Ayush Pandey, Maddaka Reddeppa, Zetian Mi. Recent progress on micro-LEDs[J]. Light: Advanced Manufacturing 4, 41(2023). doi: 10.37188/lam.2023.031
Citation:

3D printed multicore fiber-tip discriminative sensor for magnetic field and temperature measurements


  • Light: Advanced Manufacturing  5, Article number: 18 (2024)
More Information
  • Corresponding author:
    Limin Xiao (liminxiao@fudan.edu.cn)
  • Received: 05 December 2023
    Revised: 14 March 2024
    Accepted: 18 March 2024
    Accepted article preview online: 23 March 2024
    Published online: 27 March 2024

doi: https://doi.org/10.37188/lam.2024.018

  • Miniaturized fiber-optic magnetic field sensors have attracted considerable interest owing to their superiorities in anti-electromagnetic interference and compactness. However, the intrinsic thermodynamic properties of the material make temperature cross-sensitivity a challenging problem in terms of sensing accuracy and reliability. In this study, an ultracompact multicore fiber (MCF) tip sensor was designed to discriminatively measure the magnetic field and temperature, which was subsequently evaluated experimentally. The novel 3D printed sensing component consists of a bowl-shaped microcantilever and a polymer microfluid-infiltrated microcavity on the end-facet of an MCF, acting as two miniaturized Fabry-Perot interferometers. The magnetic sensitivity of the microcantilever was implemented by incorporating an iron micro ball into the microcantilever, and the microfluid-infiltrated microcavity enhanced the capability of highly sensitive temperature sensing. Using this tiny fiber-facet device in the two channels of an MCF allows discriminative measurements of the magnetic field and temperature by determining the sensitivity coefficient matrix of two parameters. The device exhibited a high magnetic field intensity sensitivity, approximately 1805.6 pm/mT with a fast response time of ~ 213 ms and a high temperature sensitivity of 160.3 pm/℃. Moreover, the sensor had a low condition number of 11.28, indicating high reliability in two-parameter measurements. The proposed 3D printed MCF-tip probes, which detect multiple signals through multiple channels within a single fiber, can provide an ultracompact, sensitive, and reliable scheme for discriminative measurements. The bowl-shaped microcantilever also provides a useful platform for incorporating microstructures with functional materials, extending multi-parameter sensing scenarios and promoting the application of MCFs.
  • III-V semiconductor optoelectronic devices have been widely used in a variety of fields, such as in illumination, displays, data communication, horticulture, and biological detection. To date, however, these applications required individual devices to be relatively large, of the order of a few hundred of microns or more, often driven by the need to maximize the output power1-3. Now, the development of advanced displays with ultrahigh resolution (e.g., >2,000 pixel per inch (PPI)), especially those for augmented and virtual reality (AR and VR), makes it necessary to use smaller device sizes for improving the resolution4. Biomedical and visible-light communication applications of microscale light sources have also been of great interest recently4-7. The micro-LED devices would have the benefits of higher self-emissive brightness, ultra-high integration density, robustness and stability as compared to existing technologies, such as liquid crystal displays (LCDs) and organic light emitting diodes (OLEDs)8. A comparison of these technologies is shown in Table 1.

    Table 1.  Comparison between some candidate technologies for future display applications9.
    PropertiesTechnology
    Liquid CrystalOrganic LEDQuantum Dot BasedInorganic Micro-LED
    EmissionBacklitSelf-emittingSelf-emitting/BacklitSelf-emitting
    LuminanceLowMediumMediumHigh
    ContrastPoorHighHighHigh
    LifetimeLongMediumShort/mediumLong
    ToxicityLowLowHighLow
    Response Timemsμsnsns
    Power ConsumptionHighMediumMediumLow
    CostLowMediumMediumHigh
     | Show Table
    DownLoad: CSV

    The pioneering technology for modern displays was LCDs, using liquid crystals to block emission from a backlight, and colors were attained through color filters. However, it had some limitations including color saturation, slow response times and poor conversion efficiency8, 10. More importantly, a large fraction of the optical power generated from the backlight was being wasted in these displays as they were not self-emitting. LCD displays are also difficult to scale down to small sizes with high resolution. Therefore, displays moved towards more energy-efficient self-emitting displays comprised of LEDs, which could be either inorganic (e.g., III-nitride based) or organic LEDs. While OLEDs have been widely adopted in displays, they are not without their limitations. Primary among them is the limited brightness of OLEDs (<1000 cd/m2) which is a major drawback. OLEDs also suffer from an efficiency roll-off at higher injection currents, resulting in significantly lower output power than inorganic LEDs11. Furthermore, OLEDs need longer burn-in times, and they are highly resistive, usually operating at current densities on the order of mA/cm2 from Refs. 12-14. Moreover, it has remained challenging to achieve high PPI OLED displays due to the shadow effect of the fine metal mask used to define pixels. It has also been difficult to achieve high efficiency OLEDs at shorter emission wavelengths (higher energy photons), severely limiting the performance of blue-emitting devices15-19. Generally speaking, the maximum operational temperature of OLEDs is limited to 50-70 ˚C, and they have much shorter lifetimes than inorganic devices. These issues can be readily solved by using inorganic LEDs, which have excellent stability, robustness, brightness, and long lifetimes. Inorganic LEDs typically reach their peak efficiencies at current densities of ~0.1-100 A/cm2, making them extremely bright with output luminance greater than 100,000 cd/m2 from Ref. 20. – a necessity for high-power applications. These features have led to their widespread adoption in diverse applications such as general lighting, automotives, horticulture and medicine21.

    However, the above applications of inorganic LEDs are primarily suited for large-area devices, and at present the efficiencies of visible-emitting inorganic III-nitride micro-LEDs are extremely low, as compared to their larger area counterparts. This is of particular concern for future micro-LED display applications, which need small LED chip sizes. The approximate maximum size requirements for individual micro-LEDs in some major applications are summarized in Table 28, 22-25. Further reductions in device area would be beneficial in terms of lowering the cost of the displays25.

    Table 2.  Micro-LED size requirements for different applications8, 22-25.
    ApplicationLED Dimensions (in μm)
    AR/VR1-5
    Wearables5-30
    Phones5-50
    Televisions20-80
    Automotive Displays50-100
    Digital Displays80-100
     | Show Table
    DownLoad: CSV

    While the efficiency is inherently higher for shorter wavelength blue-emitting devices26, it is significantly lower for devices having longer wavelength green and red emissions, due to the generation of defects when growing layers with high In composition and the stronger polarization fields, that together deteriorate radiative recombination. This difference is magnified for devices in the micron and nanoscale, wherein even blue-emitting devices struggle to attain high efficiencies.

    The low device efficiency has spawned great interest in improving micro-LED performance, with many research groups actively working in the field27-33. This has resulted in a marked improvement in the device efficiency for small-area blue InGaN LEDs, with most research work focusing on techniques to effectively passivate the sidewalls of the devices so as to reduce non-radiative surface recombination. D. Hwang et al. reported an external quantum efficiency (EQE) of 40.2% for a device with a size of 10 µm × 10 µm in the blue spectrum34. A high EQE of 20.2 ± 0.6% for blue micro-LEDs was reported very recently by a top-down approach, wherein a conventional multiple-quantum well epilayer LED heterostructure was etched to form nano-rods that were then passivated using an SiO2-based sol-gel35. However, the successful realization of high-efficiency small-area devices has not been replicated for green and red InGaN LEDs. A peak EQE of 14% for green micro-LEDs was demonstrated with a device size of 40 × 40 µm36. J.M. Smith et al. have studied the effect of surface recombination on device size for micro-LEDs with diameters between 1 µm and 30 µm. They measured a maximum EQE of ~7% for green micro-LEDs with device dimensions of 6 × 6 µm37. For devices with red emission, P. Li et al. have examined in detail the temperature dependent properties of red micro-LEDs with an area of 60 µm × 60 µm, and a peak EQE of 3.2%38. They have also demonstrated a red-emitting micro-LED with a tunnel junction contact, having a maximum EQE of 4.5%39. Recently, Y.M. Huang et al. reported a 6 µm × 25 µm sized red micro-LED with a peak EQE of 5.02%, with a focus for visible light communication applications40. A peak EQE of 1.75% has also been demonstrated for a device with 2 µm diameter41. The EQE of some III-nitride based LEDs from literature, of different emission colors, are plotted in Fig. 1 for varying device active areas26, 29, 32-101. The EQE of the LEDs shows a drastic reduction when the area of the LEDs becomes smaller for all wavelengths. This reduced efficiency greatly inhibits the commercialization of micro-LED technology. The causes for this efficiency cliff will be discussed in the next section.

    Fig. 1  EQE of some III-nitride blue, green and red LEDs reported in the literature, plotted against the device active area26, 29, 32-98, showing a drastic reduction of the efficiency with reducing device dimensions. Our previous work is indicated by the stars64, 65, 98-101.

    Regarding the emission wavelength of devices, conventionally, the III-nitrides (AlInGaN) are more generally used for shorter wavelength blue-green emission, whereas the smaller bandgaps of AlInGaP alloys make them better suited for yellow-red emission. However, specifically for micro-LEDs, several factors have motivated the investigation of using InGaN to cover long-wavelength red devices as well. Firstly, the bandgaps attainable through the InGaN material system can cover the entire visible spectrum, which could enable full-color red-green-blue (RGB) devices made from a single material. It has also been shown that InGaN-based LEDs are significantly less impacted by temperature due to the better quantum-confinement of charge carriers, as compared to AlInGaP devices, which improves their usability in applications where heating can affect performance102. Finally, and most importantly, a lower surface recombination velocity has been measured in the III-nitrides, as compared to AlInGaP, making them a preferable alternative for small-area devices103, 104. This is due to the increase in the dominant effect of surface recombination for micro-LEDs especially those with dimensions < 5 µm, which have a high surface-area to volume ratio.

    Instead of using a combination of III-V semiconductor materials for emitting at different wavelengths, phosphors and quantum dots have been proposed for color conversion. While this method might simplify the production of displays, there are several technological challenges. To date, color converters suffer from low conversion efficiency, poor/non-uniform color due to cross-talk, and size limitations8, 105, which greatly limit their application for micron-scale devices. While phosphors can be created with nanoscale particles, that could potentially be used in micro-LEDs, the reduced size of the particles results in a decrease of their color conversion efficiency106. Quantum-dot based color conversion technologies are a promising alternative, especially to a mass-transfer approach for micro-LED displays, however their incorporation is also complicated with an increase in fabrication steps, and hence cost. Quantum dots are also known to suffer from saturation and degradation when under illumination, which can also lead to leakage of the light that is used for exciting them, thereby affecting the output color107, 108. Furthermore, toxicity is a major concern with quantum dots, as several of the compounds used for making them are comprised of heavy metals, such as Pb and Cd9. Active research is being undertaken to address these problems, however, fundamentally, the fact that both quantum dot and phosphors involve the down-conversion of short-wavelength blue light to attain longer wavelength green and red emission means that they would be ultimately less efficient than light sources that intrinsically produce that color.

    The above discussion highlights that despite their low efficiencies, the prospect of creating monolithic RGB LEDs makes III-nitride inorganic devices the foremost approach for future micro-LED display technology. This review article firstly discusses the unique challenges associated with III-nitride micro-LEDs, including their fabrication, the difficulties in attaining longer emission wavelengths such as green and red, as well as their integration into displays. We then discuss the novel nanostructure-based approach by which several of these outstanding issues can be addressed. Our work on high-efficiency green micro-LEDs is then presented, including initial Ga-polar tunnel-junction devices, to more recent high-efficiency excitonic N-polar green submicron LEDs. This is followed by our work on developing N-polar sub-micron scale red InGaN LEDs. Subsequently, the use of photonic crystals to address the color purity of InGaN micro-LEDs is considered, with a demonstration for green micro-LED devices, along with their use for creating photonic crystal surface emitting lasers. The versatility of nanostructure devices is then further exemplified through a demonstration of green-emitting devices directly on non-native silicon substrates, as well as the realization of multi-color pixels monolithically grown in a single epitaxy step. Finally, we conclude with a summary of the major roadblocks and outlook for nanostructure-based III-nitride optoelectronic devices.

