Fraunhofer-Institut für Zuverlässigkeit und Mikrointegration IZM

Filters

22
Reset filters

Settings
Now showing 1 - 10 of 22
  • Publication
    Experimental verification and analysis of analytical model of the shape of bond wire antennas
    ( 2017)
    Ndip, I.
    ;
    Huhn, M.
    ;
    Brandenburger, F.
    ;
    Ehrhardt, C.
    ;
    Schneider-Ramelow, M.
    ;
    Reichl, H.
    ;
    Lang, K.D.
    ;
    Henke, H.
    In this Letter, the authors present the experimental verification of an analytical model, which captures the realistic shape of bond wire antennas (BWAs) in dependence on the wire bonding and design parameters. Using the verified model, the impact of the shape of the wires on the performance of BWAs is quantified.
  • Publication
    Impact of process tolerances on the performance of bond wire antennas at RF/microwave frequencies
    ( 2012)
    Ndip, I.
    ;
    Öz, A.
    ;
    Tschoban, C.
    ;
    Schmitz, S.
    ;
    Schneider-Ramelow, M.
    ;
    Guttowski, S.
    ;
    Reichl, H.
    ;
    Lang, K.-D.
    Due to the multitude of advantages bond wire antennas have over conventional planar antennas (especially onchip planar antennas), they have received much research attention within the last four years. The focus of the contributions made so far has been on exploiting different configurations of single-element and array bond wire antennas for short-range applications at RF/microwave frequencies. However, the effects of process tolerances of bond wires on the radiation characteristics of bond wire antennas have not been studied in published literature. Therefore in this paper, we investigate the impact of up to 20% fluctuations in the parameters of bond wires on the performance of 42 GHz and 60 GHz bond wire antennas. Our results reveal that the length and radius of bond wires are the most and least sensitive parameters, respectively. Furthermore, the severity of the impact of process tolerances depends on the impedance bandwidth of the original antenna, before considering the tolerances. For example, a 10% change in the length of a bond wire causes the resonance frequency of a 42 GHz antenna to be shifted out of the specified 3GHz bandwidth (40.5 GHz-43.5 GHz) required for point-to-point communication. However, although a 10% change in length of a bond wire yields a 2.5 GHz shift in the resonance frequency of a 60 GHz bond wire antenna, it doesn't completely detune the antenna because of the original 6 GHz bandwidth available, prior to the fluctuation. Therefore, to prevent the impact of process tolerances from severely degrading the performance bond wire antennas, these antennas should be designed to have larger bandwidths than specified. For experimental verification, a bond wire antenna was designed, fabricated and measured. Very good correlation was obtained between measurement and simulation.
  • Publication
    Impact of process tolerances on the performance of bond wire antennas at RF/microwave frequencies
    ( 2012)
    Ndip, I.
    ;
    Öz, A.
    ;
    Tschoban, C.
    ;
    Schmitz, S.
    ;
    Schneider-Ramelow, M.
    ;
    Guttowski, S.
    ;
    Reichl, H.
    ;
    Lang, K.-D.
    Due to the multitude of advantages bond wire antennas have over conventional planar antennas (especially on-chip planar antennas), they have received much research attention within the last Jour years. The Jocus oj the contributions made so Jar has been on exploiting different configurations oj single- element and array bond wire antennas Jor short-range applications at RF/microwave jrequencies. However, the effects of process tolerances of bond wires on the radiation characteristics of bond wire antennas have not been studied in published literature. Therefore in this paper; we investigate the impact of up to 20% fluctuations in the parameters of bond wires on the performance of 42 GHz and 60 GHz bond wire antennas. Our results reveal that the length and radius of bond wires are the most and least sensitive parameters, respectively. Furthermore, the severity of the impact of process tolerances depends on the impedance bandwidth of the original antenna, before consideri ng the tolerances. For example, a 10% change in the length of a bond wire causes the resonance frequency of a 42 GHz antenna to be shifted out of the specified 3GHz bandwidth (40.5 GHz-43.5 GHz) required for point-to-point communication. However, although a 10% change in length of a bond wire yields a 2.5 GHz shift in the resonance frequency of a 60 GHz bond wire antenna, it doesn't completely detune the antenna because of the original 6 GHz bandwidth available, prior to the fluctuation. Therefore, to prevent the impact of process tolerances from severely degrading the performance bond wire antennas, these antennas should be designed to have larger bandwidths than specified. For experimental verification, a bond wire antenna was designed, jabricated and measured. Very good correlation was obtained between measurement and simulation.
  • Publication
    Interface formation in the US-wedge/wedge-bond process of AlSi1/CuNiAu contacts
    ( 2011)
    Geissler, U.
