EUV Reticle Management: From Mask Shop to Fab
After decades of research and development and billions of dollars in investment, Extreme Ultraviolet (EUV) Lithography is finally on a clear path for insertion into high volume manufacturing (HVM) in the 2019-2020 time frame. Given the imminent reality of volume use of EUV, numerous questions regarding the entire EUV infrastructure, especially those concerning the reticle and pellicle, must be addressed in the very near term.
Several key differences between EUV and traditional optical lithography present serious challenges for timely and economically viable insertion into HVM. From a reticle perspective, the key differences are:
- Mask substrates: Since the reticle is now reflective instead of transmissive, tiny pits or bumps in the substrate will impact the reflectivity of the multi-layer reflector.
- Multi-layer reflector: Composed of dozens of atomically perfect layers of alternating materials, any defect in the films will result in an embedded phase defect in the final reflecting surface.
- Absorber: The absorber stack is several wavelengths high, resulting in shadowing due to the 6 degree angle of incidence of the illumination and creating significant mask 3D effects.
- Inspection: For the first time in many years, the actual pitch on the reticle will shrink significantly, requiring higher resolution inspection. The mask is reflective only (no transmitted light), reducing the channels of information available for image analysis. In addition, dimensions of the same patterns in different parts of the field may be a few nanometers different due to mask 3D, variation in stray light (flare and out of band radiation), and shadowing effects, complicating die-to-die inspection.
- Defect printability: In the absence of an actinic tool, printability analysis will require extensive simulation of the off-axis illumination, mask 3D and stochastic effects to determine defect printability. The results may be a probability distribution, not a simple pass/fail result.
- Pellicle: Unclear if they will be ready, how long they will survive under high power, and whether they will fully protect the reticle surface from contamination.
- Backside: EUV reticles are electrostatically clamped so large backside particles will distort the surface and create pattern placement errors on the front side of the mask.
Each of these changes require careful attention and adoption of specific inspection solutions at each step of the process. Unlike optical masks, where a single inspection can be expected to detect all critical defects during the mask making process, EUV masks require a series of unique inspections which are optimized for each specific step of the mask making and qualification process. In addition, defect printability analysis will need to consider all of the complex mask 3D and shadowing effects related to non-normal incidence of the illumination, as well as stochastic effects due to photon shot noise.
The uncertainty over the type, availability, durability and transmission of the pellicle also raises questions regarding the optimal inspection strategy for fab qualification and periodic requalification during use in the wafer fab. The requal cycle is further complicated by the fact that the exposure tool is entirely in vacuum. Minimizing the number of times a reticle is loaded and unloaded may help limit particle adders, but the use of long “trains” of identical lots may not be possible in all types of fabs, and if the trains become too long the amount of material at risk between inspections could pose a significant risk of processing hundreds or even thousands of defective wafers. On the other hand, removing a reticle from the tool during a long exposure sequence is highly undesirable, making wafer based inspection during long train exposures a more viable solution. Balancing reticle and wafer based inspection cycles to minimize such risks will be an important part of a successful EUV adoption strategy.
In this presentation we will describe the various inspection solutions for each stage of reticle inspection, with different options depending on the pellicle strategy selected. We will show how data from multiple sources can be combined to produce an effective end-to-end solution for EUV reticle inspection from the mask blank through the fab production cycle.