    The primary reason for the efficiency cliff, i.e., the degradation in LED efficiency with decreasing lateral dimension, is the increased surface recombination when the device areal sizes are reduced. Conventional top-down processing of III-V devices requires a plasma etch step to define the device mesas, however this also results in severe surface damage along the periphery of the mesas, forming crystal defects and dangling bonds and also introducing impurities37, 42, 49, 109. The surface defects created through this process play a major role on the carrier injection properties, especially when the ratio of surface area to volume is high, as in the case of micro-LEDs. Further, the non-radiative surface recombination depends strongly on the material properties. In this regard, the III-nitrides are more promising candidates, as compared to AlInGaP devices, for long wavelength micro-LEDs due to their significantly reduced surface recombination velocities103. The large variation in the reported surface recombination velocities is likely due to different fabrication procedures and the different In compositions of the active regions in the samples studied. The plasma etch step has a detrimental impact on the p-doped layer, where exposure to plasma results in an N-deficient near-surface region. These point defects typically compensate the Mg acceptors, resulting in a low free hole concentration that can impact the charge transport properties of the device110-114. To show the effect of plasma etching, J. M. Smith et al. studied the size-dependent characteristics of blue and green InGaN micro-LEDs. Their investigation revealed that the EQE drops off drastically when the device is scaled below 10 µm in lateral size. This study emphasized the main challenge associated with top-down micro-LEDs – mitigating the impact of plasma damage induced defects in the near-surface region.

    Several methods have been investigated to recover from the plasma etch induced damage, including annealing, exposure to nitrogen plasma, wet chemical etching, and surface treatments. It has been found that thermal annealing can reduce crystallographic damage in the near-surface region, partially recovering the device characteristics115, 116. Extending the duration of the annealing was shown to have further beneficial effects on crystal quality, however it would also result in the decomposition of the active region, affecting the emission wavelength and luminescence efficiency. It was shown that through the combination of thermal annealing with a step where the etched surface was exposed to N2 plasma, the surface stoichiometry was improved and the device characteristics could be recovered117. Hydrogen plasma treatment has been shown to enhance the peak EQE by 1.4 times in InGaN-based green micro-LEDs. This was attributed to the deactivation of Mg acceptors around the device mesa, which inhibited the injection of charge carriers along the region with a high density of surface defects56. Surface treatments immediately after mesa etching have been studied, including using atomic layer deposition (ALD) for depositing a dielectric Al2O3, or an (NH4)2S treatment, which can effectively passivate surface states57, 118. Blue-emitting InGaN micro-LEDs with an EQE of ~20.2% have been obtained by sol-gel SiO2 passivation, which was shown to be over two times more effective than SiO2 deposited using plasma-enhanced ALD35. The low-temperature sol-gel deposition approach avoided exposing the device sidewalls to thermal or plasma effects that occur in ALD, thereby minimizing any additional surface or structural defect creation that could occur due to atomic reactions. Through a combination of chemical treatment and sidewall passivation, M.S. Wang et. al. reported device size-independent peak EQE of micro-LEDs119. Finally, wet etching, typically using KOH, was shown to be effective in removing leakage paths formed during the plasma etch step. It was found that the device performance could be recovered after wet etching ~50-60 nm of the semiconductor, indicating the extent of plasma damage to the crystal, however this depth would also be affected by conditions of the plasma etch120, 121. In addition, this method is not selective, attacking even non-etched regions of the device, as well as any metal contacts. However, despite these extensive studies, the efficiency of micro-LEDs fabricated utilizing the conventional top-down etching process remains quite limited, especially for green and red devices.

    There is a large lattice mismatch between the constituent binary compounds of InGaN alloys – InN and GaN have a lattice mismatch of ~10%122. This makes it extremely difficult to grow high-quality InGaN epilayers with emission in the green and red regions of the visible spectrum, due to the tendency of InGaN to form defects and dislocations123, 124. The low miscibility of InGaN alloys also causes significant phase separation for high In content alloys, which results in broad luminescence peaks that make it hard to achieve pure red emission125. Another consequence of the lattice mismatch is that the grown InGaN epilayers would also be under a large compressive strain, and the resulting strong piezoelectric field spatially separates the electron and hole wavefunctions51. The reduced overlap of the carrier wavefunctions limits radiative carrier recombination, further reducing the internal quantum efficiency (IQE). The emission color of the generated light also depends on the injected carrier density. At low carrier injections where the piezoelectric polarization fields cause severe band bending, the emission energy is lower than the bandgap of the alloy. At higher injection currents, where the injected carriers can screen the polarization fields, thereby flattening the bands, the emission wavelength shifts closer to the bandgap. This quantum-confined Stark effect (QCSE) implies a current dependance for the emission color, limiting the applicable brightness range of practical color devices126, 127.

    Various methods have been developed to enable efficient red emission of InGaN. High-efficiency large-area red and orange LEDs were demonstrated by implementing V-pits to relax compressive strain, thereby helping to increase indium incorporation80, 128, 129. The V-pits are usually formed at the start of the growth of the low temperature layer, extending through the active region. They can help enhance the injection of carriers, as holes can be transported from the semi-polar facets of the V-pit into the deeper quantum wells. With the use of this technique, an EQE of 24% was attained for a large-area 1 mm2 LED having emission at 608 nm. However, while the average size and density of V-pits can be somewhat controlled, their location cannot, which is a major detriment to their inclusion in small-area micro-LEDs, where individual devices may randomly contain a few of them.

    The large compressive strain in InGaN impacts indium incorporation into the crystal lattice, making it difficult to achieve long wavelength green and red emission130, 131. Several groups have attempted to address this challenge by deliberately relaxing the strain within the active region. A thick underlying n-GaN buffer layer was shown to reduce residual in-plane stress of the InGaN active region, improving the crystal quality52, 132. This method greatly helped in red-shifting the emission wavelength, however the efficiency remained below 2%, even for larger area 400 µm × 400 µm devices. Another similar technique to relax strain involved the growth of a superlattice beneath the active region, which also assisted in increasing the In content of the InGaN active region45, 133.

    Pseudo-substrates, created by selectively performing electrochemical etching of doped GaN layers to form porous layers, were shown to be an effective tool to relax epitaxial films46, 66, 134. While red micro-LEDs have been demonstrated with this method, the fabrication process is quite complex, requiring a regrowth of the LED on the substrate after it has been made porous. Further, the porous nature of the substrate creates problems with electrical stability, reliability, and thermal conduction. Utilizing this approach, micro-LEDs with areas of up to 6 µm × 6 µm were shown, with an on-wafer EQE of ~0.2%.

    Recently, semi-relaxed InGaN pseudo-substrates have been investigated for the growth of the LED structure, where an InGaN layer first grown on GaN is separated from the substrate using Smart Cut™ technology, and then transferred onto a sapphire substrate with SiO2-SiO2 bonding70. The subsequent device structure growth is done on the InGaN pseudo-substrate layer, which has a lattice constant more closely matched to the red-emitting InGaN active region. The quality of the transferred pseudo-substrates and their p-type doping, however, remain obstacles affecting device performance and yield.

    Strain relaxation was also achieved through the partial decomposition of an InGaN underlayer beneath the active region135-137. This method did not involve additional fabrication steps, and only used thermal annealing of an InGaN decomposition layer grown beneath a high temperature InGaN/GaN superlattice decomposition stop layer. The high temperature caused voids to form in the InGaN decomposition layer, helping to relax the InGaN/GaN superlattice, as well as the subsequent active region growth. While this technique is quite promising, the defects and surface roughness as a result of the high temperature growth and annealing are problems that need to be addressed, which have so far limited the device EQE to below 1%.

    To date, the integration of individual, multi-color micro-LEDs into a high-resolution display has remained a major challenge. A single pixel contains at least three different color LEDs (RGB), while displays contain several million pixels. VR applications require a pixel density of at least 60 pixels per degree (PPD) to match the visual acuity of a human with 20/20 visual acuity. Practical headsets would require a field of view of at least 100 degrees, which corresponds to a 6K screen resolution. Considering that typical headsets would have dimensions of the order of 2 inches, these screens would require pixel densities in excess of 3000 PPI138. Further, glasses-like AR displays would likely require even higher densities23, 24, 138, necessitating small LED chip sizes, as shown in Table 2.

    In a display, pixels are arranged in an array, and driven in either a passive or active-matrix configuration. Both configurations use row and column interconnect lines to select a line and apply the appropriate driving voltage (current) to the pixels in that line. This is done at a high frequency so that the eye perceives a single image across the display. The primary difference between the two techniques is that while the individual LEDs in a passive matrix array are switched on for a short period of time, however, the presence of a charge storage capacitor in active-matrix arrays allow for the devices to stay on until the value assigned to them is updated. This enables significantly higher brightness at similar input currents for active matrix displays. However, the additional backplane transistors needed for active-matrix displays using micro-LEDs require micro-CMOS circuitry. This is typically achieved through a transfer process. Traditional pick-and-place methods for integration would struggle to handle such large quantities of devices while maintaining high throughput, yield, and precision. Further, the large number of devices makes it imperative to develop effective methods for defect identification, testing and mitigation. While pick-and-place would remain viable in applications with low PPI requirements, for emerging technologies requiring high PPI, such as AR/VR, the placement of the devices and their yield are major challenges with pick-and-place mass transfer139.

    Direct wafer-scale mass transfer can enable wafer-to-wafer or die-to-wafer assembly, greatly simplifying the fabrication process. This monolithic integration of the LEDs onto complementary metal-oxide semiconductor (CMOS) and low-temperature polysilicon (LTPS) backplanes has been demonstrated for producing screens of different sizes and resolutions140, 141. However, so far these displays have only been in a single color, with the expectation that color-converting quantum dots could be potentially integrated into them for making a full-color display.

    Considering the challenges facing micro-LED efficiency, integration and long-wavelength operation, nanostructures offer an alternative path and solution for achieving high efficiency micro-LEDs. The bottom-up approach to growing nano or micro-LEDs precludes the need for plasma etching of the active region to define device mesas, as in the case of conventional top-down LEDs, thereby avoiding the creation of surface defects. Both metal-organic chemical vapor deposition (MOCVD) and molecular beam epitaxy (MBE) have been used for the growth of nanostructures on a variety of lattice mismatched substrates with dislocation-free crystals, greatly enhancing the quantum efficiency of charge carrier recombination3, 27, 142-162. The high surface area to volume ratio during crystal epitaxy is also beneficial for strain relaxation and promoting the incorporation of indium in InGaN layers163-165. Nanostructures have an enhanced dopant incorporation, due to strain relaxation, which is crucial to maximizing injection efficiency in LEDs166-169. The incorporation of wide-bandgap AlGaN in InGaN nanostructures also results in the spontaneous formation of a core-shell structure, with the wider bandgap material forming the shell, thereby shielding the low bandgap active region from the effects of surface recombination146.

    Selective area epitaxy (SAE) of nanostructures has been demonstrated, using both MBE and MOCVD, which allows for precisely defining the dimensions, shape, morphology, and placement of the nanostructures prior to epitaxy170-176. The growth involves a precise control of growth conditions to allow crystal formation only in areas where the substrate is exposed, with no growth occurring in the mask region covering the substrate elsewhere. The morphology of the growth has been shown to depend critically on growth conditions, as well as the condition of the substrate. K. Kishino et al. studied the selective-area growth of InGaN/GaN nanowires using MBE177-181, and demonstrated multi-color (red, green, blue, yellow) pixels monolithically grown on a substrate. By varying the dimensions of the nanowires, the emission wavelengths of InGaN/GaN nanowires could be tuned through the entire visible spectrum. Their work demonstrated how the emission wavelength of nanostructures depends on their dimensions, as well as configuration in an array – where the effect of shadowing of material during epitaxy can affect composition. L. Samuelson et al., studied selective-area growth of micron-scale and smaller platelets, pyramids and nanowires using MOCVD, especially for their application in micro-LEDs182, 183. T. Wang et al. demonstrated a peak EQE of ~9% for a green LED array, grown using selective area epitaxy, wherein the diameter of individual devices was ~3.6 µm33, 44. They further demonstrated red-emitting micro-LEDs using this method with a peak EQE of 1.75%41. In addition, they monolithically integrated a high-electron-mobility transistor (HEMT) with a micro-LED, to target high-speed visible-light communication applications184. A record modulation bandwidth of 1.2 GHz was measured from the monolithically integrated bottom-up device, showcasing the possibilities of bottom-up device epitaxy and fabrication.

    A schematic of the SAE process used in plasma-assisted molecular beam epitaxy is shown in Fig. 2a. Through this process, highly uniform arrays can be grown over relatively large areas, shown in Fig. 2b. The composition of the alloys can be readily tuned by adjusting the ratio of the metal fluxes for the constituent elements (In and Ga). Photoluminescence spectra from nanostructures grown using this method are shown in Fig. 2c, with emission covering the entire visible spectrum. As the incorporation of In depends strongly on the arrangement of nanostructures, due to effects such as adatom migration and flux shadowing from adjacent structures, through this method multiple different InGaN emission colors can be attained2, 177, 185.

