CONTINUOUS WAVE EPR: G-ANISOTROPY IN SELENIUM Motivation For the past forty years, Electron Spin Resonance (ESR) has been used to elucidate chemical structures [CITECITE]. This first study will use continuous wave (CW) ESR to analyse the interaction of selium dopants in a silicon crystal to observe the potential for this and similar dopant qubit systems. Background: Donors in silicon as a qubit system Donor impurities in bulk silicon provide ideal hosts for an associated electron spin Si has a low natural abundance of nuclear spin and a weak Isotopically enriched silicon: a spin vacuum Se^+ Deep donors At low temperature, the electron associated with the donor becomes bound. Selenium is a _deep donor_: there is a large energy gap from the band edge compared to _shallow donor_ qubit candidates such as phosphorus or bismuth. Continuous Wave Electron Spin Resonance Spectra The phenomenon of Magnetic Resonance Lamour precession Classically, a static magnetic field oriented in an axis _z_ exerts a torque on a magnetic dipole resulting in precession about this axis. }{dt} = \gamma \times {0}{0}{B_z} \\ {dt}{\mu_x} {\mu_y}{\mu_z} = {\gamma B_z\mu_y} {-\gamma B_z\mu_x}{0} \Rightarrow = {}{}{}, \quad\quad This frequency of precession ω is termed the _Lamour_ frequency. Quantum treatment and the resonance condition We may define a spin operator $} $ to relate the magnetic moment of an electron by the g-tensor, which gives information about the environment of the electron. } = {2}{\sigma_x}{\sigma_y}{\sigma_z} \\ = -\mu_B \cdot And then define a Hamiltonian operator analogous to the classical case }= -\gamma }\cdot Consider an electron with a wavefunction that will evolve due to Schrodinger’s equation under the influence of this associated Hamiltonian = \alpha(t) + \beta(t)\\ = i{dt} In the static case, this is simply a projection in time about the axis of the field as before. Now, adding a circularly polarised time dependent magnetic field with a frequency ωmw Applying a rotation unitary to our Hamiltonian, we move into the rotating frame and lose the time dependence Rabi oscillations The Bloch equations Experimental Procedure Results Discussion Angular dependent studies: Anisotropic g-factor The Landé g-tensor H_{} = \mu_B \cdot \cdot Background Experimental Procedure Goniometer Design A _goniometer_ is a device to set the position of the sample with angular precision. To do this, _Mathematica_ code was written to communicate with a _Tinkerforge_ stepper motor. This code is dynamic, allowing on the fly changes to the angle of increment or motor drive current. It also allows for feedback: when the motor fails to make it to the desired rotation, the real angle is displayed allowing for fine tuned adjustment. The line shape, resonance field and linewidth agree with previously published spectra READOUT Isotopically Enriched Silicon: A spin vacuum For readout of the qubit in a Kane system, it was proposed that the nuclear spin is transferred to surrounding electrons with spin correlated charge measured using a Single Electron Transistor However, charge fluctuations inherent in the device would couple back to the nuclear spin and introduce an additional source of coherence Optical readout of the nuclear spin state was thought infeasible in bulk silicon as the maximum limit spectral resolution was assumed to be attained. In seeking to redefine the kilogram, the Avogadro project isolated a silicon ingot with 99.995% ²⁸Si purity Karaiskaj showed ²⁹Si atoms, with a non-zero nuclear spin, were a dominant cause of spectral broadening Removing inhomogeneous isotope broadening, as achieved with the Avogadro project to gain pure ²⁸Si, allows sharp linewidth features, such as bound exciton photoluminescence (PL) lines, to be resolved. This is evident from the PL excitation spectrum of ²⁸Si compared to natural silicon by Yang et al. (see Fig. [nat]). [A comparison of natural and isotopically enriched silicon, taken from ] Bismuth donor bound excitons An _exciton_ is an electron hole pair existing in a bound state: in this case coupled to a donor present in the silicon bulk. The donor bound exciton ($}^0}$) is localised to its binding centre and possesses no net translational kinetic energy. This is why the PL peak attributed to $}^0}$ is so sharp and hard to detect in natural silicon. Other processes have kinetic energy according to the Boltzmann distribution and hence display a Maxwell-Boltzmann lineshape. Steger showed that, due to the low concentration of donors in the sample, photon emission events happen very rarely . As should be expected from an indirect bandgap material, the majority of $}^0}$ decay is non-radiative: through an Auger process. $}^0}$ forms an ionised donor and energetic free electrons. Rather than attempting optical readout, it would be much easier to measure the electrons produced