Optical Fiber‐Based Gas Sensing for Early Warning of Thermal Runaway in Lithium‐Ion Batteries

Owing to constantly increasing energy density and power density, thermal runaway is becoming the most dangerous event in lithium‐ion batteries, while gas sensing can provide the earliest and clearest signals for forewarning the advent of thermal runaway. However, gas‐sensing signals are not included in current battery management systems (BMSs). Especially for the challenging in‐cell sensing, gas sensing demonstrates significant advantages over out‐of‐cell sensing in the aspect of collecting signals precisely and in a timely manner. The integration of gas sensing with batteries is an important and valuable topic. In this review, the necessity and challenge of in‐cell sensing are expounded, the importance and promise of in‐cell gas sensing are highlighted, optical fiber‐based gas sensing technologies for in‐cell applications are summarized, and the engineering issues of integrating sensors with battery system are discussed. The topic of this article deserves more attention and efforts for the development of cell‐level BMSs and safe batteries in future.


Introduction
The invention of lithium-ion batteries (LIBs) is undoubtedly one of the most influential events of the 20th century, which drives the rapid development of mobile devices such as mobile phones and laptops, as well as the booming electric vehicles (EVs).With the fast growth of demand for energy and power density of LIBs, [1] the safety issue has become a critical concern, especially in the fields of EVs. [2]Various battery abuses, [3] such as overcharge, over-discharge, and long-term high-power operation, etc., are the main causes of shortened battery life and damaged safety.In the worst case, thermal runaway can occur, where the internal thermal energy in LIBs is so great that the external cooling system can no longer maintain a safe operating temperature, then irreversible damage will cause burning even explosion.Although thermal runaway is an extreme result, its occurrence seriously DOI: 10.1002/adsr.202300055[6] The thermal runaway is a process containing several temperature-dependent steps in which different components of LIBs decomposed or reacted with electrolyte.9][10] To avoid, or at least forewarn the arrival of thermal runaway, battery management systems (BMSs) have been introduced to keep LIBs operating under safe conditions. [11,12]Typically, a BMS consists two parts: hardware of various sensors positioned outside of cells or modules for collecting necessary electrical signals, and software of various algorithm models constructed from the collected signals for estimating the states of LIBs.
In principle, the state of charge (SOC) and state of health of LIBs can be estimated by a BMS, and then the BMS operates the LIBs properly according to the feedback of inline signals of current, voltage, temperature, etc.However, several drawbacks still plague BMSs, resulting in thermal runaway from time to time.First, gas signals are missing in most, if not all, current BMSs.Actually, a previous report [13] has demonstrated that gas signal can provide the earliest and clearest warning for thermal runaway, compared with the signals of voltage, temperature, smoke, pressure, etc.Therefore, it is necessary to integrate gas sensors into BMSs to enhance the early warning ability of algorithm models.Second, the sensors employed in BMSs are arranged outside cells or modules, causing an unavoidable time delay or even severe deviation from the real signals.For example, under high-power supply mode of LIBs, a temperature deviation as large as 10-20 °C can be recorded between internal and external sensors. [14]As a result, a significant accumulation of error can be expected during long-term operation of BMSs based only on external sensors.Replacing external sensors by internal ones in BMSs is an attractive idea, which provides direct and precise inline signals for cells, thus contributing to better BMSs and safer LIBs.Some studies showed that it is feasible to embed sensors in cells, [15][16][17] suggesting the great potential of in-cell sensing approach.
Considering the vital role of gas sensors, we would like to focus on gas sensors exclusively in this review, while other types of sensors have been reviewed previously. [17,18]Roughly, we can classify current gas sensors into two categories: electrical devices represented by resistive gas sensors, and optical devices represented by optical fiber-based gas sensors.Obviously, electrical sensors have dominated current market due to their cost-efficiency, high precision, and easy integration with the BMSs.However, they do not seem to be good choices for in-cell applications. [9]First, electrical devices generally require separate channels for each sensor, which means a considerable footprint in cell, that is, sensors inevitably occupy the space that should be filled by active materials.Second, the normal operation of electrical devices may be threatened by electromagnetic environment in cells.Third, the in-cell corrosive environment put more challenges on the long-term surviving of electrical devices because the commonly used metal-oxide sensing layer is susceptible to corrosion.In contrast, optical fiber-based gas sensors are made of silicon oxides or polymers that are immune to electromagnetic field and most corrosive components in cell, moreover, the dimension of optical fiber is also quite small, bringing negligible influence to cell operation. [19]n this review, we organize the content as follows: In Section 2, we introduce the thermal runaway process and corresponding gases generation during the process.In Section 3, we make a concise introduction to optical fiber-based gas sensing technologies, which mainly concentrate on the configuration and mechanism.In Section 4, some examples reported in literatures are demonstrated.In Section 5, the engineering issues of how to integrate optical fibers into cells are discussed according to reported cases.Finally, we make a summary and a prospect about the future of gas sensing in cell based on optical fiber-based technologies.