    Fig. 2  a Schematic of the SAE process for the growth of nanostructures. This figure has been adapted with permission of the Chinese Laser Press, from the Ref. 65: A. Pandey et al., “Strain-engineered N-polar InGaN nanowires: towards high-efficiency red LEDs on the micrometer scale”, Photonics Research, vol. 10, no. 12, pp. 2809-2815, 2022. b SEM image of an array of nanowires grown using SAE. c PL spectra measured from various InGaN/GaN nanowire arrays covering the visible spectrum.

    The arrangement of the nanostructures can also be set such that they form a photonic crystal or metasurface structure186, 187. By carefully designing the photonic crystal, the emission properties of the nanostructures formed within them can be enhanced – the light extraction efficiency can be increased, emission could be made more directional, and the emission spectral broadening can be greatly reduced. This approach has been further exploited in the design and fabrication of nanostructure-based surface emitting lasers, i.e., photonic crystal surface emitting lasers (PCSELs).

    Finally, the unique growth process of nanostructures can enable their easy integration onto a variety of substrates, which could be utilized to monolithically integrate the micro-LED devices with circuitry comprising of different materials, such as silicon and SiOx188, 189.

    Strong green emission has previously been demonstrated from nanowires grown using plasma-assisted molecular beam epitaxy (PA-MBE)100, 142, 158, 190, 191. To fabricate micro-LED devices, nanowire arrays were patterned on Ga-polar GaN-on-sapphire substrates for SAE using a thin 10 nm Ti mask layer100. Electron beam lithography was used to etch vias defining the injection openings into this layer. Then, using optimized growth conditions for high selectivity of growth (growth only in the openings where GaN is exposed), a multiple quantum disk InGaN/AlGaN LED was grown. Following the active region, a p-AlGaN electron blocking layer was grown to reduce electron overflow. High resolution TEM images show that this AlGaN layer, along with the AlGaN barriers in the active region, form a shell around the InGaN active region. The presence of the high bandgap Al-rich shell greatly reduced the impact of surface recombination on the nanostructures, which should already be low due to the absence of any plasma etching steps. An n++/p++ GaN tunnel junction contact layer was also incorporated above the p-GaN layer to improve the hole injection to the active region.

    To fabricate the nanowires into micro-LEDs, nanowire arrays were first filled with Al2O3 deposited by ALD, which was then etched back to reveal the top of the nanowires. Plasma-enhanced chemical vapor deposition (PECVD) was then used for depositing a thick SiO2 insulation layer. Stepper lithography was used to etch injection vias into the SiO2 layer to define the active area of the devices. Finally metal contacts were deposited and annealed. Fabricated devices exhibited relatively good I-V characteristics, and strong green electroluminescence (EL), as shown in Fig. 3a. As the growth of the InGaN was primarily along the semi-polar facets of the Ga-polar nanowires, the reduced polarization fields resulted in a small wavelength shift with injection current, with EL spectra at different currents plotted in Fig. 3b. For micro-LEDs having an area of 3 µm × 3 µm, a maximum EQE of ~5.5% was measured at a current density of ~3.4 A/cm2, shown in Fig. 3c. Furthermore, as the micro-LEDs were fabricated in arrays of nanowires, with individual nanowires having identical emission and morphology, there were relatively small variations in the normalized EQE of different area devices formed in the same array, shown in Fig. 3d.

    Fig. 3  a J-V of a fabricated green-emitting TJ micro-LED. The inset shows a device under operation. b EL spectra measured at different injection currents for the device. c EQE and output power vs. current density for a high-efficiency green micro-LED. d Normalized EQE vs lateral dimension for different devices. This figure has been reprinted from Ref. 100: X. Liu et al., “High efficiency InGaN nanowire tunnel junction green micro-LEDs”, Applied Physics Letters, vol. 119, no. 14, p. 141110, 2021., with the permission of AIP Publishing.

    Recently N-polar nanostructure-based LEDs have gained significant attention. Previous work on bottom-up nanostructure micro-LEDs was primarily aimed toward materials with metal polarity. In metal-polar devices with conventional structures having the p-type layer on top of the active region, the polarization fields inhibit carrier injection to the active region, resulting in severe electron overflow/leakage and nonradiative parasitic recombination outside of the device active region that diminishes device efficiency146, 149, 192-194. Further, the tip of metal-polar nanowires has a faceted morphology that complicates device fabrication, unlike N-polar nanowires that have flat top surfaces195. The reversed polarization fields present in N-polar devices are also beneficial to charge carrier injection and can greatly improve device emission characteristics at high current injection192. N-polar InGaN has been shown to have a higher decomposition temperature than its metal-polar counterpart, which can make higher growth temperatures possible, resulting in improved material quality196, 197.

    Taking advantage of these benefits, high efficiency N-polar green nanowire sub-micron scale LEDs have been demonstrated. To ensure the N-polarity of the grown nanowires, an N-polar GaN substrate was used for seeding the initial nanowire nucleation. Over a base n-GaN segment, a six-period InGaN quantum disk/AlGaN barrier active region was grown, followed by a p-type AlGaN electron blocking layer and a p-GaN contact layer. A schematic of the nanowires and their structure is shown in Fig. 4a. SEM images of the nanowires following growth confirm the uniform morphology and flat top surface, shown in Fig. 4b. Similar to studies on Ga-polar structures, an AlGaN shell was formed, protecting the active region from surface recombination. This was confirmed with elemental mapping of the active region, presented in Fig. 4c, for In and Al. High resolution atomic-scale images shown in Fig. 4d also confirmed the N-polarity of the grown nanowires.

    Fig. 4  a Schematic and device structure of N-polar green micro-LEDs. b SEM image of an array of InGaN/GaN nanowires. c Elemental mapping of the active region in the nanowires for In and Al. d High resolution atomic scale image of the nanowire, confirming the N-polar orientation. This figure has been reproduced with permission of the Chinese Laser Press, from the Ref. 99: X. Liu et al., “N-polar InGaN nanowires: breaking the efficiency bottleneck of nano and micro LEDs”, Photonics Research, vol. 10, no. 2, pp. 587-593, 2022.

    To fabricate devices using the above nanowire arrays, they were planarized using Al2O3 deposited by ALD and SiO2 deposited by PECVD. Lithography was used to define injection vias in the insulating SiO2 layer at the sub-micron scale. The inset of Fig. 5a shows an SEM image of the injection window for a sub-micron device which consisted of only four nanowires. Fabricated devices showed negligible reverse leakage current, as plotted in Fig. 5a. The turn-on voltage of the device could be reduced in future work by optimizing the device fabrication process. Fig. 5b plots the EL spectra from the device at different injection currents. The main emission peak is at ~530 nm, and it remained stable with varying injection current. Green emission was observed from a sub-micron device under operation, even under room-light illumination, shown in the inset of Fig. 5b. The variation of the measured output power with injection current is plotted in Fig. 5c, and the EQE vs. injection current in Fig. 5d. The EQE reached a maximum of ~11% at a relatively low current density of 0.83 A/cm2. The low current corresponding to the peak EQE suggests that the non-radiative SRH recombination is minimal in the devices . This confirmed both the excellent material quality, as well as the benefits of the bottom-up fabrication process which avoided exposing the active region to plasma damage.

    Fig. 5  a J-V of a fabricated N-polar green-emitting submicron scale LED. The inset shows an SEM image of an injection window that forms a submicron device. b EL spectra from a submicron LED measured at different injection currents. The inset is an optical microscope image of an operating device under room light illumination. Plots of measured c output power and d EQE with current density for the submicron LED. This figure has been reproduced with permission of the Chinese Laser Press, from the Ref. 99: X. Liu et al., “N-polar InGaN nanowires: breaking the efficiency bottleneck of nano and micro LEDs”, Photonics Research, vol. 10, no. 2, pp. 587-593, 2022.

    The structure of the quantum wells formed in the nanowire are also of extreme interest, as previous work has shown that InGaN insertions in N-polar GaN nanowires tend to form a faceted surface to facilitate strain relaxation27, 198-200. These facets are formed along the semi-polar planes, with different indium compositions, resulting in a complex geometry for the active region. The benefits of strain relaxation and highly confined InGaN active regions along the facets can localize charge carriers, promoting excitonic recombination, potentially further increasing the peak EQE of the device. This advantage has been harnessed recently to realize high efficiency excitonic green submicron scale nanowire LEDs98. Fig. 6a-c show TEM images of nanowires, along with high resolution images at the center and the facets for the InGaN active region within the nanowire. Elemental maps of the Ga and In ratio at different positions along the radius of the nanowire showed an increase in indium composition for the growth along the semi-polar facets. The extent of faceting is a consequence of the growth conditions and the dimensions of the nanowires, which must be controlled to tune the emission properties from the facets177, 198. The high indium composition and low strain in the faceted InGaN regions200 is essential for promoting carrier recombination within them, especially at low current injection. Further, these factors also improve the electron-hole wavefunction overlap, increasing the exciton oscillator strength and binding energy in the faceted regions201-203. The improved excitonic recombination can greatly reduce nonradiative Shockley-Reed-Hall recombination and therefore enhance device efficiency.

    Fig. 6  a STEM HAADF image of nanowires comprising an excitonic micro-LED. High-resolution HAADF images of the InGaN quantum disk active region from b the center of the nanowire and c near the sidewalls of the nanowire. d Atomic fraction of indium plotted along the growth direction in different regions of the nanowire. This figure has been reprinted with permission from Ref. 98: A. Pandey et al., “An Ultrahigh Efficiency Excitonic Micro-LED”, Nano Letters, vol. 23, no. 5, pp. 1680-1687, Mar 8, 2023. Copyright 2023 American Chemical Society.

    These nanowires were fabricated into submicron scale devices using the process described above. The inset of Fig. 7a shows a submicron injection window, similar to that used in the device. The designed area of the submicron LED is ~750 nm × 750 nm. The J-V characteristic of the fabricated device is shown in Fig. 7a, with low reverse current leakage. The inset of Fig. 7b shows a camera image of a submicron device at high injection current, with bright green emission. A microscope image of the device operating at low injection is shown in Fig. 7c. The variations of EQE and WPE vs. current density are plotted in Fig. 7b, c, respectively. The excitonic nature of recombination promoted a high efficiency, especially at low current injections, as measured here with a peak EQE of 25.2%. The corresponding peak WPE was 20.7%, suggesting efficient carrier injection. At higher injection currents, when the carrier density exceeds the Mott density, Coulombic screening would cause the excitons to disassociate204 and free carrier (electron and hole) recombination from the central region of the nanowire (away from the facets) would dominate, and this effect contributes to the sharp decline of efficiency above 0.3 A/cm2.

    Fig. 7  a J-V characteristic of the excitonic green micro-LED. The SEM image in the inset shows a submicron device injection window. b EQE and c WPE vs. current density for the device. The inset of b shows a camera image of an operational device at high current injection. The inset of c shows an optical microscope image of a device at low current injection. This figure has been reprinted with permission from Ref. 98: A. Pandey et al., “An Ultrahigh Efficiency Excitonic Micro-LED”, Nano Letters, vol. 23, no. 5, pp. 1680-1687, Mar 8, 2023. Copyright 2023 American Chemical Society.

    This study brought excitons to the foreground in the search for routes to overcome the efficiency bottleneck of a broad range of nanoscale optoelectronic and quantum devices including LEDs, lasers, detectors, and single photon source, to name a few. It also opens up another avenue in tuning the dimensions of the grown nanowires – which are crucially responsible for the extent of strain relaxation and faceting, thereby affecting excitonic recombination.

    As N-polar green emitting submicron LEDs showed excellent performance, the next step was extending them towards longer wavelength in the red64. Firstly, a relatively thick InGaN segment was used to minimize the QCSE. An in-situ anneal at a temperature 50°C higher than the growth temperature of the InGaN segment was incorporated to improve the emission intensity of the active region. Such an anneal step had previously been demonstrated to greatly enhance the luminescence by reducing the density of defects205, 206. Here, an order of magnitude increase in the photoluminescence emission was measured from the device active region using the in-situ anneal, as shown in Fig. 8a. TEM of grown nanowires in Fig. 8b showed a thick InGaN active region, and elemental mapping nanowires in Fig. 8c, plotted along the growth direction, confirmed the presence of graded interfaces, suggesting that there was significant In inhomogeneity due to effects such as composition pulling and In diffusion. This would explain the broad FWHM of the emission.

    Fig. 8  a PL spectra for a non-annealed sample and a sample with in-situ annealing. b TEM image showing the location of Ga and In atoms in the nanowire crystal. c Plot of the elemental map of In and Ga along the growth direction. This figure has been reproduced with permission of the Chinese Laser Press, from the Ref. 64: A. Pandey, Y. Malhotra, P. Wang, K. Sun, X. Liu, and Z. Mi, “N-polar InGaN/GaN nanowires: overcoming the efficiency cliff of red-emitting micro-LEDs”, Photonics Research, vol. 10, no. 4, pp. 1107-1116, 2022.

