Open Access
Add to BibSonomy
Abstract
Laser drop on demand jetting of Cu-base braze droplets was proven a suitable method for joining wires to electrode structures of electronic devices, particularly if the electrical contacts need to withstand high thermal loads. During joining, a braze preform of 600 µm diameter is placed inside a capillary, molten by a laser pulse and subsequently ejected from the capillary by inert gas overpressure similarly to conventional solder ball bumping processes. However, since the liquidus temperature of the used braze material of 990 °C is about 760 °C higher than of standard Sn-based solders used in electronics packaging, the system technology was modified significantly to enable jetting of CuSn alloys. In particular, the beam source emits a five times higher optical output power than standard machines designed for processing Sn-based solders. In addition, a modified capillary made from technical ceramic was machined, to withstand the significantly higher heating- and cooling rates during the process. In order to understand the influence of capillary geometry on droplet detachment, and flight trajectory, two capillary geometries were machined applying a picosecond laser ablation process. Subsequently, stereoscopic high speed videos of droplet detachment and flight phase were analyzed. Using this approach it is possible, to determine droplet flight trajectories, velocities and lateral positional deviations in dependency of relative inert gas overpressure inside the machining head, pulse power and capillary geometry. The findings indicate a significant influence of the capillary geometry and the applied overpressure on the droplet flight trajectory, whereas the role of the laser pulse power seems neglectable.
© 2018 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Particularly the automotive industry is recently facing increasingly stringent laws regarding environmental protection. One approach to decrease energy consumption and thus to increase the range of vehicles, is a reduction of component weight, which can be achieved e. g. by lightweight construction. A promising method to achieve this goal is a substitution of discrete parts by smart structural components, which are able to merge different functionalities in one structure. These composite assemblies with actuator functionality recently became focus of considerable interest of different engineering fields [1–3]. One approach to achieve active functionalities of structural components is to integrate piezo actuators e. g. into structural car or airplane body components by aluminum die casting. Such structures can provide functionalities like sensor applications, structural health monitoring or active vibration damping [4]. During Al-die casting, the modules and particularly the contacts between the electrode of the piezo actuator and the wire need to withstand significant thermal loads of > 600 °C [5,6] if standard Al-casting alloys are used. In order to obtain electrical contacts with high thermal stability, a novel laser assisted wire bonding process has been developed, and is referred to as laser drop on demand joining. It is based on a process known as solder ball bumping, which is commercially available and was investigated in [7,8] and [9]. However due to the low liquidus temperatures of tin- based solders of about 230 °C, the process needed to be significantly altered, in order to fulfill the requirements regarding thermal stability posed by the Al-die casting. To achieve the thermal stability required by the Al-die casting process, CuSn12 braze preforms with a liquidus temperature of 990 °C and a solidus temperature of 825 °C [10] are used. Using braze instead of solder, the joints are capable to endure the temperatures occurring during Al-die casting as shown recently in [11], without altering the piezoelectric properties of the piezo actuator, due to the localized thermal input, which is restricted to the interface between wire and electrode structure. Other dominant joining/contacting processes used in electronics packaging such as soldering, or the application of conductive adhesives cannot be applied, due to the low thermal stability of the joints, which do not withstand subsequent Al-casting. Micro welding and ultrasonic wire bonding on the other hand would result in damaging the 20 µm Ag-electrode structure of the piezo module, respectively the ceramic substrate, in which the piezo ceramic is embedded, resulting in a destruction of the piezo module. Also, various piezoelectric printing heads are described in literature to apply conductive droplets onto substrates, however, due to the loss of the piezoelectric effect when the Curie temperature of the piezo actuator is exceeded, such devices are currently not used to generate braze droplets with liquidus temperatures of about 1000 °C. The main application of piezo based print heads is the application of solder or dispensing conductive inks [12,13] or Gallium [14], all of which do not show sufficiently high thermal stability for a subsequent casting process.
The novel laser drop on demand joining process described in this work was developed to circumvent the deficiencies described in the section above and can be divided into four phases: first, a braze preform with a liquidus temperature of about 1000 °C is inserted into a capillary; subsequently the preform is irradiated by a laser pulse, which thoroughly melts the preform resulting in its ejection from the capillary by nitrogen overpressure. After a flight phase, the molten droplet wets electrode structure and Cu-wire, resulting in a firm joint (Fig. 1).
Fig. 1 Schematic of laser drop on demand joining process with coordinate system convention; x-axis is perpendicular to the drawing plane [15].
