Novel imaging and microscopy techniques have repeatedly revolutionized science, medicine, and industry. Recent examples include cryogenic electron microscopy, which enables us to retrieve the atomic structure of proteins [1], or interferometric scattering microscopy, which allows us to study protein interactions and dynamics on unprecedented scales [2], [3]. These advances became possible by pushing the sensitivity of established techniques, i.e., classical electron and light microscopy, respectively. Alternatively, fundamental advancements can be achieved when imaging new qualities of a specimen. Examples include nuclear magnetic resonance, which, addressing the spin of a sample, enables us to study the structure and dynamics of disordered proteins [4], or scanning tunneling microscopy, which provides atomic details of solid-state samples via coupling to its local density of states [5].
Traditional microscopes, while operating with quantum probe particles (photons, electrons), rely on intensity-based measurements that collapse the wavefunction of the probe particle, effectively disregarding information about the sample. Our vision is to replace classical state preparation and detection technology with a quantum computer inside a microscope to enable quantum-computer-enhanced measurements (QCEM). This will enable measurements at higher sensitivity beyond the standard quantum limit, and the imaging of novel quantities of interest.
Quantum measurement protocols based on manipulating individual qubits have led to advances in inertial sensing, time-keeping, and the measurement of electromagnetic fields [6]. Quantum metrology protocols become even more powerful when multiple or non-local properties of a system need to be characterized [7], [8]. Recently, one of us proved an exponential speed-up in machine learning tasks when comparing quantum-computer-assisted learning to the classical processing of measurements on single quantum states[9]. These theoretical results were experimentally verified on Google’s 53-qubit sycamore quantum computer [10].
In electron microscopy, several fundamental science problems are currently intractable:
Electron-induced sample damage limits the maximally allowed electron dose in EM. For biological specimens, the electron fluence is limited to roughly 10 electrons per square Angström. Shot noise then limits the spatial resolution to 2-4 nm. This makes it impossible to retrieve the atomic structure of one single protein, study intrinsically disordered proteins, or dynamically watch biology at work. It is thus crucial to increase the information obtained per electron-sample interaction. And while EM uniquely provides the spatial and temporal resolution required to study coherent excitations in biological systems, their signature is weak. It is currently not possible to deduce it from a few electrons.