    Red-emitting micro-LEDs were then fabricated, following the process as the green N-polar submicron devices. The J-V characteristics of a 750 nm × 750 nm area device are plotted in Fig. 9a, showing a sharp turn-on at ~2.5 V, with negligible reverse leakage current. Injection-dependent EL spectra for the device have been plotted in Fig. 9b, confirming the peak emission at ~620 nm at low injection currents. At higher injection currents a blue-shift was seen in the EL peak, as is expected due to the QCSE. The EQE vs. current density has been plotted in Fig. 9c, showing a peak of ~1.2% at a current density of 0.5 A/cm2. This was the first demonstration of a submicron scale red micro-LED, and the efficiency attained was significantly better than conventional top-down fabricated micro-LEDs with areas of 100 µm2 or smaller at the time of this work.

    Fig. 9  a J-V characteristics of a red submicron LED. b EL spectra of the InGaN/GaN micro-LED measured at different injection currents. c Variation of the EQE vs. injection current density. This figure has been reproduced with permission of the Chinese Laser Press, from the Ref. 64: A. Pandey, Y. Malhotra, P. Wang, K. Sun, X. Liu, and Z. Mi, “N-polar InGaN/GaN nanowires: overcoming the efficiency cliff of red-emitting micro-LEDs”, Photonics Research, vol. 10, no. 4, pp. 1107-1116, 2022.

    To further improve the nanowire-based red micro-LEDs, an InGaN/GaN short-period superlattice (SPSL) was incorporated beneath the device active region65. Previous work has shown the benefits in strain relaxation and achieving longer wavelength emission by incorporating a SPSL in conventional planar quantum well devices207-209. The benefits of the SPSL layer for strain relaxation can be further enhanced with the use of nanostructures, due to the increased surface area to volume ratio, thereby allowing for more efficient red emission.

    A four-period InGaN (8 nm)/GaN (8 nm) SPSL was included in InGaN/GaN nanowires through which it was possible to significantly red-shift the PL emission from an InGaN dot active region, as shown in Fig. 10a. Micro-LED devices fabricated on nanowires with the incorporated SPSL showed excellent J-V characteristics, seen in Fig. 10b. The variation of the EQE and WPE with current density for the submicron LED is plotted in Fig. 10c. The EQE reached a peak value of ~2.2%, and the WPE peaked at 1.7%. The EL spectra for the devices, plotted in Fig. 10d, showed a peak emission ~630 nm. While lower injection spectra were dominated by the emission from the InGaN dot, at higher currents the SPSL also contributed a green-yellow emission peak that could distort the spectral purity of the emitted light.

    Fig. 10  a PL spectra measured for samples containing only the InGaN/GaN SPSL, only the InGaN dot, and both the SPSL and InGaN dot. b J-V characteristics and c current-dependent EQE and WPE of a red submicron LED with an InGaN/GaN SPSL incorporated beneath the active region. d EL spectra of the InGaN/GaN micro-LED measured at different injection currents. This figure has been reproduced with permission of the Chinese Laser Press, from the Ref. 65: A. Pandey et al., “Strain-engineered N-polar InGaN nanowires: towards high-efficiency red LEDs on the micrometer scale”, Photonics Research, vol. 10, no. 12, pp. 2809-2815, 2022.

    There remains significant room for performance improvement for these red micro-LEDs. The devices shown here have relatively high turn-on voltages and low WPE, which limits their practical use. To tackle these problems, further improvement and tuning of the device is in progress, with an emphasis on improving the p-doping within the GaN contact layer, as well as by including an electron-blocking layer and tunnel junction in the device heterostructure.

    While high-efficiency long-wavelength InGaN-based micro-LEDs have been demonstrated, their spectral purity has remained a challenge125. These devices typically have large full-width half maximum (FWHM) of the emission due to the inhomogeneous In distribution as a result of phase separation, composition-pulling, local In-rich clusters and inter-diffusion of In between the InGaN active region and the surrounding layers. The QCSE also causes a current dependence of the emission peak, resulting in large shifts of emission wavelength as the device is operated at higher currents. To address these problems, the use of photonic crystal arrays has been demonstrated to tune the emission properties of devices by forming optical microcavities71, 186, 210. The Purcell effect within such photonic crystals can further increase the IQE of the desired emission186. Utilizing such photonic crystals, it has been shown that the emission can be significantly narrowed to a few nanometers in wavelength178, 186, 210 thereby greatly enhancing the emission from selected optical resonance modes.

    The dimensions, spacing and arrangement of the nanostructures that form a photonic crystal play a direct role in the emission properties. Fig. 11a shows a schematic of a hexagonal array of nanowires with their reciprocal lattice vectors. Fig. 11b plots the simulated photonic band structure of an InGaN/GaN nanowire photonic crystal, having an emission wavelength λ = 505 nm, lattice constant a = 250 nm and lateral dimension d = 0.85a. The Γ point in the photonic band structure corresponds to zero group velocity of light, which is necessary for creating a resonance mode. Fig. 11c shows the calculated profile of the electric field profile for a nanowire array having a lateral dimension of 5 µm. TM-polarized light with electric field parallel to the c-plane of the nanowire crystals dominated the band-edge mode. Through further design optimization, effective guiding of the light mode has also been demonstrated using smaller arrays of nanowire-based photonic crystals186, 211.

    Fig. 11  a Schematic of a nanowire photonic crystal array, with the lattice constant a, nanowire diameter d, and the reciprocal lattice vectors labelled. b Simulated photonic band structure for a 2D hexagonal array of nanowires designed for an emission wavelength λ = 505 nm, a = 250 nm and d = 212.5 nm. c Electric field profile of the band edge mode calculated by 3D finite-difference time-domain method for a 5 µm × 5 µm nanowire array having λ = 505 nm. This figure is reprinted with the permission of John Wiley and Sons, from the Ref. 186: Y. H. Ra, R. T. Rashid, X. Liu, J. Lee, and Z. Mi, “Scalable nanowire photonic crystals: Molding the light emission of InGaN”, Advanced Functional Materials, vol. 27, no. 38, p. 1702364, 2017.

    Photonic crystals have also been incorporated in micro-LEDs creating photonic nanocrystal (PhNC) devices210. The EL spectra for a 3 µm × 3 µm PhNC micro-LED showed a stable emission peak at ~548 nm over nearly four orders of magnitude variations of injection currents, illustrated in Fig. 12a186, 210, 212, which showed the successful mitigation of the impact of the QCSE-related blue-shift in emission45, 46, 58, 66. A relatively low efficiency droop of only ~30% was measured up to an injection current of 200 A/cm2, shown in Fig. 12b.

    Fig. 12  a EL spectra of a PhNC micro-LED with different injection currents. b Relative EQE of the PhNC micro-LED versus injection current. This figure is reprinted from Ref. 210: X. Liu, Y. Wu, Y. Malhotra, Y. Sun, and Z. Mi, “Micrometer scale InGaN green light emitting diodes with ultra-stable operation”, Applied Physics Letters, vol. 117, no. 1, p. 011104, 2020, with the permission of AIP Publishing.

    Dense, small-area lasers are of interest for applications including visible light communication, data storage and biosensing187, 213, 214. Photonic nanocrystal-based devices can also be extended for creating surface-emitting lasers which can be used for the aforementioned purposes212. Typical vertical-cavity surface-emitting lasers (VCSELs) require high quality dielectric Bragg reflectors (DBRs), which is a complex task considering the lattice mismatch between III-nitride layers and the relatively small difference in dielectric constant215-217. This can be avoided in photonic crystals, where the photonic band edge resonant effect allows the formation of standing waves211, 212 in nanocrystal surface-emitting lasers (NCSELs). Fig. 13a schematically depicts a green-emitting NCSEL, with the inset containing optical microscope images of a fabricated laser diode when switched off and under operation. Fabricated NCSELs showed a turn-on voltage of ~3.3 V at room temperature, with the J-V curve plotted in Fig. 13b. The semi-log scale plot in the inset confirmed low leakage current for the device under reverse bias. The EL spectra at different injection currents, measured under continuous-wave bias, are shown in Fig. 13c. At lower injection currents a broad emission peak was seen, however as the current increased, a lasing peak at ~523.1 nm began to dominate the spectrum with a narrow linewidth. The output power of the laser is plotted against the injection current in Fig. 13d, with a threshold current Jth ~400 A/cm2, that is significantly lower than conventional III-nitride VCSELs187, 218-222. Fig. 13d, f plot the FWHM and wavelength, respectively, of the lasing peak against injection current. As the injection crossed the lasing threshold, there was a sharp reduction in the FWHM from ~30 nm to ~0.8 nm. The lasing peak position showed little variation with increasing injection current and remained stable at ~523.1 nm above threshold.

    Fig. 13  a Schematic of an InGaN/GaN NCSEL. Optical microscope images of a green NCSEL under operation and when off are shown in the inset. b J-V characteristics of the green NCSEL, with a semi-log scale plot of the J-V curve in the inset. c Injection-dependent EL spectra measured at different injection currents under CW bias at room temperature. d Output power vs. injection current of the NCSEL, showing a threshold current of ~400 A/cm2. Plots of e spectral linewidth and f peak wavelength with injection current. This figure is from Ref. 212: Y.-H. Ra, R. T. Rashid, X. Liu, S. M. Sadaf, K. Mashooq, and Z. Mi, “An electrically pumped surface-emitting semiconductor green laser”, Science Advances, vol. 6, no. 1, p. eaav7523, 2020. © The Authors, some rights reserved; exclusive licensee AAAS. Distributed under a CC BY-NC 4.0 license http://creativecommons.org/licenses/by-nc/4.0/”. Reprinted with permission from AAAS.

    This work showcased the extreme versatility of the selective area growth method for the realization of micron-scale or smaller optoelectronic devices. The measured threshold current density is significantly lower than conventional planar devices operating at a similar emission wavelength. The selective area growth of the photonic crystal also allows lasing without the presence of a distributed Bragg reflector (DBR) – a thick, multi-layer structure that is quite complex to grow with high quality – for the III-nitrides. While the severe heating effect limited high-power operation of this device, proper packaging and thermal management can greatly benefit its performance for potential applications in projectors, optical storage and communication, as well as displays.

    As nanostructures can be grown to be nearly defect-free on a variety of substrates, even those with a large lattice mismatch, this opens up an avenue for the direct integration of micro-LEDs with their back-plane circuitry. Green-emitting N-polar nanowire micro-LEDs with stable operation were demonstrated on silicon substrates, showing stable emission by engineering the strain and polarization within the active region, which is comprised of InGaN quantum wells and AlGaN barriers223. An SEM image of the grown nanowires is shown in Fig. 14a. The AlGaN barrier in the active region of the nanowires formed a shell structure around the InGaN quantum well active region, shown in the TEM images in Fig. 14b, c. This thin AlGaN shell minimized the impact of surface recombination on the InGaN quantum wells. Further, the use of AlGaN barriers induced a tensile strain within the grown layers, that compensated for the compressive strain in the InGaN layers. The strain compensation promoted In incorporation within the nanowires and also assisted in screening of the QCSE. The J-V characteristic of fabricated sub-micron scale nanowire micro-LED devices is shown in Fig. 14d, displaying low reverse leakage current and a rectification ratio of over four orders of magnitude at voltages of ±8 V. The inset of Fig. 14d shows bright green emission from the device under operation. A stable emission peak was measured shown in the EL spectra plotted in Fig. 14e. The variation of the peak position with injection current is shown in Fig. 14f, confirmed the stable emission, with negligible change up to injection currents of over 1 kA/cm2. The potential integration of micro-LED structures epitaxially grown directly on Si wafer can significantly reduce the manufacturing cost and complexity involved in the integration for display applications.

    Fig. 14  a SEM image of an N-polar nanowire array grown on silicon. b HAADF-STEM image of the active region of the nanowires. The green lines show the AlGaN shell around the active region. c High-magnification HAADF image of the cyan-box region in b, showing the formation of a GaN/AlGaN superlattice on the sidewall. d J-V curve for a sub-micron scale green-emitting nanowire micro-LED. The inset shows an operating device. e Injection-dependent EL spectra of the device. f Variation of the emission peak with injection current for the device. This figure has been reproduced from the Ref. 188: Y. Wu et al., “InGaN micro-light-emitting diodes monolithically grown on Si: achieving ultra-stable operation through polarization and strain engineering”, Light: Science & Applications, vol. 11, no. 1, pp. 1-9, 2022.