Thermal Runaway and Vent Gas
Thermal runaway is a complex process in which fast increasing thermal energy will trigger chain reaction. [3]These reactions usually generate a large amount of gases, resulting in significant pressure and volume increase in battery cell.When the internal pressure is higher than a certain value, the gases will quickly erupt, causing permanent damage to the cell. [20]he thermal runaway process displays several stages. [21]Initially, the solid electrolyte interface (SEI) decomposition happens at 90-120 °C. [22]In this stage, the metastable component of SEI (assuming (CH 2 OCO 2 Li) 2 ) decomposes and releases gases: [23] ( As the SEI decomposition goes on, the anode gradually contacts and reacts with electrolyte, [20,24] while the main product of the reaction is still the same component of SEI.Thus, the SEI decomposes and regenerates simultaneously in a wide temperature range of 120-250 °C.In this stage, the total reactions are shown as: [23,25] 2Li When the temperature increases to 130 °C, the polyethylene separator starts to melt, [20] and the cathode begins to release oxygen. [26,27] Then the electrolyte salt LiPF 6 starts to release toxic gas [22] and the electrolyte solvent such as dimethyl carbonate (DMC) is oxidized by oxygen and produces carbon oxides: [3] LiPF 6 ⇌ LiF + PF 5 (4a) Once the temperature increases to 350 °C, the binder in electrodes starts to react with Li: [28] The amount of gas is also a critical issue.For the 38450 type pouch cell, [29] the volume of battery cell (≈10 mL) expands by 95% after baking at 80 °C for 120 h.As for the 18650 cylindrical cell, [30] the amount of generated gas is around 265 mmol at 150 °C.In another work, 18650 cylindrical cell with LPF cathode releases more than 6 g gas under thermal abuse. [7]In short, the amount of vent gas is considerable large to the sensitivity of conventional gas sensors.More information of vent gas can be found in Table 1.
Besides gas species and quantity, the time of gas venting should also be valued, which sets the maximum response time of gas sensors employed in batteries.For the thermal abuse cases, gas venting happens at 105 s in 18650 cylindrical cell, [31] in contrast, which is observed at around 60 s in the pouch cell. [13]For the overcharging abuse cases, gas venting occurs at 362 s in 26650 cylindrical cell, [10] with a surface temperature of only 53 °C, however, that happens at 1186 s in prismatic cell, much longer than other cell types. [32]Besides, a cell pack consisting 32 single cells in the BESS cabin (83 m 3 ) [33] releases CO 2 and H 2 at 130 s and 435 s, respectively.As we can see, the time of gas venting varies by the factors of cell types, components, abuse situations, etc., but for these gas sensors with a maximum response time of 1 min are sufficient to meet the pre-warning requirement for all kinds of cells.

Configurations and Mechanisms of Optical Fiber-Based Gas Sensing Technologies
Different from electrical devices, photons, rather than electrons, are used as infofrmation carriers in optical fiber-based sensing technologies.Depending on the objects of measurement, the technologies can be divided into two kinds: detecting the variation of intensity from transmitted light, and detecting the shift of wavelength or phase from reflection light.The former one contains the technologies of direct absorption, which involves direct  interaction of light with target gases, while the latter one concerns the technologies of indirect interaction, which generally requires exotic materials as an intermediary to bridge the relation between gases and light.To meet the requirements for incell gas sensing, eligible technologies should have properties as follows: 1) A response time shorter than 1 min, corresponding to the minimum gas venting time in cells; 2) A moderate limit of detection (LOD) or sensitivity.The threshold of vent gas concentration during thermal runaway can be up to 2000 ppm, [34] and the high concentration of vent gas is also expounded in Section 2. Thus ultralow LOD or ultrahigh sensitivity is not a must.
3) A small sensing head, at most in the level of millimeter scale.The dimensions of a typical 18650 and 38120 cylindrical cell are 6.5 cm and 14 cm, [35] respectively.As for the pouch cell and prismatic cell, their sizes are customizable.

Hollow Core Optical Fibers Gas Sensors
Hollow core optical fiber, is also called photonic crystal fiber (PCF).[38] For PBG-PCF, the core is occupied by air hole that is surrounded by many photonic crystal microstructures acting as cladding, and the guiding mechanism follows the PBF effect, [39] that is, the light of specific wavelengths propagating in the air core will be reflected back by the photonic crystal structures.Conversely, the core of TIR-PCF is a solid of a higher refractive index (RI), which is surrounded by air holes of lower RI.Thus, the guiding mechanism is close to normal optical fiber-based on TIR.Besides, PCF based on other guiding mechanism is being developed, for example, anti-resonant hollow core fiber (ARHCF, Figure 1c).ARHCF has an air core of large diameter with negative curvature structures surrounded by, such as capillary tubes.Its waveguide mechanism follows the anti-resonant reflecting optical waveguide principle. [40]That is, the low RI core and the high RI capillary walls can be regarded as the Fabry-Perot (FP) resonator, thus the light whose wavelength does not satisfy the resonant condition of the FP resonator will be reflected back to the core and only a small part of resonant light can propagate out of the fiber.PCF based sensors belong to the direct absorption technology.In a typical configuration, a segment of PCF generally splices with normal optical fiber and acts as absorption cell.Direct absorption technologies obey Lambert-Beer law: where I and I o are intensity of transmitted light and incident light, respectively. is the spectral absorption coefficient of gas, which is related to absorption line profile of gas, l is the path length of absorption, and N represents gas concentration.
The PCF sensors have various merits such as robust to environment disturbing, and relatively long light-gas interaction length that is benefit to achieving high sensitivity.However, the slow response time that can be attributed to slow diffusion of target gas in air holes, [36] is still the main drawback.The problem is more severe in PBG-PCF cases in which the diameter of air hole is usually 20-30 μm, leading to longer time for gas diffusion. [41] gas diffusion model [42] suggested that the length of PCF fiber should be shorter than 7 cm to ensure a response time of 1 min.However, decreasing the PCF fiber length will degrade the sensitivity at the same time.Thus, a series of structures such as inlet/outlet chamber, [43] side opening, [44][45][46] etc. are designed to speed up the response time.

