SamsPcbGuide, Part 14: Technology - Microdevelopment and Chip-On-Board Technology

In this self-isolating article, I will talk about the wire bonding of micro-leads (English wire bonding). In the context of printed circuit boards, we will focus on the technology of mounting crystals on a printed circuit board (English chip-on-board, COB). Be sure to watch the video on the links, microwelding is very beautiful!

Welding provides an electrical connection of the crystal with the findings of the case (when packaging the chip), or directly with the conductors of the printed circuit board (COB technology). An alternative way of electrical connection is the inverted mounting of the crystal (English flip-chip), both in the design of the case itself and directly on the printed circuit board (Fig. 1).

Installation using microwire leads appeared after the first integrated circuits in the early 1960s and has been successfully used to this day. Inverted installation is a modern technology that has arisen in response to the requirements of increasing the number of conclusions, increasing speed and reducing dimensions. However, it has a number of design limitations associated with ensuring reliability (I wrote about this in detail here ), and is technologically more complex.

This article will not examine in detail the classification and theory of various welding methods - this is a very voluminous material that goes beyond the scope of the issue under discussion. The fact is that over its long history, the technology of connecting microwire leads has developed in the direction of increasing stability, reliability, speed of the assembly process, expanding the capabilities of equipment to create welding loops of complex shape and high density of installation (Fig. 2). A variety of tasks and the lack of universal technology led in the search process to the development of various welding methods. Consider briefly the main points. Despite the variety of methods, a common principle for all is thatthat a welded joint is formed as a result of pressure and heating of the contacting surfaces to a high temperature until the formation of interatomic compounds (most often these are intermetallic compounds). Depending on the heating method, the weld is divided into the following main types: thermocompression (external heating), ultrasonic (friction with an ultrasonic pulse), thermosonic (a combination of external heating and ultrasonic pulse) and contact (pulsed heating with electric current) welding. The main materials for microwire terminals are aluminum, gold, copper. Copper is used instead of gold to lower costs, but it is harder and also oxidizes quickly in air, which complicates the welding process and requires more sophisticated equipment that creates an inert atmosphere in the welding zone (nitrogen or molding gas).The high conductivity of copper is a driver for replacing aluminum in the breakdown of power devices, despite the more complex manufacturing process.

The axis of the wire during unwrapping can be oriented in parallel - this is a wedge type wedge (eng. Wedge bond), or perpendicularly - it is a wedge type weld (eng. Ball bond) (Fig. 3). The loop most often has two contact points, therefore, according to the type of welding points, the welding methods are divided into “ball-wedge” and “wedge-wedge”. The most common are ultrasonic welding with wedge-wedge aluminum wire ( video ) and thermo-sound welding with gold / copper ball-wedge type ( video ). In the latter case, the tip of the wire is fused with a spark discharge ( video ) to form a ball , which only adds epicity and beauty to the welding process. For power devices, aluminum and gold tapes are used ( video with chic soundtrack).

The key parameters in ultrasonic / thermosonic welding are the welding force, power and duration of the ultrasonic pulse. Their combination for a given welding setup, a specific wire (diameter, stiffness), a specific cooking tool, specific parameters of the contact pad (size, material) should ensure the repeatability of the welding process with guaranteed connection reliability parameters. Directly controlled parameters are appearance, pull-out force (Eng. Pull test) and shear force (Eng. Shear test) (Fig. 4), indirectly - failures during thermal cycling and other tests in the product.

The selection of parameters is in some way a magical procedure, but there are a number of recommendations. In general, it is carried out by the method of scientific pokingdesign experiment (DOE), i.e. sequential enumeration of parameters in a certain range. In this search, you can build on the published results on parameter optimization (for example, articles [1, 2]), on the recommendations of the equipment manufacturer, and, of course, on the experience gained. Next, test uncoupling is performed for each set of parameters, followed by control of the appearance and efforts of separation / shift. To control the appearance, an increase of x100 with the ability to measure linear dimensions is enough (Fig. 5), specialized equipment is used to measure tear / shift forces (for laboratory tasks, for example, you can use a gram meter with a hook, and shift manually and examine the fracture shape at magnifications x100 ... 200 and more).To assess the appearance, it is important to look at as many micrographs of the beautiful and ugly weld points as possible (herehere and here, for example, there are some good microphotographs with a description), because in my experience there is a correlation between the beauty of the unwrapping and its quality. In addition, with experience, there is an understanding of how to vary them when setting parameters to obtain the desired result (there is not enough effort when welding or too much power), that is, the search becomes more and more conscious and directional. At one time, reading the articles from theories of the formation of welding and the influence of parameters on its quality helped me a lot [3, 4]. And yet these equally no less beautiful articles [5, 6], where the authors (monuments to put such) studied the formation of a loop using a high-speed camera. And so, the number of articles read to finish up to 100, the number of boiled and torn jumpers up to 10,000 and the magic in this process will be slightly less. Still very much from the installation,Of course, it depends - in a bitter struggle I squeezed the maximum out of the Belarusian EM-4450 machine.