    As has been previously shown, the emission color of nanowires is also strongly dependent on their dimensions and spacing2, 177, 224-226. During selective area growth of InGaN by molecular beam epitaxy, the incorporation of metal adatoms depends on both the incoming metal flux, along with a contribution from the adatom migration along the lateral surfaces of the nanostructures. This is especially true for In adatoms, which have a diffusion length of ~100 nm depending on the growth temperature177, that is comparable to the dimensions of nanostructures. It should be noted that there is also a difference between the epitaxy of tightly packed nanowire arrays and single nanowires with large spacing in between them, as nanowire arrays have the added effect of shadowing of the impinging metal flux. By designing nanowires with different diameters, the nanowires will have different InGaN compositions, and hence different emission wavelengths. By exploiting this unique effect, nanowires with emission from blue to red, covering the entire visible spectrum, were grown in a single epitaxial step, using selective-area epitaxy by patterning openings of different diameters on a substrate as shown in Fig. 15a224. The nanowires of smaller diameters were observed to have a red-shifted emission wavelength. A schematic of the grown nanowires is shown in Fig. 15b with the emission color of the different nanowires indicated in the diagram, along with the heterostructure of the nanowires. An SEM image of the grown nanowires is shown in Fig. 15c. The grown nanowires were fabricated into single nanowire devices, schematically shown in Fig. 16a. Probing the devices resulted in emission peaks in the blue, green, orange, and red wavelengths, attained by reducing the diameter of the nanowires. The EL spectra for the emission of the different diameter nanowire devices are shown in Fig. 16b, spanning from ~460 nm to ~660 nm. This demonstration of multi-color devices that could together form a pixel, obtained from a single growth on a single chip, highlights the unique capabilities of the selective-area epitaxy process, which could significantly simplify the integration of nano or micro-LEDs into displays for practical applications.

    Fig. 15  a Schematic of the different size openings patterned for selective-area epitaxy. b Schematic of nanowires with different diameters having different color emission wavelengths. The different layers that comprise the nanowires are shown on the right. c SEM image of the different size nanowires that were grown. This figure has been reprinted with permission from Ref. 224: Y.-H. Ra et al., “Full-color single nanowire pixels for projection displays”, Nano Letters, vol. 16, no. 7, pp. 4608-4615, 2016. Copyright 2016 American Chemical Society.
    Fig. 16  a Schematic of fabricated single-wire devices having different diameters and emission wavelengths. b EL spectra of the different diameter nanowire devices. This figure has been reprinted with permission from Ref. 224: Y.-H. Ra et al., “Full-color single nanowire pixels for projection displays”, Nano Letters, vol. 16, no. 7, pp. 4608-4615, 2016. Copyright 2016 American Chemical Society.

    From an application perspective, display technologies present the most immediate use for micro-LEDs, however micro and nanoscale optoelectronic devices will significantly impact other emerging applications as well. For example, micro-LED based optical interconnects have drawn considerable attention recently. Unlike electrical links which require high power for long distances, optical interconnects would have extremely low power consumption. The development of high-density micro-LED based optical transceivers has also enabled high bandwidths that are robust over a wide range of environmental conditions227, 228. Micro-LEDs have also attracted attention in the field of biological sensing, where they have been shown to potentially replace complicated existing sensors, as the diode of the LED can both harvest energy wirelessly (by absorbing photons) and transmit signals (by emitting photons)229. Nitride-based micro-LEDs have also been used for imaging dyes, fluorescence spectroscopy and even stimulating neurons230. Some of these unique applications may require the development of specialized micro-LED devices, that could be more easily achieved with the versatile nanostructure-based approach.

    Nanostructures have shown tremendous potential to overcome critical challenges of nano or micron-scale optoelectronic devices. The difficulties in creating long-wavelength III-nitride optoelectronic devices are greatly reduced in such nanostructures, and by utilizing them as the basis of micro-LED devices, it is possible to attain efficiencies significantly greater than conventional top-down micro-LEDs, while reducing the dimensions of the devices down to the sub-micron scale. The benefits of nanostructures are further expanded through the selective-area epitaxy process. It enables the bottom-up formation of photonic crystals, that could be used to optimize the emission properties of LEDs, and even be incorporated for low-threshold, DBR-free surface-emitting lasers. The major challenge of the integration of millions of individual, multi-color micro-LED devices into a practical display can be greatly simplified through the selective-area growth approach by allowing for the growth of complete pixels in a single epitaxy step, on a variety of substrates.

    These previous works have shown the promise of nanostructures and shone light on some active areas of research interest. The efficiency of green and red micro-LEDs can be improved through optimization of the p-doping in the nanowire heterostructures101. Furthermore, in the case of red-emitting micro-LEDs, research is ongoing to further improve the strain relaxation, as well as to inhibit the parasitic recombination at high currents from the SPSL layers. Nanowire micro-LEDs also typically peak in efficiency at relatively low injection currents, showing strong droop at higher injection currents. The use of an AlGaN electron blocking layer in these devices could reduce the droop effect, and also form an Al-rich shell around the active region. While a tunnel junction contact has been demonstrated for nanowires100, 231, 232, the use of such a structure in N-polar devices, grown using selective area epitaxy, is of ongoing interest as it can improve both the charge carrier injection, as well as reduce efficiency droop. Optimized devices could be used in photonic crystals to realize narrow-linewidth, spectrally pure high efficiency micro-LEDs, which could even be integrated monolithically on a single chip. Finally, as the properties of nanostructures grown using selective-area epitaxy is sensitive to their dimensions, it is critical to carefully control and fabricate the substrate before epitaxy, as well as the nanostructure devices afterwards.

    In polar semiconductors such as InGaN, the distinct ionic character can lead to strong electron phonon coupling effect, which may significantly impact their electronic, optical and excitonic properties. Recent theoretical and experimental studies have revealed that the exciton binding energy in nanoscale III-nitride heterostructures can be dramatically increased, compared to their bulk structures188, 233-236. As an example, the exciton oscillator strength can be enhanced by one to two orders of magnitude in InGaN nanostructures with efficient strain relaxation235. Theoretical studies have further shown that polaronic exciton contribution to the binding energy can be as large as 190 meV in GaN nanowires237. Moreover, recent studies have shown that the strong exciton-phonon interaction, i.e., the formation of polaronic excitons, can further impact the charge carrier transport, relaxation, and recombination. For example, the unique polaronic exciton effect can transform an indirect bandgap h-BN to be extremely bright light emitters in the deep UV238. As such, we envision that a fundamental study of excitons in InGaN deep nanostructures could offer a path to break the efficiency bottleneck of micro and nanoscale LEDs.

    The potential uses for micro-LED technologies have motivated considerable resources into their development. While obstacles remain in the path of nanostructure-based micro-LEDs, they can be solved by a fundamental understanding of the physics and properties of III-nitride-nanostructures and the further development and refinement of the epitaxy and fabrication methodologies. Accordingly, III-nitride nanostructures offer a very promising path to overcome the efficiency, scaling, and integration challenges of micro-LEDs for many emerging and demanding applications.

    The contents presented in this review article are based on results published in the literature, including those at the University of Michigan. For the studies performed at the University Michigan, we acknowledge the support from NS Nanotech, Inc., U.S. Army Research Office, and National Science Foundation.

    Z.M. supervised the project. A.P., M.R. and Z..M. contributed to the writing of the manuscript.

    Some IP related to the work of nanowire LEDs at McGill University and University of Michigan was licensed to NS Nanotech, Inc., which was co-founded by Z. Mi. The University of Michigan and Z. Mi have a financial interest in the company.

    Z.M. supervised the project. A.P., M.R. and Z..M. contributed to the writing of the manuscript.

    Some IP related to the work of nanowire LEDs at McGill University and University of Michigan was licensed to NS Nanotech, Inc., which was co-founded by Z. Mi. The University of Michigan and Z. Mi have a financial interest in the company.

  • [1] Ma, X. X. et al. A functionalized graphene oxide–iron oxide nanocomposite for magnetically targeted drug delivery, photothermal therapy, and magnetic resonance imaging. Nano Research 5, 199-212 (2012). doi: 10.1007/s12274-012-0200-y
    [2] Werfel, F. N. et al. Superconductor bearings, flywheels and transportation. Superconductor Science and Technology 25, 014007 (2012). doi: 10.1088/0953-2048/25/1/014007
    [3] Levchenko, I. et al. Recent progress and perspectives of space electric propulsion systems based on smart nanomaterials. Nature Communications 9, 879 (2018). doi: 10.1038/s41467-017-02269-7
    [4] Dey, T. K., Chattopadhyay, M. K. & Ghatak, S. K. Magnetic field sensor based on second harmonic response of polycrystalline YBCO pellets. Sensors and Actuators B:Chemical 55, 222-226 (1999). doi: 10.1016/S0925-4005(99)00089-1
    [5] Muhammad, F. et al. Measurement of magnetic field components using a single passive SAW magnetic sensor. Sensors and Actuators A:Physical 352, 114163 (2023). doi: 10.1016/j.sna.2023.114163
    [6] Niekiel, F. et al. Highly sensitive MEMS magnetic field sensors with integrated powder-based permanent magnets. Sensors and Actuators A:Physical 297, 111560 (2019). doi: 10.1016/j.sna.2019.111560
    [7] Liu, H. et al. An overview of sensing platform-technological aspects for vector magnetic measurement: a case study of the application in different scenarios. Measurement 187, 110352 (2022). doi: 10.1016/j.measurement.2021.110352
    [8] Li, Y. X. et al. All-fiber-optic vector magnetic field sensor based on side-polished fiber and magnetic fluid. Optics Express 27, 35182-35188 (2019). doi: 10.1364/OE.27.035182
    [9] Pu, S. L. et al. Microfiber coupling structures for magnetic field sensing with enhanced sensitivity. IEEE Sensors Journal 17, 5857-5861 (2017). doi: 10.1109/JSEN.2017.2734908
    [10] Liu, Q., Li, S. G. & Wang, X. Y. Sensing characteristics of a MF-filled photonic crystal fiber Sagnac interferometer for magnetic field detecting. Sensors and Actuators B:Chemical 242, 949-955 (2017). doi: 10.1016/j.snb.2016.09.160
    [11] Yin, J. D. et al. All-fiber-optic vector magnetometer based on nano-magnetic fluids filled double-clad photonic crystal fiber. Sensors and Actuators B:Chemical 238, 518-524 (2017). doi: 10.1016/j.snb.2016.07.100
    [12] Wang, J. et al. Magnetic field and temperature dual-parameter sensor based on magnetic fluid materials filled photonic crystal fiber. Optics Express 28, 1456-1471 (2020). doi: 10.1364/OE.377116
    [13] Gu, S. F. et al. Highly sensitive magnetic field measurement with taper-based in-line Mach-Zehnder interferometer and Vernier effect. Journal of Lightwave Technology 40, 909-917 (2022). doi: 10.1109/JLT.2022.3141450
    [14] Zhang, D. W. et al. Highly sensitive magnetic field microsensor based on direct laser writing of fiber-tip optofluidic Fabry–Pérot cavity. APL Photonics 5, 076112 (2020). doi: 10.1063/5.0012988
    [15] Xiong, C. et al. Fiber-tip polymer microcantilever for fast and highly sensitive hydrogen measurement. ACS Applied Materials & Interfaces 12, 33163-33172 (2020).
    [16] Liao, C. R. et al. Design and realization of 3D printed fiber-tip microcantilever probes applied to hydrogen sensing. Light:Advanced Manufacturing 3, 5 (2022).
    [17] Huang, H. Q. et al. Four-dimensional printing of a fiber-tip multimaterial microcantilever as a magnetic field sensor. ACS Photonics 10, 1916-1924 (2023). doi: 10.1021/acsphotonics.3c00347
    [18] Wang, F. M. et al. Three-dimensional printed microcantilever with mechanical metamaterial for fiber-optic microforce sensing. APL Photonics 8, 096108 (2023). doi: 10.1063/5.0159706
    [19] Zou, M. Q. et al. 3D printed fiber-optic nanomechanical bioprobe. International Journal of Extreme Manufacturing 5, 015005 (2023).
    [20] Zou, M. Q. et al. Fiber-tip polymer clamped-beam probe for high-sensitivity nanoforce measurements. Light:Science & Applications 10, 171 (2021).
    [21] Shen, Z. H. et al. Dual self-growing polymer microtips on a multicore fiber for humidity and temperature discriminative sensing. Journal of Lightwave Technology 41, 4322-4330 (2023). doi: 10.1109/JLT.2022.3166353
    [22] Yariv, A. & Yeh, P. Photonics: Optical Electronics in Modern Communications. 6th edn. (New York: Oxford University Press, 2006).
    [23] Gómez-Pastora, J. et al. Analysis of separators for magnetic beads recovery: from large systems to multifunctional microdevices. Separation and Purification Technology 172, 16-31 (2017). doi: 10.1016/j.seppur.2016.07.050
    [24] Crangle, J. & Goodman, G. M. The magnetization of pure iron and nickel. Proceedings of the Royal Society A:Mathematical,Physical and Engineering Sciences 321, 477-491 (1971).
    [25] Li, Q. et al. Softening hard stool using magnetically controlled Fe3O4 nanoparticles. Sensors and Materials 29, 1749-1761 (2017).
    [26] Li, M. et al. Ultracompact fiber sensor tip based on liquid polymer-filled Fabry-Perot cavity with high temperature sensitivity. Sensors and Actuators B:Chemical 233, 496-501 (2016). doi: 10.1016/j.snb.2016.04.121
    [27] Gissibl, T. et al. Two-photon direct laser writing of ultracompact multi-lens objectives. Nature Photonics 10, 554-560 (2016). doi: 10.1038/nphoton.2016.121
    [28] Li, B. Z. et al. Femtosecond laser 3D printed micro objective lens for ultrathin fiber endoscope. Fundamental Research 4, 123-130 (2024). doi: 10.1016/j.fmre.2022.05.026
    [29] Woods, L. M. et al. Materials perspective on Casimir and van der Waals interactions. Reviews of Modern Physics 88, 045003 (2016). doi: 10.1103/RevModPhys.88.045003
    [30] Ji, W. et al. Spacing-tailored multicore fiber interface for efficient FIFO devices. Journal of Lightwave Technology 40, 5682-5688 (2022). doi: 10.1109/JLT.2022.3177622
    [31] Ma, J. et al. High-sensitivity and fast-response fiber-tip Fabry–Pérot hydrogen sensor with suspended palladium-decorated graphene. Nanoscale 11, 15821-15827 (2019). doi: 10.1039/C9NR04274A
    [32] Li, W. B. et al. Determination of the temperature and concentration dependence of the refractive index of a liquid mixture. The Journal of Chemical Physics 101, 5058-5069 (1994). doi: 10.1063/1.467428
    [33] Xia, J. et al. A magnetic field sensor based on a magnetic fluid-filled FP-FBG structure. Sensors 16, 620 (2016).
    [34] Li, X. G. et al. Measurement of magnetic field and temperature based on fiber-optic composite interferometer. IEEE Transactions on Instrumentation and Measurement 66, 1906-1911 (2017). doi: 10.1109/TIM.2017.2670067
    [35] Zhang, R. et al. Mach-Zehnder interferometer cascaded with FBG for simultaneous measurement of magnetic field and temperature. IEEE Sensors Journal 19, 4079-4083 (2019).
    [36] Liu, H. F. et al. Ultra-sensitive magnetic field sensor with resolved temperature cross-sensitivity employing microfiber-assisted modal interferometer integrated with magnetic fluids. Applied Physics Letters 109, 042402 (2016). doi: 10.1063/1.4959838
    [37] Li, X. G. et al. Multi-modes interferometer for magnetic field and temperature measurement using Photonic crystal fiber filled with magnetic fluid. Optical Fiber Technology 41, 1-6 (2018). doi: 10.1016/j.yofte.2017.12.002
  • Cited by(14)

