What fascinates me the most is the ability to bring the different fragile systems together to build a larger system that can do none of the components can do on its own. That's why I spent a lot of time playing with different physical systems over the past years.
The following is an overview of these systems and brief summaries of my contribution in these fields. For a full list see Google Scholar or NASA ADS (so far both seem to capture citations differently)
Certain types of electrical circuit made of superconducting material, when cooled down past its superconducting critical temperature down to the milliKelvin regime, become the so-called superconducting circuit. Its macroscopic parameters such as current and voltage do not behave like the classical parameters we are used to in our daily life, instead they behave quantum mechanically. Such system turns out to be one of the most promising platforms to realise a quantum processor.
In this work at Chalmers, we were working towards building a highly performing multi-qubit processors based on superconducting circuits.
Link to the paper: https://arxiv.org/abs/2403.00285
Published at PRX Quantum with an accompanying feature story by the APS Physics magazine, check it out!
People used to say (and I think it is still the case today) that superconducting qubits are nice because they can be individually controlled and (if needed) tuned to have the parameters we need. This means that imperfection in nanofabrication and naughty little defects that keep our qubits from getting the nice coherence they are supposed to have, can be avoided during the qubit operation. While this is nice for physics experiment, this is slightly problematic for scaled-up quantum processor. If you have N qubits, without any additional smart tricks, you will need more than N control lines to control the system. This is a problem because we don't have much space within a chip, and putting as many control lines as needed within the already tiny space we have in the chip might lead to increased crosstalk, which is scary. But before we start tackling this issue, we need to know what is the expected level of crosstalk within such system and if it is actually any good at all to start with. In my latest paper, I answered this question for a small-scale quantum processor.
Link to the paper: https://arxiv.org/abs/2112.02717
To realise the full promise of a quantum computer, i.e., being able to solve useful problems that are intractable by even the most powerful classical computers we can make, we need a big quantum computer. It is estimated that with the current physical error rate in superconducting circuits, with the aid of a conventional quantum-error correcting scheme, one needs around 100K - 1M qubits. That's a LOT of qubits. This is not even mentioning that we need to be able to bring in control and readout signals to each of these qubits and not to mention possibly needing signal lines to create the interaction between them. So to cut the story short, we don't just have to deal with many qubits but also the signal delivery lines.
Conventional quantum chips dealing with few-qubits devices have the signal lines and the qubits on the same layer, so one can imagine that you'd need to find ways for the signal to 'jump' over the quantum elements on the chip if the signal lines are to reach qubits that are deep within the array. This is not to mention that the qubits are really prone to interactions and thus if careless routing will lead to an increased crosstalk.
So learning from the semiconductor packaging industries, we leverage the space available on top and below the chip, separating the system into multiple layers, thus the name 3D integration technologies (in contrast to the so-called 2D planar technologies). This allows us to separate the signal layers with the qubit layers, potentially alleviating the 'signal criss-cross' issue and the crosstalk issue as well. As a first step, we separate the system into two layers, which in this paper, we call the Control chip and the Qubit chip. We fabricated them on two chips, and flip one of them, and do the so-called 'flip-chip' bonding, in our case thanks to our Finnish collaborator at VTT Finland. Chalmers and VTT are not the first one to do this. The UCSB group and the MIT Lincoln Lab group are the two groups that explored this technology for quantum compatibility earlier.
In this paper, we demonstrated a performance-preserving flip-chip packaging technology, showing qubits with T1>90us, F(1Q)> 99.9%, F(2Q, CZ) > 98.6%, i.e., the introduction of a secondary chip and the additional fabrication process does not degrade the quality of the qubit system. We talked in a lot of details about our design strategies and how we think we will attempt to scale this system to a larger device.
Other works in which I've secondary contributions:
Simplified Josephson-junction fabrication process for reproducibly high-performance superconducting qubits
(Link to paper: https://arxiv.org/abs/2011.05230)
Qubit-compatible substrates with superconducting through-silicon vias
(Link to paper: https://arxiv.org/abs/2201.10425)
Transmon qubit readout fidelity at the threshold for quantum error correction without a quantum-limited amplifier
(Link to paper: https://arxiv.org/abs/2208.05879)
Mitigation of frequency collisions in superconducting quantum processors
(Link to paper: https://arxiv.org/abs/2303.04663)
Fast analytic and numerical design of superconducting resonators in flip-chip architectures
(Link to paper: https://arxiv.org/abs/2305.05502)
We all know of magnets, they are stones that can attract or repulse each other. They have been object of fascinations for centuries, and are highly useful in various parts of our life and industry. Perhaps the nearest magnets we can get hold of are the magnets we put on our fridge's door. Who might have guessed that there's so much more than simply being a fridge magnet. Like any other material, magnetic materials are composed of molecules / atoms that each possesses a magnetic property called spin. Turns out that certain types of magnetic materials, when put under the right condition, can exhibit a phenomenon called spin waves or magnons. Essentially, when an electromagnetic radiation is sent into such system, the spins react to this disturbance, and because of the sizeable interactions between neighbouring spins, this disturbance is propagated across the whole magnetic material, forming so called spin waves.
