Counting atoms faster: policy-based nuclear magnetic resonance pulse sequencing for atomic abundance measurement

Rohan Shenoy, Evan Austen Coleman, Hans Gaensbauer, Elsa Olivetti

TL;DR: We train a policy which counts atoms of a given element in a sample more quickly than traditional approaches, by controlling a magnetic field to align their nuclear spins.

Abstract: Quantifying the elemental composition of a material is a general scientific challenge with broad relevance to environmental sustainability. Existing techniques for the measurement of atomic abundances generally require laboratory conditions and expensive equipment. As a result, they cannot be deployed *in situ* without significant capital investment, limiting their proliferation. Measurement techniques based on nuclear magnetic resonance (NMR) hold promise in this setting due to their applicability across the periodic table, their non-destructive manipulation of samples, and their amenability to *in silico* optimization. In this work, we learn policies to modulate NMR pulses for rapid atomic abundance quantification. Our approach involves three inter-operating agents which (1) rapidly align nuclear spins for measurement, (2) quickly force relaxation to equilibrium, and (3) toggle control between agents (1) and (2) to minimize overall measurement time. To demonstrate this technique, we consider a specific use case of low-magnetic-field carbon-13 quantification for low-cost, portable analysis of foodstuffs and soils. We find significant performance improvements relative to traditional NMR pulse sequencing, and discuss limitations on the applicability of this approach.

Lay Summary: Counting the number of atoms of a given element in a sample of material is a general scientific challenge in environmental sustainability. For example, you might want to test whether a food product is exposing you to lead or cadmium poisoning without sending it to a laboratory for testing. Current methods to count atoms generally require expensive and non-portable equipment. As a result, they cannot be used in everyday settings or in remote areas without impractical costs. Measurement techniques based on nuclear magnetic resonance (NMR) are promising to address this challenge because they can target many elements in the periodic table, and they can measure samples of material without destroying or altering them. In this work, we show how to use reinforcement learning in NMR to count atoms faster. We do so by building three inter-operating systems which (1) cause a sample to "chirp" a loud radio-wave signal which is related to the count of atoms in the sample, (2) quiet the chirp to quickly reset the system, and (3) switch control between behaviors (1) and (2) to improve overall performance. To demonstrate this technique, we build an example in simulation to demonstrate its application to low-cost, portable analysis of foodstuffs and soils. We find significant performance improvements relative to the traditional approach in NMR, and discuss the limitations of our method.

Application-Driven Machine Learning: This submission is on Application-Driven Machine Learning.

Primary Area: Applications->Chemistry, Physics, and Earth Sciences

Keywords: NMR, MRI, elemental analysis, atomic abundance, carbon measurement, MRV, RL, magnetization, nuclear spin, pulse sequencing

Submission Number: 11246