Grating Optical Fibers Gas Sensors
Grating refers to a periodic structure of different RI along the fiber, and the most reported types including fiber Bragg grating (FBG), titled fiber Bragg grating (TFBG), and long period grating (LPG), according to the variation of periodic length or spatial configuration.
FBG is the most common grating structure in which a number of reflectors are perpendicular to the longitudinal axis of fiber and distributed with a constant period (Figure 2a).Light with specific wavelength can be almost totally reflected by the grating but the rest of light with other wavelengths can pass through, and the selected wavelength by reflection can be described by: where  B is the selected wavelength, namely Bragg wavelength, n eff is effective RI of the fiber core, and Λ is period of the grating.
The FBG is widely used for in-cell temperature and strain sensing, since n and Λ are the function of temperature and strain.However, due to lacking sensitivity to the surrounding medium, a bare FBG is not suitable for gas sensing, and a gas sensitive coating layer is required.TFBG is another type of FBG, which has reflection planes titled with a certain angle  to vertical axis (Figure 2b).Thus, part of the light transfers from core mode into cladding modes, and even escapes from the fiber in a radiative mode. [47]The cladding modes propagate backward and interfere with forward core modes, resulting in a series of peak drops in the transmission spectrum.The interference wavelengths can be written as: where n eff,core and n i eff ,clad are the effective RI of core mode and ith cladding mode, respectively. is the tilted angle.The transmission spectrum of the TFBG consists of many downward peaks where the longest Bragg wavelength is caused by core mode and the rest of wavelengths are caused by cladding modes. [48]When the RI of cladding mode is smaller than that of surrounding medium, light would escape from cladding in the radiative mode due to the breakdown of TIR.Thus, the TFBG is highly sensitive to the surrounding environment.
LPG has a much longer period of grating than the FBG (Figure 2c), which usually ranges from 100 to 1000 μm. [49]Similar to the TFBG, there is a part of light can transfer from core mode into cladding modes in the LPG.Unlike in TFBG, the cladding modes in LPG interfere with core mode at the same propagating direction. [50]Thus, the core mode and cladding modes in the LPG match at the wavelengths that can be described by equation: where, n eff,core is the effective RI of core mode and n v eff ,clad is the effective RI of vth cladding mode.The parameter n eff,core and n v eff ,clad are functions of temperature, strain, indices of fiber core/cladding/environmental medium, and light wavelength. [51]oth the TFBG and the LPG are of high sensitivity to surrounding environment, however, for gas sensing, an extra gas sensitive coating is still necessary.Thus, grating optical fiber sensors were largely used as indirect interaction gas sensing technology.

Optical Fiber FP Interferometer Gas Sensors
Optical fiber FP interferometer (FPI) is the simplest interferometer consisting of two reflectors along the core of fiber.The space between reflectors defines a resonant cavity.The walls of the cavity, namely the reflectors, allow only part of the incident light transmitting, leading to an interference within the FP cavity.The walls do not have to be vertical planes, various configurations are acceptable. [17]Usually, according to the medium of the FP cavity, the FPI is classified into intrinsic (Figure 3a) and extrinsic (Figure 3b) types.For intrinsic FPI, the cavity medium is the same as the fiber.In contrast, the cavity is made of materials other than the fiber in extrinsic FPI.The intensity modulation spectrum of the FPI depends on phase difference between reflection light and transmission light. [52]The phase difference can be calculated by equation: where  FP is the phase difference, n is the RI of FP cavity medium, L is the cavity length, and  is the incident light wavelength.Thus, variation of the RI or the FP cavity length can lead to a shift of the interference wavelength.For gas sensing, gas sensitive materials are coated on the wall of FP cavity to cause cavity length change [53] or even fabricated into FP cavities. [54]

Optical Fiber Evanescent Wave Gas Sensors
Evanescent wave is a special phenomenon that occurs at the boundary of attenuated total reflection.Part of the electromagnetic field of incident light extends into optically thinner media region which is extremely close to the TIR interface, [55] called evanescent field (Figure 4).The amplitude of the evanescent field decreases exponentially along the direction perpendicular to the interface, which is described as: where E o is the amplitude at the interface and d p is the penetration depth that can be calculated by following formula.
where  is the wavelength of incident light, n 1 and n 2 is the RI of optically thicker and thinner media respectively and  res is the incident angle.Usually, the penetration depth is a length merely close to a single wavelength.To release evanescent wave from optical fiber, different fiber geometries have been adopted according to the ways of cladding-reducing treatment, including the de-cladding fiber, [56] D-shape cladding, [53,57] and U-shape fiber. [58]However, these treatments degrade greatly the mechanical strength of optical fiber in the meantime.
In principle, this technology belongs to direct absorption but evanescent wave instead of the total incident light is used as signal carrier.

Optical Fiber Plasmon Resonance Gas Sensors
Surface plasmon resonance (SPR) and localized surface plasmon resonance (LSPR) are widely used for enhancing evanescent field, thus a general configuration of SPR/LSPR sensor is based on an evanescent wave optical fiber.SPR (Figure 5) concerns a continuous metal film, but LSPR requires incontinuous metal nanoparticles.In both cases, the real part of permittivity for the metal, such as Au or Ag, should reverse to that of surrounding dielectric such as air.The main difference between SPR and LSPR is that resonance wave can propagate along metal film surface but be localized at the very nearby space of metal nanoparticles.
The propagation constant of surface plasmons,  SP , follows equation: [59] where  is the wavelength of light,  d is the permittivity of dielectric and  m is the permittivity of metal.
The propagation constant of the evanescent wave is: where  p is the permittivity of optical fiber and  res is the incident angle of light.When the propagation constants of SPs and evanescent wave match each other, then a resonance occurs: During this process, the energy of incident light transfers into evanescent wave thus the intensity of evanescent wave is enhanced.Besides, the parameter of  m and  p in Equations ( 13) and ( 14) are the functions of wavelength, therefore a wavelength  res that satisfies the Equation ( 15) is selected, namely resonance wavelength.For gas sensing, variation of gas concertation will induce a change in environmental dielectric permittivity  d , resulting in a shift of the resonance wavelength.
Therefore, SPR or LSPR gas sensors can be classified either to direct absorption or indirect interaction sensing mechanism, depending on whether the measured parameter is the intensity or the wavelength.