Since computational fluid dynamics (CFD) simulations carried out in [16] indicate a significant impact of the capillary geometry on the gas flow velocities, modifying the capillary geometry seems to be a suitable approach, to enhance the droplets positioning accuracy and the process window. In order to be able to review this hypothesis, two different capillary geometries were machined by means of laser ablation, one with a conic shape and another with a diffuser type geometry. Subsequently the droplet detachment from both capillaries was investigated by stereoscopic high-speed imaging, enabling a three-dimensional tracing of the droplet flight trajectory and thus a quantitative comparison of the capillary geometries impact on the process result.
2. Process, system and methods
2.1 Experimental setup
The beam source used for the experiments is a IPG YLR-200-SM-WC Ytterbium fiber laser with a wavelength of 1070 nm and a maximum optical output power of 200 W, which is about five times higher than standard solder ball bumping systems commercially available and used in electronics packaging. This modification is necessary, due to the significantly higher melting temperature of the CuSn-braze material in comparison with tin based solder, since the liquidus temperature of the braze alloy exceeds the liquidus temperature of standard tin based solders by about 800 °C, resulting in pulse powers required to be about four times higher than for tin based solder material. In addition, the reflectivity of Cu-alloys is generally higher than the reflectivity of Sn [17], which must also be ruled out by scaling up the optical output power of the beam source. Additionally, since the tungsten carbide capillaries used in the standard solder ball bumping system are prone to wetting with Cu-based braze due to the formation of intermetallic phases [18], novel capillaries were machined by laser ablation, which consist of zirconia toughened alumina (ZTA), which show poor wetting with the used Cu-based braze material. Further, the higher optical output power of the system requires a process control, capable of interrupting the laser pulse after droplet detachment, to avoid electrode/substrate perforations. Therefore a photodiode is utilized to determine the precise moment of braze detachment by detecting scattered laser radiation. The photodiode signal is used to trigger the interruption of the laser pulse. Since the interruption of the laser pulse needs to be carried out within less than 2 ms to avoid substrate perforation, a real time operation system based on a NI cRIO-9066 embedded controller was implemented, which enables the interruption of the laser pulse within about 20 µs. The process control is described in more detail in [15]. The setup used in the scope of this paper thus represents a new development and is not commercially available. It is shown schematically in Fig. 2.
During and after melting of the braze preform, the surface tension of the braze melt, exceeds gravity force, which prevents droplet detachment, if no additional force is applied [19]. Therefore nitrogen overpressure is applied, in order to enforce detachment of the braze droplet from the capillary. This is ensured by attaching the machining head to a pressure regulator, which enables the application of a defined gas overpressure to the machining head. The applied overpressure is monitored in situ by a pressure sensor. The coaxial CCD camera enables an alignment of the machining head to the joint position and further can be used to detect capillary failure. The CuSn12 braze preforms were custom made by the company Ecka Granules, to particularly suit the demands of the laser drop on demand joining process. The braze spheres were produced by atomization in argon atmosphere to avoid oxidation. A subsequent sieving step was applied to obtain preform batches with the desired diameter of 600 µm.
2.2 Capillary material and geometry
In order to understand the influences of the capillary geometry on the gas flow velocity fields at the orifice and in the joining area below the capillary, computational fluid dynamics (CFD) simulations were carried out in [16], which indicated a significant influence of the capillary geometry on the gas flow velocity. First, several different geometries have been simulated in order to determine the geometry with the highest reduction of the gas flow velocity. Figure 3 shows a standard conic capillary in comparison with a diffuser type capillary, yielding the highest reduction in gas flow velocity of all investigated geometries. The droplet position, where it impinges and wets the electrode structure, is indicated by the dashed white line.
Fig. 3 CFD-Simulations of stationary gas velocity fields for conic (geometry 1, left) and diffuser type capillary shapes (geometry 2, right) after applying an overpressure of 125 mbar; cross sectional view [16]. The droplets final position and geometry is indicated by the dashed line.