and hence collate this to the resonant creation of $}^0}$. Measurement of quantised charge, achieved over one hundred years ago by Millikan, is an accurate and stable technique . In this report, we propose using Auger electron detected Magnetic Resonance (AEDMR) as described by Steger to improve measurement of a bismuth spin qubit system through conversion from spin to electrical charge . [Levels in bismuth and the qubit scheme ] [Levels in bismuth and the qubit scheme ] The bound exciton is a simple system from a spin perspective. The electron is forced antiparallel to the donor electron and forms a spin singlet with no overall contribution. Under the influence of an external magnetic field Bz both the donor and exciton electron will be polarised Sz = −1/2. During the capture of free excitons onto a donor atom to create the bound exciton, it is necessary to perform an electron flip and create a singlet electron pair. The hole is seen as a P-type species with a non-zero orbital angular momentum component giving a total angular momentum of 3/2 J = L + S = 1 + 1/2 = 3/2 Bismuth bound excitons in silicon have been studied since the 1960s (second figure in ) and recently suggested for QIP purposes by Sekiguchi et al. . In bismuth, the hyperfine coupling is a large 1.5GHz and the hyperfine doublet produced at zero-field can be resolved in even natural silicon, unlike other donors (see second figure of ). At higher fields, the six hyperfine splitting of D⁰ ground-state in transition between D⁰ and $}^0}$ are visible. The bismuth donor in this system has nuclear spin I = 9/2 with the exciton splitting we have a system with 60 possible transitions (See Fig. [lev]). Steger et al. have developed a non-contact AEDMR approach, avoiding sample strain and a need for ohmic contacts (see supp. info. ). The device is put between two capacitive plates and a 114 kHz sine wave applied to one plate (see Fig. [aedmr]). The signal coupled through the sample in absence of a resonant signal must be cancelled for small impedance changes in the sample to be monitored. To do this, an impedance bridge is constructed using a phase shifter, an attenuator, and a small capacitor. If the sample impedance is equal to the capacitor, a 180∘ phase shift is applied so that the total signal at the summing point is nil. The sample impedance is not purely imaginary as there will be a free carrier background hence the attenuation stage. The voltage applied to the capacitor must be adjustable in phase and amplitude in order to always obtain a null in absence of a resonance. A lock in amplifier is then used to extract the resonant signal from the carrier. Readout works by applying a tunable single frequency laser at the frequency for each qubit transition. If the donor bound exciton is not present in the |n⟩ state, no decay is produced, and hence there is no current contribution. By comparing the current contribution of different states we may find the relative populations and hence measure the state fidelity compared to the relative populations at initialisation. [Experimental setup for AEDMR, taken from ] CTRL SYSTEM: CHEAP TEC REGULATED LASER SYSTEM FOR HYPERPOLARISATION OF DONOR QUBITS IN SILICON Motivation Rapid accurate qubit initialisation has been reported with the use of optical _hyperpolarisation_. In bismuth systems, a 1080nm wavelength donor bound exciton transition is required to realise this method. The aim of this work is to show hyperpolarisation through the design, fabrication and testing of a purpose-built controller to reliably heat and control a 1060nm laser to enable it to perform this transition. This chapter develops a cost effective controller from the powerful Arduino platform, a plug-and-play Arduino shield, and off the shelf components to provide robust wavelength and optical output. The development is motivated by the price of proprietary laser controllers, and a lack of information about the requirements of such a device. The final design combines a laser and TEC controller in a single box, each priced upwards of a thousand pounds in manufacturer catalogues. Aims and Objectives Background A well characterised initial state can be obtained by _hyperpolarisation_. This selective optical pumping mechanism obtains nuclear spin polarisation in excess of thermal equilibrium The optical pumping mechanism is explained by Steger et al. in an earlier paper: a polarising laser is pressed to one hyperfine line to convert D⁰ into D⁰X which then decays to repopulate both this and the opposite electron spin state By selectively pumping |11⟩ for several milliseconds, we can completely initialise the system in state |10⟩, for example. The optically enhanced electron rate W and cross-relaxation rate R are labelled on the right picture in Fig. [lev] Use of these relaxation processes complete initialisation. The