Magnet-Assisted GaN Monolithically Integrated Device for Optical Three-Axis Motion Sensing

photonics1
Aug. 31, 2024

Abstract

Motion sensors, capable of detecting acceleration and inclination along the three axes, are essential in emerging domains of robotics, autonomous vehicles, and virtual reality technologies. To meet the growing demands for precise and reliable motion tracking in cutting-edge applications, here, a compact three-axis motion sensor is developed utilizing a monolithically integrated GaN-based optoelectronic device with split-type magnets. The sensor demonstrates a linear response across an acceleration measurement range of ±10 g for the X- and Y-axes and ±5 g for the Z-axis. It also exhibits fast response and recovery times of 1.50 and 1.03 ms, respectively, along with a minimum detection limit of 0.035 g. The sensor is capable of precisely detecting the tipping angle and acceleration of the dump truck as well as assessing the posture and motion status of unmanned aerial vehicles. The developed sensor offers promising solutions for advancements in automated vehicle operation and robotic motion.

The rapid technological advancements in autonomous driving, robotics, and virtual reality have increased the demand for enhanced multiaxis motion sensing capabilities. (1) Acceleration sensors are often used for motion detection, (2) capable of not only detecting changes in velocity and displacement of an object based on acceleration information but also facilitating the measurement of its angular tilt and posture. (3,4) Across diverse applications, from robotics (5−8) and virtual reality to autonomous driving, (9,10) the precise detection of acceleration variations along multiple axes is essential.

Currently, accelerometers are being developed based on a variety of principles, including piezoelectric, (11,12) piezoresistive, (13) capacitive, (14,15) optical, (16,17) and triboelectric nanogenerators. (18,19) Significant progress has been made in expanding the measurement range and improving sensitivity through the advanced combination of structural designs and nano/microfabrication techniques. (20,21) Among them, optical sensors, known for their high precision and rapid response, have been extensively researched. (22−25) However, optical systems frequently necessitate the assembly of light sources, sensing components, and spectrum analyzers, resulting in bulky and costly configurations.

To meet the growing demand for miniaturized devices in diverse portable and wearable applications, it is crucial to develop compact motion sensing devices that possess a wide measurement range, easy integration, cost-effective manufacturing, and multiaxis sensing capability. (26,27) This has triggered increased attention toward the chip-scale integration of optoelectronic devices. Notably, GaN-based semiconductors emerge as promising candidates due to their high efficiency, stability, and longevity. (28) Several GaN-based integrated devices have been demonstrated, exhibiting exceptional sensing functionality in cell activity monitoring, (29) gas detection, (30,31) and respiratory monitoring. (32) However, the potential of motion and acceleration detection based on GaN-based integrated devices remains unexplored.

In this work, a novel magnet-assisted three-axis optical motion sensor is introduced. The sensor incorporates a GaN-based optoelectronic device with split-type magnets. The GaN-based device integrates a light emitter and five photodetectors onto a chip-scale platform through a monolithic design, forming the foundation for multiaxis motion perception. With the aid of split-type magnets and a deformable polydimethylsiloxane (PDMS) film, the motion signal can effectively convert into an optical signal sensed by the on-chip detectors. Among the available elastomeric materials, PDMS offers the advantages of reasonable cost, simplicity of handling, remarkable structural flexibility, and rapid manufacturing. The properties of the emitter and detector are investigated, and the sensing responses of the integrated device to acceleration changes are characterized to verify the effectiveness of the proposed motion sensor.

Experimental Section

Click to copy section linkSection link copied!