In order to quantify the impact of the capillary geometry on the droplet flight trajectory, which was suggested by the simulation, both geometries were machined from CeramTec 950 Zirconia toughened Alumina (ZTA), Al2O3/ZrO2, a by laser ablation, applying picosecond laser pulses. To generate the desired geometries, an ablation strategy described in [20] was used, which enables the generation of cavities of defined dimensions. Two geometries were generated, geometry 1 represents a standard conic capillary, geometry 2 a diffuser type capillary. The conic capillary has an lower diameter of 580 µm and an upper diameter 780 µm, which results in a convex shape, to hold the spherical braze preform with a diameter of 600 µm in place during the melting process. The diffuser type capillary has a conic upper half, consisting of a conic shaped cavity with an upper diameter of 780 μm and a waist diameter of 580 μm in the center of the ceramic plate. The bottom of the diffuser type orifice exhibits a radius of curvature of 1.12 mm, to broaden the orifice and thus to obtain a diffusing effect on the passing process gas. The diffuser type geometry was machined in two steps. First the upper part of the cavity was generated, subsequently the capillary was rotated 180° followed by another ablation process with modified parameters to obtain the desired geometry. This procedure was necessary, since the diffuser type capillary represents an undercut for the laser ablation process. ZTA was chosen as capillary material, due to its advantages in comparison with other technical ceramics such as its high thermal stability, low heat conductivity and poor wetting behavior with the braze melt [16,21]. The surface roughness inside the cavity of the capillary after laser ablation was measured, using a laser scanning microscope. The values for Rz respectively Ra were found to be 0.89 ± 0.16 µm (n = 5), respectively 0.19 ± 0.03 µm (n = 5) as stated in [16].
In order to understand, how capillary geometry, gas overpressure and pulse power influence the droplet flight trajectory and velocity, investigations were carried out in the scope of this work, utilizing stereoscopic high speed imaging and image processing, to enable three-dimensional tracing of the droplet during its flight phase to quantify the influence of the capillary geometry on the droplet flight trajectory.
2.3 High-speed camera system
For tracing the droplet flight trajectory, two Vision Research Phantom V1210 high-speed cameras were used, which were aligned at a 31.5 degree angle, to form a convergent stereo camera system (see section 2.4, Fig. 4). The cameras complementary metal-oxide-semiconductor (CMOS) sensor has a pixel size of 28 μm and a bit depth of 12 bits. The dimension of the sensor is 35.8 mm x 22.4 mm with a maximum resolution of 1280 x 800 pixels, at which an image rate of 12 000 fps can be obtained. The maximum framerate of 103 400 fps can be obtained at a resolution of 256 x 256 pixels or below [22]. In the experiments, a framerate of 30 000 Hz at a resolution of 512 x 512 pixel was chosen, which represents a comparably large field of view while maintaining a high recording frequency and thus is a good balance of required spacial- and temporal resolution. Aforementioned parameters were kept constant for all experiments. In order to protect the CMOS sensor from scattered laser radiation and overexposure, two edge pass filters were mounted in front of the objectives, which are transparent sole at a narrow wavelength band of the pulsed illumination laser with a wavelength of 807 nm. The frequency of the illumination laser is matched to the recording frequency of the cameras of 30 000 Hz.
Fig. 4 Detected edges of checkerboard pattern used for calibration (Left) and visualization of the extrinsic parameters of the used convergent stereo camera system with the respective detected calibration pattern planes drawn to scale (A) and magnified (B). The calibration pattern planes are drawn to scale and thus appear small in the visualization.
2.4 Tracking of droplet flight trajectory by stereoscopic high-speed imaging
This chapter explains the fundamentals of the applied stereoscopic image reconstruction procedure in a condensed manner. To obtain 3D information of the droplet flight trajectories, the following steps are necessary to be carried out:
- • Camera alignment and synchronization
- • Recording the calibration pattern for different angles and distances
- • Calibration by edge detection which is divided into
- • Rectification of the recorded frames
To avoid exceeding the scope of this work, the authors refer to [23] and [24] for a profound description of the underlying principles of stereoscopic image reconstruction.
After the alignment of the cameras, single- and stereo calibration is carried out by applying the method proposed by [25] and [26], using the software environment Matlab. In order to obtain high-quality calibration images and thus accurate results, a precisely manufactured calibration pattern is necessary, since the process observation is carried out in microscopic scale. Therefore, a 6 × 7 aluminum oxide checkerboard pattern with 2 mm edge length of each square and an accuracy of ± 0.001 mm was used (see Fig. 4 left). Sintered aluminum oxide was chosen as substrate material of the calibration pattern, since it suppresses reflections of the illumination laser better than e.g. glass calibration patterns.
During recording of the calibration images, the checkerboard pattern is positioned at capillary location. It is moved laterally, translationally and being rotated to generate a large number of calibration images. Image 4 shows the visualization of the stereoscopic camera system and the respective optical axis of both cameras indicated by the dashed lines. Camera 1 represents the coordinate origin, the sizes of the camera symbols were scaled. Further, in Fig. 4, 20 exemplary calibration pattern planes are shown in unscaled (A) and scaled manner (B) for better visibility.
