Kuljit Virk Email & Phone Number

Fremont, Ca, Us

Milpitas, California, Us

The scientific challenge in my work is to develop optical detection methods for CMOS device defects that are far below the diffraction limit at wavelengths yielding strong enough interaction to generate signals. My charter in BBP is to develop fundamental understanding of the sensitivity challenges in inspection as device technology evolves. Through in-depth modeling of optical scattering in upcoming devices, I and my team find potential solutions in both the optical and algorithm space and advise management in strategic technology development.I direct a team of PhD physicists that engages with advanced chipmakers in the early phase of each technology node and guides the optical inspection strategies based on the modeling and simulation paradigm I developed in the past 6 years. My team is also actively developing statistical learning-based anomaly detection technology that works by modeling the probability distribution of optical images of patterned wafers.Some published examples:Theoretical techniques that accelerate by 100X - 1000X the computation of scattering matrices of defects embedded in complex periodic media. Published in JOSAB 38, 1763 (2021)Ensemble of neural networks with false detection rate below 1 out of 100,000 while maintaining a true positive rate above 90%. This invention is based on the theory of weak learning. Patent Pending.

Milpitas, California, Us

• Developed and applied techniques for Monte-Carlo modeling of signal to noise ratio in all inspection layers for advanced logic, and new technologies such as gate all around devices.• Established a modeling paradigm within BBP by constructing an HPC ecosystem to aggregate large scale scattering simulations into probabilistic predictions of sensitivity.• Grew a deep multi-year modeling collaboration with Intel’s inspection R&D through technical engagements that demonstrated value of simulation to identify solutions and shorten time to results.• Invented deep neural net ensembles to achieve extremely low failure rate (less than 1E-5) defect detection in SEM images of patterned wafers, eliminating a key risk in the development of SEM review technology to process multi-million sites per hour.

Milpitas, California, Us

Uncovered the fundamental limitations of signal-to-noise ratio in optical inspection of 3DNAND devices. This work played a key role in steering the company's technology development decisions for this market.

San Jose, Ca, Us

In its Fremont facility, Western Digital produces wafers with magnetic read/write heads for hard disk drives. Everyday, the fab produces large amounts of metrology and process data as the wafers move through the manufacturing steps. My work at Western Digital was to create a physics-based data fusion system to predict writer head shapes and enable early detection of process excursions. The writer-head shape is critical to achieving high density storage.At the heart of the data-fusion system was a model that produced the surfaces 3D writer-heads as a function of parameters that could be discerned from analysis of metrology data. An automated analysis of CDSEM images extracted the head shape at the wafer surface, and combined that with metrology data from Ellipsometry, Atomic Force Microscopy, Focused Ion Beam Microscopy, to construct the full 3D writer shape below the wafer surface. I developed and integrated the system into the manufacturing line to create a quasi real-time prediction of the writer shapes over the entire wafer as it passed the key process steps.The new analysis methods and tools I developed for characterizing device variability led to process improvements and enabled detection of process excursions missed in standard metrology on test sites.

Santa Clara, Ca, Us

At Silvaco, I was responsible for advancing its simulation capabilities in quantum electronics, quantum well lasers, photo-detectors, and passive photonic devices. I added to Silvaco's portfolio: * Ab-initio simulations of quantum well lasers with self-consistent band structure computation, electrostatic potentials, and charge density. * Modeling of distributed feedback lasers.* Simulators for organic light emitting diodes, with an effective model of Forster transfer between chromophores. * 2 dimensional Frequency Domain Finite Difference (FDFD) models along with Coupled Mode Theory to mode light propagation through structured waveguides, and electro-optic modulators. I also validated calculations based on analytical results where possible or corroboration with published literature. I prepared documentation of the implemented models and trained the application engineers for their use.

New York, Ny, Us

• Developed and published an ab-intio (Density Functional Theory and Master Equations) approach to modeling nanostructures coupled to metallic surfaces, with applications to photovoltaics.• Created an exact and efficient method to calculate interfacial polarization in surface coupled nanocrystals, and identified novel charge transfer characteristics based on the approach.• Modeled quantum transport using NEGF in molecular junctions to analyze experimental data• Modeled terahertz light propagation and emission in layered semiconductors, and explained experimental data on radiation emitted by shift currents in quantum wells.

Toronto, Ontario, Ca

• Developed a Keldysh Green functions formalism to derive non-linear optical susceptibility and ultrafast spectroscopy of many body systems. • Invented an ultrafast optical technique to probe Berry phases in semiconductors.• Performed an early study of the decoherence of qubits in Si:P system due to conduction electrons.• Developed a new approach to compute high-field excitation of semiconductors in length gauge based on the concept of Wilson loops in lattice gauge theory. • Primary author of 6 peer-reviewed journal articles and presented at international conferences

Ph.D, Physics, Semiconductors, Optics

Master Of Science, Physics, Semiconductors, Optics

Basc, Engineering Physics