Fabrication of GaN Devices

An epitaxial structure was grown on a 4 in. c-plane sapphire substrate using metal–organic chemical vapor deposition, comprising a 2 μm-thick unintentionally doped GaN, 2 μm-thick n-GaN with a doping level of 10¹⁸ cm^–3, 182 nm-thick In_0.25Ga_0.75N/GaN multiple quantum wells (MQWs), and 160 nm-thick p-GaN with a doping level of 10¹⁷ cm^–3. A 120 nm-thick indium tin oxide layer (ITO) was deposited on the p-GaN surface as a current spreading layer. The mesas of a light-emitting diode (LED) and five photodetectors (PDs) were patterned through photolithography, and the unmasked areas were revealed using inductively coupled plasma etching. A 360 nm-thick SiO₂ passivation layer was coated via plasma-enhanced chemical vapor deposition, followed by the addition of a 3.2 μm-thick SiO₂/TiO₂ distributed Bragg reflector using an optical thin-film coater. Both 2.4 μm-thick electrodes were fabricated using a combination of photolithography, electron-beam evaporation, and lift-off processes. The fabricated wafer was then diced into small chips, each with 2.5 × 2.5 × 0.15 mm size, using laser micromachining. A detailed description of the fabrication process is provided in Supporting Information S1.

Preparation of Pressure Sensing Films

Prepolymer and curing agents were mixed in a 10:1 ratio to formulate the polydimethylsiloxane (PDMS) gel (Sylgard 184, Dow Corning). The mixture was spread over a sapphire template patterned with a microlens array structure. To eliminate air bubbles, the coated samples were subjected to a degassing process in a vacuum chamber. After curing at 80 °C for 2 h, the PDMS film with dimensions of approximately 5 mm × 5 mm × 0.1 mm was gently peeled off. The detailed preparation process is provided in Supporting Information S2.

Assembly of Three-Axis Motion Sensors

The fabricated GaN device was packaged onto a printed circuit board (PCB) to establish electrical connections. Subsequently, the patterned PDMS film was affixed over the device using PDMS as the adhesive. Magnetic sheets were affixed to the substrate with PDMS and precisely aligned with the center of the GaN device. Finally, a magnetic ball was placed on the PDMS film to complete the assembly process.

Experimental Measurement and Calibration

The sensor was driven by a sourceMeter (Keithley 2450), providing a constant current to the LED. The photocurrents generated by the PDs were collected by using multimeters (Keithley DMM6500), which provide a measurement resolution of 0.05 nA. For calibration purposes, the developed sensor was mounted within a 3D-printed mold alongside a commercial accelerometer (ADXL345) as a reference. During the experimental measurements, various magnitudes of acceleration were provided by using a linear reciprocating machine, while angular variations were introduced through a rotating motorized platform.

Results and Discussion

Click to copy section linkSection link copied!

Structural Layout and Principle

Figure 1a shows an exploded view of the proposed design of the motion sensor. The split-type magnets consist of multiple magnet sheets, each with a thickness of 1 mm and radius of 4 mm, which are fixed beneath a 1.6 mm-thick PCB. A magnetic ball with a radius of 2.5 mm is positioned on top. Due to the magnetic attraction between the fixed magnet sheets and the magnetic ball, the magnetic ball exhibits a dynamic response to acceleration. Specifically, the ball shifts its position in response to acceleration, and in the absence of acceleration, the ball returns to its initial, centered position.

Figure 1. Configuration of the magnet-assisted optical motion sensor. (a) Exploded view showing the schematic layout of the sensor structure. (b) Schematic diagram and (c) optical image of the GaN device containing an LED and PD_1–5. (d) scanning electron microscopy (SEM) image of the patterned PDMS film. (e) Optical image of the resultant sensor.

Figure 1b shows the schematic layout of the LED and the five PDs formed on a GaN-on-sapphire device. As depicted in Figure 1c, the device, which integrates an LED and five PDs in a monolithic design, was fabricated on a single chip with dimensions of 2.5 mm × 2.5 mm through wafer-scale microfabrication techniques. The central circular LED has a radius of 335 μm. The ring-shaped PD₅ surrounding this LED has an inner diameter of 345 μm and an outer diameter of 700 μm. The four PDs (PD_1–4) are positioned around the perimeter. Both the LED and PDs have the same diode structure incorporating InGaN/GaN MQWs. Upon current injection, the LED emits light as a result of carrier recombination within the MQWs. The emitted light is partially reflected by the sapphire boundary and directed toward the PDs. The MQWs within the PDs absorb the reflected light, generating electron–hole pairs and, subsequently, photocurrent.