Photoacoustic Spectroscopy Gas Sensors
Photoacoustic spectroscopy (PAS) is a powerful sensing method and had been widely utilized in many fields.For PAS gas  The structure of photoacoustic cell.Reproduced according to the terms of the CC BY license. [150]Copyright 2022, the authors, published by MDPI.
sensor, it holds impressive advantages for example, low LOD independent with light-gas interaction length, large response extent varying linearly as a function of incident light power, zerobackground, etc., over conventional light adsorption methods whose response extent large scale with the light-gas interaction length.Therefore, the LOD in PAS gas sensors can be as low as ppb or even ppt level. [60,61]he sensing mechanism in PAS is based on detecting gas pressure wave in PA cell where the magnitude of wave corresponds to the energy absorbed by target gas from incident light.Thus PAS works in the direct absorption sensing mechanism.Generally, the incident light is applied in a modulated or a pulsed mode, and then is absorbed by target gas molecules.The absorbed energy is then emitted in a non-radiative relaxation way from the excited gas molecules, which locally and periodically raises temperature, thus changing gas pressure regularly.Usually, the gas pressure wave is collected by a microphone (Figure 6), and the response signal can be depicted as following equation: [62] S = PMC cell C m (16)   where S is the response voltage signal (mV) collected by microphone; P is the power (W) of incident light; C cell is the cell constant (Pa•cm•W −1 ), which is a scaling factor related to modulation frequency, geometry, and measuring condition;  is the molar optical absorption coefficient (cm −1 •mol −1 •dm 3 ) of gas;  is the conversion efficiency of absorbed light energy to heat with no units; C m is the density of target gas molecules (mol −1 •dm 3 ).However, microphone has distinct defects in many application areas, for instance, environments of high temperature, corrosiveness, electromagnetic interference, etc. Optical fiber-based acoustic detectors are good alternatives because of their inertness to above environments.These optical acoustic detectors can be roughly classified into three kinds: [62] 1) intensity-modulated optical acoustic sensing where gas pressure wave is reflected by the vibration magnitude of a cantilever or a diaphragm that is directly contacted with gas in PA cell.2) FBG-based optical acoustic sensing where gas pressure wave-induced vibration is displayed by the shift of FBG wavelength.3) interferometer-based optical acoustic sensing where gas pressure wave-induced light path change is measured by optical interferometers.The response time for all-optical PAS is usually in the level of seconds.
The PA cell is a necessary for all PAS gas sensors, in which gas pressure wave is generated by the modulated or pulsed incident light.Among the above-introduced optical acoustic sensors, interferometer-based sensor showed high sensitivity and good compactness in structure.For example, extrinsic FPI-based optical acoustic sensing can be miniaturized into millimeter scale, [63] in which the incident light and read-out light share the same fiber and the FP cavity has a hole on side wall as inlet/outlet for gas, acting as the PA cell at the meantime.

In Comparison with Other Kinds of Gas Sensing Technologies
Acoustic wave gas sensors are strong competitors to the optical fiber gas sensors.There are two typical acoustic wave gas sensors that have been extensively studied, namely, bulk and surface acoustic wave devices.In bulk acoustic wave devices, the piezoelectric material is sandwiched by two metallic electrodes, and the acoustic wave is propagated vertically through the whole material.In surface acoustic wave devices, the acoustic wave traveled only the on surface of piezoelectric material, on which two metallic interdigitated electrodes are deposited, with a gap between them working as the sensing area.The acoustic wave sensing mechanism is based on the vibration frequency shift of piezoelectric materials due to a mass change resulting from absorbed gas molecules on surface. [64]Generally, LOD down to ppm or ppb can be reached, and the response/recover time is often less than one minute. [65]Similar with optical fiber gas sensors, the acoustic wave gas sensors can be operated at room temperature and miniaturization of device to millimeter scale is also feasible.Even though no electrons are concerned with the sensing process, metal electrodes are still required to stimulate piezoelectric materials for high frequency vibration by high frequency AC voltage, which might be eroded due to the corrosive environment in cell.Besides, the sensing mechanism relies on mass change; therefore, poor selectivity to gas is inevitable and a coating sensitive layer to specific gas is required for achieving gas selectivity.Moreover, each acoustic device requires an individual channel for signal collection, leading to difficulty in distributed application. [66,67]ifferential electrochemical mass spectrometry (DEMS) is also a gas sensing method that has been widely used in electrochemical system to facilitate the electrochemical reaction mechanism study. [68]Simply speaking, DEMS is a vacuum test equipment based on a mass spectrometer. [69]With the help of mass spectrometer, DEMS holds excellent selectivity to gas species; therefore, it can be used to precisely identify gas molecules in cells.However, its drawbacks are also distinct, for instance, a vacuum environment is required for sampling that means a vacuum pump is a necessity; the mass spectrometry is subtle and its sizable volume makes the system not suitable for outdoor use.Thus, although DEMS is a powerful tool for gas sensing, but it can hardly be used in practical cell level sensing applications.

Gases Sensing Based on Optical Fiber Technologies
In principle, optical fiber sensing technologies have no limits on the types of detecting gases.However, according to the thermal runaway process introduced in Section 2, only a few gases, for instance, carbon dioxide, methane, hydrogen, etc., are the main products.Therefore, in this part, we will concentrate on these gases' detection exclusively.
Moreover, it is necessary to emphasize the fact that in-cell gas sensing based on optical fiber technology is still empty in literatures, thus here we just make a summary from the aspect of technical feasibility.

Direct Absorption
The main absorption peaks of carbon dioxide locate at nearinfrared (NIR) and middle-infrared (MIR) region.In a PCF based sensor work, [70] a 74 cm long PBG-PCF was employed as absorption cell for CO 2 detection, and a 2003 nm light source was used to avoid interference from moisture, which achieved a LOD of 2% and a response time of 10 min.
In contrast, ARHCF showed a much faster response time and a lower LOD due to its large-diameter hollow core accelerating airflow and low coupling between cladding modes and core mode.Nikodem et al. [71] reported that a 1.35 m long ARHCF sensor can achieve a LOD of 5 ppmv and a response time of 5 s.Besides, Jaworski et al. further utilized 1 m long ARHCF for multigas detection, [72] in which CO 2 and CH 4 were detected simultaneously by wavelength modulation spectroscopy at 1.574 and 3.334 μm, respectively.A LOD of 114 ppmv and a 1.5 s response time for CO 2 , and a LOD of 24 ppbv and a 40 s response time for CH 4 were achieved.The low LOD and fast response time of ARHCF sensor render a rich room for user to balance the response time and sensitivity by tailoring the length of sensing head to match the dimension of cells.
PAS is widely employed for CO 2 sensing as well.The mostly used acoustic sensor is still microphone, which is cheap and mature for vibration detection.In a recent work, [73] a sphere-tube coupled PA cell was designed to enhance the response of CO 2 with a 2004 nm distributed feedback (DFB) laser, which showed a LOD of 23 ppm in 5 s.In a similar work, [74] a sphere PA cell with multi-resonance property was designed, and the location of microphone for achieving maximum response has been theoretically simulated and verified by experiment, in which 767 ppm CO 2 can be detected in 1 s.For all-optical photoacoustic sensing, 2.4 ppm in 1 s and 318 ppb in 116 s have been achieved by quartzenhanced PA cell based on a tuning fork. [75]Besides, based on a FP pressure sensor and a digital virtual lock-in amplifier, a low LOD of 60 ppb CO 2 has been realized within 100 s. [76]

