Spatial Targetting with Magnets
When an electric current flows through a wire, it creates a magnetic field around the wire. By looping wire into a coil, it is possible to concentrate these magnetic fields within the coil such that the magnetic field strength is high inside and much lower outside. It is this principle that allows us to spatially target gene activation.
Solenoid Magnetic Field (Credit: Hyperphysics)
First, we will design and build thin solenoids for different field types (oscillation frequencies). When we power these solenoids, this means that there will only be a thin region inside where the field strength is high. If calibrated correctly, a magneto sensitive pathways built to be receptive to the applied field type, will only be activated within this thin planar region. Thus within our target volume only cells within a plane have this pathway activated.
Plane Of Activation
This process is then repeated with solenoids and different field types in the two other spatial orientations, each time only activating a single target magneto-sensitive pathway.
2 Plane Activations
3 Plane Activations
Cells can be genetically engineered so that each of these magneto-sensitive pathways triggers the production a distinct biological signal. We can then use a genetic AND gate to combine these signals such that cells only with all 3 magneto-sensitive pathways activated our target gene. Given that these pathways are activated in 3 different planes, only cells at the intersection of these planes (a single point) will have this gene activated.
E. Coli AND Gate Design
Heating Cells with Magnets
Background
For our thermal pathways, we require a way of triggering localised heating within a specific region of cells. This can be achieved using tiny magnetic nanoparticles (MNPs) which are an effective heating agent when subjected to a high frequency alternating magnetic field (AMF). Magnetite (Fe₃O₄) nanoparticles are an ideal choice due to their magnetic properties, biocompatibility and ease of functionalisation for dispersion or targeting to specific tissues.
Néel Relaxation and Brownian Relaxation are two primary mechanisms by which magnetic nanoparticles dissipate energy as heat when exposed to an AMF. Néel Relaxation occurs when the magnetic moment flips within the nanoparticle without any physical rotation of the particle itself. Brownian Relaxation involves the physical rotation of the entire nanoparticle in response to the AMF, leading to energy dissipation due to viscous drag from the surrounding medium. The rate of energy dissapation in these nanoparticles is a function of the nanoarticle size and proportional to oscillation frequency, f and B² (the square of field strength). The graph below shows the contribution of the two mechanisms to the specific heat absorption rate at different particle sizes.
Theoretical estimation of heating rates as a function of particle diameter (Suto et al.)
Stanley et al 2012, found that if you apply a 480kHz, 5mT magnetic field to a 1mg/ml concentration of 20nm Fe₃O₄ nanoparticles, it is possible to achieve a specific absorbtion rate, SAR of 0.63W/g. Based on this, we decided to test two different sources of magnetite nanoparticles: commercial hydrocarbon based ferrofluid from Ferrotec [#EFH1] and PVP coated 15-20nm Fe₃O₄ nanoparticles suitable for water dispersion from US Research Nanomaterials [#US7568].
Ferrotec MNPs (credit: Ferrotec)
US Nano MNPs (credit: US Nano)
Design
AMFs are generated using coils or inductors through which an alternating current (AC) flows. The design and driving circuitry of these coils influences the frequency and magnitude of the resulting magnetic field. Commercial devices for MNP heating are available but we were unable to source one for our experiments due to their high cost. Instead, we designed our own using custom coils, tank capacitors and a low cost zero voltage switching (ZVS) circuit used for induction heaters.
Commerical Magnetogenetics Device (credit: MSI Automation)
ZVS Induction Heater (credit: Cuifati)
A coil, or inductor, combined in parallel with a capacitor is a resonant circuit. When AC is applied to the circuit at an appropriate frequency, the circuit naturally oscillates creating high currents through the coil that generates the magnetic field.
ZVS Circuit Model in LTSpice
The ZVS driver consists of two switching devices (MOSFETS) and control circuitry to manage the switching synchronisation so that when connected to our LC resonator, the resonant frequency is achieved. We initially selected a ZVS driver circuit using IRFR120 MOSFETs however, after some initial design iterations it became clear that these would not be powerful enough to achieve the desired magnetic field. Instead we chose IRPN260 based drivers theoretically capable of 10x higher powers. In reality, these higher power limits cannot be reached due to thermal instability.