It is of particular importance to cover the entire image plane by moving the calibration pattern, to be able to derive a precise lens distortion model of the optical system composed by camera and objective, which is necessary for subsequent rectification of the high-speed recordings. Since the experiments were carried out in different series, before each series of experiments, a new calibration was carried out, since marginal misalignments of the cameras can result in large deviations of the results. After calibration and determination of camera and lens distortion, rectification, triangulation, derivation of a disparity map and 3D-reconstruction can be carried out.
2.5 Stereoscopic 3D-reconstruction of droplet flight trajectory
In order to reconstruct the flight trajectory of the braze droplet, the spatial position in x-, y- and z-direction of the droplet must be determined. To achieve this, the images are binarized and rectified. Subsequently, the center points of the braze droplets are identified sequentially for each frame respectively time step. The binarization is carried out by applying a threshold value for the left and right image. Subsequently, the centroid of the detected region is calculated, to be able to describe the droplet as one discrete coordinate. The respective 2D-coordinates of the droplet are subsequently stored in a matrix, from which the 3D-coordinates of the braze droplet can be derived by comparing the 2D-coordinates of the droplets in each frame pair with the disparity map obtained during stereo camera calibration. After this step, the 3D-coordinates of the centers of the braze droplet in the camera coordinate system are known for all time steps. In order to simplify and unify the evaluation of the flight trajectory of the droplet, the camera coordinate system is shifted by a coordinate transformation in such a way that its origin corresponds to the orifice location of the capillary. In Fig. 5, an exemplary droplet flight trajectory is shown as a projection in x-y-plane (left) and in pseudo 3D (right). The color bar indicates the calculated transient velocities for two consecutive time steps respectively frame pairs.
Fig. 5 Exemplary droplet flight trajectory is shown as a projection in x-y-plane (left) and in pseudo 3D (right). PLaser = 160 W; tPulse = 40 ms; PGas = 20 mbar, capillary geometry: diffuser type; mean droplet velocity: 1.346 m/s
From the 3D-coordinates of the braze droplet during the flight phase, the time resolved and mean droplet velocity, as well as the lateral positional deviation of the droplet position relative to the capillary orifice can be derived as a function of capillary geometry, machining head overpressure and pulse power. For instance, Fig. 5. indicates the droplet being accelerated after detachment, which is to be attributed to the impulse of the gas molecules emanating from the capillary orifice and therefore accelerating the droplet. This effect is superimposed by the droplets gravitational acceleration, resulting in an steadily increasing droplet velocity during flight phase.
3. Experimental design
The scope of this work is to describe both, droplet velocity and the lateral deviation of the droplet from the center point of the capillary cavity, in dependence of the laser pulse power, the gas overpressure and the capillary geometry. The parameters were measured for PPulse from 60 to 200 W, which represents the maximum optical output power of the beam source with increments of 10 W, whereas the overpressure was varied from 20 to 110 mbar with increments of 10 mbar. First, the complete parameter window was established for both capillary geometries. Subsequently, an additional set of experiments within the established parameter window was carried out, in order to give information about the statistical significance of the findings. The results of the investigations are discussed in the following sections.
3.1 Lateral position deviation
The parameter window was established by gradually increasing both, pulse power and overpressure, until droplet atomization occurred the first time. Figure 6 shows the effect of droplet atomization (right) in comparison with the formation of a discrete droplet within the valid parameter window (left). At higher pressures, the gas impulse on the braze melt results in its atomization if it exceeds the surface tension of the droplet. This effect determines the right boundary of the parameter window shown in Figs. 7-10.
Fig. 6 Freeze frame of two videos: Left: Discrete droplet after detachment from the machining head (Conic capillary, PPulse = 130W; Overpressure = 70 mbar); Right: droplet atomization by exceedingly high overpressure (Conic capillary, PPulse = 130W; Overpressure: 110 mbar)
Fig. 9 Results of the stereoscopic evaluation of the average droplet velocity for the conic capillary shape.
Fig. 10 Results of the stereoscopic evaluation of the average droplet velocity for the diffuser shaped capillary.
