Stability of Flip-Chip Interconnects Assembled with Al/Ni(V)/Cu-UBM and Eutectic Pb-Sn Solder During Exposure to High-Temperature Storage

The thermal stability of flip-chip solder joints made with trilayer Al/Ni(V)/Cu underbump metalization (UBM) and eutectic Pb-Sn solder connected to substrates with either electroless Ni(P)-immersion gold (ENIG) or Pb-Sn solder on Cu pad (Cu-SOP) surface finish was determined. The ENIG devices degraded more than 50 times faster than the Cu-SOP devices. Microstructural characterization of these joints using scanning and transmission electron microscopy and ion beam microscopy showed that electrical degradation of the ENIG devices was a direct result of the conversion of the as-deposited Ni(V) barrier UBM layer into a porous fine-grained V₃Sn-intermetallic compound (IMC). This conversion was driven by the Au layer in the ENIG surface finish. No such conversion was observed for the devices assembled on Cu-SOP surface finish substrates. A resistance degradation model is proposed. The model captures changes from a combination of phenomena including increased (1) intrinsic resistivity, (2) porosity, and (3) electron scattering at grain boundaries and surfaces. Finally, the results from this study were compared with results found in a number of published electromigration studies. This comparison indicates that degradation during current stressing in the Pb-Sn bump/ENIG system is in part due to current-crowding-induced Joule heating and the thermal gradients that result from localized Joule heating.     

The corrosion behavior of BK-7 glasses for use in non-hermeticelectro-optic devices

Examines the stability and corrosion resistance, of four different commonly used optical glasses, including BK-7 made by three different manufacturers in hot humid ambients. We found that although all three BK-7's are considered industrial equivalents, they have slightly different chemical compositions, and exhibit significantly different corrosion behaviors. The corrosion rates varied by at least a factor of two. This improvement appears to be related to the ZnO content in the glass. We found one borosilicate optical glass with similar optical properties to BK-7 which has at least a factor of 3 times better corrosion resistance than the best BK-7. We propose a six stage corrosion model to explain the corrosion behavior of borosilicate glasses in hot humid environments: 1) adsorption of “bulk” like water at defective sites on the surface of the glass; 2) ion exchange between the alkali ions in the glass and protons in the “bulk” like water 3) growth of the “bulk” like water droplet followed by more ion exchange until the entire surface of the glass is covered with a “bulk” like water film; 4) dissolution of the silicate network by the basic “bulk” like water film and formation ionic SiO^$species in the “bulk” like water; 5) reaction between the ionic SiO^$ species and the hydrolized alkali in solution; 6) precipitation and growth of the alkali silicate reaction product. Finally, we determine the acceleration factor for device failure due to corrosion of one type of BK-7 glass. The temperature (T) dependence of failure is assumed to take an exponential form with an activation energy between 0.64-0.81 eV. The relative humidity (RH) dependence of failure is estimated to be of the form Rate ∝ exp {constant(RH²)}. The fitting constant is estimated to fall between 4.8×10^-4 /%² and 6.0×10^-4/%²     

Temperature-humidity-bias behavior and acceleration factors fornonhermetic uncooled InP-based lasers

The stability of uncooled InP-based laser diodes in humid ambients was studied. Nonhermetic devices were aged at two different temperatures and humidities at a constant current and at one temperature and humidity at six different drive currents. For all nonhermetic devices failure occurred as a result of a large increase in the threshold current. The reverse leakage current for the failures did not increase when the threshold current increased, indicating that the change in threshold was a result of a change in reflectivity of one or both facets. The hermetic control group of devices aged under many of the same conditions showed a gradual increase in both the threshold current and slope efficiency. The median lifetimes as determined by assuming a device was a failure when the threshold current increased by 50% was strongly dependent upon humidity temperature and drive current. The lifetime data was fit to and equation of the form lifetime exp(-E_α/kT) exp(-B_RH [RH²]). The values of E_α and B_RH were 0.52 eV and 4.9×10^-4/%², respectively. The current drive data was fit to and expression of the form lifetime a exp(I^αI_op) where I_α as 0.09 h/mA. The lifetime dependence on current drive was modeled by assuming that the drive current caused a local temperature rise through thermal resistance. This local temperature rise then caused a decrease in the local humidity at the diode surface through an expression of the form %RH_diodeα exp (-5990[1/(T_r+T_ambient)-1/T_ambient ])where where T_r is the local temperature rise due to thermal impedance. Finally, we present our preliminary results on the reliability of nonhermetic SiO_x passivated lasers. These results indicate that such lasers can be made with sufficient reliability for use in telecommunications application     