Our primary design challenge for thermal activation was to construct a suitable electromagnetic coil. The magnetic field strength and an estimate for the inductance for a given solenoid can be calculated using the following equations:
Plots of the B field equation with different parameters demonstrate the inherent design tradeoff:
As coil radius increases our field strength and spatial precision decreases
As coil length increases our field strength increases but spatial precision decreases
For maximising the field strength, we want to increase the the number of turns and current through our electromagnet and minimise it’s radius. However, the electromagnet must contain a plate for our cells so there is a lower bound for the radius. Similarly, as we scale the number of turns the width of our coil expands reducing the spatial precision of our activation. Using this analysis and further characterisation experiments outlined below we designed, built and tested 7 different coil designs to find one suitable for our purpose. Shown below are iterations of our coil design:
Hover over the designs to see more details:
Characterisation
For each resonant coil design, we first connected an oscilloscope to determine the inductance of the coil and the current. Given the amplitude, \(V_{peak} \) and frequency, \(f\), of the measured output, we can accurately calculate the inductance and current flow:
Picoscope Measurement
In addition to using the simple analysis for magnetic field in a solenoid (Eqn 1), we also used finite element analysis (FEA) to determine the spatial distribution of field strength across our target volume. This could then be used for downstream analysis of heating distribution and pathway activation:
Initial Coil FEA
Final Coil FEA
Final Coil FEA - field strength at r=0
To test the heating efficacy of our designs, we placed a PCR tube containing 0.2ml of each of our nanoparticles at 100mg/ml into our coils and measured the temperature rise every minute. The results are shown below:
Different MNP heating in coil V4
PEFH1 heating in different coils
EFH1 heating with different input currents
Note for the second and third figures, ΔT is relative to water in the same conditions.
By comparing the heating efficacy of MNPs in the same coil it is clear that the EFH1 MNPs are significantly more effective at heat generating than the US Nano water based. We hypothesise this may be due to the different particle sizes between the two samples. Coil V5 was a significant reduction in heating efficacy relative to the tight coil V4. We managed to approach the performance by transitioning to copper tubing based coils like coil V6. As expected, increasing the current through the coils causes a comparable increase in heating efficacy.
Following the process described by Ring et al, we fit our data to the curve below using least squares regression:
\[ T(t) = A(1- e^{-Bt})+C \]We can then calculate the SAR as:
\[ SAR = A_{fit}*B_{fit}*c_p*c_{Fe} \]Where \( c_p \) is the specific heat capacity of the solute and \(c_{Fe}\) is the concentration by mass of MNPs. In Coil V4 this yields the following results:
| MNP | Water (None) | EFH1 | US Nano |
|---|---|---|---|
| SAR (W/g) | 25.5 | 394.9 | 83.1 |
Our concentrations in these experiments were 100x higher than that of Stanley et al and our field strength was 2.5x higher. Scaling down linearly with concentration and quadratically with field strength would yield a SAR of 0.63W/g for the EFH1 which agrees with their findings.
By combining our known distribution of magnetic field field strength and our calculated specific absorption rate, we can model how the temperature distribution across our target volume will change over time using thermal diffusion equations:
Testing
Next, by using our final iteration of coil design, we tried to demonstrate targetted heating. 3ml of 100mg/ml MNPs was placed in a 35mm petri dish. The intial temperature distribution was imaged using an MLX90640 thermal camera. Then the sample was placed within the coil targetting a specific region for two minutes. The sample was then thermally imaged again. The results are shown below:
Targetted Heating Experiment
This clearly demonstrates that designed electromagnetic coils can be used to heat specific regions of a media using disperse magnetic nanoparticles. With more time, we would have liked to complete further experiments to quantify the precision of heating using this process.
In order to make this aproach valuable for biological applications, it would be necessary to significantly reduce the overall concentration of nanoparticles in our media and, as a result, increase the field strength of the coil. This is possible, but would require significantly higher power sources and more effective cooling methods (such as water cooling). Given the initial safety parameters we laid out at the start of the project, we felt it was not appropriate for us to increase the power of our designs further at this stage.
Electromagnets for Force and homogeneous field
Background
In addition to high-frequency alternating magnetic fields, direct and very low-frequency alternating magnetic fields play a crucial role in biological systems(Krylov & Osipova, 2023). To simulate conditions similar to the Earth's natural magnetic field (MF) or to generate low-frequency MFs, a device capable of producing strong, homogeneous fields is necessary to effectively characterize biological responses.
The Earth's natural MF strength ranges from 25 to 65 µT(Finlay et al., 2010), which can be easily replicated using solenoid coils. While this would theoretically be sufficient for EPG-related constructs, as species like Kryptopterus bicirrhis use the Earth's MF as guidance, our system likely requires much stronger fields due to the different mechanisms driving EPG.
Furthermore, low-frequency alternating magnetic fields are essential for manipulating 250 nm magnetic nanoparticles(Del Sol-Fernández et al., 2021). The Wsc1 mechanism operates by applying magnetic field gradients on these magnetic nanoparticles, creating a pulling force that acts on both the nanoparticles and the associated cellular receptors.
This setup is versatile and can also be applied to investigate the behavior of organisms under high magnetic field conditions, offering a range of possibilities for biological studies.
Theory
Superparamagnetic Nanoparticles
MF gradient for driving MNP with induced force
The magnetic nanoparticle (MNP) can be approximated as having a dipole moment when placed in an external magnetic field. The magnetization of the MNP is given by:
\[ m = V_m \Delta \chi \frac{B}{\mu_0} \]
where:
- \( m \) is the magnetic moment of the nanoparticle,
- \( V_m \) is the volume of the MNP,
- \( \Delta \chi \) is the difference in magnetic susceptibility between the MNP and its surroundings,
- \( B \) is the external magnetic field strength, and
- \( \mu_0 \) is the permeability of free space.
