Welcome to…
QuantumControl
at the Institute for Gravitational Physics in Hannover!
Our scientific mission:
The control of complex
(quantum) systems
We are part of the Institute for Gravitational Physics at Leibniz University Hannover.
Our team works mainly on the continuous measurement and stabilisation of (mostly quantum optical) systems exhibiting noise at or below the quantum level.
We apply these techniques in the field of interferometric gravitational wave detection and quantum technologies.
We are part of the Institute for Gravitational Physics at Leibniz Universität Hannover, which collaborates closely with the Max Planck Institute for Gravitational Physics (Albert Einstein Institute). We also have lab space in the HITec building.
Our group was founded in July 2010 within the Centre for Quantum Engineering and Space-Time Research (QUEST), which now has the status of a faculty (QUEST-LFS).
We are part of the DFG-funded clusters of excellence QuantumFrontiers and PhoenixD.
We are involved in the German Center for Astrophysics (DZA) and the Quantum Valley Lower Saxony (QVLS).
Our team is involed in the LIGO Scientific Collaboration (LSC) and the Einstein Telescope (ET).
Our research is mainly on squeezing and optomechanics
as well as metasurfaces and machine learning.
Coherent Quantum-Noise Cancellation (CQNC)
Optomechanical systems, such as the LIGO gravitational wave detectors, are fundamentally limited by the Standard Quantum Limit (SQL), caused by the interplay of quantum backaction and shot noise. Coherent Quantum-Noise Cancellation (CQNC), as proposed by M. Tsang and C. Caves [Phys. Rev. Lett. 105, 123601 (2010)]., offers a way to surpass the SQL across all measurement frequencies.
Based on this idea, we have carried out a detailed analysis of an all optical CQNC scheme and assessed its advantages, requirements and limitations [Maximilian H. Wimmer et al., Phys. Rev. A 89, 053836 (2014), Jakob Schweer et. Al, Phys. Rev. A 106, 033520 (2022)]. The conclusion of these analyses is that even under realistic (non-ideal) requirements CQNC is possible.
Our approach introduces an all optical “effective Negative Mass Oscillator” (eNMO) placed in series with an optomechanical system. The eNMO will be operated in series with the optomechanical oscillator and cancel the quantum backaction noise by mimicking the optomechanical system’s dynamics with an opposite phase, thus demonstrating coherent quantum noise cancellation. To achieve perfect CQNC, the characteristics (such as transfer functions and their coupling strength) of the optomechanical system and the eNMO have to be matched precisely as possible [Jakob Schweer et. Al, Phys. Rev. A 106, 033520 (2022)]. To match this evaluated CQNC criteria as closely as possible the first steps towards CQNC are to build and characterise the two subsystems independently before they are combined.
Optomechanical System
To cancel quantum backaction noise via CQNC, we first need a system susceptible to such noise. We will use a MatE cacvity (membrane at the edge), in which a high-quality soft-clamped, high-quality silicon nitride membrane is placed in a Fabry-Pérot cavity. One advantage of the MatE system is that it allows precise tuning of the optical linewidth by adjusting the membrane’s position, making it possible to match the eNMO’s linewidth. When the MatE system is operated in high vacuum and at cryogenic temperatures (<4 K) gas damping and thermal noise are sufficiently reduced so that the soft clamped membrane is susceptible to quantum radiation pressure noise.
In our group, the characterization of optical and mechanical parameters via optomechanically induced transparency (OMIT) and dynamical backaction (DBA) is already well established at room temperature and can be easily transferred to the cryogenic operation of the MatE system. Once this is achieved, the quantum backaction limitation of the MatE system will be verified through a measurement of ponderomotive squeezing. After demonstrating the QBA limitation, the system will be coupled with the eNMO, enabling coherent quantum backaction noise cancellation.
The “effective negative mass oscillator” (eNMO)
The eNMO cancels out the QBA noise by creating a tailored, frequency-dependent squeezed state called “inversely squeezed state”. This is achieved by mimicking the Hamiltonian of an optomechanical system consisting of a beam-splitting process and a down-conversion process.
In our group, we have done this with a novel all-optical approach using the “Squidy” cavity. This is a coupled cavity system with a two-mode squeezing crystal. We have also developed quantum tomography to visualize the frequency-dependent squeezed ellipse for each frequency. Matching the parameters and coupling the Squidy cavity with the optomechanical system will be the next step towards CQNC.
Metasurfaces
The motivation for our work on highly reflective metasurfaces stems from gravitational wave astronomy. Here, interferometers are used as measuring devices whose performance is partially limited by the thermal noise of the individual components, especially in the low-frequency range. Even at cryogenic temperatures, as planned for the Einstein Telescope, the thermal noise of conventional multilayer coatings of the silicon mirrors represents a significant limitation. At the same time, the absorption of these coatings may only be in the sub-ppm range for the wavelength of 1550 nm under consideration. For this reason, we have developed highly reflective coatings based on meta-surfaces as part of PhoenixD.
Our simulations are based on periodically arranged silicon nanospheres on a sapphire substrate, which behave like a highly reflective meta-surface. In our publications, we have shown the dependence of the reflective properties on the manufacturing tolerances, as well as the influence of the substrate and the polymer protective film in which the three-dimensional nanostructures are embedded. A theoretical reflectivity of 100% for the design wavelength can be achieved by assuming the exact arrangement of the nanostructures and neglecting material losses. The interaction between the electromagnetic field and the electric or magnetic dipole resonances of the Si nanoparticles in the metasurface is utilised.
The nanospheres in question can e.g. be produced by laser printing. However, with this method it is difficult to ensure a uniform size and precise positioning of the individual nanoparticles. We have, therefore, investigated electron beam lithography as an alternative. Compared to laser printing, the freedom of geometry is limited here, so we are investigating structures with vertical walls, such as cylinders, as an alternative to spheres. By successively adjusting the diameter and height of the nanocylinders and the periodicity of their arrangement in the array, we were able to design the metasurface based on simulations in such a way that the electrical and/or dipole moment of the cylinders is in resonance with the wavelength of the light. The geometric degree of freedom, which is increased by one compared to nanospheres, allows us to adjust the phase of the reflected light from -π to π by tuning the electrical and magnetic resonance. Even when embedding the nanostructures in a protective layer with a refractive index of 1.4, perfect (100%) reflection can be achieved according to our simulations. The approach we have developed can be used to tune the perfect mirror effect at different wavelengths in the spectral range in which the absorption of the starting material is negligibly small.
Machine learning for (quantum) optical experiments
Optical experiments typically require various control loops to stabilize the experimental degrees of freedom against perturbations. These locks significantly increase the complexity of the experiments. Furthermore, precise alignment is needed, which can be very time-consuming. As machine learning can be a useful tool to tackle complex and time-intensive problems, we are researching how to implement these techniques to improve our experiments.
We investigate, for example, how reinforcement learning can help streamline alignment work and, therefore, free up time. One goal here is also to make academic work less stressful. Another topic is to investigate how machine learning can help to find new control strategies where traditional methods fail, for example, when we want to use unconventional error signals or have unpredictable noise sources.