Chip-scale atomic magnetometer

  • Savino Piccolomo

Student thesis: Doctoral Thesis


This thesis describes the physics and technology adopted to implement a Bell-Bloom type atomic magnetometer at Strathclyde University. We show that the device is a functional experimental prototype, useful as the primary reference for future chip-scale miniaturization. The sensor head and core of the device is a small (approximatively 20 mm3) cubic cell derived from a silicon wafer structure and filled with a caesium azide (CsN3) compound. When the wafer is exposed to UV light, the azide is dissociated in its components; in this way the cells are filled with caesium atoms in the vapour form, which constitute the magnetically sensitive elements of the device, and nitrogen gas, used to optimize the performances of the sensor head. We are able to characterize a whole wafer in a relatively short period of time, in terms of caesium vapour and nitrogen gas pressures, and for this purpose, we developed a data analysis tool based on caesium spectroscopy. In the Bell-Bloom scheme the magneti c resonance is not excited by RF coils but by optical modulation of the input laser light: the laser could be modulated in frequency, amplitude (i.e. intensity) or polarization but we adopted the amplitude modulation scheme because it requires the use of only one beam and an overall simpler set-up; it also allows the possibility of directly detecting Larmor oscillations. In the basic experimental set-up, a light beam from a diode laser, tuned to the D1 line of the caesium spectrum, is elliptically polarized and amplitude modulated; the light is then focused in and out the cell, through its opposite glass apertures, and the transmitted light is detected. The sensor head is enclosed inside a magnetic shield where the desired (lownoise and spatially uniform) magnetic field is generated by two orthogonal pairs of Helmholtz coils. The cell is heated to reach the desired caesium vapour density with a non-magnetic resistor: its current is switched on and off and the signal is detected onduring the off part to avoid undesired magnetic noise. The laser intensity is cycled from a short and intense impulse, called "pump", to a longer and weaker beam, named "probe". The first is used to align the spins of the caesium atoms, so that the macroscopic magnetization precess around the magnetic field, while the second detects this so-called Larmor precession. The two orthogonal polarization components of this oscillation are found to be almost out of phase which justifies the use of a polarimeter to separately analyse their properties and investigate the possibility of signal subtraction to improve the signal-to-noise ratio. In order to describe and optimize the experiment, the main experimental parameters (polarization, pump and probe intensities, sensor head temperature, laser frequency, magnetic field intensity and direction) have been analysed extensively in their effect on the oscillating signals. We observed that the Larmor oscillations are damped and a pseudo magnetic field is generated by the laser, and that both are proportional to the probe intensity. This research also presents a model that successfully explains the main experimental observations and extends the common representation in current literature by further predicting the best laser frequency and input polarization and the out of phase phenomena previously described. The model uses the multi-pole expansion of the density matrix, truncated to its first order, and gives a satisfactory vectorial representation of the relationships between the main experimental parameters. Finally, we have reached a complete noise characterization of the apparatus in the frequency domain. A data analysis tool is able to fully describe the performances of the magnetometer in terms of its sensitivity relatively quickly. A sensitivity lower then 5 pT/ p Hz at 850 Hz has been achieved which, if reproduced on a miniaturise
Date of Award3 Oct 2016
Original languageEnglish
Awarding Institution
  • University Of Strathclyde
SponsorsUniversity of Strathclyde & EPSRC (Engineering and Physical Sciences Research Council)
SupervisorErling Riis (Supervisor) & Aidan Arnold (Supervisor)

Cite this