Since we are working with a single type of MNP at a constant temperature, we can consider \( V_m \), \( \Delta \chi \), and \( \mu_0 \) as constants.
The force acting on the MNP (\( F_m \)) is given by the gradient of the magnetic field interacting with the dipole moment:
\[ F_m = \nabla (m \cdot B) \]
Substituting the expression for \( m \):
\[ F_m = \nabla \left( V_m \Delta \chi \frac{B \cdot B}{\mu_0} \right) \]
This simplifies to:
\[ F_m = V_m \Delta \chi \frac{1}{\mu_0} \nabla B^2 \]
Therefore, if we want to know the force acting on the MNP, we only need to know the magnetic field gradient within the region. This also guides us to design the coil which can produce such a gradient.
Since gradients are generally produced by the interface between two materials with different magnetic permeability, especially for those with sharp edges, magnetic flux will tend to move across the path with the lowest reluctance. Consequently, the magnetic flux will be concentrated at the very end of a pointed surface, which creates a strong magnetic field. This guides us in designing our coil to generate force on our MNPs.
Alternating Low frequency magnetic field
MNPs are attached to the surface of cells by chemical interactions (Ni-NTA with His tag), which means we have no control over the position of MNP attachment. Therefore, it is very likely that only part of the MNP will be in the right position activating the sensory receptors.
Additionally, the MNP is meant to be working on living cells, which are relatively mobile compared to a solid surface. Yeast cells growing on solid media are living in a biofilm produced by themselves. The biofilm is a viscous material which will produce drag force on cells moving through, but will not stop the cells. The free moving nature of cells within the matrix may cause problems, especially when we are testing in liquid culture.
In order to overcome this issue, we choose to work on solid media and directly attach MNP onto cells growing in the colony.
Another measure we took is to use a regulated magnetic field applying on the cells. We want to reverse the direction of the magnetic field, which could potentially create two-directional pull and push force on the receptors. This may better activate the receptor proteins, since the exact pattern of activation is unknown.
However, since we do not want unfavored side effects by reverting the magnetic field too quickly, we decide to use a very low frequency alternating magnetic field at a frequency of \(1 \, \text{Hz}\).
Magnetic coil properties
The coil used in our design can be considered a large solenoid. To calculate the magnetic field (\(B\)) within a simple solenoid, the equation is:
\[ B = \frac{\mu_0 N I}{l} \]where:
- \(B\) is the magnetic flux density,
- \(\mu_0\) is the permeability of free space,
- \(N\) is the number of turns of the solenoid,
- \(I\) is the current passing through the coil, and
- \(l\) is the length of the solenoid.
However, to achieve either a more uniform magnetic field or a greater magnetic field gradient, we incorporate a soft iron core. The iron core has a much higher magnetic permeability than air, allowing us to shape and concentrate the magnetic field effectively. This enhances the field's strength and directs it where needed.
Some fundamental concepts are necessary for understanding this design:
- Magnetic Flux (\(\Phi\)): Analogous to electric current or water flow, magnetic flux represents the total "flow" of the magnetic field through a given area. It is measured in Weber (Wb).
- Magnetic Flux Density (\(B\)): The magnetic flux per unit area, given by: \[ B = \frac{\Phi}{A} \] where \(A\) is the cross-sectional area. \(B\) is measured in Tesla (T).
- Magnetomotive Force (MMF): This is similar to electromotive force in electric circuits, representing the "driving force" behind the magnetic flux. It is calculated as: \[ \text{MMF} = N \cdot I \] where \(N\) is the number of turns in the coil, and \(I\) is the current through the coil.
- Reluctance (\(R\)): Reluctance is the opposition to magnetic flux in a magnetic circuit, analogous to resistance in electrical circuits. It is given by: \[ R = \frac{l}{\mu \cdot A} \] where \(l\) is the length of the magnetic path, \(\mu\) is the permeability of the material, and \(A\) is the cross-sectional area.
For a long air-cored solenoid:
- The MMF is \(N \cdot I\),
- The reluctance is \(R = \frac{l}{\mu_0 \cdot A}\),
- Therefore, the magnetic flux is: \[ \Phi = \frac{\text{MMF}}{R} = \frac{N \cdot I \cdot \mu_0 \cdot A}{l} \]
- The flux density then becomes: \[ B = \frac{\Phi}{A} = \frac{N \cdot I \cdot \mu_0}{l} \] which matches the classical solenoid equation.
When an iron core is used, its high magnetic permeability (\(\mu_{\text{iron}}\)) significantly reduces the reluctance, as:
\[ R_{\text{iron}} = \frac{l}{\mu_{\text{iron}} \cdot A} \]Since \(\mu_{\text{iron}}\) is much greater than \(\mu_0\), the reluctance becomes much smaller, allowing for a significantly higher magnetic flux (\(\Phi\)). This in turn increases the flux density (\(B\)), resulting in a much stronger magnetic field.