Indirect Interaction
Integration of metal organic frame (MOF) with optical fiber received a lot of attention in recent years due to its good gas selectivity [77] and extreme high surface to volume ratio that can concentrate gas within MOF's pores. [78]Chong et al. [79] made a single mode fiber CO 2 sensor with a LOD of 20 ppm working at NIR of 1.57 μm, in which a 5 cm long cladding has been removed, and followed by a layer of nano porous Cu-benzene-1,3,5tricarvoxylate (BTC) grown on the fiber surface.Similarly, Kim ) Dynamic response at ≈242 nm of ZIP-8 coated optical fiber after exposure to various gases.Reproduced with permission. [80]Copyright 2019, American Chemical Society.et al. [80] developed a multimode fiber evanescent wave sensor based on a MOF coating made of zeolitic imidazole framework-8 (ZIP-8, Figure 7a), which showed a selective absorption for CO 2 among H 2 , O 2 , CO, and CO 2 in a balance of N 2 (Figure 7b) and a response time of tens of seconds.The adsorption of CO 2 by ZIP-8 leads to the RI of ZIP-8 approaching to that of the optical fiber, resulting in an enhancement of evanescent field.To avoid the interference from moisture, the same group further introduced a hydrophobic alkylamine with a long hydrocarbon chain such as oleylamine on the top of MOF, as a layer of protective material. [81]Other MOF coatings for CO 2 sensing have been investigated [82] as well.Selective detection of CO 2 among various organic gases has also been realized by an LSPR sensor, [83] combined with a platinum nanowire structure and carbon nanotubes (CNTs), reaching a sensitivity of 6200 nm RIU −1 .Besides, another sensor based on NiO/rGO applying for CO 2 detection [84] showed a response time of 16 s.
The interference sensors are powerful tools for CO 2 detection.In an FPI sensor work, [85] a sensitive film comprised of thymol blue and TMAH encapsulated Ormosil was used to detect CO 2 and relative humidity (RH).CO 2 and RH in the range of 0-6% and 0-90% respectively were detected with negligible crosssensitivity.

Direct Absorption
Similar to carbon dioxide, methane has relative strong absorption peaks in NIR and MIR region. [86]Hu et al. [87] designed a PBG-PCF with free space coupling method for CH 4 detection to minimize the mode interference in the PBG-PCF, which showed a LOD of 360 ppbv and a response time of 75 s at 1653 nm.The relative slow response time can be further speeded up by enhancing the gas flow in fiber.A fast response CH 4 sensor was obtained by using Kagome PCF with core diameters more than 100 μm. [88]uch a large channel allows CH 4 to diffuse through 1.3 m long fiber in less than 10 s, and the LOD of CH 4 reaches sub-ppm level at MIR sensing wavelength.
In addition to PCF, tunable diode laser absorption spectroscopy (TDLAS) is a classic method for gas sensing, which renders a highly reliable identification to gases, taking advantage of ultraprecise laser wavelength modulation for specific gas.Wang et al. [89] designed a multipoint TDLAS CH 4 sensing system based on a pseudo differential detection technique employing high frequency square modulation, which showed a response time of several seconds in the concentration range of 0-3.61%.
Similar to CO 2 sensing by PAS, actually, in most multi-gas PAS sensing cases, CH 4 as a co-existed gas is detected with CO 2 simultaneously.In the above works introduced in the CO 2 sensing part, the LOD of CH 4 has been reported as 509 ppb in 5 s, [73] 69.2 ppm in 1 s, [74] and 37 ppb in 100 s, [76] respectively.

Indirect Interaction
Choosing proper sensitive intermediaries is the key for high performance gas sensing.TIR-HCP integrated with a LPG was reported for CH 4 sensing as well, [90] in which all channels inside of HCP were coated by ultraviolet curable fluoro-siloxane nanofilm incorporating with cryptophane A, except for one coated by Ag to eliminate cross-sensitivity.
The sensitivity of sensor can be improved by integrating with graphene.An LPG-SPR sensor based on graphene (Figure 8) achieved a high sensitivity to CH 4 . [91]A layer of graphene was coated on an LPG fiber that has a layer of Ag coating on surface in advance to enhance the intensity of the evanescent field, resulting in a sensitivity of 0.344 nm% −1 , 2.96 times to that of the traditional LPG sensor.Furthermore, a graphene-CNT was used to improve the sensitivity of sensor by introducing more active sites for sensing. [92]inc oxide can be used as CH 4 sensitive material, [93] however, high temperature is generally required to activate the material.An SPR sensor with zinc oxide/Pt matrix was reported, which showed a sensitivity of 1.8 × 10 5 nm RIU −1 to CH 4 under room temperature, [94] with a response time shorter than 1 min.
Cryptophane was also used for CH 4 sensing.Yang et al. [95,96] designed a PCF sensor with a cryptophane A coating in the air holes, which demonstrated a response time of 1 min.Further, an extra layer of Au coating was inserted to generate SPR, resulting in a high sensitivity (1.99 nm % −1 ). [97]Copyright 2017, the authors, published by MDPI.

