Here are the highlights from Session Five “New Probe Card and Contact Technologies” of the 22nd annual IEEE Semiconductor Wafer Test Workshop (SWTW) from Tuesday June 12, 2012.
Tsutomu Shoji (Japan Electronics Materials Corp. ‐ Japan) and Takashi Naito (Advantest ‐ Japan), “Full Wafer Contact Breakthrough with Ultra‐High Pin Count”:
Awarded Best Overall Presentation
As the number of probes on probe cards increase due to greater parallelism, driven by the desire for one touchdown testing and the future transition to 450 mm wafers, the total force required to probe a wafer increases if there is no reduction in the force per probe. This wafer prober chuck needs to apply the required force by pushing the wafer against the probe card typically held in place by the structure of the prober. With 200K probes on a 450 mm wafer each requiring 5 gF this is approximately equal to 1 ton (2205 lbF) of applied force. To reduce these force requirements wafer chuck and prober structure, Advantest and JEM have switched to a vacuum based system. The Vacuum Probe Contact System (VPCS) develops the required force to draw the wafer and the “probe card” together to make the electrical connections required to test the wafer. This approach has successfully been used in wafer level burn-in applications.
For the contact technology, micro-electrical-mechanical system (MEMS) probes were selected due to their high repeatability and accuracy. This system requires pre-heating of the probes and the wafer to ensure proper alignment. As the VPCS was developed for platforms such as the HA5100CELL test cell, which combines the essentials of four testers and a prober to test four wafers in parallel, numerous challenges related to the vacuum were identified. The largest challenges included eliminating vacuum leakage, preventing over compression of probes, and avoiding excessive tilt when bringing the probes and wafer into contact. A new mechanical structure called a “Unit Holder” was created to replace the structure of a typical probe card and includes an improved sealing area, mechanical stops to prevent over compression, and a pre-load spring to provide tilt correction.
Test results shared showed successful full 300 mm wafer contact at 85 °C, ambient, and -30 °C based upon consistent Cres and probe mark analysis. They are working to extend this solution to 450 mm wafers.
Questions:
- Is the wafer hot or cold before they move the wafer in contact to the probes which has been pre-heated? The system preforms a pre-heating step before contacting the wafer.
- How does the system accommodate different thicknesses of back ground wafers? The overdrive is controlled by the vacuum pressure. So if the particular wafer thickness changes, the system can change the vacuum pressure to accommodate.
Takehisa Sakurai and Masahiro Kato (Hitachi Chemical Co., Ltd.), “High Density and Low Cost Approach for the PCB of Semiconductor Tester”:
High speed, high density, fine pattern, and lower cost are all competing requirements placed upon printed circuit boards (PCBs). These conflicts usually require a trade-off discussion with a customer based upon the functional requirements of the customer’s end product.
An overview of Hitachi’s unique multi wire board (MWB) technology was provided. (MWB was also presented at SWTW last year.) MWB is based upon automated placement of individual insulated wires instead of routing metal traces on sheets of dielectric (found in standard PCBs). By using this method Hitachi is able to increase the signal density of a PCB by 2.5x while improving signal quality due to the greater uniformity of the conductors. The placement of wires is a serial process typically costing more than traditional PCB fabrication using photolithography to process an entire layer at a time.
To increase performance and reduce costs, they have created a hybrid PCB consisting of a MWB core sandwiched between high performance “traditional” PCB layers. This process provides high signal density in the MWB core while reducing the cost of the overall PCB by fabricating the outer layers using traditional fabrication methods. Since the outer layers are regular PCBs, they provide low inductance metal power planes which MWBs lack and when needed higher speed materials can be used.
Question:
- What is the temperature performance of these materials when used as probe cards? A full characterization of the material and structures at extreme temperatures has not yet been completed.
Frank Schonig (Rosenberger), “LIGA and its Application to Electrical Interconnects”:
The Lithographie, Galvanoformung, Abformung (Lithography, Electroplating, and Molding = LIGA) process is used to electroplate or electroform microstructures. LIGA is used to create MEMS structures and other small parts with high aspect ratio features such as watch parts. Currently Rosenberger uses LIGA to build several products including the Cascade Microtech Z-Probe which operates at up to 67 GHz.