    ;
    Funck, J.
    ;
    Schneider-Ramelow, M.
    ;
    Engelmann, H.-J.
    ;
    Rooch, I.
    ;
    Müller, W.H.
    ;
    Reichl, H.
    Interface formation between 25-m AlSi1 wire and flash-Au substrate metallization during a bonding time of 50 ms has been investigated. Only a few milliseconds after the ultrasonic power is switched on, intermetallic phase growth starts, continuing until the end of the wire-bonding process. During the entire bonding time, the fraction of the interface covered with Au 8Al3 increases, and at the end of the bonding time, the interface is nearly completely covered with that phase. Finite-element modeling (FEM) of the temperature in the interface region indicates maximum temperatures well below 100°C, thus making solely thermally activated intermetallic phase growth impossible. However, it is demonstrated that the phase growth observed during the ultrasonic wire-bonding process could result from an accelerated diffusion process caused by a higher vacancy concentration. The accelerated diffusion process would have an activation energy Q of 0.36 eV.
  • Publication
    Modeling and optimization of bond wires as transmission lines and integrated antennas at RF/microwave frequencies
    ( 2011)
    Ndip, I.
    ;
    Tschoban, C.
    ;
    Schmitz, S.
    ;
    Ostmann, A.
    ;
    Schneider-Ramelow, M.
    ;
    Guttowski, S.
    ;
    Reichl, H.
    ;
    Lang, K.-D.
    In this contribution, the authors present a systematic approach for optimizing the RF performance of bond wires. First of all, a comparative analysis between two of the most commonly used bond wire signal configurations, the two conductor and coplanar configurations, is done. Their results reveal that although the partial self-inductance of the signal wires is the same in both configurations, the partial mutual inductance of the coplanar configuration is higher, resulting in a smaller loop inductance. Consequently, the return and insertion losses are smaller. By reducing the distance between the signal and return currents, they further reduced the loop inductance, and significantly optimized the coplanar configuration. For example, considering a 1 mm long bond wire with a diameter of 25 micron, they successfully kept the power lost through the coplanar configuration below 10 % at 15 GHz, in comparison to the 70 % power lost through the two-conductor configuration at the same frequency. However, more than 30 % of the entire power is lost through the optimized coplanar configuration at 40 GHz. At such frequencies where bond wires are unsuitable to be used as transmission lines, they demonstrate that they are very efficient as antennas by designing a half-loop integrated bond wire antenna having a bandwidth of 3 GHz. For experimental verification, test samples were designed, fabricated and measured. An excellent correlation was obtained between simulation and measurement.
  • Publication
    Modeling and analysis of coplanar bonding wires for high-speed applications
    ( 2010)
    Tschoban, C.
    ;
    Ndip, I.
    ;
    Schmidt, S.
    ;
    Guttowski, S.
    ;
    Schneider-Ramelow, M.
    ;
    Lang, K.-D.
    ;
    Reichl, H.
    In this contribution, two-conductor and coplanar bond wire configurations are modeled, analyzed and compared. The impact of the pitch of the bond wires and the distance of separation between the signal wire and a nearby reference plane, on RF performance are quantified for GSM 900/1800, 2.4 GHz/5GHz WLAN and high-speed applications above 10 GHz. Considering a bond wire diameter of 50 m, length of 1mm and pitch of 100 m, it is found that almost 3 times more power is lost through the two-conductor model than coplanar configuration at 10 GHz. This makes the coplanar configuration more suited for high-speed applications.
  • Publication
    The ultrasonic wedge/wedge bonding process investigated using in situ real-time amplitudes from laser vibrometer and integrated force sensor
    ( 2010)
    Gaul, H.
    ;
    Shah, A.
    ;
    Mayer, M.
    ;
    Zhou, Y.
    ;
    Schneider-Ramelow, M.
    ;
    Reichl, H.