polarising laser used by Steger at al pumps D⁰ from |2⟩ to |3⟩. Simultaneously, a radio frequency NMR pulse is applied to pump the D⁰ population in |1⟩ to |2⟩. A low power _readout_ laser pumps the population of |4⟩ into |1⟩. The technique takes around 100ms to completely populate state |3⟩. System Design The Cheap TEC Regulated Laser (CTRL) system develops a device to monitor and control the power and wavelength of a laser diode through the use of a MATLAB graphical interfaced with an Arduino AVR microcontroller. The section describes the entire system and design decisions taken throughout the prototyping process. The entire unit has been designed, in this instance, to interface with a Thorlabs TCLDM9 thermoelectric cooler (TEC) on which is mounted a L1060P200J 1060 nm laser diode. Two 9-pin plugs control the TEC and laser power. The recommended TEC and laser controller _starter set_ from THOR labs costs over £1.5k in addition to the £300 TEC mount unit. In contrast, the Arduino board is £30. This alternative, including circuitry and case, should cost around £100. System Requirements It is crucial this £500 laser diode is powered with a stable controlled current. This current should be stabilised before drawn through the device and transient signals suitably suppressed. Modular Design The software will be written in MATLAB as the. The Arduino will be used as a serial object with data taken from the input pins and written to the output pins. Thermoelectric Cooler Heating the laser diode causes the cavity to expand inducing a positive shift in the lasing wavelength. The unit has two sensors to monitor the temperature: a traditional NTC thermistor and an AD592 chip. The specification sheet for the TCLDM9 indicate two 13W thermoelectric coolers providing up to 20 W of heating or cooling power. Thermoelectric coolers operate through the Peletier effect. A semiconductor junction will dissipate heat across the junction dependent on the direction of incident current. Hence, a bidirectional current source is required. The amplitude of this source ought to be varied to control the heating/cooling power. The THOR lab recommended controller, TED200C provides a maximum ± 2 A supply. A H-bridge is commonly used to supply a variable current supply to a Peletier element. The idea between this element is depicted below. The LM592 H-bridge is used in the Arduino Motor Shield. Shields are plug-and-play modules that sit astride the Arduino chip. A digital pin on the Arduino controls the heating/cooling direction and a PWM sets the intensity of this action. The shield is externally powered by the same source as the laser driver to supply higher currents than the Arduino is capable. Laser driver Voltage controlled current sink An external circuit must be constructed to power the laser safely. The specifications of the laser state that it must have an operating current between 280mA and 300mA. This circuit will be designed to provide a variable current up to 300mA. This design is constrained by the grounding of one side of the laser and the integration of the photodiode. FIG laser pins To power this device, a current sink must be built. The initial prototype is shown below. The current through the laser is set by adjusting the voltage drop over a 10Ω resistor. The op-amp works to ensure the transistor remains in the forward active region VBE = 0.7V. Control signal The _Arduino_ can output a PWM signal that may be smoothed to an equivalent DC level through a low pass filter with a large capacitance. = 15 \Omega \times 10\mu F This PWM signal is sent through an initial op-amp stage to apply the appropriate gain and DC offset. This was calculated using the formula from REF DESIGNING OFFSET AND BIAS IN 30SEC OFFSET AND GAIN FORMULA Safety precautions To operate the laser safely and accurately, several precautions were added. Firstly, the addition of a capacitor in parallel to the laser allows transients signals to be be passed through. Secondly, a Zener diode was added parallel to the resistor to prevent a higher than intended voltage drop over the current setting resistor.. The diode is rated at 3.3 V and will short the resistor if this voltage is exceeded. Thirdly, a single pole double throw switch was attached to ground the laser diode until the rest of the circuit reaches a steady state. External Casing The Arduino and laser driver circuit are mounted in a metal box with holes drilled for the switches (I/O and laser ground) and external connections (USB, D-PLUG, power supply). Both the Arduino and the laser driver are mounted on posts away from the ground plate and screwed fast on four sides. Testing and Refinements It was seen in testing this circuit, a current below the threshold could not be attained. It was assumed this was because the 741 could not take an input as low as negative supply. The LM302 