A new product line of Monolithic Compliant Interconnects (MCI) has been developed using LIGA. Example applications included interposer connections at 0.8 mm pitch and wafer level chip scale packages (WLCSP) contactors at 0.4 mm pitch. Since the shape of the parts are defined using photolithography, this allows creation of arbitrary shapes which optimize the device based upon material characteristics. The interposer connections allow greater compression (approximately 31% of the height versus a typical 20%) which increases compliance at lower force than existing connectors.
The WLCSP contactors are designed to displace traditional micro-spring pins for testing devices at 0.4 mm pitch and below. Unlike micro-spring pins, especially barrel-less designs, the signal runs along the shortest nearly linear path from end to end. This is achieved by pre-loading the spring when the contactor is placed in the housing. The design also ensures there is always a “wiping action” to improve contact reliability. Unlike a traditional spring pin design, the force can be asymmetric with a different force on the PCB side versus the device under test (DUT) side.
Designs have also been completed for .8 and .4 mm pitch spring pin replacements. Spring pins at .3 and .2 mm pitch are in process.
When building parts with features of this size, material properties are no longer linear and cannot be scaled from bulk values. At this scale, there are no substitutes for direct measurement to obtain values for the design and simulation. Mr. Schonig devised a system of tensile stress test coupons at the feature sizes that interested Rosenberger. Based upon their research, Rosenberger appears to be the only LIGA company that has achieved direct measurements of material values. And by closing the loop between design and fabricated parts they are able to optimize performance. An example of this is to increase the stored energy in the spring structure while reducing the material stress.
The non-recurring expense of the LIGA process is less than one-half the cost of a single stamping die. Not only does LIGA prototyping cost less than traditional stamping, it is significantly faster in returning parts in six weeks where it can take eighteen weeks to obtain to obtain a stamping die.
Innovative products with greater performance and functionality can be fabricate by understanding materials behavior and establishing design rules combined with process knowledge and stability, In addition to using this process for Rosenberger’s own products, they are making these interconnect products available to others.
Questions:
- What is the contract resistance (Cres) of the interposers? The small interposer have around 50 to 55 mOhms of Cres.
- On the measured force curves for multiple samples there appears to be a variation between the parts? There is a +/- 1 µm manufacturing tolerance resulting in a little more material in the cross section of the spring. However since the springs are independent this should not be a problem.
Raffaele Vallauri (Technoprobe Spa – Italy) “TPEG: a New Vertical MEMS Solution for High Current, Low Pitch Applications”:
There is an increased demand for probe card applications with fine pitch probes that reduce pad damage but require increased current carrying capacity (CCC). The challenge is to optimize the solutions against these conflicting requirements at the lowest cost.
To meet these challenges, Technoprobe realized the need for custom materials and advanced fabrication processes. The Technoprobe Etching and Galvanic (TPEG) process, similar to LIGA, was developed and implemented in-house. Nickel, palladium, and copper based alloys were evaluated. For each alloy mechanical (stress-strain), electrical (resistivity), and thermal (co-efficient of thermal expansion and other changes) properties were characterized. Similar to Mr. Schonig’s experience, material properties need to be directly measured at this scale. The characterization data was fed back to their design simulations to optimize materials and design.
Development started on a full finite element analysis (FEA) non-linear model of a buckling beam (vertical) type probe and supporting structure for the mechanical analysis. At the same time a simplified multi-physics model was built to study the electro-thermal interactions. As their characterization and development work continues, these models will be further refined. Once they correlate to experiment results the multi-physical and mechanical FEA will be combined. Then they hope to study the interactions such as how higher friction between the probe needle and guide plate reduces CCC due to contact stability. CRes remains the dominant factor in determining CCC.
A portfolio of different MEMS springs is now under development using TPEG. Currently the T1 probe is available and supports a minimum pitch of 55 µm with CCC > 750 mA and a force of 2 – 2.5 gF. Additional probes such as the T3 and T4 will have higher CCC while others (T0) will be targeted at tighter pitch. The T4, which uses different alloys, is targeted for extreme temperature probing from -55 to 200 °C for automotive applications.
By developing the TPEG process in-house they are able to accelerate cycles of learning. External MEMS processing was taking over thirteen weeks for a design, whereas internal processing is less than four weeks.
Question:
- What is the production capability of TPEG? Current it is 50K probes / month and it will be increased to 150K / month in the fourth quarter of 2012.
High Technology Business Development - Ira Feldman 