Hydrogen
Since the intensities of absorption peaks of hydrogen in NIR is too weak to be applied to direct absorption test, current works for hydrogen detection are based on indirect interaction technologies.
Palladium (Pd) can absorb hydrogen by more than hundreds times of its own volume, [99] leading to significant change in morphology and optical properties.Cao et al. [100] reported an intrinsic FPI for multipoint H 2 detection based on Pd, in which a layer of 500 nm Pd was coated on the FP cavities to generate strain change.The sensor works in a concentration range from 0.25% to 10% with a LOD of 0.25% and a response time of hundreds of seconds.The slow response time can be improved by thinning the thickness of Pd.Zhang et al. [101] fabricated a H 2 sensor probe based on an FPI with a nano Pd film acting as the FP cavity, in which an ultrathin Pd film of 27.42 nm leads to a short response of 10 s.The deformation of Pd can be further amplified by combining with graphene.Ma et al. [102] presented a sensor with Pd-decorated graphene based on an FPI, in which the lattice expansion of Pd can be transferred to the graphene film, resulting in a wavelength shift.With a configuration of 5.6 nm Pd and 3 nm multilayer graphene, the sensor achieved a LOD of 20 ppm and a sensitivity of 0.25 pm ppm −1 with a response time of 18 s.By the similar sensing principle, Luo et al. [103] made an FPI sensor with graphene/Au/Pd film on the FP cavity and FBG in the fiber core, which achieved a high sensitivity and a fast response time of 4.3 s.
In addition to the expansion of volume, there is also evanescent wave [104] and plasmon resonance [105] H 2 sensors designed based Transmission spectrum of the sensor after exposure to H 2 concentrations of 0% and 1.02%.Reproduced with permission. [107]Copyright 2022, Elsevier.
on the modulation of optical properties of Pd.Perrotton et al. [106] designed an SPR sensor with 1 cm length Au/SiO 2 /Pd coating and realized a sensitivity of 30 nm/90 ppm.The resonant wavelength shift comes from the hydrogenation of Pd, and the initial resonant wavelength can be modulated by the thickness of SiO 2 .Zhang et al. [107] designed an SPR-TFBG structure combined with Au and Pd layers (Figure 9a).The cladding mode of 1558.4 nm in the reflection spectrum (Figure 9b) was used to achieve a good sensitivity of 1.597 dB % −1 for H 2 detection, in which a temperature compensation relying on the relative wavelength shift between cladding mode and core mode has been done.Besides, a similar sensor structure with only Pd film was developed to reach a LOD of 180 ppm. [108]Instead of the relative wavelength shift, herein core mode was utilized to eliminate temperature interference.Furthermore, similar temperature compensation method was also reported by Shen et al., [109] in which the ghost mode was used to detect H 2 and achieved a better sensitivity of 4.83 dB % −1 .
Not limited to Pd, metal oxides also have been widely utilized in H 2 sensing and demonstrated good performance, including WO 3 , [110,111] Ta 2 O 5 , [112,113] etc.
Similarly, MOFs were widely used for H 2 sensing.Miliutina et al. [114] reported an SPR sensor with an IRMOF-20 MOF sensitive layer (Figure 10a).The sensor showed a good selectivity to H 2 among CO, CO 2, or NO 2 and achieved a sensitivity of 9 and 22 nm wavelength shift at 2% H 2 and 4% H 2 , respectively (Figure 10b).Meanwhile, a relative short response time of 10 s and a recovery time of 5 s were recorded.The cross sensitivity from temperature and humidity (Figure 10c,d) was tested as well, in which temperature only showed a small interference  [114] Copyright 2019, American Chemical Society.
in low H 2 concentration and the humidity proved almost no interference in the whole range of test concentration.
A summary of all the reviewed works is listed in Table 2.

Integration of Optical Fibers with Systems
The integration strategies vary with battery cell types, but one common rule is that the integration should not interfere with the battery cells or sensors.In this section, we are going to discuss the integration of optical fiber with battery system from the aspects of cells, modules, and packs, respectively.

Types of Battery Cells
The cell is the basic unit of a battery system. [115]Usually, according to their package methods, the most common cell types are cylindrical cell, pouch cell, and prismatic cell.The cylindrical cell is the sturdiest one because the shell of cell is usually made of aluminum or even hard steel. [116]Besides, the cylindrical cell has the highest power density as well.Therefore, cylindrical cell is widely used in EVs as the power battery.Similar as cylindrical cell, the prismatic cell has a hard shell and the package is compact.The prismatic cell has a big surface area, benefitting to heat dissipation. [117]For the pouch cell, its shell is usually made of several polymer layers and aluminum layers. [118]As a result, pouch cells are susceptible to mechanical actions.For all cell types, the main components are similar, including adhesive, electrolyte, and a repeating stacked structure of cathodes, anodes, and separators.Generally, there are four ways to stack electrodes: single sheet stacking, Z-stacking, cylindrical winding, and prismatic winding, [119] as shown in Figure 11.The electrode stacking method of cylindrical cells is usually based on cylindrical windings, often called jelly rolls.For the pouch  Reproduced according to the terms of the CC BY license. [119]Copyright 2019, the authors, published by IOP Publishing.
or prismatic cells, all stacking methods are available.To the best of our knowledge, there is no optical fiber sensing works about prismatic cells so far, thus in the next part, we are going to discuss the possible sites for optical fiber embedding in the cylindrical and pouch cells.

Optical Fiber Embedded in Cylindrical Cell
There is a cylindrical void in the center of jellyroll structures, which is an ideal space for embedding optical fiber.Huang et al. [19,120] inserted an optical FBG sensor into the 18650 cylindrical cell by drilling a hole at the negative pole, prior to filling the electrolyte.Similarly, Yu et al. [121] embodied optical fiber from the negative pole of 21700 cell (Figure 12a).Besides, the positive pole of the battery cell can be another entrance [122,123] (Figure 12b,c).In addition, the tightness of the cell integrated with optical fiber after sealing with epoxy was tested by weighting the cell.After several electrochemical cycles, there is no weight change has been detected. [122]Such a simple integration ensures negligible mutual interference between the optical fiber and the cells.

