Sean Eichenlaub

Particle removal challenges of EUV patterned masks for the sub-22nm HP node

The particle removal efficiency (PRE) of cleaning processes diminishes whenever the minimum defect size for a specific technology node becomes smaller. For the sub-22 nm half-pitch (HP) node, it was demonstrated that exposure to high power megasonic up to 200 W/cm2 did not damage 60 nm wide TaBN absorber lines corresponding to the 16 nm HP node on wafer. An ammonium hydroxide mixture and megasonics removes ≥50 nm SiO2 particles with a very high PRE. A sulfuric acid hydrogen peroxide mixture (SPM) in addition to ammonium hydroxide mixture (APM) and megasonic is required to remove ≥28 nm SiO2 particles with a high PRE. Time-of-flight secondary ion mass spectroscopy (TOFSIMS) studies show that the presence of O2 during a vacuum ultraviolet (VUV) (λ=172 nm) surface conditioning step will result in both surface oxidation and Ru removal, which drastically reduce extreme ultraviolet (EUV) mask life time under multiple cleanings. New EUV mask cleaning processes show negligible or no EUV reflectivity loss and no increase in surface roughness after up to 15 cleaning cycles. Reviewing of defect with a high current density scanning electron microscope (SEM) drastically reduces PRE and deforms SiO2 particles. 28 nm SiO2 particles on EUV masks age very fast and will deform over time. Care must be taken when reviewing EUV mask defects by SEM. Potentially new particles should be identified to calibrate short wavelength inspection tools. Based on actinic image review, 50 nm SiO2 particles on top of the EUV mask will be printed on the wafer.

A method to determine the origin of remaining particles after mask blank cleaning

Extreme ultraviolet lithography (EUVL) is a strong contender for the 32 nm generation and beyond. A defect-free mask substrate is an absolute necessity for manufacturing EUV mask blanks. The mask blank substrates are, therefore, cleaned with different cleaning processes to remove all defects down to 30 nm. However, cleaning suffers from the defects added by various sources such as the fab environment, chemicals, ultra pure water, and the cleaning process itself. The charge state of the substrate during and after cleaning also contributes to the number of adder defects on the substrate. The zeta potentials on the substrate surface and the defect particles generated during the cleaning process determine whether the particles get deposited on the surface. The zeta potential of particle or substrate surfaces depends on the pH of the cleaning fluids. Therefore, in this work, pH-zeta potential maps are generated for quartz substrates during the various steps of mask cleaning processes. The pH-zeta potential maps for defect particles commonly seen on mask substrates are measured separately. The zeta potential maps of substrate and contaminant particle surfaces are used to determine whether particles are attracted to or repulsed from the substrate. In practice, this technique is especially powerful for deriving information about the origin of particles added during a cleaning process. For example, for a known adder with a negative zeta potential, all cleaning steps with a positive zeta potential substrate could be the source of added particles.

Cleaning of MoSi multilayer mask blanks for EUVL

Extreme ultraviolet lithography (EUVL) is being considered as the enabler technology for the manufacturing of future technology nodes (30 nm and beyond). EUV mask blanks are Bragg mirrors made of Mo and Si bilayers and tuned for reflectivity at a wavelength λ ~13 nm. Implementation of EUVL requires that the mask blanks be free of defects at 30 nm or above. However, during the deposition of MoSi multilayers and later during the handling of blanks, defects are added to the blank. Therefore, the cleaning of EUV mask blanks is a critical step in the manufacturing of future devices. The particulate defects on the multilayer-coated mask blanks can either be embedded in or under the MoSi layers or adhered to the top capping layer during the deposition process. The defects can also be added during the handling of photomasks. Our previous studies have shown successful removal of the handling-related defects at SEMATECHs Mask Blank Development Center (MBDC) in Albany, NY. However, cleaning embedded and adhered defects presents new challenges. The cleaning method should not only be able to remove the particles, but also be compatible with the mask blank materials. This precludes the use of any aggressive chemistry that may change the surface condition leading to diminished mask blank reflectivity. The present work discusses the recent progress made at SEMATECHs MBDC in cleaning backside Cr-coated mask blanks with a MoSi multilayer and a Si cap layer on the top surface. Here we present our data that demonstrates successful removal of sub-100 nm particles added by the deposition process. Surface morphology and defect composition on the surface of the MoSi multilayer are discussed. EUV reflectivity measurements and atomic force microscopy (AFM) images of the mask blank before and after cleaning are presented. The present data shows that no measurable damage to the EUV mask blank is caused by the cleaning processes developed at the MBDC.