The other boundary of the parameter window is represented by parameters, at which no droplet detachment is observed due to a lack of pulse power to thoroughly melt the preform or insufficient overpressure to ensure droplet detachment from the capillary (right and bottom in Figs. 7-10). For the conic capillary, subsequently, 123 additional measurements were carried out within the valid parameter window for statistical evaluations. In former joining experiments (see [11,15,16] and [27]), the electrode structure was placed 1 mm below the capillary orifice, thus the lateral deviation of the droplet in x- and y- plane was calculated 1 mm below the capillary orifice in this work, using the Euclidean distance in accordance to Eq. (1).
The area between the discrete measured points was interpolated in Figs. 7-10, using a linear interpolation, in order to enhance comprehensibility of the diagrams. The discrete measured parameters are indicated by black dots.
The average lateral positioning deviation for the conic shaped capillary was found to be 170 µm ± 130 µm (n = 215). The maximum optical output of the beam source represents the upper limit of the investigated parameters. Below 20 mbar no droplet detachment was observed for any pulse power, which must be accounted to the fact, that the surface tension of the molten droplet inside the capillary exceeds the cumulative force of gravity and gas overpressure and thus prevents droplet detachment as explained in [19]. If the pulse power is chosen below 60 W, no detachment is observed. This must be explained by the pulse power to be insufficient to completely melt the preform. Thus it must be concluded, that the absorbed energy on the preform surface is not sufficiently high to overrule the preforms loss of thermal energy due to convection, heat conduction into the capillary and heat radiation. In summary, the average lateral position deviation of 170 µm for the conic capillary proved to be sufficient to reproducibly generate joints of 1 x 1 mm electrode structures with Cu-wires as shown in [11]. Thus the average lateral displacement of the droplet represents just 17% of the total electrode structure area. The lateral displacement of the droplet should not exceed 30% of the electrode structures diameter, respectively 300 µm, which is fulfilled for all investigated parameters. The significant standard deviations must be attributed to the influence of varying braze preform mass and geometry, which is to be attributed to their manufacturing process of atomization, which can result in preform porosity or adherent satellite articles on its surface. Nonetheless it must be stated, that, even if the standard deviation seems high compared with the mean value, both, standard deviation and mean value are sufficiently low to allow reproducible droplet detachment on the substrate, since the droplet diameter is 600 µm, resulting in the average lateral position deviation to be 28 percent of the mean droplet diameter. In order to be able to compare the capillaries performance, the same procedure was carried out for the diffuser type capillary. The results are shown in Fig. 8:
The number of valid measurements to establish the parameter window were 111, again, additional 102 parameter were tested for statistical evaluation, resulting in a total number of 213 parameter sets to be evaluated by stereoscopic high-speed videos for the diffuser type capillary. The average lateral position deviation 1 mm from the capillary orifice was found to be 150 µm ± 120 µm (n = 213). Therefore, the positioning deviation of the conic capillary is 11.8 percent higher than the positioning deviation of the diffuser shaped capillary, which is particularly remarkable, since the valid parameter window was increased by 20.65 percent by replacing the conic with the diffuser type capillary. Even with this larger parameter window in which joints can be generated at higher pressures, the positioning accuracy of the diffuser type capillary is yet higher than the positioning accuracy of the conic shaped capillary. As droplet atomization occurs at higher overpressures when comparing diffuser type capillary with the conic capillary, a the stabilizing effect as indicated by the simulation was confirmed in the experiment and must be attributed to the reduced impulse of the gas molecules on the braze droplet after its detachment from the diffuser type capillary (see Fig. 3 and [16]). Thus the numerical results obtained by CFD-simulation could be confirmed by the experiments. Again, the standard deviation is high in comparison with the mean lateral position deviation, nevertheless as stated above, comparing the average lateral position deviation with the mean droplet diameter of 600 µm, it represents only one quarter of the droplet diameter and thus reproducible joints can be generated, using the described system at a working distance of 1 mm.
3.2 Droplet velocity
In order to understand, how the process parameters influence the droplet flight velocity, the information contained in the stereoscopic videos can be used as well. The velocity was calculated, applying Eq. (2):
Whereas x0, y0 and z0 indicate the respective coordinates of the capillary orifice center and thus the coordinate origin. xEnd, yEnd and zEnd indicate the spacial coordinate of the droplet before escaping the image section of the cameras. nFrames is the total number of frames, in which the droplet is visible for both cameras, and FPS indicates the framerate. Using this approach, the mean droplet velocity is obtained, since it integrates the velocity values over the total flight trajectory. Of course the process boundaries are the same, since the same videos as in section 3.1 have been evaluated. The results of the investigation for said parameters and capillaries is shown in Fig. 9. Again, in order to enhance comprehensibility of the diagrams, the area between the discrete measured points was interpolated, using a linear interpolation. The discrete points of measurements in the parameter field are indicated by black dots.