In our design, we utilize a soft iron core not only to increase the magnetic field strength but also to control and shape the field. This approach enhances the magnetic field gradient and ensures optimal force on our MNPs. The high permeability of the soft iron core effectively channels the magnetic flux, creating the desired distribution of the field.
Design
Electric design
Box diagram of the system, consisting of five parts.
Actual image of driver board.
Whole system built in reality.
The hardware system consists of five main modules:
-
Electromagnet Driver Module:
Controls the electromagnet by driving current through the coil according to the required signal pattern using an H-bridge circuit. This enables precise regulation of the magnetic field's strength and direction as dictated by the microcontroller.
-
Signal Processing Module:
Processes signals from sensors, converting PWM signals into DC signals suitable for the microcontroller. It also filters noise to ensure accurate data is transmitted.
-
Power Management Module:
Converts and regulates power from a 60V input to 12V, 5V, and 3.3V for different system components. It ensures stable power delivery and protection for all hardware modules.
-
Peripheral Module:
Manages all input/output operations, facilitating data exchange between the microcontroller and connected components like sensors, actuators, and communication ports.
-
Microcontroller Module:
Acts as the system’s control center, coordinating all other modules. It handles signal generation, data processing, power distribution, I/O management, and external communication to maintain overall system functionality.
This modular design provides efficient control and integration of all system components for optimal operation.
H bridge circuit
We implemented an H-bridge setup to drive the coil, allowing control over the current direction through the coil.
The H-bridge works like an array of switches, which can control the direction of the load connected between the Vcc and GND nodes. This allows us to control the direction of current, therefore controlling the polarity of the electromagnet. More details here.
H bridge circuit work by switching two pairs of switches, usually MOSFETs. It is capable to produce two way current with potential difference equals to EMF of the power supply. Image from Here
We chose to use a full solid-state H-bridge driver instead of an H-bridge driver made from relays. This allows us to use a PWM signal to finely control the magnet strength and prevent circuit breakdown due to frequently switching the circuit at high voltage.
H-bridge driver in our system. with input isolated by optoisolator, and output have fuse protection.
For this purpose, we used the MDA48D10AA driver module with the following specifications:
- Input Voltage: 11-75V
- Max Output Current: 10A
- Turn-on Time (Ton): 2.8 µs
- Turn-off Time (Toff): 638 ns
These specifications make the module capable of driving most coils with a resistance greater than 1 ohm, providing sufficient power for our application. This module is also suitable for other electromagnetic devices.
To control the voltage across the electromagnet and thereby its magnetic strength, we use Pulse Width Modulation (PWM). By modulating the duty cycle of the PWM, we adjust the average voltage applied to the coil. Since the coil has significant inductance, it resists rapid changes in EMF, acting as a low-pass filter that smooths the high-frequency PWM signal, resulting in the desired low-frequency output.
Simulated waveform of the PWM driving signal, and the resultant current through the inductor. Red curve is the PWM wave, green curve is the current through inductor. The inductor can effectively remove the high frequency bit of the PWM signal, equals to calculating average to the signal.
The simulated LTSPICE schematics for the circuit, representing the minimal system of our coil
To prevent interference (crosstalk) between the PWM signal and the desired low-frequency signal, the PWM frequency is set to 100 kHz, which is much higher than the operational signal frequency (not exceeding 200 Hz). This large difference allows effective filtering of the high-frequency PWM while preserving the low-frequency control signal.
The system is controlled using a Raspberry Pi Pico microcontroller, which offers more memory and ease of use compared to the Arduino Mega series. We recommend the Raspberry Pi Pico for new projects requiring efficient PWM control and higher performance.
Coil Design
We utilized a vintage iron-cored electromagnet provided by the Cavendish Laboratory, Cambridge. This coil features an adjustable air gap, allowing for flexible control over the magnetic field strength and distribution.
Coil we retrieved from the Cavendish Laboratory. It is 60 year’s old now!
Coil Specifications
- Current Capacity: 0.75 A per coil
- Coil Construction: 4900 turns of 24 s.w.g. enameled copper wire
- Cold Resistance: 90 Ω
- Hot Resistance: 126 Ω
- Maximum Field Strength: 1.3 T with a 5 mm air gap when both coils are powered
To optimize and predict the coil's magnetic field behavior, we performed a finite element analysis using the software FEMM (Finite Element Method Magnetics). Given the adjustable air gap, we explored four main configurations to suit different experimental applications:
Large Air Gap
This is the standard setup, which produces a strong, homogeneous magnetic field between the air gaps. When the gap is set to 25.4 mm, the coil can generate a field strength of up to 360 mT. The uniformity of the field makes this setup suitable for experiments requiring consistent magnetic exposure across the sample.
The photo for the large airgap coil. The top coil is removed as the level of magnetic field strength can be achieved through existing coil.
FEMM simulated magnetic field distribution of this setup.
Distribution of magnetic field along the central axis
Distribution of magnetic field along the horizontal axis
Anti-Polarity Coil
By reversing the current direction in one of the coils, we create an opposing magnetic field within the air gap, effectively canceling out the field and achieving near-zero magnetic flux. This setup acts as a negative control, maintaining the heating effect of the coil without exerting a magnetic force on the sample. It allows the assessment of thermal effects independently of magnetic influence.