can do this. However, still this chip showed a lower limit of 200 mA. Breadboarding this chip in a simple inverting amplifier showed the output of the chip was always 0.7 V below the value of the negative supply. An alternative solution was implement. A diode was added to sweep up this excess voltage. The diode, once forward biased, takes a voltage of 0.7 V regardless of the current and significantly improved the lower limit to 60 mA. However, the specifications [cite specifications] state a threshold of 40 mA. A second diode resulted in a slight loss of resolution from the PWM but this is a suitable tradeoff for the improved 6 mA lower limit. FIG Ideal diode characteristic. External Casing The Arduino and laser driver circuit are mounted in a metal box with holes drilled for the switches (I/O and laser ground) and external connections (USB, D-PLUG, power supply). Both the Arduino and the laser driver are mounted on posts away from the ground plate and screwed fast on four sides. User interface A Graphical User Interface (GUI) was built in Matlab for fast intuitive control of the ctrl system parameters. The interface is split into three parts, the first to select the COM port and initialise the board, then monitoring for both TEC and lasers. Sliders can increment the PWM values or manually set through text entry. Graphs for the temperature and laser intensity are updated by polling the values of these sensors at second intervals. The interface was designed by using a series of handle calls to update values in simultaneity to the timer object that polls the sensors. Final system cost Optics Beam collimation

loading page

Continuous wave EPR: g-anisotropy in Selenium Motivation For the past forty years, Electron Spin Resonance (ESR) has been used to elucidate chemical structures [CITECITE]. This first study will use continuous wave (CW) ESR to analyse the interaction of selium dopants in a silicon crystal to observe the potential for this and similar dopant qubit systems. Background: Donors in silicon as a qubit system Donor impurities in bulk silicon provide ideal hosts for an associated electron spin Si has a low natural abundance of nuclear spin and a weak Isotopically enriched silicon: a spin vacuum Se^+ Deep donors At low temperature, the electron associated with the donor becomes bound. Selenium is a deep donor: there is a large energy gap from the band edge compared to shallow donor qubit candidates such as phosphorus or bismuth. Continuous Wave Electron Spin Resonance Spectra The phenomenon of Magnetic Resonance Lamour precession Classically, a static magnetic field oriented in an axis z exerts a torque on a magnetic dipole resulting in precession about this axis. \[\begin{aligned} \frac{d\mathbf{\mu}}{dt} = \gamma\mathbf{\mu} \times { \global\colveccount3 \begin{pmatrix} \colvecnext }{0}{0}{B_z} \\ \frac{d}{dt}{ \global\colveccount3 \begin{pmatrix} \colvecnext }{\mu_x} {\mu_y}{\mu_z} = { \global\colveccount3 \begin{pmatrix} \colvecnext }{\gamma B_z\mu_y} {-\gamma B_z\mu_x}{0} \Rightarrow \mathbf{\mu} = { \global\colveccount3 \begin{pmatrix} \colvecnext }{\sin{\omega t}}{\cos{\omega t}}{\mathbb{C}}, \quad\text{defining}\quad \boxed{\omega = -\gamma B_0}\end{aligned}\] This frequency of precession $\omega$ is termed the Lamour frequency. Quantum treatment and the resonance condition We may define a spin operator $\mathbf{\hat{S}} $ to relate the magnetic moment of an electron by the g-tensor, which gives information about the environment of the electron. \[\begin{aligned} \mathbf{\hat{S}} = \frac{\hbar}{2}{ \global\colveccount3 \begin{pmatrix} \colvecnext }{\sigma_x}{\sigma_y}{\sigma_z} \\ \boldsymbol{\mu} = -\mu_B \mathbf{g} \cdot \mathbf{S} \end{aligned}\] And then define a Hamiltonian operator analogous to the classical case \[\begin{aligned} \huge{\mathscr{H} }= -\gamma \mathbf{\hat{S}}\cdot \mathbf{B} \end{aligned}\] Consider an electron with a wavefunction that will evolve due to Schrodinger’s equation under the influence of this associated Hamiltonian \[\begin{aligned} \Ket{\psi} = \alpha(t) \Ket{0} + \beta(t)\Ket{1}\\ \mathscr{H} \Ket{\psi} = i\frac{d}{dt}\Ket{\psi}\end{aligned}\] In the static case, this is simply a projection in time about the axis of the field as before. Now, adding a circularly polarised time dependent magnetic field with a frequency $\omega_\text{mw}$ Applying a rotation unitary to our Hamiltonian, we move into the rotating frame and lose the time dependence Rabi oscillations The Bloch equations Experimental Procedure Results Discussion Angular dependent studies: Anisotropic g-factor The Landé g-tensor \[\begin{aligned} H_{\text{Zeeman}} = \mu_B \mathbf{B} \cdot \mathbf{g} \cdot \mathbf{S}\end{aligned}\] Background Experimental Procedure Goniometer Design A goniometer is a device to set the position of the sample with angular precision. To do