray diffraction analysis were performed for phase identification, and thermal stability was evaluated by means of thermogravimetric measurements. The results showed that after the first step, chromizing, the structure of the products was a mixture of several Cr–Si phases, depending on the donor (Cr) concentration during the deposition process. After the second step, siliconizing, the pure CrSi₂ phase was formed as a result of Si enrichment of the initial Cr–Si phases. It was also revealed that this compound has thermoelectric properties similar to those reported elsewhere. Moreover, it was found to have exceptional chemical stability even at temperatures up to 1273 K.     

Electrical properties of ZnO Nano-particles embedded in polyimide

The ZnO nano-particles were made in the polyimide dielectric matrix by using the chemical reaction between the zinc metal film and polyamic acid. The concentration of the ZnO particle is about 1.5×10¹² cm⁻², with average size below 10 nm, and its shapes are almost spherical. Then, the polyimide layer is a stable dielectric material with a dielectric constant of 2.9. To investigate the electrical properties of ZnO particles in the polyimide insulator film, we fabricated a metal-insulator-semiconductor (MIS) structure and measured capacitance-voltage (C-V) with temperature modulation. At room temperature, C-V hysteresis with a voltage gap of 2.8 V appeared in the MIS structure using SiO₂/Si substrate. As the measuring temperature decreased, the C-V curves were shifted slightly to the accumulation region with gate bias. It was considered that the electrical charging may occur dominantly in nanoparticles, having only a few defects at the interface of the polyimide/SiO₂ and the polyimide/ZnO.     

Effect of conditions of deposition and annealing of indium oxide films doped with fluorine (IFO) on the photovoltaic properties of the IFO/<Emphasis Type="Italic">p</Emphasis>-Si heterojunction

In₂O₃:F (IFO) films were deposited onto crystalline silicon and glass by the pyrosol method. The effect of temperature and oxygen in the course of deposition and also of subsequent annealings in various media on the photovoltaic properties of the IFO/Si structure was measured. It is found that IFO forms a rectifying contact to p-Si, makes it possible to obtain a high photovoltage U _p = 586 mV and an internal quantum yield higher than 97% for the IFO/(pp ⁺)Si structure, and features a low (0.3–0.4 Ω cm) resistivity. An increase in U _p is stimulated by an increase in the temperature of the IFO deposition, a low content of oxygen in the carrier gas, and annealing in argon with methanol vapors. It is concluded that oxygen profoundly affects the surface of the IFO grains and that the transition layer greatly affects the photovoltaic properties of the IFO/(pp ⁺)Si structures.     

Warpage simulation for the reconstituted wafer used in fan-out wafer level packaging

Fan-out packaging technology involves processing redistribution interconnects on reconstituted wafer, which takes the form of an array of silicon dies embedded in epoxy molding compound (EMC). Yields of the redistribution interconnect processes are significantly affected by the warpage of the reconstituted wafer. The warpage can be attributed to the crosslinking reaction and viscoelastic relaxation of the EMC, and to the thermal expansion mismatch between dissimilar materials during the reconstitution thermal processes. In this study, the coupled chemical-thermomechanical deformation mechanism of a commercial EMC was characterized and incorporated in a finite element model for considering the warpage evolution during the reconstitution thermal processes. Results of the analyses indicate that the warpage is strongly influenced by the volume percentage of Si in the reconstituted wafer and the viscoelastic relaxation of the EMC. On the other hand, contribution from the chemical shrinkage of the commercial EMC on warpage is insignificant. As such, evaluations based on the comprehensive chemical-thermomechanical model considering the full process history can be approximated by the estimations from a simplified viscoelastic warpage model considering only the thermal excursion.     

Monitoring extent of curing and thermal–mechanical property study of printed circuit board substrates

Precise control of curing conversion for epoxy-based printed circuit board (PCB) substrates and clarification of curing–property relationship are critical for the performance and reliability assessment, and for the design optimization of electronic systems. In this article, various epoxy composites for PCB substrates were analyzed by infrared spectroscopy (IR), differential scanning calorimetry (DSC), rheometry, dynamic mechanical analysis (DMA), and scanning electron microscope (SEM). Compared with mid-IR and DSC, near-IR (NIR) is found to be a reliable method for the characterization of curing conversion process by detecting the consumption of epoxy groups. And DMA is a powerful method for measuring the conversion of PCB materials by testing glass transition temperatures (T_g) and viscoelastic properties. The curing behaviors of a variety of epoxy composites show distinct differences in both curing rate and activation energy, and the growth tendency of T_g with curing conversion also changed depending on the material compositions. Correlation of curing conversion versus thermal properties shows that the activation energy of curing at different stage by DSC resembles the tendency of T_g transitions tested by DMA. Mechanical properties of the composites show close relationship with the curing conversions. Peel strength, the indicator of adhesion strength between copper foil and epoxy composites, was tested on all the specimens of different curing conversions, and the results showed a maximum value at curing conversion between ca. 90 and 95%.     

Paradoxical Enhancement of the Power Factor of Polycrystalline Silicon as a Result of the Formation of Nanovoids

Hole-containing silicon has been regarded as a viable candidate thermoelectric material because of its low thermal conductivity. However, because voids are efficient scattering centers not just for phonons but also for charge carriers, achievable power factors (PFs) are normally too low for its most common form, i.e. porous silicon, to be of practical interest. In this communication we report that high PFs can, indeed, be achieved with nanoporous structures obtained from highly doped silicon. High PFs, up to a huge 22 mW K^?2 m^?1 (more than six times higher than values for the bulk material), were observed for heavily boron-doped nanocrystalline silicon films in which nanovoids (NVs) were generated by He⁺ ion implantation. In contrast with single-crystalline silicon in which He⁺ implantation leads to large voids, in polycrystalline films implantation followed by annealing at 1000°C results in homogeneous distribution of NVs with final diameters of approximately 2 nm and densities of the order of 10¹⁹ cm^?3 with average spacing of 10 nm. Study of its morphology revealed silicon nanograins 50 nm in diameter coated with 5-nm precipitates of SiB _x . We recently reported that PFs up to 15 mW K^?2 m^?1 could be achieved for silicon–boron nanocomposites (without NVs) because of a simultaneous increase of electrical conductivity and Seebeck coefficient. In that case, the high Seebeck coefficient was achieved as a result of potential barriers on the grain boundaries, and high electrical conductivity was achieved as a result of extremely high levels of doping. The additional increase in the PF observed in the presence of NVs (which also include SiB _x precipitates) might have several possible explanations; these are currently under investigation. Experimental results are reported which might clarify the reason for this paradoxical effect of NVs on silicon PF.