Magnetic field within the same coil setup, with the polarity of two coils inverted. Creating a space with minimal magnetic field.
Small Air Gap Coil
Reducing the air gap to 5 mm enables the coil to produce a maximum field strength of 1300 mT, providing a strong activation suitable for applications that require intense magnetic fields. This setup is particularly useful for characterizing biological parts under extreme magnetic conditions. However, due to the small gap, only limited sample types can be placed within the coil.
Magnetic field within the small airgap setup. The magnetic field have been greately strengthened because of the lowered magnetic reluctance.
Top Core Removed Coil
In cases where sample manipulation or space is a priority, the top coil and iron core can be removed. This modification increases the operational space around the sample but reduces the overall magnetic field strength. However, it creates a high magnetic field gradient near the edge of the remaining iron core, as the flux concentrates along the path of least reluctance through the pointed core edges.
This is the main coil setup used in most of our experiments.
Electromagnet with top ironcore and coil removed.
The magnetic field distribution with top core removed
The magnetic field strength and greadient have been greately strengthed around the edge of the core.
Home made Cored coil
To make our results more accessible to those without a ready-made electromagnet, we have designed a coil that can be easily manufactured for use in biological applications. This coil can be created by 3D printing a spool to wind the copper wire around, allowing for simple replication.
The coil we designed for easy manufacturing
For optimal performance, the coil should be wound with 5000 turns of 24 SWG (Standard Wire Gauge) enameled copper wire. The winding should be tight and uniform to ensure consistent magnetic properties.
Simulated magnetic field distribution aroudn custom hardare. It can be the substitute of the top removed coil
Additionally, to achieve maximum magnetic field strength, a soft iron bar with a diameter of 38.1 mm should be used as the core material. The soft iron serves as an effective guide for the magnetic field, concentrating and enhancing its strength. It is crucial to use soft iron specifically, as its high permeability and low coercivity are ideal for efficiently channeling the magnetic flux generated by the coil.
Solenoid for Spatial Activation
To achieve spatial activation of our samples, a specially designed coil is required—one that can accommodate the size and shape of the samples effectively.
: The solenoid we wound to fit in 30 mm petri dish.
We choose to demonstrate the spatial activation on our 30 mm petri dish, so the coil is designed to fit the petri dish in, and have a steep magnetic field (MF) distribution, which allows spatial activation.
Distribution of magnetic field around solenoid
: Distribution of magnetic field around Y direction.
: Distribution of magnetic field around X direction.
In order to create a more homogeneous field in one direction but a steeper distribution in the other direction, we attempted to add two separate coils at the edge of our original coil. This gave the following results:
Altered solenoid design creating more homogeneous field.
Unfortunately, it did not provide satisfactory results, so we turned to a backup plan involving the use of a machined iron core. By shaping the core appropriately, it is possible to concentrate the magnetic field along a specific line. This focused field would enable precise spatial activation of the samples, targeting specific regions as needed. The machined core design allows better control over the magnetic field distribution and provides a solution where precise spatial gradients are required for the experiment.
Sensor Design
ACS712 Current Sensor: This sensor is specifically designed to accurately measure the current passing through the coil. Since the coil's resistance can vary with its temperature, monitoring current rather than voltage is crucial. Given that our system's maximum current is approximately 5 Amps, we selected the 5 Amp range variant of the ACS712 to match this requirement.
By monitoring the current value, we can accurately calculate the magnetic flux. Additionally, by combining current measurements with voltage readings, we can determine the coil's resistance. This allows us to implement a safety cutoff mechanism to prevent overheating.
AH49E Linear Hall Effect Sensor: The AH49E is employed to detect the polarity of the coil and to monitor the magnetic field strength, particularly when it is below 100 mT. Despite searching for a Hall sensor with a higher dynamic range, we found that no sensors exceeding 100 mT were available, indicating that our magnet's strength surpasses conventional requirements.
Two-Stage Low-Pass Filter: To convert a PWM signal into a format readable by a microcontroller, a low-pass filter is necessary. We designed a second-order low-pass filter using an RC circuit based on the following schematic. This filter has a cutoff frequency of 1.4 kHz, which is sufficient for our application, allowing clean signal processing.
The design of the second order lowpass filter.
OPT101 Photodiode: For real-time measurement of chemiluminescence and fluorescence when the magnetic field acts on the EPG-NanoLuc and EPG-TEV systems, a photosensitive device is required. The OPT101 photodiode was chosen due to its built-in amplifier, which simplifies signal processing. At its highest gain level, it offers a sensitivity of 2.3 V/µW, making it suitable for detecting low light levels generated in our experiments.
User interface design
To facilitate control over the electromagnet, we developed software to run on the host machine, providing an interface for easy operation. This software allows users to monitor key magnet parameters, control the magnet's operation, and adjust the waveform settings as needed.
GUI for controlling the electromagnet
For further details, please refer to the software repository.