this, Mathematica code was written to communicate with a Tinkerforge stepper motor. This code is dynamic, allowing on the fly changes to the angle of increment or motor drive current. It also allows for feedback: when the motor fails to make it to the desired rotation, the real angle is displayed allowing for fine tuned adjustment. The line shape, resonance field and linewidth agree with previously published spectra Readout Isotopically Enriched Silicon: A spin vacuum For readout of the qubit in a Kane system, it was proposed that the nuclear spin is transferred to surrounding electrons with spin correlated charge measured using a Single Electron Transistor \cite{Kane1998} However, charge fluctuations inherent in the device would couple back to the nuclear spin and introduce an additional source of coherence \cite{Fu2004} Optical readout of the nuclear spin state was thought infeasible in bulk silicon as the maximum limit spectral resolution was assumed to be attained. In seeking to redefine the kilogram, the Avogadro project isolated a silicon ingot with 99.995% $^{28}$Si purity \cite{Becker2010} Karaiskaj showed $^{29}$Si atoms, with a non-zero nuclear spin, were a dominant cause of spectral broadening \cite{Karaiskaj2001} Removing inhomogeneous isotope broadening, as achieved with the Avogadro project to gain pure $^{28}$Si, allows sharp linewidth features, such as bound exciton photoluminescence (PL) lines, to be resolved. This is evident from the PL excitation spectrum of $^{28}$Si compared to natural silicon by Yang et al. (see Fig. \ref{nat}). \label{nat}A comparison of natural and isotopically enriched silicon, taken from \cite{Yang2006} \label{ae}Bismuth donor bound excitons An exciton is an electron hole pair existing in a bound state: in this case coupled to a donor present in the silicon bulk. The donor bound exciton ($\text{\textnormal{D}}^0\text{\textnormal{X}}$) is localised to its binding centre and possesses no net translational kinetic energy. This is why the PL peak attributed to $\text{\textnormal{D}}^0\text{\textnormal{X}}$ is so sharp and hard to detect in natural silicon. Other processes have kinetic energy according to the Boltzmann distribution and hence display a Maxwell-Boltzmann lineshape. Steger showed that, due to the low concentration of donors in the sample, photon emission events happen very rarely \cite{Steger2012}. As should be expected from an indirect bandgap material, the majority of $\text{\textnormal{D}}^0\text{\textnormal{X}}$ decay is non-radiative: through an Auger process. $\text{\textnormal{D}}^0\text{\textnormal{X}}$ forms an ionised donor and energetic free electrons. Rather than attempting optical readout, it would be much easier to measure the electrons produced and hence collate this to the resonant creation of $\text{\textnormal{D}}^0\text{\textnormal{X}}$. Measurement of quantised charge, achieved over one hundred years ago by Millikan, is an accurate and stable technique \cite{Mil}. In this report, we propose using Auger electron detected Magnetic Resonance (AEDMR) as described by Steger to improve measurement of a bismuth spin qubit system through conversion from spin to electrical charge \cite{Steger2012}. The bound exciton is a simple system from a spin perspective. The electron is forced antiparallel to the donor electron and forms a spin singlet with no overall contribution. Under the influence of an external magnetic field $B_z$ both the donor and exciton electron will be polarised $S_z = -1/2$. During the capture of free excitons onto a donor atom to create the bound exciton, it is necessary to perform an electron flip and create a singlet electron pair. The hole is seen as a P-type species with a non-zero orbital angular momentum component giving a total angular momentum of $3/2$ \[\begin{aligned} J = L + S = 1 + 1/2 = 3/2\end{aligned}\] Bismuth bound excitons in silicon have been studied since the 1960s (second figure in \cite{Haynes1960}) and recently suggested for QIP purposes by Sekiguchi et al. \cite{Sekiguchi2010}. In bismuth, the hyperfine coupling is a large 1.5GHz and the hyperfine doublet produced at zero-field can be resolved in even natural silicon, unlike other donors (see second figure of \cite{Sekiguchi2010}). At higher fields, the six hyperfine splitting of $\text{D}^0$ ground-state in transition between $\text{D}^0$ and $\text{\textnormal{D}}^0\text{\textnormal{X}}$ are visible. The bismuth donor in this system has nuclear spin $I = 9/2$ with the exciton splitting we have a system with 60 possible transitions (See Fig. \ref{lev}). Steger et al. have developed a non-contact AEDMR approach, avoiding sample strain and a need for ohmic contacts (see supp. info. \cite{Steger2012}). The device is put between two capacitive plates and a 114 kHz sine wave applied to one plate (see Fig. \ref{aedmr}). The signal