The average velocity for the conic shaped capillary was found to be 1.70 ± 0.65 m/s (n = 215). Again, the large standard deviation must be attributed to the fact, that significant braze preform volume and mass deviations are occurring, due to the manufacturing process of the preform, which are generated by atomization [11], which can result in preform porosity. A lower droplet mass thus results in an increased droplet acceleration due to the impulse of the gas particles on the droplet during the flight phase. In order to be able to compare both capillary geometries, the same approach was pursued for the diffuser shaped capillary (Fig. 10):
The average droplet velocity was found to be 1.83 m/s ± 0,71 (n = 213), which is 7.6% higher than the average velocities obtained with the conic capillary. This is mainly attributed to the rarer occurrence of droplet atomization at higher gas pressures, which must be accounted to the diffusing effect of the capillary geometry on the gas velocity fields (Fig. 3) as discussed above. Thus it can be concluded, that using the diffuser shape capillary not only the positioning accuracy can be increased by 13.3 percent, but also the parameter window can be increased by 20.65 percent. In addition, a 7.6 percent higher droplet velocity can be obtained using the diffuser shaped capillary, without resulting in droplet atomization. In summary, the diffuser shaped capillary outperforms the conic capillary in every respect, investigated in the scope of this work, which was expected from the CFD-simulations. As a conclusion, the pursued approach of first modelling and simulating a capillary geometry, which is capable to reduce gas flow velocities and subsequently machining it by means of laser ablation proved to be a suitable measure to enhance both, process parameter window and positioning accuracy of the braze droplet. The diffusing effect of the altered capillary geometry enables to apply overpressures which would result in droplet atomization for the conic type capillary. Thus the impact of the gas impulse on molten braze droplets in dependence of the capillary geometry was successfully quantified for the first time using stereoscopic high-speed imaging.
4. Summary
In the scope of this paper, a novel drop on demand joining technology was investigated by means of stereoscopic high speed imaging. The system is capable of applying CuSn12 droplets with a liquidus temperature of 990 °C and a solidus temperature of 825 °C [10] onto electrode structures of electronic circuit carriers, in order to join the electrode structure with Cu-wires. In order to quantify the influence of machining head overpressure, pulse power and capillary geometry on the droplet velocity, stereoscopic high-speed videos were evaluated to derive three-dimensional information of the droplet flight trajectory. In order to do so, the Euclidean distance of the droplet projected in the x- y- plane was calculated. The mean lateral position deviation of the diffuser type capillary was found to be 150 µm which is 13.3 percent lower than the mean lateral position deviation of the conic capillary of 170 µm. Nevertheless both are small in comparison with the 600 µm diameter of the braze droplet, which indicates high reproducibility of the process in terms of positioning repeatability. In addition, the results show an increasing droplet flight velocity by increasing the machining head overpressure for both capillary geometries. However the pulse power does not show a large influence on the droplet flight velocity. Additionally, the diffuser type capillary does broaden the parameter window significantly by 20.65%, since the diffusor type capillary suppresses droplet atomization at higher overpressures. This must be explained by the reduced impulse of the gas jet onto the molten droplet after detachment and during flight phase due to the diffusing effect of the capillary shape (see Fig. 3). Therefore the maximum droplet velocity obtainable with the diffuser type capillary is 3.189 m/s whereas the maximum droplet velocity which was obtained with the conic capillary is 2.948 m/s, which is 7.56% lower. Overpressure and pulse power does not seem to significantly influence the positioning accuracies, since the measurements do not show a clear trend. This is true for both capillary geometries. Overall the diffuser type capillary shows superior process performance than the conic shaped capillary. In summary, laser drop on demand joining is a novel brazing method, which offers significant advantages over soldering and wire bonding which are currently the dominating joining methods in electronics packaging. The LDJ process is particularly suitable for applications, which require high thermal stability of the generated joints such as high power electronics or electronics for harsh environments such as sensors for deep drilling heads or thermoelectric generators to be attached to jet engines. A particular advantage of the described contact free joining process is the fact, that the thermal energy input into the joining area is determined mainly by the used preforms geometry, respectively its mass, which enables the generation of braze joints on thin substrates such as 20 µm Ag-foils or sintered Ag-layers without the risk of electrode perforation. By modifying the capillary geometry as shown in the present paper, the joints lateral position deviation in respects to the capillary orifice could be increased, which further enhances the applicability for electronics packaging.