Given the inherent structural rigidity of GaN devices, their response to the magnet ball movement triggered by acceleration changes is severely limited. To overcome this, a deformable PDMS film featuring honeycomb-shaped patterns with a 1 μm diameter, shown in Figure 1d, is fabricated and utilized as a sensing medium to convert motion-induced pressure variations into optical signals. As illustrated in Figure 1e, the assembled device shows an effective size of 5 × 5 × 8 mm, highlighting its compactness and portability.

Figure 2 illustrates the sensing principle of the motion sensor. In Figure 2a, the motion sensor is in a static state, where the PDMS film experiences slight stress due to magnetic attraction. Since the sapphire with a refractive index of 1.77 is mostly exposed to air, a substantial portion of light undergoes total internal reflection, partially reaching the PDs. Upon acceleration, the magnetic ball shifts from its initial position due to inertia, exerting pressure on the PDMS film. Knowing that the refractive index of PDMS is 1.4, the compressed PDMS micropatterns increase their contact with the sapphire, leading to a reduced amount of totally reflected light, as depicted in Figure 2b,c. Therefore, the photocurrent change in the PDs serves as an indicator, reflecting both the direction and magnitude of the acceleration. A detailed analysis of the influence of the ball position on the photocurrent is provided in Supporting Information S3.

Figure 2. Sensing principle. Schematic diagrams showing the changes in structure and light distribution at (a) static and with accelerations along (b) lateral and (c) vertical directions. (d) Simulation results showing the pressure distribution on the 100 μm-thick PDMS film with acceleration applied in the lateral and vertical directions.

To investigate the pressure variations across the PDMS film induced by the motion of the magnetic ball, simulations utilizing a finite element method were carried out, as shown in Figure 2d. When acceleration is applied in the X-/Y-axis direction, the ball shifts toward the opposite side, exerting pressure on the corresponding segment of the PDMS film. The simulation results also demonstrate that the pressure exerted by the ball on the PDMS film is primarily a ring-like pattern when acceleration along the Z-axis occurs. These results serve as a crucial reference for positioning the LED and PDs within the GaN devices. In particular, the on-chip design incorporates a central circular LED, four symmetrically positioned PDs around the perimeter for the X- and Y-axis sensing, and a central annular PD for the Z-axis detection. Thinning the PDMS film facilitates enhanced force conduction, potentially improving the sensitivity of the sensor (Supporting Information S4). However, the preparation of the thin, patterned PDMS films poses a challenge, as the demolding process can readily introduce film damage. To strike a balance between preparation reliability and device performance, a film thickness of 100 μm is selected.

Properties of GaN Device

The electrical and optical properties of the GaN device were investigated. Figure 3a shows the emission spectrum of the LED at different driving currents, with the wavelength peak ranging from 526 to 529 nm and a full width at half-maximum from 28 to 26 nm. The optical image in the inset of Figure 3a shows the device with an operating LED. Figure 3b plots the current–voltage (I–V) curve of the LED, indicating a forward-biased voltage of 2.63 V at a current of 10 mA. The resistance, determined from the reciprocal slope of the linear region, is around 32.2 Ω. The inset in Figure 3b indicates that the light output power of the LED is linearly proportional to the bias current.

Figure 3. Performance of GaN devices. (a) Emission spectra of the LED under different current biases. The inset shows an optical image of the devices with LED under operation. (b) I–V curve of the LED. The inset shows the driving current–light output power curve of the LED. (c) I–V curve of the PD_1/2/3/4 and PD₅ under different LED currents. (d) Plot of the LED drive current versus photocurrent of the PD_1/2/3/4 and PD₅.

As plotted in Figure 3c, the photocurrents of the PD_1/2/3/4 with the same structure and the ring-shaped PD₅, measured under reverse bias voltage, remained at a low level below 10^–8 A. When a constant current of 10 mA was applied to the LED, the photocurrents of the PD₅ increased to the order of 10^–5 A. This demonstrates that the PDs respond effectively to the LED light intensity. The PD₅ shows a greater photocurrent response than PD_1–4 due to its positioning in close proximity to the LED. The stable and low photocurrent at an increased reverse bias voltage indicates that the leakage current of the device was negligible. The linear correlation between the photocurrent of the PDs and the drive current of the LED is shown in Figure 3d.