Characterisation
Electric Characterisation
Rise and Fall Time:
- Pico pin rise time: 34.8 ns
- Pico pin fall time: 33.6 ns
- Coil voltage rise time: 33.84 ns
- Coil voltage fall time: 46.56 ns
The rise and fall times were measured to define the maximum PWM frequency that the driver can handle. Based on these values, the maximum achievable PWM frequency is approximately 1 MHz, allowing for high-frequency activation of the coils.
PWM Frequency: The PWM frequency output from the Raspberry Pi Pico was measured at 101.97 kHz.
Maximum Current: While we did not test the absolute maximum current to avoid blowing the fuse (rated at 10 A), the custom coil was tested at 8.5 A. Under these conditions, the driver generated a significant amount of heat but continued to operate properly.
Stability: For certain applications requiring extended activation times (up to 10 hours), stability is crucial. The coil was tested over a 5-hour period without experiencing any overheating or malfunctions, demonstrating reliable performance for long-term use.
The resistance of the coil was measured to be 90.9 Ω when cool. This baseline resistance is important for calculating power dissipation and monitoring temperature-dependent changes during operation.
Magnetic Characterization
The rise and fall times of the magnetic field were measured using a linear Hall sensor. Due to the sensor's saturation at B = 100 mT, accurate measurements of the magnetic field (MF) above this value were not possible. The sensor has a sensitivity of 66.67 mT/V and a zero-field offset of 2.5 V.
To characterize the magnetic field waveform produced by the electromagnet, a 100% duty cycle, 1 Hz square wave driving signal was applied. The magnetic response of the coil was then monitored at different driving voltages to evaluate waveform characteristics, including rise and fall times, which provide insights into the system's control precision.
Rise Times at Different Voltages:
-
At 60.00 V: Rise time is 144.9 ms (Note: Hall sensor saturation leads to inaccuracies at this voltage).
-
At 40.00 V: Rise time is 188.1 ms.
-
At 20.00 V: Rise time is 307.3 ms.
Magnetic Field Strength at Different PWM Levels
To demonstrate precise control over magnetic field (MF) strength using PWM driving, we varied the duty cycle of the PWM signal from -100% to 100%, where negative values correspond to reversed polarity. The test was conducted at 22.85 V to avoid saturating the Hall sensor.
Waveform of magnetic field for PWM signal sweeping across -100% to 100%
The resulting magnetic field waveform showed high linearity and minimal jitter, indicating that our system can reliably generate varying levels of magnetic field strength across different PWM duty cycles. This linear relationship confirms the ability to accurately control the magnetic field using PWM adjustments.
Inductance
The inductance of the coil is a critical factor in our electromagnet's performance, especially considering its large size and iron core, which significantly increases the inductance.
To measure the inductance, we opted not to use the resonance method, as the resonant frequency would be too low for accurate measurement. Instead, we constructed an LR circuit to measure the fall time (\(\tau\)), which allows us to calculate the inductance.
Waveform of large coil driven by square wave. The peak voltage is resulted by instanetaneous change in potential differece cause increase in impedance.
For the standard coil setup, we measured a fall time of \(\tau = 3.248\) ms using a 15 kΩ resistor. This yielded an inductance value of approximately 48.7 H, indicating a substantial inductance, which accounts for the observed waveform characteristics.
- Resistor (R): 15 kΩ
- Fall Time (\(\tau\)): 3.248 ms
- Inductance (L): \(L = R \times \tau = 48.7 \, \text{H}\)
This large inductance value explains the gradual transitions in the magnetic field waveform and the overall behavior of the electromagnet.
Residual Magnetization
It was observed that the iron core retains some residual magnetization, which may cause issues when applying a DC field to the coil. The typical residual magnetic field is around 3 mT, stemming from the soft iron's natural tendency to maintain some level of magnetization.
To mitigate this effect, we apply an alternating magnetic field to the coil, effectively "flushing" the residual magnetization and restoring the core to a neutral state before applying the desired DC field. This process ensures more accurate control over the magnetic field during operation.
Testing
Testing on Wsc1
The Wsc1 activation was characterised using a 30.1 V driving voltage, generating a magnetic field strength of 113 mT with a 1 Hz square wave. This resulted in a zigzag waveform in the magnetic field, indicating dynamic behaviour during activation.
The setup we used to test yeast cells on our low frequency coil.
The activation was sustained for one hour. For further details on the process and results, please refer to the Wsc1 page.
Testing on EPG
To evaluate the activity of EPG under the influence of a magnetic field, we designed a luminometer using the OPT101 sensor, specifically created to fit within the gap between the electromagnets and hold a PCR tube as a reaction vessel. However, the sensor readings exhibited a significant amount of noise, which was likely caused by electromagnetic interference present in the building.
The setup we used to test E.coli cells with EPG-Nanoluc construct.
After consulting with the electrical workshop in the engineering department, the interference was confirmed as the primary source of noise. Unfortunately, due to time constraints, we were unable to develop a more noise-resistant and sensitive sensor to mitigate this issue. Future improvements to the setup will need to focus on enhancing sensor performance in electromagnetically noisy environments.