Optical Fiber Embedded in Pouch Cell
Unlike cylindrical cells, the pouch cells are usually taken prismatic winding or Z-stacking mode, without an inner void structure for embedding optical fiber.Considering the tight inner structure of the pouch cell, optical fiber is inevitable to contact the electrode materials. [124,125]To protect the internal structure of the battery cells, some designs have been proposed to avoid such direct contact.Miele et al. [126] employed two separators to sandwich a PCF sensor in the pouch cell.Besides, it is important to assess the mutual inference between the optical fiber and the electrodes materials.A pouch cell with an inserted bare optical fiber evanescent wave sensor laying on the surface of cathode [127] showed a great stability over cycles for both the cell and the optical fiber.However, the stability of optical fiber sensor obviously degrades after having a coating layer.Gardner et al. [128] inserted an SPR optical fiber sensor with an Au coating into the pouch cell between the separator and the anode (Figure 13a).After 50 Reproduced according to the terms of the CC BY license. [121]Copyright 2018, the authors, published by Elsevier.b) FBG sensor was inserted into the central void of jellyroll from the positive pole to monitor temperature.c) The complete sensing system.Reproduced according to the terms of the CC BY license. [122]Copyright 2018, the authors, published by Elsevier.
times cycling test, the columbic efficiency of the cell was maintained at 99.8%, but the signal of sensor decreased with cycling times and approached to zero in final, and obvious corrosion can be observed on the Au coating layer.Apart from the electrode surface, a more radical site, inside of the electrode, was reported as well.Bae et al. [129] investigated FBGbased optical strain sensors that were embedded in different sites of cell, including laid on the surface of the anode (Figure 13b), and implanted inside the anode (Figure 13c).The optical fiber implanted into the anode was able to simultaneously measure longitudinal and transverse strains, thus having a better strain response than the fiber laid on the electrode surface.In both cases, the cell and the senor showed negligible mutual interference.
Sealing of the pouch cell integrated with optical fiber has also been studied. [127,130]Raghavan et al. [130] found that it is very likely to have voids at the entrance of the pouch cell (Figure 13d,e), and proposed a method to tightly seal the entrance with a thermal sealing film (Figure 13f).

Integration of Optical Fiber into Modules and Packs
Cells with similar performances are pre-assembled in parallel or series to build modules in which control components such as cell supervisory circuit board and cooling plates are mounted on a frame to achieve the primary balancing of individual cells.The modules can further be assembled to form packs with integrated BMS. [131]ne of the biggest challenges for optical fiber sensor integration is the cost.Typically, a battery pack that consisted of cylindrical cells has 16 modules, and each module has 444 cells.As for battery pack consisting of the pouch cells, the number of cells is 100. [132]The cost of optical fiber sensor was estimated to be as high as 1/4 of the price of the vehicle itself, even if every five cells share one sensor. [18]ultiplexing or distributed sensing technology renders a solution that enables the integration of multiple sensing heads in a shared optical system. [12,18,130,133]Usually, wavelength-division multiplexing (WDM) and time-division multiplexing (TDM) are the most used methods. [134]WDM adopts different wavelengths for each sensing head, thus the maximum number of the battery cells it can withstand depends on the bandwidth that can be used.TDM introduces a time delay between sensors to separate them in time domain. [135]Typically, WDM can support tens of sensors while the number is up to 1000 for TDM. [136]Moreover, a WDM/TDM hybrid optical fiber network [137] was designed to speed the system, [138] by developing the network with several wavelength channels, and each wavelength was divided into several time domain channels.As for gas sensing technologies, an external multi-gas sensing network is reported by connecting dozens of sensor probes on the fiber. [139]Although there is a lack of the research of internal gas sensing network, some of the technologies related to multiplexing gas sensing based on optical fiber [100,140] can be applied in the battery sensing field.
From the aspect of practical application, the optical fiber sensing prefers the battery system with small cell number but large capacity rather than large cell number but low capacity. [18]This tendency is showing in the market now, for example, the 18650 cylindrical cell is gradually replaced by the larger 48600 cylindrical cell. [12]part from the cost issue, there are other issues regards to redesign of the battery module or pack.For example, an extra space in the battery module or pack is required for the whole optical system besides sensing heads.Furthermore, considering the optical fiber arrangement, additional fixtures are anticipated The fiber was integrated between anode and separator.Reproduced according to the terms of the CC BY license. [128]Copyright 2022, the authors, published by MDPI.b,c) FBG sensor inserted on the surface of anode and implanted inside anode.Reproduced with permission. [129]Copyright 2016, Wiley.d) Schematic of FBG integrated with pouch cell.e,f) SEM images of pouch cells without and with heat-sealing film surrounded respectively.Reproduced with permission. [130]Copyright 2017, Elsevier.
in module assembly, but will not have much impact on the module's design. [130]

Integration of Optical Fiber into BMS
The integration of optical fiber sensors into BMS is mainly focused on how to the information from the optical gas sen-sors and how to deal with the gas signal.In order to measure the real-time spectra produced by various optical sensors, a spectrometer should be equipped in BMS.Taking the FBG sensor for an example, two wavelength interrogation setups have been developed: one is based on a broadband source and a spectrometer, and the other is based on a fast scanning tunable laser and a photodetector. [141]Both two setups support WDM and TDM.Besides, another interrogation system based on array waveguide grating (AWG) to optimize EVs' battery pack is reported. [142]This device divides optical signal from the FBG into corresponding channels to present the discrete FBG spectrum. [143]The AWGbased interrogation system has the advantages of low cost, compact size, and multiplexing ability. [142,143]n addition to the hardware updating, developing corresponding algorithms for processing sensing signal is imperative.According to Section 3, there are two types of signals for optical fiber sensing in general: light intensity change, and wavelength or phase shift.Light intensity change is easy to tackle.As for wavelength or phase shift, the algorithm of peak-tracking techniques should be added to the BMS, which has been well-reviewed in other work. [141][146] There were only a few of simple battery modeling researches based on vent gas, [34,147] but the internal gas signalbased model is still lacking.