The fascinating property of spin waves don't just stop there, as a matter of fact, it has one of the most interesting dispersive properties with frequencies extending from GHz to THz, it is geometry- and bias-field dependent (bulk, films). This means that one can also think of a variety of applications for spin waves. As a matter of fact, it was one of the leading candidates to implement classical computer back in the 60s, although the industry eventually chose Silicon as the way to go. Recently, magnon-based computing has been resurfaced as one of the candidates to implement beyond CMOS computing technology.
In these works, we are interested in the interface between such microwave magnonic systems and superconducting circuit. In particular interest is the use of planar architecture in both the magnetic and the photonic counterparts. Basically, can we bring some of the interesting magnon-based applications in the microwave regime for the superconducting circuit system? Another possible path would be ask if we can use the quantum measurement toolbox that has been developed for superconducting circuit platform to probe the physics of magnons at the quantum level.
The works performed in this field formed the bulk of my DPhil thesis in the group of Alexy Karenowska at the University of Oxford.
Link to paper: https://arxiv.org/abs/1610.09963
Phenomenon of strong coupling in the context of circuit- or cavity- QED is the idea that two systems that are compatible but not necessarily the same are able to exchange excitations or energy faster than the rate at which the energy leak out to other channels/environment. It has led to a huge amount of interesting works spanning different fields for the last few decades. While it is not necessarily a quantum-mechanical phenomena, it is of special importance for the circuit- and cavity-QED community as demonstrating strong coupling between two quantum systems opens up a way to, for instance, perform coherent control of a quantum system via the other system. While it is not a sufficient condition, for many, this is usually one of the first goals when trying to demonstrate the feasibility of hybridizing two very different systems.
Prior to this paper, most of the works in demonstrating strong magnon-photon coupling have been performed with a YIG sphere inside a 3D cavity. The reason why this is preferable especially in the context of demonstrating the so-called quantum magnonics regime, where the magnonic and the photonic counterparts are able to exchange a quantum excitation, is because it is possible to separate the region with high magnetic field (needed to excite a magnon) with the region of high electric field (needed to excite the superconducting qubit). The separation is crucial as standard superconducting qubit that is made of Aluminium structure can only remain superconducting up to a very low magnetic field threshold and only under certain field orientation.
In the spirit of our pursuit for anything related to the planar architecture, we were interested in putting both the superconducting circuit structure and the magnonic systems in a region with high magnetic field. This is a very challenging experiment as superconductivity is easily destroyed by the presence of magnetic field. In this work, we successfully demonstrated the use of a resonator in a planar architecture made out of Niobium film strongly coupled to a YIG sphere under the presence of a sizeable in-plane magnetic field (~100mT). We also demonstrated that the highly non uniform field of the resonator is able to excite higher order magnetostatic modes of the YIG sphere, a feat which is not easily achieved in the context of a 3D cavity. This work also opens up the possibility to demonstrate the direct coupling between a planar superconducting qubit structure to various geometries of YIG system.
Link to paper: https://arxiv.org/abs/1711.00958
One of the basic building blocks of the computing systems we can find all around us today is a transistor, typically made out of Silicon. Silicon has the so-called energy bandgaps - electrons with energy in this range are unable to propagate in pure Silicon crystal. The concept of bandgap soon found itself re-discovered in different contexts, for instance in photonics. The idea is that excitations in a periodic structure are coherently scattered by the periodic potential of the structure. These scatterings interfere with each other, and become prominent when the energy of the excitation lies within the bandgaps. Physicists often called them artificial crystal structures in reference to the crystalline structure in which this effect was first discovered.
It turns out that it is possible to create artificial crystal structure for magnons whose bandgaps are in the microwave regime. It is called the magnonic crystal. Due to the highly tunable dispersion relation of magnons, it is also possible to shift the bandgap in frequency and to switch it on or off on demand. For instance, the periodicity can be created by periodically corrugating the magnetic film thickness, width, saturation magnetisation, or essentially any physical parameter that can influence the dispersion relation. In addition to this, the position of the bandgap can be shifted across a wide range of frequency in real time by tuning the magnetic field. It should be mentioned that such ability is harder to implement in the case of a photonic crystal.
Such device can be a useful addition to the control and measurement toolbox of microwave quantum circuits at millikelvin temperatures. This motivated us to pursue this line of research, by performing the first measurement of such magnonic crystal at such temperature environment in order to investigate any possible problem that might arise. Through systematic investigations, we were finally able to measure the magnonic bandgap. However, our measurements immediately revealed an insight into an issue that can prevent such device from being practical at such low temperature environment. This became the subject of investigation for the next paper.
