Nanonex NX-2500 Nanoimprint Lithography Update (Cornell)
Vincent Genova, Process Engineer, CNF
Posted Oct. 2012
Nanoimprint lithography (NIL) has the advantage of high throughput with sub-10nm resolution. NIL is included on the ITRS roadmap for 45nm and below nodes for advanced electronic devices. In addition to electronics, NIL is a benefit to many applications including displays, nanophotonics, biotechnology, and MEMS.
The NX-2500 has both thermal imprint (T-NIL) and photocurable imprint (P-NIL) capabilities. The thermal imprint module can reach temperatures up to 300C with rapid heating and cooling rates. The photocuring module uses a narrow band 200W UV lamp with automatic control. It has submicron overlay alignment accuracy and has the ability to handle irregular shaped and sized substrates up to 100mm diameter.
Recently the efforts of CNF Fellow Carol Newby and CNF technical staff members Vince Genova and John Treichler have resulted in the development of an established baseline process for photocurable nanoimprint lithography (P-NIL) on the NX-2500. For P-NIL, the starting template material is quartz. The quartz wafer is blanket sputter deposited with 20nm of chrome. The established process uses the ASML DUV stepper for patterning features to less than 200nm on the template. The applied resist is UV210 along with an anti-reflective layer AR3 yielding a combined thickness of around 660nm. After selectively etching the ARC layer in the Oxford 80 RIE system, the pattern is transferred into the chrome using Cl2/O2 based chemistry in the Trion ICP etch system. Once the remaining resist is removed from the chrome, the pattern is then transferred into the quartz to a depth of around 80nm using CF4 chemistry in the Oxford 80. The resulting etch profile is critical for successful imprinting. After wet etching the chrome, the quartz template is coated with a fluorosilane anti-stiction layer FOTS in the MVD 100. The anti-stiction layer enables the easy removal of the template after the imprint and eliminates the need for rigorous cleaning of the template.
|Figure 1. AFM of 1um features etched into the quartz template using different etch chemistries. The CF4 etch shows the lowest roughness and is therefore most suited to the template preparation.|
The P-NIL process can be applied to many types of substrates, but has been demonstrated on silicon. The P-NIL process utilizes a bi-layer resist system in which the first resist layer (200nm) is purely organic and the upper UV resist layer (90nm) contains silicon. Following the replication of template features into the upper layer of resist, a very critical pattern transfer process must occur in the residual and transfer layers of resist. The amount of residual UV resist remaining is a function of the applied imprint pressure. The pattern transfer consists of a selective fluorocarbon etch chemistry for the residual layer and the use of oxygen plasma to clear the organic resist layer in the Oxford 80. These etches must preserve the critical dimensions (CD) defined by the imprint process. The patterned imprint resist is then used as a mask to transfer the pattern into the silicon using an established “photonics etch” of combined SF6/C4F8 chemistry in the Unaxis 770 ICP system. Linewidths around 180nm have been etched into 500nm of silicon with perfect anisotropy and line edge resolution.
|Figure 2. SEM cross-section images of a 1um feature a) in the bilayer resist after imprinting, b) after descum and underlayer etches, and c) after transfer into silicon.|
|Figure 3. SEM cross-section images of sub-200nm features transferred into silicon.|
This process will be further demonstrated with pattern transfer into silicon oxide and silicon nitride. The process is directly applicable to SOI based devices. Further development of the imprint process will be applied to features patterned by electron beam lithography.
For further information on this process, please contact Vince Genova, firstname.lastname@example.org