    1. Li, Y., Zhang, K., Zhi, T. et al. 3175 PPI active-matrix Micro-LED device array towards full high-definition light engine. Materials Science in Semiconductor Processing, 2025. doi: 10.1016/j.mssp.2024.109178
    2. Li, P., Wu, W., Shi, Y. et al. Integrated Broadband THz Switching using Photoconductivity Modulation in Si-on-Sapphire Substrates. IEEE Transactions on Terahertz Science and Technology, 2025. doi: 10.1109/TTHZ.2025.3537460
    3. Jiang, Y., Jiang, Z., Mo, M. et al. Impact of Mg-Doped AlGaN Electron Blocking Layer on Micro-LEDs: A Comparative Analysis of Carrier Transport Versus Chip Size and Current Density. IEEE Transactions on Electron Devices, 2025, 72(1): 306-311. doi: 10.1109/TED.2024.3509822
    4. Veeramuthu, V., Kim, S.-U., Lee, S.-W. et al. Scalable InGaN nanowire µ-LEDs: paving the way for next-generation display technology. National Science Review, 2025, 12(1): nwae306. doi: 10.1093/nsr/nwae306
    5. Tsai, C.-Y., Chiu, H.-S., Cheng, Y.-C. et al. Nanoparticle-Enhanced Ultrathin and Ultrahigh-Definition Color Converted Panchromatic Microarrays for Microlight-Emitting Diode Displays. ACS Applied Electronic Materials, 2024, 6(12): 8879-8888. doi: 10.1021/acsaelm.4c01590
    6. Wang, D., Wang, P., Mondal, S. et al. Nanoscale Engineering of Wurtzite Ferroelectrics: Unveiling Phase Transition and Ferroelectric Switching in ScAlN Nanowires. ACS Applied Nano Materials, 2024, 7(23): 26756-26764. doi: 10.1021/acsanm.4c04549
    7. Gridchin, V.O., Shugabaev, T., Lendyashova, V.V. et al. Growth of long core-shell InGaN nanowires by plasma-assisted molecular beam epitaxy with gradually increasing substrate temperature | [Выращивание длинных нитевидных нанокристаллов InGaN со структурой ядро-оболочка методом плазменно-активированной молекулярно-пучковой эпитаксии при постепенном повышении температуры подложки]. St. Petersburg State Polytechnical University Journal: Physics and Mathematics, 2024, 17(32): 143-147. doi: 10.18721/JPM.173.228
    8. Ji, L., Zhang, J., Zhang, G. et al. A polyamide-facilitated soldering approach for Mini LED precise alignment leveraging 3D interfacial networks. Journal of Science: Advanced Materials and Devices, 2024, 9(4): 100817. doi: 10.1016/j.jsamd.2024.100817
    9. Qian, Y., Hsiang, E.-L., Huang, Y.-H. et al. High-Efficiency Vertical-Chip Micro-Light-Emitting Diodes via p-GaN Optimization and Surface Passivation. Crystals, 2024, 14(6): 503. doi: 10.3390/cryst14060503
    10. Shugabaev, T., Gridchin, V.O., Kuznetsov, A. et al. Influence of the wet-chemical treatment on the optical and structural properties of core-shell InGaN nanowires | [Влияние химического травления на оптические и структурные свойства нитевидных нанокристаллов InGaN ядро-оболочка]. St. Petersburg State Polytechnical University Journal: Physics and Mathematics, 2024, 17(3.1): 306-309. doi: 10.18721/JPM.173.162
    11. Zhao, S.. III-nitride nanostructures: Selective area epitaxy, surface emitting lasers, and emerging photodetectors. Semiconductors and Semimetals, 2024. doi: 10.1016/bs.semsem.2024.10.001
    12. Xiao, Y., Wu, Y., Reddeppa, M. et al. Spectrally Pure, High Operational Dynamic Range, Deep Red Micro-LEDs. Nano Letters, 2024. doi: 10.1021/acs.nanolett.4c03262
    13. Haupt, O., Griffel, T., Fatahilah, M. et al. Placement Accuracy of MicroLEDs in the 5μm Size Range Being Laser Mass Transferred. Digest of Technical Papers - SID International Symposium, 2024, 55(1): 1282-1284. doi: 10.1002/sdtp.17778
    14. Bosch, J., Durand, C., Alloing, B. et al. Fabrication strategies of flexible light sources based on micro/nano III-nitride LEDs. Journal of Information Display, 2024, 25(1): 61-73. doi: 10.1080/15980316.2024.2310638
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Figures(9) / Tables(2)

Research Summary

Miniaturized discriminative sensing: 3D printed multicore fiber-tip probes

Multicore fiber contains multiple optical transmission channels in a single fiber, providing the advantages of high integration and space division multiplexing. The 3D printing technique can efficiently fabricate multiple customized sensing elements on the end facet of multicore fiber, making it widely utilizable in ultracompact fiber-tip devices with an exquisite structure incorporating functional materials. Li-Min Xiao from China’s Fudan University and colleagues now report a 3D printed multicore fiber-tip composite probes for magnetic field and temperature discriminative sensing. The multi-probes can provide an ultracompact, sensitive, fast, and reliable scheme for discriminative measurement, especially the size of sensing space is extremely limited. The bowl-shaped microcantilever can provide a useful hybrid platform for incorporating 3D printed microstructures with fruitful functional materials.


show all

Article Metrics

Article views(2947) PDF downloads(901) Citation(14) Citation counts are provided from Web of Science. The counts may vary by service, and are reliant on the availability of their data.

3D printed multicore fiber-tip discriminative sensor for magnetic field and temperature measurements

  • 1. Advanced Fiber Devices and Systems Group, Key Laboratory of Micro and Nano Photonic Structures (MoE), Key Laboratory for Information Science of Electromagnetic Waves (MoE), Shanghai Engineering Research Center of Ultra-Precision Optical Manufacturing, School of Information Science and Technology, Fudan University, Shanghai, China
  • 2. Zhongtian Technology Advanced Materials Co., Ltd. Nantong, Jiangsu, China
  • 3. Key laboratory of Specialty Fiber Optics and Optical Access Networks, Joint International Research Laboratory of Specialty Fiber Optics and Advanced Communication, Shanghai University. Shanghai, China
  • 4. Department of Electrical and Electronic Engineering, The Hong Kong Polytechnic University, Hong Kong, China
  • Corresponding author:

    Limin Xiao, liminxiao@fudan.edu.cn

doi: https://doi.org/10.37188/lam.2024.018

Abstract: Miniaturized fiber-optic magnetic field sensors have attracted considerable interest owing to their superiorities in anti-electromagnetic interference and compactness. However, the intrinsic thermodynamic properties of the material make temperature cross-sensitivity a challenging problem in terms of sensing accuracy and reliability. In this study, an ultracompact multicore fiber (MCF) tip sensor was designed to discriminatively measure the magnetic field and temperature, which was subsequently evaluated experimentally. The novel 3D printed sensing component consists of a bowl-shaped microcantilever and a polymer microfluid-infiltrated microcavity on the end-facet of an MCF, acting as two miniaturized Fabry-Perot interferometers. The magnetic sensitivity of the microcantilever was implemented by incorporating an iron micro ball into the microcantilever, and the microfluid-infiltrated microcavity enhanced the capability of highly sensitive temperature sensing. Using this tiny fiber-facet device in the two channels of an MCF allows discriminative measurements of the magnetic field and temperature by determining the sensitivity coefficient matrix of two parameters. The device exhibited a high magnetic field intensity sensitivity, approximately 1805.6 pm/mT with a fast response time of ~ 213 ms and a high temperature sensitivity of 160.3 pm/℃. Moreover, the sensor had a low condition number of 11.28, indicating high reliability in two-parameter measurements. The proposed 3D printed MCF-tip probes, which detect multiple signals through multiple channels within a single fiber, can provide an ultracompact, sensitive, and reliable scheme for discriminative measurements. The bowl-shaped microcantilever also provides a useful platform for incorporating microstructures with functional materials, extending multi-parameter sensing scenarios and promoting the application of MCFs.

Research Summary

Miniaturized discriminative sensing: 3D printed multicore fiber-tip probes

Multicore fiber contains multiple optical transmission channels in a single fiber, providing the advantages of high integration and space division multiplexing. The 3D printing technique can efficiently fabricate multiple customized sensing elements on the end facet of multicore fiber, making it widely utilizable in ultracompact fiber-tip devices with an exquisite structure incorporating functional materials. Li-Min Xiao from China’s Fudan University and colleagues now report a 3D printed multicore fiber-tip composite probes for magnetic field and temperature discriminative sensing. The multi-probes can provide an ultracompact, sensitive, fast, and reliable scheme for discriminative measurement, especially the size of sensing space is extremely limited. The bowl-shaped microcantilever can provide a useful hybrid platform for incorporating 3D printed microstructures with fruitful functional materials.


show all
    • Magnetic field sensing plays a pivotal role in numerous fields such as medicine, transportation, and aerospace1-3. Among the various types of magnetic field sensors that have been developed4-7, optical fiber-based sensors possess outstanding characteristics such as compactness, long-distance interrogation, low cost, and high sensitivity, thus attracting significant interest8-14. In recent years, various optical fibers integrated with magnetic functional materials have been extensively investigated for performance optimization. Typically, side-polished fibers8, tapered fibers9, photonic crystal fibers10-12, fiber Mach-Zender interferometers (MZIs)13, and fiber Fabry-Perot interferometers (FPIs)14, can be used to detect the magnetic field when incorporated with magnetic fluids (MFs). In particular, the sensitivity of the side-polished fibers filled with MFs has reached 2370 pm/mT8. However, a sensor based on the magneto-optical effect of MFs may be affected by temperature perturbation. In addition, the packaging of fluids leads to bulky implementations that require laborious operations. Temperature crosstalk can be effectively eliminated by integrating multiple sensing elements12, at the expense of increasing the size of the entire sensing component; however, the different spatial locations of multiple elements can cause measurement errors in multi-parameter discriminative sensing.

