Technology News
05/01/2003
Process enhances spin-on low-k dielectric processing
A new processing scheme allows for the solid-phase processing of spin-on porous low-k materials, overcoming the issues of barrier layer integrity, mechanical polishing, process-induced k value increases, and other integration concerns that have delayed porous low-k implementation.
Shipley Co. describes its Solid First process (see figure) as an important development for integrating the company's Zirkon low-k materials into manufacturing processes. "This will enable use of ultra-low-k materials to be considered for next-generation devices," said a company spokesperson.
Widespread adoption of spin-on dielectrics has been blocked by process integration problems. The new process uses Zirkon LK dielectric materials, described by Shipley as a "tough, uniform spin-on dielectric film in a nonporous state that withstands demanding process steps associated with advanced interconnect fabrication." Past problems like etching, polymeric residue removal, and chemical mechanical polishing (CMP) can be done on this robust film, according to Shipley.
By using the Zirkon low-k materials, spin-on hard mask, and etch stop materials, the Solid First process enables conventional interconnect processing to be done without increased copper resistivity and CMP-induced lamination. As part of the process, 2.5nm pores are formed during a curing step that follows CMP.
Researchers at Shipley worked with sister company Rodel on the spin-on dielectric research. Both companies are part of the Rohm and Haas Electronic Materials Group. Shipley supplied Rodel with wafers that had undergone the Solid First process, in nonporous form with some inlaid structures. Rodel researchers investigated how well the films withstood CMP process pressures and environment.
"The data that came back confirmed there was improvement, and the Solid First process was a marked benefit in the area of CMP," said the Shipley spokesperson.
The Solid First process holds the potential to enable ultra-low-k dielectrics to be used in next-generation devices, according to Shipley, which developed the process. |
null
The company reports it is working with industry consortia and major tool companies to confirm that its process works well with their specific tools. Work is progressing on barrier thickness and barrier film uniformity.
"One of our equipment suppliers has done some work with ashing and wet chemical cleaning," says Michael Gallagher, ILD program manager at Shipley. "They have looked at the infrared spectrum of the material, and they see no damage to the film."
Gallagher also reports that in CMP the material has been processed by both industry consortia and Rodel. Shipley estimates that it will take 18 months to two years before the Solid First process gains broad industry acceptance.
Diamagnetic levitation may enhance ultra-clean manufacturing
SRI International, Menlo Park, CA, has received a US patent for a frictionless transport technology based on diamagnetic levitation, a technique that can be used to levitate objects. The new technology offers the potential to improve various ultra-clean manufacturing processes that use microdevices.
Unlike the other two magnetic phenomena that can be used to levitate objects —ferromagnetism and paramagnetism — diamagnetic levitation is intrinsically stable and operates at room temperatures. It requires no energy supply.
A major benefit to this frictionless transport system is that it eliminates the possibility of contamination from particles that are rubbed off of mechanical devices such as gears and pulleys that are used to move materials between processing stages in clean manufacturing environments. SRI now has claims on both linear and rotational uses of diamagnetic levitation. The technology can be used in a broad range of temperatures.
Diamagnetism is a weak, repelling magnetic force exhibited by many materials that are commonly thought not to have magnetic properties. SRI's patents cover the use of this passive magnetism supplied by permanent magnets stabilized with diamagnetic materials, thereby requiring no additional power to provide lift. A commercial client has built this diamagnetic transport technology into a pilot production line.
OLED flexibility on display
Making flexible displays using organic LEDs (OLEDs) is a relatively recent undertaking, and making them able to withstand real-world usage conditions requires a moisture barrier that has been elusive.
Universal Display Corp.'s and Vitex Systems' combined efforts resulted in a prototype flexible OLED, 4096-pixel display that is <0.7mm thick, has a resolution of 80dpi (dots/.inch) and can be flexed to a curvature with a radius <1 in.
The key to enabling a commercial solution for flexible OLED displays is keeping moisture out. According to Vitex Systems' VP of sales and marketing, John McMahon, in order to produce an impermeable barrier, the coating on a flexible display must be continuous and defect-free.
The choices, however, have been problematic. PVD films have limited smoothing ability; organic films are not good moisture barriers; polycrystalline films have propagating grain boundaries; and amorphous films are not fully densified near room temperatures. In particular, when such coatings are applied to a plastic surface, they are rough and riddled with holes. Each area of roughness produces a grain boundary defect — and a path for moisture. As the defects grow, cracks and holes go through to the inorganic layer beneath, thus getting through and damaging the display.
Vitex Systems, a commercial spin-off of Battelle Memorial Institute, uses its own in-line monomer condensation and ceramic deposition system (Guardian) to deposit alternating layers of polymer and ceramic films the company calls Barix. McMahon explains that the vacuum deposition uses a liquid monomer that is flash evaporated to a monomer gas.
"The gas is then sprayed onto the top of the OLED in a vacuum where it condenses back to the liquid phase, completely covering the OLED display and forming an atomically smooth liquid surface," continues McMahon. "The monomer is then exposed to UV light, which converts it to a solid polymer, still atomically smooth."
By using multiple sputtered inorganic barrier layers combined with the highly nonconformal vacuum-deposited polymer layers, defects between layers are decoupled (any holes that might develop in one layer would not be co-located with a hole in the adjacent layer so moisture can't get through).
Additionally, the polymer layers are conformal and very smooth. The water permeation rate of the Barix layer ranges from 1¥10-4 - 1¥10-6/m2 /day at 25°C (depending on the number of layers) vs. 0.1 for PECVD inorganic coatings and 10 for an uncoated PET (polyester). The Barix coating used was five layers thick.
KLA-Tencor takes "AIM" at overlay control
A new optical overlay metrology solution uses a grating technology that addresses problems associated with traditional box-in-box (or frame-in-frame) overlay targets as the industry moves to the 65nm node. At that node, the overlay requirement is projected at 26nm (3s) and metrology precision at 2.6nm (3s).
The new solution from KLA-Tencor is called AIM, which operates on the company's Archer platform. The function of an overlay target is to measure how well a layer is registered to the previous layer. The difference between the grating target and the box-in-box target is that the grating target can help measure not only how well a target layer is registered to the previous layer, but also how well a device is registered to the previous device layer. "The grating target allows for a true representation of the device in terms of its pitch and CD," explains Ady Levy, senior director of marketing and applications in the Optical Metrology Division at KLA-Tencor. "Because the target behaves similarly to the device, it captures scanner aberrations and focus errors" (see figure).
The grating target is far more dense than a box-in-box target. "With the grating target, more than 4¥ the number of line edges are used than with a box-in-box target," says Levy. "With more points to match up to the line edges, there is a significant reduction in random and processing errors using the grating method. The more information, the better grainy and low-contrast layers can be measured."
null
Another problem with box-in-box targets is that they degrade during the CMP process. Because the grating by design has less open area (that is, less chance for damage to occur), the total measurement uncertainty is reduced. "Box-in-box targets are more sensitive to over- and under-polishing," states Levy. "As a result, they are more sensitive to specific process conditions. But the grating target — being similar to the device density — results in a more accurate overlay measurement."