Link to paper: https://arxiv.org/abs/1903.02527
If magnon-based quantum technologies at millikelvin temperatures are to be useful and practical, it is necessary that the medium in which the magnons are excited possess the lowest damping possible. This is even more important in the context of quantum technologies as a quantum state is prone to dissipation, and maintaining the lowest dissipation is therefore in our utmost interest. Yttrium Iron Garnet (YIG) is the usually the go-to material to study magnon dynamics and prototyping devices, this is due to YIG possessing by far the lowest magnon damping among all materials.
In our prior works, we noticed the increased magnon damping in YIG films at millikelvin temperatures. In this work, we performed systematic studies of magnon damping in YIG films at millikelvin temperatures. The studies revealed the presence of two additional damping channels that are present in YIG films at millikelvin temperatures. The first one is due to the two-level-system (TLS) impurities that are ubiquitous among other solid state systems at low temperatures, including the superconducting circuit system. The TLS-related damping has also been observed in bulk YIG system. The second source is due to the substrate on which high quality YIG films are grown, which is Gadolinium Gallium Garnet (GGG). In this work we have also shown that by removing the substrate, it is possible to reduce the damping in YIG down to be much lower than the one at room temperature, and comparable to what is observed in bulk YIGs. The results and insights in this work are hoped to provide direction towards getting high quality magnonic quantum technologies based on YIG films.
We were taught at school that the physical matter we can see and touch around us is actually a huge collection of different atoms. Atoms are extremely tiny, typically about few Angstroms in size, and therefore is very hard to isolate, let alone manipulating it, and not to mention that it jiggles around because it is hot. It took many generations of PhDs and decades of advances in various fronts of research to finally being able to cool down an atom or an ensemble of atoms to the temperature so low that we can start isolating and manipulating them. They are called cold atoms.
It turns out to be a clean and useful platform to test various predictions of theories and models, such as the Two Level System model, which is what we call the qubit these days. It is undoubtedly one of the systems that have motivated this idea of quantum technology. Today, one of the leading candidates for implementing a quantum computing system is the trapped ion system, coming from the cold atom family.
There are generally two types of cold atoms: ions and neutral molecules. In my work I have dealt with trapped neutral atoms, specifically a single atom system trapped in a dipole trap and an atomic cloud system. They have been the subject of collaborative works for about two years where we demonstrated an 'effective' non linear interaction between single photons generated by two disparate atomic systems, as well as a velocity sensor based on atomic cloud. The former demonstrated one of the main requirements for the so-called Linear Optical Quantum Computing paradigm. The latter is what we would categorise as a quantum sensor in today terminology.
Link to paper: https://arxiv.org/abs/1504.00818
Circa 2013, the idea of seriously pursuing the quantum computation route is not as widespread as today, and it was relatively unclear which physical system (photon? trapped ion? neutral atom? quantum circuit? nmr system?, etc. ) would be the best one to use. The idea of implementing quantum computational system based on photons is very interesting because we can make photon travels over very long route with relatively low decoherence, and they interact rather weakly with other photons in vacuum. So one can imagine the situation where a network of different quantum systems around the world being used to in a way that makes the most out of their own advantages to perform quantum computation/communication. However, its extremely weak non linearity also makes it very hard to generate interactions that are required to entangle systems, this is why photons were somewhat not that attractive to implement quantum computation, perhaps until very recently where a number of companies became more serious with this idea.
There is a quantum computational paradigm called the linear optical quantum computing (LOQC) using photons. It proposes to generate a an "effective" interaction between two single photons using the so-called Hong-Ou-Mandel effect/the two-photon interference. It's a quantum interference phenomena. In this work, together with three other PhD students, we demonstrated the single photon generations from two different atomic systems: a trapped single atom, and a cold atomic ensemble. Both systems have a different photon generation scheme and are located in two different rooms. Therefore we have to tailor the shape of each photons, transport one of the single photons through a long low-loss optical fibre to another room, and set the timing so precisely that the two single photons can meet each other. Because the single photon qualities from both systems were really good, we were able to successfully demonstrate the two-photon interference effect between both single photons, and thus the compatibility between the two systems.
This work is the subject of my MSc thesis, in the group of Christian Kurtsiefer at the National University of Singapore.