      Two-photon polymerization (TPP), a 3D printing technique with high fabrication accuracy and flexibility, can effectively fabricate micro/nanostructures on an optical fiber platform, achieving superior results in realizing complex high-performance optical, mechanical, and biological structures. A variety of fiber-tip devices such as microcantilevers15-19 and clamped-beam probes20 have been implemented on the fiber tip, using the TPP technique for hydrogen, nanoforce, magnetic field, and biosensing applications. However, 3D printed fiber-tip devices are generally implemented based on an ordinary single-core fiber; and the manufactured tip element is based on a single channel, making it difficult to integrate functional multi-elements. A multicore fiber (MCF) contains multiple optical transmission channels in a single fiber, providing the advantages of high integration and space division multiplexing. We implemented an MCF-tip temperature and humidity discriminative sensor by fabricating dual microtips21; however, the polymer self-growing fabrication method cannot customize different structural patterns without restraint. The TPP 3D printing technique can efficiently fabricate multiple customized sensing elements on the end facet of MCF based on various application scenarios, making it widely utilizable in ultracompact MCF-tip devices with an exquisite structure incorporating functional materials.

      In this study, we propose, design, and experimentally demonstrate ultracompact MCF-tip probes for magnetic-field and temperature-discriminative sensing (Fig. 1). Using the TPP technique, a bowl-shaped microcantilever and microfluid-infiltrated microcavity were printed on two different cores of an MCF, acting as two miniaturized FPIs. An iron ball was incorporated inside the bowl-shaped tip of the microcantilever to render it magnetically sensitive, and the microfluid-infiltrated microcavity acted as a highly sensitive temperature-sensing element. Discriminative measurements of the two parameters can be realized using a sensitivity coefficient matrix. By attaching an iron ball into a 3D-printed bowl of the microcantilever with 2 μm thickness, a high magnetic field intensity sensitivity of 1805.6 pm/mT with a fast response time of ~ 213 ms was obtained. The microfluid-infiltrated microcavity exhibited a high temperature sensitivity of 160.3 pm/℃ in the range of 25℃ to 55℃. The condition number of the sensor was as low as 11.28, indicating that the sensor has high reliability in the discriminative measurement of the magnetic field and temperature. The spatial location accuracy in discriminative measurements was significant due to the highly integrated sensing components on the MCF-tip with core-to-core spacing of 50 μm. The proposed 3D printed MCF-tip probes have the capability to detect multiple signals on a tiny fiber tip through multiple channels within a single fiber, providing an ultracompact, sensitive, and reliable scheme for discriminative measurements when the sensing space is extremely limited, as in medical, transportation, and aerospace applications.

      Fig. 1  Schematic diagram of the MCF-tip probes for magnetic field and temperature discriminative sensing.

    Design and Principle
    • The principle of the MCF-tip dual FPIs for discriminative measurement of the magnetic field and temperature is shown in Fig. 2. A polymer microcantilever and microfluid-infiltrated microcavity are printed on the end facet of the MCF to reflect the light emerging from the two different cores. Light from the fiber core is partially reflected back to the fiber core under the action of Mirror-1. The transmitted light is then partially reflected by Mirror-2 and collected by the fiber core. Thus, when the two beams of reflected light with a phase difference meet in the fiber core, they generate an interference resonance. The reflected light intensity of the two-beam interference in the fiber FPI is expressed as22,

      Fig. 2  Principle of FPIs with dual-tip MCF for discriminative measurement of magnetic field and temperature.

      IR=I0[R1+R2η2R1R2ηcos(Δδ+δ0)]I0[1γcos(Δδ+δ0)]
      (1)

      where IR is the intensity of the interference light; I0 is the intensity of the light discharged into the FPI; R1 and R2 are the reflectivities of the two mirrors; η is the transmission coefficient of the FP cavity; Δδ is the phase difference between the two beams of light; δ0 is the initial phase of the incident light; and γ is the extinction ratio of the reflection spectrum. Here, the phase difference can be written as:

      Δδ=2πλΔQ=2πλ2nL
      (2)

      where λ is the wavelength; ΔQ is the optical path difference (OPD) between the two light beams; n is the effective refractive index (RI) of the medium in the FP cavity; and L is the cavity length. Variations in n and L can lead to variations in OPD, which is represented by a spectral shift of the FPI interference resonance wavelength.

      In FPI-1, Mirror-1 is the interface between the fiber end facet and air, and Mirror-2 is the interface between air and the polymer microcantilever. A magnetic field can cause the microcantilever to deform when incorporated into the iron ball. The variation in the deflection of the microcantilever is equivalent to the variation in the cavity length of FPI-1. The relationship between the magnetic force F (kN) acting on the microcantilever and deflection ΔL (mm) can be described as20,

      ΔL=FL33EI
      (3)

      where L (mm) is the microcantilever length; E (GPa) is Young’s modulus of the polymer; and I is the second moment of the area of the microcantilever. In the experiments, the printed microcantilever was rectangular, and the second moment of the area Ire was calculated as20,

      Ire=bh312
      (4)

      where b (mm) and h (mm) are the width and thickness of the microcantilever, respectively.

      The magnetic force F attracting the iron ball can be calculated using the “effective” dipole moment method in which the magnetized particle is replaced by an “equivalent” point dipole with a moment. Briefly, the F acting on the dipole is calculated as23,

      F=μ0V(M)H
      (5)

      where μ0 is the permeability of free space; V is the volume of the iron ball; M is the magnetization of the iron ball; and H is the intensity of the magnetic field. Because the measured magnetic field intensity was between 30-90 mT in our experiment, which is far less than the magnetic field intensity when pure iron has a saturated magnetization24, the unsaturated magnetization of the iron ball can be expressed as23,

      M = 3χχ+3H
      (6)

      where χ denotes the susceptibility of the iron balls. In this case, Eq. 5 can finally be written as25,

      F=μ0V3χχ+3HHx
      (7)

      where H/x is the magnetic field gradient. Therefore, the deflection of the microcantilever can be demodulated by monitoring the shift in the traced dip wavelength. The force acting on the microcantilever can be calculated using Eq. 3, which is consistent with the magnetic force acting on the iron ball calculated by Eq. 7.

      In FPI-2, Mirror-1 and Mirror-2 are the upper and lower interfaces of the microfluid-infiltrated microcavity, respectively. Similarly, variations in the volume and RI of the microfluid-infiltrated cavity with temperature can cause a shift in the FPI-2 interference resonance wavelength. The two FPIs with different properties can obtain different spectral shifts in terms of magnetic field intensity and temperature variations. Therefore, the wavelength shifts of FPI-1 (Δλ1) and FPI-2 (Δλ2) can be characterized by the sensitivity coefficient matrix of the two-parameter two-equation system21:

      [Δλ1Δλ2]=[SH1ST1SH2ST2][ΔHΔT]
      (8)

      where ΔH is the variation in magnetic field intensity; ΔT is the variation of temperature; SH1,2 and ST1,2 are the sensitivity coefficients of magnetic field intensity and temperature of FPI-1 and FPI-2, respectively. Thus, the magnetic field intensity and temperature encoded in the cavity length or RI of the medium can be extracted by tracking the shift in the interference resonance wavelength.

    Device fabrication and characterization
    • The MCF-tip probe fabrication process is shown in Fig. 3. Prior to polymerization, the microcantilever and microfluid-infiltrated microcavity were 3D modeled according to design, considering the characteristics of the models in planning fabrication parameters such as slicing, line spacing, and scanning path (Fig. 3a). A bowl-shaped structure was added to the top of the polymer microcantilever to carry the iron ball. The microcavity on another core end facet solidifies the surface profile during polymerization, allowing the microfluid polymer to be enclosed in the microcavity26. To balance fabrication quality and efficiency, slicing and line spacings were optimized to 500 nm and 300 nm, respectively, and an S-shaped scanning path was adopted. A 3D laser lithography system for micro/nanofabrication (Nanoscribe GmbH)27,28 was used to print the microcantilever and microcavity onto the MCF end facet.

      Fig. 3  Flowchart illustrating device fabrication. a Modeling and scanning path design. b Polymerization of the micro/nanostructures on the MCF end facet. c Sample development (removal of unpolymerized photoresist. d Coating of the adhesive. e Microcantilever is incorporated with an iron ball. f Photoresist is solidified by the light emitted from the fiber core to reinforce the sample.

      The geometric arrangement of fiber tip (four-core MCF), photoresist (IP-L), coverslip (170 mm thickness), matching oil (RI=1.518), and an objective lens (63×, NA=1.4) are shown in Fig. 3b. The laser power (5 mW) and scanning speed (50 mm/s) were optimized according to the planned scanning path during TPP. Scanning of the laser exposure point was implemented by moving the piezoelectric stage, which has the advantages of high precision and repeatability. After polymerization, the sample was developed using Propylene Glycol Methyl Ether Acetate (PGMEA) and isopropyl alcohol solutions successively to remove the unexposed photoresist, as shown in Fig. 3c, which retains the cured microcantilever and liquid polymer encapsulated inside the microcavity. To ensure that the bowl-shaped structure adhered to the iron ball, an adhesive was applied to the hollow bowl, as shown in Fig. 3d. A small amount of green light-cured adhesive collected by a fiber tip was placed into the hollow bowl by slowly moving the displacement stage. Fig. 3e shows the process of assembling the iron ball into the hollow bowl. Under the action of van der Waals forces29, the iron ball is captured by the fiber and assembled using the slowly moving displacement stage. Finally, the adhesive inside the hollow bowl was solidified to reinforce the sample by irradiating it with a 532 nm laser from the fiber core (Fig. 3f).

      The four-core MCF and printed micro/nanostructures were characterized using scanning electron microscopy (SEM), as shown in Fig. 4. The multicore fiber (SM-4C1500) from Fibercore Ltd. had four cores arranged in a square pattern, with diameters for cladding, core, and mode field of 125 µm, 8 µm, and 8.4 µm, respectively, and core-to-core spacing of 50 µm. The large core spacing provides space for the fabrication of long microcantilever with high sensitivity and ensures the independence of the two FPIs. Fig. 4bd show the printed micro/nanostructures obtained from different angles. Both the microcantilever and microcavity can be clearly distinguished. The microcantilever with a length of 40 μm, a width of 20 μm and a thickness of 2 μm and the hollow bowl with a diameter of 30 μm were completely printed onto the fiber end facet. Moreover, the smooth surface of the microcantilever and the good parallelism between the microcantilever and the fiber end facet contributed to the efficient excitation of the interference spectrum. It is worth noting that there is a 10 × 10 μm hollow region fabricated in the middle of the microcantilever to improve mechanical deformation sensitivity16. Another FPI is a microfluid-infiltrated microcavity, which has a bottom area of 30 × 30 μm, a side-wall thickness of 5 μm, and a top wall thickness of 2 μm. The box-shaped microcavity was completely printed and tightly enclosed on all sides, successfully enclosing the microfluidic polymer. Both micro/nanostructures were printed on different fiber core end facets at designed positions, reflecting the light from different fiber cores, thus forming the FPIs of air and liquid polymer mediums, respectively.

      Fig. 4  Scanning electron micrographs. a Four-core MCF. b-d Microcantilever and microfluid-infiltrated microcavity.

      The reflection spectra of the sensor were investigated by connecting a broadband source (BBS), optical spectrum analyzer (OSA), 3 dB coupler, optical switch, and a sample with an MCF fan-in/out device30, as shown in Fig. 5a. The reflection spectrum of microcantilever illustrates a dip wavelength with a fringe visibility of 4.27 dB and a free spectrum range (FSR) of 31.30 nm at 1550.8 nm, whereas the reflection spectrum of microfluid-infiltrated microcavity shows a dip wavelength with a fringe visibility of 6.28 dB and an FSR of 23.65 nm at 1531.3 nm (Fig. 5b). The reflection spectra of the dual FPIs had a large FSR, which is conducive to a wide range of measurements.

      Fig. 5  a Scheme of spectral measurement. b Reflection spectra of the dual FPIs.