The fabricated GaN devices possess dual functionality, capable of behaving as either emitters or detectors. While the PD₁ and PD₄ remained detectors, swapping the role of LED and PD₅ would significantly degrade the performance of acceleration sensing in the Z-axis, as detailed in Supporting Information S5. This was consistent with the simulation that the geometric center of the PDMS film experiences minimal force variation during the Z-axis movement. Moreover, the field strength of split-type magnets and the size of the magnetic ball were optimized to achieve a balanced performance in sensitivity, linearity, and measurement range (See Supporting Information S6).

Performance of Three-axis Acceleration Sensor

The acceleration sensing performance of the device was studied. The experimental setup is depicted in Figure 4a. The photocurrent values along the X-, Y-, and Z-axes, used to indicate the acceleration of the corresponding axis, denoted by I_Δ, are defined as follows:

I_{Δ X} = I_{P D_{1}} - I_{P D_{2}}

$I_{Δ X} = I_{P D_{1}} ? I_{P D_{2}}$

(1)

I_{Δ Y} = I_{P D_{3}} - I_{P D_{4}}

$I_{Δ Y} = I_{P D_{3}} ? I_{P D_{4}}$

(2)

I_{Δ Z} = I_{P D_{5}}

$I_{Δ Z} = I_{P D_{5}}$

(3)

where I_PD₁ to I_PD₅ represent the photocurrents of PD₁ to PD₅, respectively.

Figure 4. Acceleration sensing performance of the sensor. (a) Schematic diagram of the experimental setup for acceleration measurement. Δ Photocurrent response measured under different (b) X-axis, (c) Y-axis, and (d) Z-axis accelerations. The solid line represents a linear fit to the data. (e) Dynamic response of the sensor when changing acceleration over ±1, ±2.5, and ±5 g. (f) Transient response of the device. The yellow- and green-shaded areas represent the response time and recovery time based on the T90 method, respectively. (g) Minimum limit of detecting acceleration.

When the photocurrent response of the sensor to acceleration was measured, the LED was driven by a constant current of 10 mA, and the PDs were kept unbiased. Figure 4b illustrates the correlation between I_ΔX and the acceleration range of ±10 g along the X-axis. Linear regression analysis revealed an R² of 0.985 and a responsiveness (K) of 61.34 nA/g. Similarly, for the Y-axis, under a sensing range of ±10 g, an R² of 0.997 and a K of 67.86 nA/g are determined, as shown in Figure 4c. The sensing range of the Z-axis, as depicted in Figure 4d, spans ±5 g, with an R² of 0.989 and a K of 43.23 nA/g. The linear response simplifies data processing by allowing for direct proportionality between the output signal and input acceleration. The sensor exhibits a narrower response range along the Z-axis than along the X-/Y-axis due to the initial pressure exerted by the magnetic ball on the PDMS film. This subsequently reduces the deformation ability of the film when the sensor is subjected to acceleration. However, if the acceleration exceeds 15 g, the magnetic ball may detach from the chip, leading to a device failure.

Figure 4e shows the dynamic response of the sensor to varying acceleration levels of ±1, ±2.5, and ±5 g. The corresponding photocurrent changes are approximately ±65, ±150, and ±350 nA, respectively, which closely match the actual applied acceleration. The slight instability of the curves is attributed to the motion deviation of the linear reciprocating machine. Figure 4f illustrates the rapid response of the sensor when an acceleration of 3.91 g is applied along the X-axis, exhibiting a response time of 1.50 ms and a recovery time of 1.03 ms. The response times under varying accelerations along the lateral and vertical directions can be found in Supporting Information S7. Figure 4g shows the minimal detection limit derived from the gravitational acceleration component. The response was measured at Z-axis offsets of 2, 4, and 6° from the axis of the earth, yielding the limit of detection of approximately 0.035 g (sin 2° ≈ 0.035).

Compared to previously reported acceleration sensors, the sensor developed in this work exhibits superior performance in terms of multiaxis detection ability, size, measurement range, response time, and limit of detection (See Supporting Information S8). Notably, the sensor possesses a wide acceleration sensing range on all three axes coupled with a minimal detection limit. Most importantly, the compact size of the sensor facilitates the ease of operation and integration, making it a highly practical and versatile tool. To further enhance the sensing performance, it would be beneficial for future studies to investigate the dependencies of LEDs and PDs, along with the refractive index of sapphire and PDMS, on the operational wavelength.