New requirements for the cleaning of EUV mask blanks

Extreme ultraviolet (EUV) substrates have stringent defect requirements. For the 32 nm node, all particles larger than 26 nm must be removed from the substrate. However, real defects are irregularly shaped and there is no clear dimension for an irregular particle corresponding to 26 nm. Therefore, the sphere equivalent volume diameter (SEVD) for a native defect is used. Using this definition and defect detection measurements, all particles larger than 20 nm must be removed from the substrate. Atomic force microscopy (AFM) imaging and multiple cleaning cycles were used to examine the removal of particles smaller than 50 nm SEVD. Removal of all particles larger than 30 nm was demonstrated. Particles that required multiple cleaning processes for removal were found to be partially embedded. The best cleaning yield can be obtained if the cleaning history of the substrate is known and one can choose the proper cleaning processes that will remove the remaining particles without adding particles. Ag, Au, Al2O3, Fe2O3, and CuO particles from 30 nm to 200 nm were deposited on quartz surface. It was shown that these deposited defects are much easier to remove than native defects.

Dependency of EUV mask defects on substrate defects

Extreme ultraviolet (EUV) mask blanks must have nearly zero defects larger than 30 nm. Mask blank defects are an accumulation of defects present on the substrate, defects added during the multilayer (ML) deposition process, and defects added by handling the mask blank. A majority of the detectable defects are already present on the substrate before the ML deposition. However, very few of the defects present on the substrate before the ML deposition are detectable. This raises the question of whether the substrates surface condition contributes to the total number of defects on the mask blank. Here the results of investigations on the relation between the total number of defects on the multilayer and the substrate surface condition are presented. The final surface condition is determined by the mask cleaning process. Correlation studies between defect maps before and after multilayer deposition are presented, and the relation between final defect size on the multilayer and substrate are discussed. SEMATECHs Mask Blank Development Center (MBDC) has a unique capability to characterize the surface of EUV glass substrates by atomic force microscopy (AFM), scanning electron microscopy (SEM), surface energy measurement, and zeta potential metrology. A series of experiments were performed in which different cleaning processes were used to modify the substrate surface condition before multilayer deposition. The effect of the cleaning process on the number of pits and particles after ML deposition was examined. The results indicate that although there is a direct relationship between the number of defects remaining on the substrate and mask blank defects after multilayer deposition, the variation in the total number of defects on the mask blank mainly corresponds to pits and particles already present on the substrate before cleaning and are not the result of the cleaning processes that were used before multilayer deposition.

Nanopit smoothing by cleaning

Defect smoothing is a critical need for improving defects. There are different methods such as using a smoothing layer or multilayer deposition; however, smoothing processes tend to add defects of their own to the surface. This paper presents a novel pit smoothing method based on an anisotropic substrate etch process. Smoothing power is defined as a metric for comparing the smoothing capability of different smoothing processes. Defect smoothing by cleaning is a surface modification technique with a smoothing power <10 that does not add defects to the surface. This is demonstrated by comparing total defects on the mask blank and mask blank substrate for two processes: a standard ozone-based cleaning and a smoothing cleaning. The smooth/clean methods led to fewer defects on the blank and substrate surfaces than the standard clean while still meeting extreme ultraviolet (EUV) blank roughness requirements. Finally, it is shown that smoothed pits are still printable. Therefore, further improvements to the smoothing power of smooth/clean processes are needed. SEMATECH is currently working to improve smooth/clean processes for low thermal expansion material (LTEM) EUV substrates.