    Sensing performance and discussion
    • The magnetic-field intensity of the sensor was measured in a magnetic-field environment generated by a magnet and calibrated using a Gaussian meter. Fig. 6 shows the response of the microcantilever incorporated with a ~30 μm diameter iron ball to various magnetic field intensities. As shown in Fig. 6a, as the magnetic field intensity increases from 30 to 90 mT, the red shift in the reflection spectrum is 7.6 nm, indicating that the cavity length of FPI-1 increases. This can be attributed to the increase in the magnetic force acting on the iron ball when the magnetic field intensity increases, resulting in a larger upward deformation of the microcantilever. In Fig. 6b, the variation in the traced dip wavelength is plotted as a function of magnetic field intensity, showing low- and high-sensitivity areas. The performance of the sensor was relatively stable, and the error bar of each measurement point was calculated through two testing cycles, all less than 0.25 nm. Notably, the slope of the dip wavelength shift is not a linear function of the magnetic field intensity over the entire measurement range; however, this sample can be fitted with a slope of 154.5 pm/mT with good linearity (R2 = 0.97) in the range of 50–90 mT (Inset of Fig. 6b).

      Fig. 6  Magnetic field response of microcantilever incorporated with a ~30 μm diameter iron ball. a Reflection spectra vs. magnetic field intensity. b Variation of dip wavelength with magnetic field intensity. c Repeatability test in two cycles of magnetic field intensity measurement. d Response time measurement.

      The repeatability of the sensor was investigated using two cycles of magnetic-field intensity measurements. Fig. 6c shows the variation in the dip wavelength with magnetic field intensity during cycling. The dip wavelength at different magnetic field intensities is relatively stable regardless of whether the magnetic field intensity increases or decreases. The results indicate that the magnetic field sensor has good repeatability, attributed to the excellent recovery of the polymer microcantilever.

      The response time of the sensor was investigated by recording the light intensity variation in the reflection spectrum at 1530 nm at varying magnetic field intensities. Fig. 6d illustrates the light intensity variation of the scanning wavelength when the magnetic field intensity increases from 30.8 to 63.3 mT. The response time is defined as the time interval for the sensor to reach 90% of the steady-state response31. The response time of the sensor was estimated to be 213 ms, which indicates that the sensor can respond quickly to changes in magnetic field intensity.

      To further verify the repeatability and stability of the sensor, a microcantilever incorporated with a ~43 μm diameter iron ball was implemented to test five cycles of magnetic field measurements. The variation in the dip wavelength with magnetic field intensity during the five testing cycles is shown in Fig. 7a. The dip wavelength is stable during each measurement, and the wavelength dip points in the five cycles are almost coincident, indicating that the sensor has good repeatability in the magnetic field range of 30-60 mT. Fig. 7b illustrates the low- and high-sensitivity areas of the sample. The error bars of the dip wavelengths at different magnetic field intensities in the five testing cycles were calculated to be less than 0.28 nm. The inset of Fig. 7b shows the dip wavelength as a linear function of the magnetic field intensity over 30–50 mT, exhibiting a sensitivity of 322.3 pm/mT.

      Fig. 7  Magnetic field response of microcantilever incorporated with a ~43 μm diameter iron ball. a Five cycle repeatability test of magnetic field intensity measurements. b Variation of dip wavelength with magnetic field intensity.

      To investigate the influence of iron ball size on the sensitivity of the sensor, a microcantilever incorporated with a ~61 μm diameter iron ball was implemented to measure the magnetic field intensity. As shown in Fig. 8a, the reflection spectrum redshift value of the sample is 17.81 nm when the magnetic field intensity increases from 30 to 40 mT. In Fig. 8b, the shift of the traced dip wavelength is plotted as the magnetic field increases from 30 to 48 mT. The trend of magnetic field sensitivity variation is consistent with that of the sample incorporated with 30 μm diameter iron ball, that is, the sensitivity increases with an increase in magnetic field intensity. In the inset of Fig. 8b, the traced dip wavelength is a linear function of the magnetic field intensity in the range of 30–40 mT, and the sensitivity reaches 1805.6 pm/mT. Therefore, the magnetic field intensity sensitivity of the sensor can be improved by incorporating a larger iron ball, which causes the microcantilever to be subjected to a stronger magnetic force in the same magnetic field.

      Fig. 8  Magnetic field response of microcantilever incorporated with a ~61 μm diameter iron ball. a Reflection spectra vs. magnetic field intensity. b Variation of dip wavelength with magnetic field intensity.

      Temperature is an important physical parameter affecting the measurement accuracy of sensors. To investigate the temperature characteristics of the sensor, the temperature was measured in the range of 25–55℃ in increments of 5℃ steps by placing the sensor in an electric oven. Each measurement point was maintained for 10 min to ensure an adequate response. Fig. 9a shows the reflection spectra of the microcantilever at various temperatures; a redshift in the reflection spectrum is observed as the temperature increases. The dip wavelength at each measuring point was linearly fitted, obtaining a sensitivity of 77.4 pm with a standard error of 3.2 pm (Fig. 9b). Similarly, a red shift in the reflection spectrum of the microfluid-infiltrated microcavity is observed as temperature increases, as shown in Fig. 9c. The linear fit in Fig. 9d of the experimental data at each temperature shows a high sensitivity of 160.3 pm/℃, with a standard error of 3.8 pm. The spectral redshift of the FPI as a function of temperature can be expressed as15,

      Fig. 9  Temperature response of the microcantilever and the microfluid-infiltrated microcavity. a Reflection spectra of microcantilever vs. temperature. b Linear fitting of the dip wavelength of microcantilever as a function of temperature. c Reflection spectra of microfluid-infiltrated microcavity vs. temperature. d the linear fitting of the dip wavelength of microfluid-infiltrated microcavity as a function of temperature.

      dλdT=2k(dndTL+dLdTn)
      (9)

      where (dλ)/(dT) is the shift in dip wavelength with a variation in temperature; k is the interference order number; (dn)/(dT) is the thermo-optic coefficient of the medium in the FP cavity; and (dL)/(dT) is the thermo-expansion coefficient of the FP cavity.

      In FPI-1, because the RI of air changes with temperature are indistinguishable, the redshift of the spectrum is attributed primarily to the increased cavity length from the thermal expansion of the polymer base of the microcantilever. In FPI-2, the redshift of the spectrum is caused by a combination of the thermo-optic and thermo-expansion effects of the liquid polymer in the microcavity. As temperature increases, the RI of the liquid decreases32, which leads to the blue shift in the spectrum. However, this blueshift is weak compared to the redshift caused by the thermal expansion of the liquid; thus, the spectrum eventually appears as a spectral redshift. Compared with FPI-1, the increased temperature sensitivity of FPI-2 is due to the fact that the liquid polymer has a higher thermal expansion coefficient than the cured polymer microcantilever base of FPI-126. Because the photoresist-infiltrated microcavity may solidify over long-term light exposure, the temperature-sensing element should be further packaged light-free, or another stable microfluid can be embedded inside the hybrid microfluidic cavity.

      As the microfluid-infiltrated microcavity was not incorporated into the iron ball, it was not sensitive to the magnetic field, and the magnetic field intensity sensitivity was 0 pm/mT. Therefore, for the microcantilever sample incorporated with a ~61 μm diameter iron ball, Eq. 8 can be expressed as

      [Δλ1Δλ2]=[1805.6077.4160.3][ΔHΔT]
      (10)

      The magnetic field intensity and temperature can be measured simultaneously by monitoring the spectra of the dual FPIs based on Eq. 10. The condition number of the two-parameter sensor21 based on the calculated measured sensitivity was 11.28, which indicates that the MCF-tip multi-parameter sensor has reliable stability and high tolerance to the measurement error of the dip wavelength shift. Various previous typical fiber-based magnetic field and temperature-discriminative sensors are summarized for comparison in Table 1. The proposed MCF-tip dual FPIs sensor is smaller and has a lower condition number, making it superior to most similar sensors.

      Sensor typeStructure size
      (mm)
      Magnetic field
      sensitivity (pm/mT)
      Temperature sensitivity
      (pm/℃)
      Condition numberRef.
      PCF-FBG>10924.6123.194.3812
      FPI-FBG>1340−9228.0733
      PCF-FPI>1330−2361.6234
      MZI-FBG>15407.8−362.661.7335
      Microfiber-MZI>0.263−11930195035.5536
      PCF multimodal interferometer>1720−8020.7937
      MCF-tip dual FPIs0.1251805.6160.311.28Our work
      PCF: Photonic crystal fiber; FBG: Fiber Bragg grating; FPI: Fabry-Perot interferometer; MZI: Mach-Zender interferometer

      Table 1.  Performance comparison of fiber-based magnetic field and temperature discriminative sensors.

      To verify the reliability of the dual-parameter discriminative sensor, its performance was tested under different environments for random magnetic field intensities and temperature settings. For the magnetic field intensity and temperature measurements, our fiber tip probe and a commercial electric probe were placed in a temperature oven with a magnetic field to create a two-parameter measurement environment with an adjustable magnetic field and temperature. The magnetic field intensity and temperature in the initial environment were 30 mT and 25℃, respectively. The spectra of the dual probes in the initial environment were used as reference to calculate the variation in the spectra resulting from environmental changes. Based on Eq. 10, the measured two-parameter variation was calculated according to the measured wavelength shift of the dual FPIs, as shown in Table 2. The results of five random measurement points demonstrated that the relative measurement errors of the magnetic field intensity and temperature were less than 0.71% and 5.06%, respectively, confirming that the proposed MCF-tip dual-probe sensor with low condition number is reliable. More importantly, without temperature discriminative measurements, the magnetic field measurement error goes up to 53% (measurement of magnetic field variations over actual magnetic field variations) as the actual magnetic field changes from 30 to 32 mT with the temperature changing from 25–50 ℃ and increases by only 5% in the case of discriminative measurement.

      Random magnetic field
      and temperature setting
      Wavelength
      shift of FPI-1
      Wavelength
      shift of FPI-2
      Measured parameters variationMeasured magnetic
      field and temperature
      Relative measurement errora
      H = 32 mT
      T = 40℃
      4.83 nm2.26 nmΔH = 2.07 mT
      ΔT = 14.10℃
      H′ = 32.07 mT
      T′ = 39.10℃
      0.22%
      2.25%
      H = 34 mT
      T = 55℃
      8.94 nm4.45 nmΔH = 3.76 mT
      ΔT = 27.76℃
      H′ = 33.76 mT
      T′ = 52.76℃
      0.71%
      4.07%
      H = 38 mT
      T = 35℃
      14.61 nm1.32 nmΔH = 7.74 mT
      ΔT = 8.23℃
      H′ = 37.74 mT
      T′ = 33.23℃
      0.68%
      5.06%
      H = 32 mT
      T = 50℃
      5.52 nm3.61 nmΔH = 2.09 mT
      ΔT = 22.52℃
      H′ = 32.09 mT
      T′ = 47.52℃
      0.28%
      4.96%
      H = 40 mT
      T = 25℃
      18.50 nm0.09 nmΔH = 10.22 mT
      ΔT = 0.56℃
      H′ = 40.22 mT
      T′ = 25.56℃
      0.55%
      2.24%
      a Relative measurement error = (|H' − H|/H) × 100% or (|T' − T|/T) × 100%

      Table 2.  Performance of MCF-tip dual-probes in different environments with random magnetic field intensity and temperature settings.

    Conclusion
    • In summary, we demonstrated an ultracompact MCF-tip probe for the discriminative measurement of magnetic field and temperature. The microcantilever and microfluid-infiltrated microcavity integrated on the MCF end facet were 3D-printed by femtosecond laser-induced TPP. The microcantilever incorporated with iron ball demonstrated a magnetic field intensity sensitivity of 1805.6 pm/mT and a temperature sensitivity of 77.4 pm/℃. The microfluid-infiltrated microcavity exhibited a temperature sensitivity of 160.3 pm/℃. The two functional probes exhibited different sensitivities to the magnetic field and temperature; thus, the two parameters can be discriminatively measured by calculating the multi-parameter sensitivity coefficient matrix. The MCF-tip dual-FPIs had the smallest size and exhibited a low condition number of 11.28, which is superior to that of most similar sensors, demonstrating the importance of discriminative measurements during external temperature changes. The finesse of the microcantilever can be improved by depositing high-reflectivity coatings onto the fiber end facet and polymer microcantilever surface, which further enhances the performance of the discriminative sensor. Therefore, the bowl-shaped microcantilever can provide a useful hybrid platform for incorporating 3D printed microstructures with functional materials, thereby breaking the restrictions on the range of detectable parameters. MCF-based fiber-tip probes can also be powerful tools for multi-parameter sensing in extremely limited sensing spaces.

    Acknowledgements
    • This work was supported by the National Natural Science Foundation of China (No. 62275052, No.62275148), Shanghai 2021 Science and Technology International Cooperation Project “Program of Action for Science and Technology Innovation” (21530710400), the Jiangsu Province's Industry Outlook and Key Core Technologies-Key Projects (BE2022055-4), the Open Fund of Laboratory of Science and Technology on Marine Navigation and Control, China State Shipbuilding Corporation (2023010102).

Reference (37)

Catalog

/

DownLoad:  Full-Size Img PowerPoint
Return
Return