5. Outlook
The suitability of the process for generating joints suitable to withstand high temperatures e. g. occurring during Al-die casting have been shown by the authors in [11], which opens a wide range of possible applications of the described process for high temperature electronics packaging applications e. g. in the aeronautic industry, for harsh environment such as sensor for bore drilling heads or for high power electronics. Further possible applications of the process could be contacting thermoelectric power generators, to enhance their yield in high temperature environments such as rocket engines [28]. In addition, the process could be utilized to print three-dimensional structures and thus become an additive manufacturing process with the capability to generate structures with tailored alloy composition, since each preform can potentially consist of a different alloy. In the scope of future works, stereoscopic imaging could be utilized to evaluate physical properties of the droplets such as viscosity and surface tension, by evaluating the droplets oscillation during the flight phase as proposed in [29]. This is a particular interesting approach, since the described system technology is suitable to quantify melt viscosities of high melting materials such as Cu-, Au- or Ag- based alloys, which are particularly challenging to investigate with conventional methods due to the high liquidus temperatures, which the measurement setup needs to be able to withstand.
Funding
Deutsche Forschungsgemeinschaft (DFG) in context of the Collaborative Research Centre/Transregio 39 “Großserienfähige Produktionstechnologien für leichtmetall- und faserverbundbasierte Komponenten mit integrierten Piezosensoren und -aktoren“ PT-PIESA, subproject A04.
Acknowledgement
The authors gratefully acknowledge cooperation with the Erlangen Graduate School in Advanced Optical Technologies (SAOT) by the Deutsche Forschungsgemeinschaft (DFG) in the framework of the German excellence initiative.
References and links
1. U. Aridogan and I. Basdogan, “A review of active vibration and noise suppression of plate-like structures with piezoelectric transducers,” J. Intell. Mater. Syst. Struct. 26, 1455–1476 (2015).
2. V. Giurgiutiu, Structural Health Monitoring with Piezoelectric Wafer Active Sensors (Elsevier/Academic Press, 2008).
3. A. Preumont, Vibration Control of Active Structures: An Introduction (Kluwer Academic Publishers, 2002).
4. V. Bräutigam, Gießtechnische Integration piezokeramischer Module in Aluminiumdruckgussbauteile.Dissertation der Technischen Fakultät der Universität Erlangen-Nürnberg (2008).
5. M. Schwankl, S. Kimme, C. Pohle, W.-G. Drossel, and C. Körner, “Active Vibration Damping in Structural Aluminum Die Castings via Piezoelectricity - Technology and Characterization,” Adv. Eng. Mater. Advanced Engineering Materials 17, 969–975 (2015).
6. M. Schwankl, M. Rübner, M. Flössel, S. Gebhardt, A. Michaelis, R. F. Singer, and C. Koerner, “Active functionality of piezoceramic modules integrated in aluminum high pressure die castings,” Sens. Actuators A Phys. 207, 84–90 (2014).
7. M. Mäusezahl, M. Hornaff, T. Burkhardt, and E. Beckert, “Mechanical Properties of Laser-jetted SAC305 Solder on Coated Optical Surfaces,” Phys. Procedia 83, 532–539 (2016).
8. P. Ribes-Pleguezuelo, A. M. Inza, M. G. Basset, P. Rodríguez, G. Rodríguez, M. Laudisio, M. Galan, M. Hornaff, E. Beckert, R. Eberhardt, and A. Tünnermann, “Assembly processes comparison for a miniaturized laser used for the Exomars European Space Agency mission,” Opt. Eng Optical Engineering 55, 116107 (2016).
9. E. Beckert, T. Burkhardt, R. Eberhardt, and A. Tünnermann, “Solder Bumping — A Flexible Joining Approach for the Precision Assembly of Optoelectronical Systems,” in Micro-Assembly Technologies and Applications: IFIP TC5 WG5.5 Fourth International Precision Assembly Seminar (IPAS’2008) Chamonix, France February 10–13,2008, S. Ratchev, S. Koelemeijer, eds. (Springer US, 2008), pp. 139–147.
10. Deutsches Kupfer-Institut Düsseldorf, “Kupfer-Zinn-Knetlegierungen (Zinnbronzen)”.
11. S. Stein, J. Wedler, S. Rhein, M. Schmidt, C. Körner, A. Michaelis, and S. Gebhardt, “A process chain for integrating piezoelectric transducers into aluminum die castings to generate smart lightweight structures,” Results Phys. 7C, 2534–2539 (2017).