Temperature-humidity-bias-behavior and acceleration model for InPplanar PIN photodiodes

The reliability of InP planar PIN photodiodes in humid ambients has been studied. The dependence of the degradation rate in temperature, humidity and bias was determined by aging devices at temperatures between 50 and 150°C, humidities between 50 and 85% RH, and biases between 2 and 25 V. Failure occurred as a result of a sudden increased dark current. This increase in dark current had little affect on the device responsivity. At all aging conditions the failure distributions were well represented by a lognormal distribution in time with a dispersion of less than 0.3. From these data, an acceleration factor (AF) was developed and is given by AF=exp(E/kT)exp[A(RH)²]exp[B(V)] where E=-0.42 eV, A=-4.6×10^-4, and B=-6.7×10^-2/V. Extensive failure mode analysis was done on more then 500 failed devices. Based on this, a failure mechanism was proposed. The model requires ingress of moisture to the InP surface. The moisture is reduced at the p-contact region of the device region, giving off hydrogen. Under negative bias In and P react with the hydrogen to form gaseous IH and PH ₃. This lead to the semiconductor erosion. The erosion continues until device failure occurs. The worst case concentrate of PH ₃ which if created would produce less than 0.1 ppb PH₃ per cm³ of air. Thus it is not a fire or health hazard. Finally, we present reliability prediction for nonhermetic PIN's encapsulated in an optically clear silicone based polymer. We estimate a 20 year hazard rate of less than 100 FIT's for devices operating at an ambient of 45°C/50% RH. For comparison, the hermetic counterparts of similar design have a 20 year hazard rate of less than 10 FIT's     

Tin Whisker Test Development—Temperature and Humidity Effects Part II: Acceleration Model Development

The incubation time for both whisker growth and corrosion in thin Sn platings (3–10 $mu{hbox {m}}$ thick) on Cu-based alloys have been found to be well represented by an exponential function of humidity and an Arrhenius function of temperature for both as-deposited and reflowed tin platings. Furthermore, whisker growth was found to follow the same functionality in both corroded and non-corroded regions of the plating. The effective activation energies and humidity coefficients were found to depend upon plating thickness, exposure to reflow, and presence of corrosion. The effective activation energies ranged from 0.23 eV to 0.41 eV and the humidity coefficients ranged from $-$0.012% to $-$0.031%. Corrosion enhanced whisker growth occurred by lowering the effective activation energy for whisker growth. A theory based on excess, non-creep relaxed, oxidation induced strain was developed to explain the corrosion induced energy barrier lowering. The data showed that 60 $^{circ}{hbox {C}}/{hbox {87}}%{hbox {RH}}$ appears to be the optimal high temperature/high humidity test condition at this time for Sn over Cu substrates. Within the limits of the whisker and corrosion (incubation) acceleration functions developed in this study, it is concluded that the JEDEC tests can be used to indicate behavior at other temperature/humidity points that could be relevant storage or service conditions.      