    The ultrasonic transversal force transmitted to a chip during ultrasonic bonding is derived from measurements of the vibration amplitude at the tool tip and the die edge. To proof the derivation, the transversal force is measured as well by means of a microsensor, which is sensitive to the stress field in the silicon die. The force measured by the microsensor is further referred to as "y-force" To Al-metalized test pads with the integrated microsensors, AlSi1 wire of 25 mu m diameter was bonded using a wedge/wedge auto-bonder. Measurements of the vibration amplitudes and the y-force during bonding were conducted for nine different bonding parameter settings of force and ultrasound (us) amplitude. They confirm a theory for the friction cleaning phase as it was described earlier and will be partially presented here. Compared to earlier measurements of Au-ball-bonds, the results largely show the same behavior and imply that us wedge bonding and thermosonic ball bonding are similar processes. Furthermore, the data approves former interpretations of the bonding process starting with a stiction phase. A clear break off point was found in all pad amplitude measurements, which is followed by a friction plateau that implicates the need of a minimum friction cleaning power. The discussion made in this paper is interesting fora bond process control system. The transversal force reflects the important stages of the bond process and contains the information to Suit as a control signal. But it is impractical to measure the transversal force in situ under the wedge in industrial production, where chip, Substrate and bonding table create a complex setup with a high geometric variety. An indirect measurement of the transversal force via the tool tip amplitude opens up new possibilities for gaining an efficient control variable, because the geometry and the properties of the bonding machine are well defined. As a first step it is shown by correlating vibration measurements with microsensor signals, that the tool tip amplitude measured by laser vibrometer contains all of the necessary information needed to control the bond process. From that point, process integrable measurement systems - which are cheaper, more handy and more fail safe than the laser vibrometer - might be developed.
  • Publication
    Analytic model verification of the interfacial friction power in Al us w/w bonding on Au pads
    ( 2010)
    Gaul, H.
    ;
    Schneider-Ramelow, M.
    ;
    Reichl, H.
    To close the gap between the development of wire bonding equipment and the knowledge about the welding process itself, a friction power model is presented. The model can be used to calculate the welding quality of a wedge in dependence to the bonding and material parameters. Based on theories for Au ball/wedge bonding, the model is enhanced for us wedge/wedge (us w/w) bonding with aluminum wire. Therefore, the deformation of the wire is described by the von Mises stress, and the geometry changes due to the decreasing wedge height during bonding are taken into consideration. Furthermore, the paper introduces the minimum friction amplitude needed for any frictional cleaning of the interface and takes the precleaning during touchdown into consideration. To prove the model, four equations are highlighted to theoretically predict the shear force, the tool tip and pad amplitude, and the characteristic of the deformation during the us stage of w/w bonding. These magnitudes are measured for different bonding parameters while bonding 25 ?m AlSi-1 wire to Au metalized Si test structures. The model parameters were then fit to the experimental results for all bonding parameters.
  • Publication
    Hardening and Softening in AlSi1 Bond Contacts During Ultrasonic Wire Bonding
    ( 2009)
    Geißler, U.
    ;
    Schneider-Ramelow, M.
    ;
    Reichl, H.
    Ultrasonic wedge-wedge bonding of 25- mum AlSi1 wires is characterized as a dynamic process of hardening and softening. In the first phase of wire bonding, the wedge is predeformed and cold worked by the bondforce. After ultrasonic energy has been switched on, recrystallization starts at the interface. During the bonding time hardening and softening processes alternate and a maximum in hardness is measured. Hardening and softening processes correlate well with the grain structure, the measured grain sizes and a typical plateau in the z-deformation curve of the contact. At the end of the wire bonding process the wedge is recrystallized and softer than the predeformated wedge, but harder than the as-received wire.
  • Publication
    Ultrasonic friction power during Al wire wedge-wedge bonding
    ( 2009)
    Shah, A.
    ;
    Gaul, H.
    ;
    Schneider-Ramelow, M.
    ;
    Reichl, H.
    ;
    Mayer, M.
    ;
    Zhou, Y.
    Al wire bonding, also called ultrasonic wedge-wedge bonding, is a microwelding process used extensively in the microelectronics industry for interconnections to integrated circuits. The bonding wire used is a 25 mu m diameter AlSi1 wire. A friction power model is used to derive the ultrasonic friction power during Al wire bonding. Auxiliary measurements include the current delivered to the ultrasonic transducer, the vibration amplitude of the bonding tool tip in free air, and the ultrasonic force acting on the bonding pad during the bond process. The ultrasonic force measurement is like a signature of the bond as it allows for a detailed insight into mechanisms during various phases of the process. It is measured using piezoresistive force microsensors integrated close to the Al bonding pad (Al-Al process) on a custom made test chip. A clear break-off in the force signal is observed, which is followed by a relatively constant force for a short duration. A large second harmonic content is observed, describing a nonsymmetric deviation of the signal wave form from the sinusoidal shape. This deviation might be due to the reduced geometrical symmetry of the wedge tool. For bonds made with typical process parameters, several characteristic values used in the friction power model are determined. The ultrasonic compliance of the bonding system is 2.66 mu m/N. A typical maximum value of the relative interfacial amplitude of ultrasonic friction is at least 222 nm. The maximum interfacial friction power is at least 11.5 mW, which is only about 4.8% of the total electrical power delivered to the ultrasonic generator.