12. K. Maekawa, K. Yamasaki, T. Niizeki, M. Mita, Y. Matsuba, N. Terada, and H. Saito, “Drop-on-Demand Laser Sintering With Silver Nanoparticles for Electronics Packaging,” IEEE Trans. Compon. Packaging Manuf. Technol. 2, 868–877 (2012).
13. L. Zhang, Y. Zhu, X. Cheng, C. Wang, J. W. Gu, X. L. Chen, and E. H. Liu, “The Simulation Study of Fluid Physical Properties on Drop Formation of Drop-on-demand Inkjet Printing,” MATEC Web of Conferences 25, 03011 (2015).
14. Z. Luo, G. Zheng, and L. Wang, “A Study on the Influence of the Nozzle Lead Angle on the Performance of Liquid Metal Electromagnetic Micro-Jetting,” Micromachines 7.
15. S. Stein, J. Heberle, M. Suchy, F. Tenner, F. Hugger, S. Roth, and M. Schmidt, “High Temperature Laser Based Drop on Demand Micro Joining of Thin Metallic Layers or Foils using Bronze Braze Preforms: LAMP 2016,” J. Laser Micro Nanoeng. 11, 111–116 (2016).
16. S. Stein, M. Dobler, T. Radel, M. Strauß, H. Breitschwerdt, F. Hugger, S. Roth, and M. Schmidt, “Experimental and numerical investigations regarding laser drop on demand jetting of Cu alloys,” Prod. Eng. Res. Devel. 26, 1455 (2017).
17. T. Graf and H. Hügel, Laser in der Fertigung. Strahlquellen, Systeme, Fertigungsverfahren, 2nd ed. (Vieweg + Teubner Verlag / GWV Fachverlage GmbH, Wiesbaden, 2009).
18. S. Stein, J. Heberle, F. J. Gürtler, K. Cvecek, S. Roth, and M. Schmidt, “Influences of Nozzle Material on Laser Droplet Brazing Joints with Cu89Sn11 Preforms: 8th International Conference on Laser Assisted Net Shape Engineering LANE 2014,” Physics Procedia, 709–719 (2014).
19. A. Jeromen, C. Held, E. Govekar, S. Roth, and M. Schmidt, “Modelling of droplet detachment in the laser droplet brazing process,” J. Mater. Process. Technol. 214, 737–749 (2014).
20. T. Häfner, L. Hofmann, and S. Eiselen, “Einsatz der Simulation zur Optimierung der Abtragsrate beim Strukturieren mittels Pikosekundenlaser,” Tagungsband Laser in der Elektronikproduktion und Feinwerktechnik, 51–55 (2013).
21. S. Stein, J. Dippert, S. Roth, and M. Schmidt, “Laser Drop on Demand Micro Joining for High Temperature Wire Bonding Applications – System Technology And Mechanical Joint Performance,” JLMN 12, 239–247 (2017).
22. Vision Research Inc, Phantom v1210. Online: “Manual: Phantom v1210,” https://www.phantomhighspeed.com/Products/High-Speed-Cameras/v1210.
23. G. R. Bradski and A. Kaehler, Learning OpenCV. Computer vision with the OpenCV library, 1st ed. (O'Reilly, 2008).
24. O. Schreer, Stereoanalyse und Bildsynthese. Mit 6 Tabellen (Springer, 2005).
25. Z. Zhang, “A flexible new technique for camera calibration,” IEEE Trans. Pattern Anal. Mach. Intell. 22, 1330–1334 (2000).
26. J. Heikkila and O. Silven, A four-step camera calibration procedure with implicit image correction (1997).
27. S. Stein, S. Roth, N. Modler, M. Gude, T. Weber, and A. Winkler, “Investigations on laser based joining of novel thermo-plastic compatible piezoceramic modules,” in Proceedings of 5th Scientific Symposium Lightweight Design by Integrating Functions (2015).
28. G. J. Snyder and E. S. Toberer, “Complex thermoelectric materials,” Nat. Mater. 7(2), 105–114 (2008). [PubMed]
29. M. Perez, Y. Brechet, L. Salvo, M. Papoular, and M. Suery, “Oscillation of liquid drops under gravity: Influence of shape on the resonance frequency,” Europhys. Lett. Europhysics Letters 47, 189–195 (1999).