Putting it together: MagentaBOX
Background
With hardware designs developed to activate each of our constructs, we then set about developing our final aim - a new research platform for spatial targeting magnetogenetics, called MagentaBOX.
In order to be able to activate specific planar regions reliably, we required our electromagnets to be motorised and electronically controllable. This means we could precisely move our coils to specific locations and then power them using software. Additionally we needed a number of ways of measuring the spatial distribution of response of our constructs to the applied stimulus. This means imaging capability for colour, chemiluminescence, fluorescence and temperature. No currently available devices met all of these criteria so we set about building our own.
Initial Design
Mechanical
Before deciding the device dimensions and structure, we needed to determine a suitable layout of components within. Our coils needed to be able to move around our cells without colliding and minimising the number of moving parts. We came up with a design with the plate mounted on an extended stand and the coils mounted on precision motorised stages:
To minimise the cost and number of components necessary to measure all of our outputs we came up with a design that used just two cameras and a servo motor. We chose a low cost, raspberry pi camera sensor (OV5647) with an adjustable focus lens for our primary imaging. By using different lighting/filtering arrangements this would mean we could capture colour, chemiluminescence, fluorescence using a single camera and several custom light sources:
Imaging Type White Diffuse Light Blue Filtered Diffuse Light Orange Camera Filter Colour ✅ ❌ ❌ Fluorescence ❌ ✅ ✅ Chemiluminescence ❌ ❌ ❌ With these components designed we modelled them in Fusion 360 and designed a box structure that we could mount each of our components to. We laser cut and 3D printed the necessary parts for this design:
Initial CAD Design
Parts Collected
Assembled
Electronics
Our electronics design was based on a Raspberry Pi controller due to wide availability and standardisation. This then interfaces to an arduino via serial communication to control our hardware components. A list of the required electronics components to interface is below:
- 2 x Linear Stage Servo Motors
- 2 x Limit Switches
- 4 x Control Relay
- Electromagnet Design A
- Electromagnet Design B
- 16x02 Display
- Motor Driver Boards
- Diffuse White Lighting
- Diffuse 470nm Blue Lighting
- Servo Motor
- Raspberry Pi Camera (OV5647)
- Thermal Camera (MLX90640)
We developed a block diagram for how each of these components would connect together and then built the design.
Block Diagram
To be able to control the device, we developed a python script to control the cameras and send instructions to the arduino and an arduino program to interface to the control hardware.
Control Scripts
Working Device
User Testing
Firstly we demoed it to team members unfamiliar with hardware. They suggested adding a graphical user interface (GUI) to make the device easy to use without programming skills.
We took some time to implement this before running some user testing with our advisors. Without any instruction they were able to operate the device using the GUI. They suggested adding clearer labelling to our graphical interface and making a user guide.
We then took MagentaBOX to a Biomaker meetup to demo the device and have some researchers and hobbyists try it out. They suggested that our exposed wiring meant that we may have reliability issues and that we should make a build guide and open source all our material to make it easy for future teams to reproduce the device.
Presenting our project
Demoing the device
Final Design
Building on what we learned from our initial design and the feedback we received, we then designed and built a new version with many improvements.
MagentaBOX - Complete
MagentaBOX - Internals
MagentaBOX - Fluorescent Imaging
Key Improvements
- Improved Diffuse Light Sources to improve image quality:
- Improved Motor Mounts to reduce unwanted flex:
- Lighting position changed to box top to improve image quality:
Original
New
- Graphical interface for operating/testing the device with support for custom protocols/experiments:
Original (left), new (right)
Original (bottom), new (top)
Added GUI
Build One!
We've made a full instruction guide on how to build a MagentaBOX for future teams and researchers as well as a setup and user guide:
Commercial Magnetogenetics equipment is expensive. As an example, the lowest cost power supply we could find to drive a magnetic hyperthermia induction coil costs $1925 (Accel Instruments). Our total cost to build the deivce was £324.80 ($430.80 USD). We hope that cost reduction significantly lowers the barrier to entry for new teams and researchers interested in this area.