coupled through the sample in absence of a resonant signal must be cancelled for small impedance changes in the sample to be monitored. To do this, an impedance bridge is constructed using a phase shifter, an attenuator, and a small capacitor. If the sample impedance is equal to the capacitor, a $180^\circ$ phase shift is applied so that the total signal at the summing point is nil. The sample impedance is not purely imaginary as there will be a free carrier background hence the attenuation stage. The voltage applied to the capacitor must be adjustable in phase and amplitude in order to always obtain a null in absence of a resonance. A lock in amplifier is then used to extract the resonant signal from the carrier. Readout works by applying a tunable single frequency laser at the frequency for each qubit transition. If the donor bound exciton is not present in the $|n\rangle$ state, no decay is produced, and hence there is no current contribution. By comparing the current contribution of different states we may find the relative populations and hence measure the state fidelity compared to the relative populations at initialisation. \label{aedmr}Experimental setup for AEDMR, taken from \cite{Steger2012} CTRL System: Cheap TEC Regulated Laser system for hyperpolarisation of donor qubits in silicon Motivation Rapid accurate qubit initialisation has been reported with the use of optical hyperpolarisation. In bismuth systems, a 1080nm wavelength donor bound exciton transition is required to realise this method. The aim of this work is to show hyperpolarisation through the design, fabrication and testing of a purpose-built controller to reliably heat and control a 1060nm laser to enable it to perform this transition. This chapter develops a cost effective controller from the powerful Arduino platform, a plug-and-play Arduino shield, and off the shelf components to provide robust wavelength and optical output. The development is motivated by the price of proprietary laser controllers, and a lack of information about the requirements of such a device. The final design combines a laser and TEC controller in a single box, each priced upwards of a thousand pounds in manufacturer catalogues. Aims and Objectives Background A well characterised initial state can be obtained by hyperpolarisation. This selective optical pumping mechanism obtains nuclear spin polarisation in excess of thermal equilibrium \cite{McCamey2009}The optical pumping mechanism is explained by Steger et al. in an earlier paper: a polarising laser is pressed to one hyperfine line to convert $\text{D}^0$ into $\text{D}^0X$ which then decays to repopulate both this and the opposite electron spin state \cite{Yang2009} By selectively pumping $|11\rangle$ for several milliseconds, we can completely initialise the system in state $|10\rangle$, for example. The optically enhanced electron rate W and cross-relaxation rate R are labelled on the right picture in Fig. \ref{lev} Use of these relaxation processes complete initialisation. The polarising laser used by Steger at al pumps $\text{D}^0$ from $|2\rangle$ to $|3\rangle$. Simultaneously, a radio frequency NMR pulse is applied to pump the $\text{D}^0$ population in $|1\rangle$ to $|2\rangle$. A low power readout laser pumps the population of $|4\rangle$ into $|1\rangle$. The technique takes around 100ms to completely populate state $|3\rangle$. \cite{Steger2012} System Design The Cheap TEC Regulated Laser (CTRL) system develops a device to monitor and control the power and wavelength of a laser diode through the use of a MATLAB graphical interfaced with an Arduino AVR microcontroller. The section describes the entire system and design decisions taken throughout the prototyping process. The entire unit has been designed, in this instance, to interface with a Thorlabs TCLDM9 thermoelectric cooler (TEC) on which is mounted a L1060P200J 1060 nm laser diode. Two 9-pin plugs control the TEC and laser power. The recommended TEC and laser controller starter set from THOR labs costs over £1.5k in addition to the £300 TEC mount unit. In contrast, the Arduino board is £30. This alternative, including circuitry and case, should cost around £100. System Requirements It is crucial this £500 laser diode is powered with a stable controlled current. This current should be stabilised before drawn through the device and transient signals suitably suppressed. Modular Design The software will be written in MATLAB as the. The Arduino will be used as a serial object with data taken from the input pins and written to the output pins. Thermoelectric Cooler Heating the laser diode causes the cavity to expand inducing a positive shift in the lasing wavelength. The unit has two sensors to monitor the temperature: a traditional NTC thermistor and an AD592 chip. The specification sheet for the TCLDM9 indicate two 