Link to paper: https://arxiv.org/abs/1610.03203
In this work, we were playing with a cloud of cold atoms, instead of a single atom that was the subject of the previous paper. Sensing gravitational acceleration is a relatively established technique for the atomic cloud, and has been used to measure gravity with very high precision. In this work, we had a commercial cold atom system, in house optics, and were interested to use the atomic cloud to sense its own motion. The atomic cloud is put in the so-called electromagnetically-induced transparency (EIT) state, which is essentially state where the atomic cloud appears transparent to optical light that would otherwise be absorbed. When an optical pulse is sent into an atomic cloud that is then put in an EIT state, you can effectively slow down the light. If the atomic cloud is moving, the light is also dragged along with it. It is this light dragging effect that is used to sense the motion of the atomic cloud as demonstrated in this work.
If you only think of a diamond as an expensive jewellery gift, than you have not seen it all. Diamond is one of the hardest (if not the most) materials in the world that is extremely useful in basic industrial manufacturing processes for cutting, polishing, etc. But on top of that, the inside of a diamond holds little secrets that become extremely useful for quantum technology. Turns out that a diamond is not a perfect arrangement of Carbon atoms, as with other materials, there are defects that come in various forms, one of the most popular defects in the quantum world is the Nitrogen-Vacancy (NV) center, which consists of a Nitrogen atom impurity (instead of a Carbon atom) in one of the lattice sites next to a vacant sites, surrounded by Carbon atoms in its surrounding. The electronic distribution around this defect acts as if as a whole it is an 'artificial atom' with its own energy level. Turns out that the energy level distribution is such that: 1) it can be manipulated with microwave and optical radiation at room temperature, 2) there is an addressable two-level system (a.k.a a qubit), 3) it possesses respectably 'long' relaxation and coherence times (compared to similar systems at room temperature). Apart from being useful for quantum information processing applications, NV centers (and in fact other similar defect centers) can be useful as a quantum sensor (e.g. magnetic field, electric field, mechanical pressure, temperature, etc.) because it's such a simple system.
Link to paper: https://arxiv.org/abs/1511.08175
If one is to benefit from the use of NV centers as a basic quantum information processing element or as a sensitive quantum sensor, it is imperative to understand how its properties change when subjected to various environmental conditions. In this work, we study the coherence property of native and implanted NV centers in various diamonds (bulk and nanodiamond) at low magnetic field and electric field. By elucidating the competition between both fields in influencing the coherence properties of an NV center, this work sheds new light into the performance optimisation of an NV center for both quantum information processing, metrological, and sensing applications.
A part of this work is the subject of my research internship work in the group of Vincent Jacques at Ecole Normale Supérieur de Cachan. The work was awarded the Prix du Stage de Recherche by Ecole Polytechnique in 2013.
Link to paper: https://arxiv.org/abs/1406.5077
An NV center, if properly conditioned, can be used as a very sensitive magnetic field sensor. it can also be used to probe the nuclear spin of surrounding Carbon atoms, which are usually really weak and are thus often buried within the myriad noises present in the system. In a diamond lattice, the environment around a native NV center is 'quite clean' to the point that the NV center can sense the surrounding Carbon atoms nuclear spins even if the signals are really weak. In this work, we demonstrated the use of an NV center as a probe for these fields at room temperature.
One-time-pad encryption is able to provide truly secure communication between two parties, provided that the secret key used in the encryption is truly secure. Classical implementations of one-time-pad involve generating random key through schemes whose securities are not guaranteed and are typically only based on certain hard-to-compute assumptions. Quantum Key Distribution (QKD) can be used to generate a 'truly' random key for the one-time-pad. Another way to say it is that the generated randomness is guaranteed by quantum physics. One of the first and well-known QKD protocol is BB84 by Bennett-Brassard; it encodes quantum states in either one of the two non-orthogonal basis: Z and X. In my early days of undergraduate (circa-2010) when I was first introduced to this subject by Valerio Scarani, he was interested in whether the X and Z basis are the most 'optimal' configuration to give secure key, and I was tasked to investigate this using the information-theoretic security proof protocol that he and others have developed. The final conclusion was that the X and Z basis are indeed the most optimal configuration to use, in accordance with Valerio expected in the first place. I have also extended the calculation for the SARG protocol and reached a similar conclusion.
Circa-2012, the only commercially-available QKD machines are based on transmission via fiber optics. The operating distance (between two parties) are limited to below ~100km, in order to still have non-zero (and non-negligible) secure key rate generation. This is partly due to the non-zero attenuation in the standard fiber optics used in those days. Another issue is the fact that well-known and standard QKD protocols like BB84 require a true single photon source, which is a stringent requirement on the machine itself. The latter exposes the system to possible quantum attack that may make QKD not secure as promised. One of the remedies is to use the so-called decoy state technique, which apart from extending the operating distance by few tens of percent, can also protect the QKD system from certain quantum attacks. I was tasked at that time to evaluate the existing decoy state techniques available at that time, and examined the possibility to implement it into an existing commercial QKD product. I was able to provide certain evaluations and took the first step into integrating it with the existing QKD product.