Cost Breakdown
For Heating CoilPart Qty Source Cost (£) Laser Cut MDF (600x300mm) 6 Engineering Department 12 Blue Transparent Perspex (600mmx400mm) 1 Rapid Electrnoics 11.55 Orange Transparent Perspex (600mmx400mm) 1 Rapid Electronics 11.66 3D Print Parts, PLA 240g Engineering Department 12 White LED Strip 1 Amazon 17.99 Blue LED Strip 1 Amazon 8.99 Black Primer Spray Paint 1 Amazon 6.99 Black Spray Paint 1 Amazon 4.95 1602 Display 1 AliExpress 1.74 180deg Servo Motor 1 AliExpress 1.40 80mm Linear Stage Actuator 2 AliExpress 20.98 Micro Limit Switch 2 Aliexpress 2 12V 4 Channel Relay Board 1 AliExpress 2.92 Raspberry Pi 5 1 Raspbery Pi 55 Arduino 1 AliExpress 3.10 Micro SD Card 32Gb 1 Amazon 5 15-22 Pin RPi camera cable 1 Raspberry Pi 3 7/02 Wiring 1.4A 1 Engineering Department 6.99 MLX90640 Thermal Camera 1 Pimoroni 39 OV5647 Adjustable Focus Camera 1 AliExpress 4.45 M2.5 Standoff Set 1 AliExpress 5.20 M2.5 Heat Set Insert 50 AliExpress 1.55 M2.5x8mm Brass Screws 50 Engineering Department 1.50 M2.5x16mm Brass Screws 20 Engineering Department 1.20 M2.5x20mm Brass Screws 12 Engineering Department 0.80 M2.5x25mm Brass Screws 20 Engineering Department 1.20 M2.5 Brass Nuts 100 Engineering Department 1.50
For Low Frequency/DC CoilPart Qty Source Cost (£) ZVS Induction Heating Circuit 1 Amazon 5.50 MKP 330nF Capacitors 10 RS Electronics 0.80 12A Litz Wire 3m RM Cybernetics 22.50 3mm Copper tubing 1m Engineering Department 5 Part Qty Source Cost (£) H Bridge Driver 1 AliExpress 15 Current Sensor 1 Amazon 1 Hall Sensor 1 Amazon 1 Photodiode 1 Amazon 5 33nF Capacitors 2 Amazon 0.5 60V->12V Converter 1 Amazon 2 DC Breakout Board 1 Amazon 2 OLED Display 1 Amazon 2 Switch 1 Amazon 0.5 Breadboard 1 Amazon 0.5 Dupont Wires Various Amazon 1 Connectors 5 Amazon 2 24AWG Enameled Copper Wire 500m Amazon 11.34 Heat Insulating Materal 1xA4 Amazon 1.34 Calibration
In order to use MagnetaBOX to take meaningful fluorescence readers, we created a method to calibrate the system using fluorescein.
MagentaBOX Fluorescence Calibration
To ensure that the camera setup produced relevant and accurate data, it was essential to calibrate both the camera and the photo analysis system using standard fluorescein solutions. A serial dilution of fluorescein was photographed, and the images were analyzed using ImageJ.
Fluorescence calibration photo
Method
Nine wells were filled with varying concentrations of fluorescein via a serial dilution method. These wells were imaged using the MagentaBOX, and the camera settings were saved to ensure consistent replication in future experiments. Repeated measurements were taken in ImageJ to calculate the average mean grey value for each well. These values were then normalized to PBS (phosphate-buffered saline) and plotted against the fluorescein concentration to create a calibration curve.
Fluorescein concentration camera calibration curve
The resulting fluorescein concentration curve demonstrated a clear relationship between fluorescence intensity and concentration, which can now be used to standardize future measurements with the MagentaBOX system.
We carried out calibration of fluorescence and chemiluminescence signal on our device, to allow it be compariable with experiments carried out in other labs.
See our Measurement page for more.
References
- Makoto Suto, Yasutake Hirota, Hiroaki Mamiya, Asaya Fujita, Ryo Kasuya, Kazuyuki Tohji, Balachandran Jeyadevan, Heat dissipation mechanism of magnetite nanoparticles in magnetic fluid hyperthermia. Journal of Magnetism and Magnetic Materials. DOI: 10.1016/j.jmmm.2009.02.070
- Ring, H. L., Sharma, A., Ivkov, R., & Bischof, J. C. (2020). The impact of data selection and fitting on SAR estimation for magnetic nanoparticle heating. International Journal of Hyperthermia, 37(3), 100–107. DOI: 10.1080/02656736.2020.1810332
- Sarah A. Stanley et al., Radio-Wave Heating of Iron Oxide Nanoparticles Can Regulate Plasma Glucose in Mice. Science, 336, 604–608 (2012). DOI: 10.1126/science.1216753
- Del Sol-Fernández, S., Martínez-Vicente, P., Gomollón-Zueco, P., Castro-Hinojosa, C., Gutiérrez, L., Fratila, R. M., & Moros, M. (2021). Magnetogenetics: remote activation of cellular functions triggered by magnetic switches. Nanoscale, 14(6), 2091–2118.
- Finlay, C. C., Maus, S., Beggan, C. D., Bondar, T. N., Chambodut, A., Chernova, T. A., Chulliat, A., Golovkov, V. P., Hamilton, B., Hamoudi, M., Holme, R., Hulot, G., Kuang, W., Langlais, B., Lesur, V., Lowes, F. J., Lühr, H., Macmillan, S., Mandea, M., ... Zvereva, T. I. (2010). International Geomagnetic Reference Field: the eleventh generation. Geophysical Journal International, 183(3), 1216–1230.
- Krylov, V. V., & Osipova, E. A. (2023). Molecular biological effects of weak Low-Frequency magnetic fields: Frequency–Amplitude efficiency windows and possible mechanisms. International Journal of Molecular Sciences, 24(13), 10989.