Now back to the printed circuit boards and some features of their design using COB technology. The technology is used to reduce costs or in case of micro-mini-utilization, the creation of multi-chip modules and assemblies (in particular, LED). In fig. 6 shows an image from one of the presentationsWurth Electronics on this topic with design recommendations. The presented size restrictions can serve as a guideline, it is additionally recommended to use the 3D model of the cooking tool to check the availability of all KPs in order to avoid problems already in fact. It is important to note that the mask is removed on the printed circuit board in the weld zone, so as not to interfere with the working plane of the cooking tool. It is better not to use the area under the crystal for tracing, but to place the mounting pad there (as in the figure), especially if the base of the crystal has potential or needs to be mounted on a conductive adhesive to increase heat dissipation. Mounting the crystal on the mask is possible, especially in the case of further compounding the crystal. The crystal can also be mounted in a cut-out in a printed circuit board.pcb cavity) with the location of the CP in the same topological layer or higher.

The next nuance regarding tracing is the recommendation to orient the control unit on the printed circuit board in the direction of the welding loops for the gold wire. I understood its validity only when writing a decompression program. The bottom line is that the second cooking point is formed on the edge of the capillary (Fig. 7), and when writing a program, its center is indicated, which at large angles leads to the need to take this into account and shift the location of the point in the program. In other words, it’s more convenient, but this is not a fundamental limitation, and the convenience of tracing has a higher priority.

There is a consensus regarding the finish coating of printed circuit boards in the industry: ENIG is enough for welding with aluminum wire, ENEPIG or galvanic gold for gold wire (Fig. 8). Why can’t you use the cheaper and more affordable ENIG for welding with gold wire? The answer that could be found is that nickel drift leads to degradation of the welded joint with a significant decrease in its reliability. And in ENEPIG, palladium serves as a barrier layer that prevents this drift. It is quite acceptable to use ENIG for debugging samples, all the more so since the welding parameters, all other things being equal, are close for these coatings. ENEPIG is directly indicated as a recommended coating in many sources, and it provides data on reliability tests [7, 8].

Much attention is also paid to the problem of the formation of undesirable Au-Al intermetallic compounds (“purple plague” and other scary words) that occur when welding with a gold wire on an aluminum crystal CP or with aluminum wire on ENIG. This question is quite complex and for my full understanding of the necessary knowledge of chemistry, I, unfortunately, do not. The conclusion is that the decompression in the Au-Al system is a potential source of failures, especially at high temperatures, and should be thoroughly tested for reliability. Maximizing the pull-out force is one of the strategies, since the strength of welding and long-term reliability are connected (a thinner intermetallic layer with a larger coating area).

Due to the thin layer of gold, the weak point of the ball-wedge unwinding on ENEPIG is the second weld point. The task of obtaining high-quality unwinding is also complicated by contamination of the KP after the stage of surface mounting of components on a printed circuit board. There are two methods to increase reliability: with a reinforced bump after welding (English security bump / ball, SB) and with preliminary bumping (English ball stich on ball, BSOB, or stand-off stich, SOS) (Fig. 9). An additional optimization parameter in these technologies is the displacement of the bump relative to the welding point [9, 10]. From my own experience I can say that BSOB has shown itself well for COB. By the way, BSOB is also good because it allows you to put the second welding point on the crystal (English reverse bonding) and carry out the welding between the crystals directly in multi-chip assemblies. I would be glad if in the comments share your experience with SB / BSOB and ENIG / ENEPIG.

, , - . ( ), , . , , . « » – .

, #SamsPcbGuide! – #, . #SamsPcbCalc. , . , , . , , - !

[1] J. Gomes, M. Mayer, B. Lin, ”Development of a Fast Method for Optimization of Au Ball Bond Process”, 2015
[2] Byung-Chan Kim, Seok-Jae Ha, etc. “Process Capability Optimization of Ball Bonding Using Response Surface Analysis in Light Emitting Diode (LED) Wire Bonding”, 2017
[3] Hui Xu, Changqing Liu, etc, “Initial bond formation in thermosonic gold ball bonding on aluminum metallization pads”, 2010
[4] Hui Xu, Changqing Liu, etc. “The role of bonding duration in wire bond formation: a study of footprints of thermosonic gold wire on aluminum pad”, 2010
[5] Fuliang Wang, Yun Chen, Lei Han, ”Experiment study of dynamic looping process for thermosonic wire bonding”, 2012
[6]Fuliang Wang, Yun Chen, Lei Han, “Effect of Capillary Trace on Dynamic Loop Pro fi le Evolution in Thermosonic Wire Bonding”, 2012
[7] Chun-Hsien Fu, Liang-Yi Hung, etc. “Evaluation of New Substrate Surface Finish: Electroless Nickel / Electroless Palladium / Immersion Gold (ENEPIG)”, 2008
[8] Kuldip Johal, Sven Lamprecht, Hugh Roberts, ”Electroless nickel / electroless palladium / immersion gold plating process for gold-and aluminum -wire bonding designed for high temperature applications ”, 2000
[9] Chunyan Nan, Michael Mayer, etc. “Golden bump for 20 micron diameter wire bond enhancement at reduced process temperature”, 2011
[10] Young K. Song and Vanja Bukva, “Challenges on ENEPIG Finished PCBs: Gold Ball Bonding and Pad Metal Lift”, 2017