Motion Detection in Land and Aerial Vehicles

The ability to detect acceleration on multiple axes of the sensor suggests its potential for inclination detection. To evaluate this, the sensor was mounted on a rotating platform alongside a commercial accelerometer for precise angle control. A detailed description of the measurement setup can be found in Supporting Information S9. Figure 5a,b shows the measured photocurrent response of the sensor with various pitch and elevation angles. The photocurrent changes caused by pitch angles of 20, 60, and 90° are approximately 20, 45, and 55 nA, respectively. These values correspond proportionally to the change in the value of g × sin?θ, thereby demonstrating the sensitivity of the sensor to inclination changes.

Figure 5. Inclination and attitude detection. Δ Photocurrent response measuring (a) pitch angles and (b) elevation angles of 20, 60, and 90°. (c) Δ Photocurrent response at X- and Y-axes during the lifting process of the dump trucks. 2500 cycle tests on (d) X-axis, (e) Y-axis, and (f) Z-axis. (g) Direction and magnitude of the acceleration on the dump truck along the X- and Y-axes. Changes in three-axis acceleration of the UAV while (h) flying forward left and right, and (i) normal flight and flipping flight.

The stability of the sensor is evaluated by conducting repetitive tests on the three-axis inclination. The measured results, as shown in Figure 5d–f, indicate the consistency of the sensor output over 2500 cycles for each axis. The photocurrent curves also reveal a high level of repeatability over multiple cycles (Supporting Information S10), indicating consistent sensing performance of the sensor. Although minor baseline deviations are observed, which is a common occurrence during practical use, these can be mitigated through algorithmic adjustments.

The applicability of the sensor is demonstrated by mounting it on a 3D-printed dumper truck to simulate the dynamic movement of a land vehicle. Figure 5c plots the I_ΔX and I_ΔY values recorded during the tilting motion of the dumper. As the Y-axis is perpendicular to the plane of the dump movement, no significant photocurrent variations in the I_ΔY were observed. The tilting action of the dumper causes a change in a photocurrent of 5 nA along the X-axis. Moreover, the figure clearly shows gradual and rapid tilting movements, revealing that the sensor accurately captures both the tilt angle and the speed of the dumper during unloading. The sensor can also track the acceleration applied to the truck at every 45° interval, corresponding to eight different directions (Supporting Information S11).

Figure 5g shows the direction and magnitude of the acceleration identified by the I_ΔX and I_ΔY signals collected by the sensor, which are found to be consistent with the directions of the applied acceleration, with minor relative errors (see Supporting Information S12). Subsequently, the sensor was mounted on a UAV for posture detection, as shown in Figure 5h. When the UAV flew toward the left front, there was apparent negative acceleration on the X-axis and Z-axis and positive acceleration on the Y-axis. As the UAV altered its trajectory to the right front, negative acceleration was detected on the Y-axis. Continuous negative acceleration on the X- and Z-axes represented forward motion without flipping, while a transition from positive to negative acceleration on the Y-axis signified a directional change from left to right. It is worth noting that the detection of flipping motion is crucial for assessing the current posture and guiding subsequent flight adjustments. Figure 5i displays the three-axis acceleration signals of the UAV during routine flight and flipping, and a notable opposite shift in acceleration on the Z-axis indicates an attitude reversal.

Conclusions

Click to copy section linkSection link copied!

In summary, an optical motion sensor is presented, comprising an integrated GaN-based device with split-type magnets. The compact design integrates an LED and five PDs on a single GaN-on-sapphire chip, forming the core element for detecting acceleration along three axes. The sensor exhibits a highly linear response over an acceleration range of ±10 g in the X- and Y-axes and ±5 g in the Z-axis. In addition, the sensor shows a response/recovery time of 1.50/1.03 ms and a low detection limit of 0.035 g. Moreover, experimental applications in vehicle posture sensing and automation are demonstrated. The developed sensor holds great promise for motion sensing across diverse applications.

News