Sub-30-nm defect removal on EUV substrates

Naturally occurring sub 30 nm defects on quartz and Low Thermal Expansion Material (LTEM) substrates were characterized by using Atomic Force Microscope(AFM). Our data indicates that a majority of defects on the incoming substrate are hard defects including large, flat particles with a height less than 5 nm, tiny particles with a size of 10 nm to 30 nm SEVD and pits with a depth of about 9 nm. All the soft particles added by handling with sizes of >50 nm can be removed with a single cleaning process. At least four cleaning cycles are required to remove all of the remaining embedded particles. However, after particle removal in their initial location a shallow pit remains. Based on detailed characterization of defect and surface by AFM, we propose that these hard particles are added during the glass polishing step and therefore it is important to revisit the glass Chemical Mechanical Polishing (CMP) processes and optimize them for defect reduction. A qualitative value for particle removal efficiency (PRE) of >99% was obtained for 20 nm Poly Styrene Latex Sphere (PSL) deposited particles on surface of glass.

Sub 50 nm cleaning-induced damage in EUV mask blanks

Defects are still one of the main challenges of extreme ultraviolet (EUV) mask blanks. In particular, a majority(~75%) of substrate defects are nanometer size pits. These pits are usually created during final surface polishing of the synthetic, quartz glass substrates. This study presents data that indicates cleaning may also induce pits in the substrate surface. These pits are typically 20 nm and larger, and are contained in a circular area on the surface, which is scanned by a megasonic nozzle during cleaning. Concentrated collapse of cavitation bubbles in the areas scanned by megasonic is expected to be one of the main mechanisms of pit creation. The data indicates the existence of a hard surface layer with an estimated thickness of approximately 30 to 60 nm, which is resistive to pit creation. After this layer is removed, the number of pit defects present on the substrate increases dramatically with megasonic cleaning. It is also demonstrated that, within the detection limits of the atomic force microscope (AFM), the size of a pit does not change due to cleaning.

Hydrogenated water application for particle removal on EUV mask blank substrates

The capability of hydrogenated water to clean EUV blank substrates was examined. The hydrogenated water cleaning process was compared with an H2O2/NH4OH/H2O mixture (SC1) and ozonated water cleaning processes. A small amount ammonia added to the hydrogenated water improved the particle removal efficiency. The concentration of hydrogen and the method used to dispense the water had little effect. The use of ozonated and hydrogenated water together gave high particle removal efficiencies, which were similar to those obtained using SC1. Additionally, the use of ozonated water with hydrogenated water further reduced the amount ammonia required to achieve high particle removal efficiencies. With further process optimization hydrogenated and ozonated water has the potential to replace SC1 in cleaning EUV substrates.

The effect of size and shape of sub-50 nm defects on their detectability

The capability of SEMATECHs Lasertec M7360 inspection tool to detect particles of different sizes and composition was studied on the surface of fused silica and MoSi multilayers (MLs) with a Si cap layer. Particles of Au, Ag, SnO 2 , Fe 2 O 3 , and Al 2 O 3 were deposited and inspected 10 times with the M7360. Tool pixel size histograms were used to calculate the average pixel size per particle category. The calibration curves of pixel size for polystyrene latex (PSL) spheres were used to convert the average pixel size to the optical size of the defects as detected by the M7360. Selective sets of each category of particles then were reviewed by atomic force microscope (AFM) to calculate the sphere equivalent volume diameter (SEVD) of the particles. The contribution of the surface on which particles were deposited and defect composition and shape were studied. Our results indicate that for Fe 2 O 3 and SnO particles, size distribution on the surface of fused silica and MLs is similar and no effect of the substrate was observed. The AFM-measured SEVD size of particles were close to the nominal size of particles specified by the particle supplier. Optical size of particles were found to be larger or smaller than SEVD size for the different particles. In the case of the Au particles, the PSL equivalent optical size was found to be larger than the SEVD in good agreement with the modeling. By using prefabricated rectangular defects on a fused silica surface, we showed that the M7360 differentiates between the PSL and SEVD size of prefabricated defects. The PSL size is smaller than the SEVD size of prefabricated defects for particle sizes below 100 nm.