13W thermoelectric coolers providing up to 20 W of heating or cooling power. Thermoelectric coolers operate through the Peletier effect. A semiconductor junction will dissipate heat across the junction dependent on the direction of incident current. Hence, a bidirectional current source is required. The amplitude of this source ought to be varied to control the heating/cooling power. The THOR lab recommended controller, TED200C provides a maximum $\pm$ 2 A supply. A H-bridge is commonly used to supply a variable current supply to a Peletier element. The idea between this element is depicted below. The LM592 H-bridge is used in the Arduino Motor Shield. Shields are plug-and-play modules that sit astride the Arduino chip. A digital pin on the Arduino controls the heating/cooling direction and a PWM sets the intensity of this action. The shield is externally powered by the same source as the laser driver to supply higher currents than the Arduino is capable. Laser driver Voltage controlled current sink An external circuit must be constructed to power the laser safely. The specifications of the laser state that it must have an operating current between 280mA and 300mA. This circuit will be designed to provide a variable current up to 300mA. This design is constrained by the grounding of one side of the laser and the integration of the photodiode. FIG laser pins To power this device, a current sink must be built. The initial prototype is shown below. The current through the laser is set by adjusting the voltage drop over a $10 \Omega$ resistor. The op-amp works to ensure the transistor remains in the forward active region $V_{BE} = 0.7V$. Control signal The Arduino can output a PWM signal that may be smoothed to an equivalent DC level through a low pass filter with a large capacitance. \[\begin{aligned} \tau_{RC} = 15 \text{k} \Omega \times 10\mu F\end{aligned}\] This PWM signal is sent through an initial op-amp stage to apply the appropriate gain and DC offset. This was calculated using the formula from REF DESIGNING OFFSET AND BIAS IN 30SEC \[\begin{aligned} OFFSET AND GAIN FORMULA\end{aligned}\] Safety precautions To operate the laser safely and accurately, several precautions were added. Firstly, the addition of a capacitor in parallel to the laser allows transients signals to be be passed through. Secondly, a Zener diode was added parallel to the resistor to prevent a higher than intended voltage drop over the current setting resistor.. The diode is rated at 3.3 V and will short the resistor if this voltage is exceeded. Thirdly, a single pole double throw switch was attached to ground the laser diode until the rest of the circuit reaches a steady state. External Casing The Arduino and laser driver circuit are mounted in a metal box with holes drilled for the switches (I/O and laser ground) and external connections (USB, D-PLUG, power supply). Both the Arduino and the laser driver are mounted on posts away from the ground plate and screwed fast on four sides. Testing and Refinements It was seen in testing this circuit, a current below the threshold could not be attained. It was assumed this was because the 741 could not take an input as low as negative supply. The LM302 can do this. However, still this chip showed a lower limit of 200 mA. Breadboarding this chip in a simple inverting amplifier showed the output of the chip was always 0.7 V below the value of the negative supply. An alternative solution was implement. A diode was added to sweep up this excess voltage. The diode, once forward biased, takes a voltage of 0.7 V regardless of the current and significantly improved the lower limit to 60 mA. However, the specifications [cite specifications] state a threshold of 40 mA. A second diode resulted in a slight loss of resolution from the PWM but this is a suitable tradeoff for the improved 6 mA lower limit. FIG Ideal diode characteristic. External Casing The Arduino and laser driver circuit are mounted in a metal box with holes drilled for the switches (I/O and laser ground) and external connections (USB, D-PLUG, power supply). Both the Arduino and the laser driver are mounted on posts away from the ground plate and screwed fast on four sides. User interface A Graphical User Interface (GUI) was built in Matlab for fast intuitive control of the ctrl system parameters. The interface is split into three parts, the first to select the COM port and initialise the board, then monitoring for both TEC and lasers. Sliders can increment the PWM values or manually set through text entry. Graphs for the temperature and laser intensity are updated by polling the values of these sensors at second intervals. The interface was designed by using a series of handle calls to update values in simultaneity to the timer object that polls the sensors. Final system cost Optics Beam collimation

Joe Smith