High-reliability 1.3-μm InP-based uncooled lasers in nonhermeticpackages

We report the first uncooled nonhermetic 1.3-μm InP-based communication lasers that have reliability comparable to their hermetically packaged counterparts for possible applications in fiber in the loop and cable TV. The development of reliable nonhermetic semiconductor lasers would not only lead to the elimination of the costs specifically associated with hermetic packaging but also lead the way for possible revolutionary low-cost optoelectronic packaging technologies. We have used Fabry-Perot capped mesa buried-heterostructure (CMBH) uncooled lasers with both bulk and MQW active regions grown on n-type InP substrates by VPE and MOCVD. We find that the proper dielectric facet passivation is the key to obtain high reliability in a nonhermetic environment. The passivation protects the laser from the ambient and maintains the proper facet reflectivity to achieve desired laser characteristics. The SiO facet passivation formed by molecular beam deposition (MBD) has resulted in lasers with lifetimes well in excess of the reliability goal of 3,000 hours of operation at 85°C/90% RH/30 mA aging condition. Based on extrapolations derived experimentally, we calculate a 15-year-average device hazard rate of <300 FITs (as against the desired 1,500 FITs) for the combination of thermal-and humidity-induced degradation at an ambient condition of 45°C/50% RH. For comparison, the average hazard rate at 45°C and 15 years of service is approximately 250 FITs for hermetic lasers of similar construction. A comparison of the thermal-only degradation (hermetic) to the thermal plus humidity-induced degradation (nonhermetic) indicates that the reliability of these nonhermetic lasers is controlled by thermal degradation only and not by moisture-induced degradation. In addition to device passivation for a nonhermetic environment, MBD-SiO maintains the optical, electrical, and mechanical properties needed for high-performance laser systems     

Failure mechanism of avalanche photodiodes in the presence of watervapor

Lower cost optoelectronic modules (transmitters and receivers) could be produced if components could operate reliably in a nonhermetic package. Experiments (operation in humid ambient as a function of temperature, relative humidity (RH), and voltage) were carried out on InP-based avalanche photodiodes (APDs). The objectives were to reveal any moisture-related failure modes, to understand the failure mechanisms, to determine the relevant acceleration factors, and to propose design or material modifications to prevent the failures. Leakage current vs. time was measured for nine cells of 20 devices each. All devices were examined by optical microscopy, and representative samples were studied by scanning electron microscopy. A failure mechanism related to corrosion of the passivating SiN and the underlying semiconductor was identified. Based on the voltage and RH dependence of the degradation rate, a model for the failure process was developed. The mechanism involves RH-dependent transport of charge across the surface of the SiN and Poole-Frenkel current flow through the SiN. This process eventually produces a short circuit     

Sn Corrosion and Its Influence on Whisker Growth

The microstructure and crystal structure of condensation-induced corrosion products, vapor phase induced oxidation products, Cu-Sn intermetallics, and Sn whiskers that formed on electroplated matte Sn on Cu-alloy after exposure 2500 h in a 60 degC/93%RH ambient were characterized with scanning electron microscopy, (SEM), focused ion beam (FIB) microscopy, energy dispersive spectroscopy (EDS), transmission electron microscopy (TEM), and selected area electron diffraction (SAD). The corrosion product was identified as crystalline SnO₂. The oxidation of Sn in condensed water was at least four orders of magnitude larger than that in moist vapor at 60 degC. All Sn whiskers were found to be within 125 mum of the corrosion product. Based on these observations, a theory was developed. The theory assumes that oxidation leads to the displacement of Sn atoms within the film. Because the grain boundaries and free surfaces of the film are pinned, the oxidation-induced excess Sn atoms are constrained within the original volume of the Sn-film. The trapped excess Sn atoms create localized stress, excess strain energy, in the Sn-film. If and when the pinning constraint is relaxed, as for example would occur when the surface oxide on the film cracks, then the Sn atoms can diffuse to lower energy configurations. When this occurs, whisker nucleation and growth begins. The theory was tested by detailed measurements and comparison of the corrosion volume and the whisker volume in two different samples. The volume comparisons were consistent with the theory     

Degradation of InGaAsP-InP lasers in hot-humid ambients:determination of temperature-humidity-bias acceleration factors

Reliable nonhermetic lasers are expected to both reduce the cost of InP-based optoelectronics and lead the way for the development of revolutionary low cost packaging technologies. To date, the functional dependence of the degradation rate of lasers on temperature and humidity in noncondensing ambients is not known, Without knowledge of these dependencies, degradation rates measured in real time, tens to hundreds of hours, at high temperatures, T>60°C, and high relative humidities, RH>60%, cannot be used to estimate the degradation rate which occurs at nominal use conditions, T>50°C, and RH<50%, Therefore, the field reliability of nonhermetic InP-based lasers is not presently known nor can it be calculated. In this paper, we report the first temperature and humidity acceleration factors for InP-based lasers in nonhermetic, noncondensing ambients. These acceleration factors will, for the first time, provide a tool for estimating the reliability of nonhermetic InP-based lasers in the field