Conclusion and Prospect
Gas sensing signals are important for monitoring the states of LIBs, which provide the earliest and clearest warning to the thermal runaway but are missing in current BMSs.Integrating gas sensors into battery cells is a topic of great challenging.In this review, we elaborate on the potential of optical fiber technologies for gas sensing in cells, including direct absorption technologies such as hollow core optical fiber sensor, evanescent wave sensor, and photoacoustic spectroscopy, as well as indirect interaction technologies such as fiber of Bragg grating, and optical interferometers.Following that, we discuss the integration of optical fiber with battery system, from the levels of cell, module, and pack, respectively.In this part, we will continue to give an assessment to all the introduced technologies above.
For sensing technologies based on direct absorption mechanism, such as PCF sensors, evanescent wave sensors, and PAS sensors, their advantage is simple setup due to the simple sensing mechanism.No sensitive intermediary materials are required, thus the system is highly resistive to harsh operating conditions both inside and outside the cells.
For PCF and evanescent wave sensors, the sensing mechanism requires that the sensing head should be as long as possible to enhance the light-gas interaction length.Thus their considerably long sensing head, generally tens or hundreds of centimeters, is the main obstacle to their application in cells.Such a long sensing head not only falls outside of the dimension of battery cell but also results in a response time (only for hollow-core fiber) much longer than the threshold of 1 min for in-cell gas sensing applications.Moreover, either hollow-core design or cladding removed/reduction structure would degrade the mechanical properties of fiber, leading to a fragile sensing system.Although the response time and robustness of the system can be improved by shortening the length of sensing head, the sensitivity of sensors attenuates with the length as well, usually resulting in a sensitivity that is too low to meet detection requirements.Besides, the liquid electrolyte may fill the hole of PCF or adsorb on the surface of cladding removed fiber, creating more uncertainty.
For the PAS sensors, as remarked above, there are no strict requirements for the light-gas interaction length.Its response originates from light absorption induced gas pressure wave in PA cell.In this case, to maximize the vibration, a typical configuration is to use an open extrinsic FPI with a wall equipped with a cantilever or a diaphragm, as the PA cell.Such a system can be minimized to millimeter scale, which has a low LOD down to ppb or ppt level, and a fast response time of several seconds.Thus, PAS is very suitable for in-cell sensing.However, considering of the complex chemical environment in cell, it is difficult to guarantee the operating conditions of the FP cavity without interference.For example, liquid electrolyte entering FP cavity or just adsorbing on the wall of FP cavity can alter cantilever or diaphragm behavior and cause errors.In addition, a subtle spectrometer is a necessity that is susceptible to environmental vibrations and temperature changes. [148]ptical fiber-based direct absorption methods can detect CO 2 and CH 4 , but they are incapable for H 2 detection due to its low absorption rate in the spectral range that can be transmitted in the fiber.So the detection of H 2 can only count on indirect interaction methods.
For sensing technologies based on indirect interaction mechanisms, such as fiber grating sensors, FPI sensors, and SPR/LSPR sensors, all of the three technologies strongly rely on sensitive intermediary materials.
The sensing mechanism of them stems from gas-induced property changes (RI or volume) in the intermediary materials, which then alters the wavelength or phase of the incident light.The merit of them is the compact sensing head that can be reduced to millimeter scale or smaller.Besides, their response time is short than 1 min in general.However, the necessary intermediary materials are the main risk point for their in-cell application.The intermediary materials are sensitive to corrosive environment in cell, which would introduce both deviations to detection and severe reduction to the service life of sensors.Similarly, spectrometer is needed to monitor the wavelength shift.
Errors caused by liquid contact are a common problem to almost all in-cell gas sensors.A possible solution is to isolate the sensing head or PA cell from liquid environment in cell, with a microporous membrane that only allows gas through, which has been widely used in DEMS technology. [69]Accordingly, the cost is a slower response time due to the obstruction of membrane to gases.Alternatively, direct detection of gas in liquid is another option.Generally, a degassing process should be added prior to sensing, [149] the size of which is incompatible to in-cell sensing.In a recent work, a PZT piezoelectric-photoacoustic spectroscopy for detecting gas in liquid without degassing [150] has been proposed, shedding light on gas sensing in liquid phase.
Besides the feasibility of technologies, there are still some issues that need to be addressed.
The most challenging one is the cost of optical fiber technologies, in addition to gas sensing head, the costly laser source, photodetectors, and even spectrometer are required as well.Even though cheap hardware is under developing and some hardware has already achieved low cost, [151,152] it is still a long wait to expect technological advances to bring newer and cheaper hardware.In contrast, multiplex or distributed sensing technologies [134,153] are currently the most valuable solution, which also demonstrate the advantage of the optical sensing over the electrical sensing technology.In these technologies, all sensors share the same hardware except sensing heads, thus significant cost reduction can be achieved and measurement differences between different systems can be minimized.
The engineering issue of integrating gas sensors with cells and BMSs is also a challenge.Although several approaches have been discussed in this review, most of them are actually based on other types of sensors like temperature or strain, which are not affected by direct contact with electrodes or electrolytes.However, direct contact with substances other than gases is likely to have unpredictable effects on gas sensing.As for integrating gas sensors with BMSs, current algorithm models of BMSs are based on electrical signals of temperature, voltage, or current, novel models including gas signals are required.Considering the temperature fluctuation in cells, temperature compensation is required to calibrate the gas sensing results.
In summary, in-cell gas sensing is a pivotal issue, but currently, only limited progress has been achieved.The success of in-cell gas sensing not only prompts BMSs for longer battery lifetime and safer daily operation, but also brings in-depth understanding of mechanism in LIBs, which obviously deserves more attention and efforts.

Figure 2 .
Figure 2. Schematic of grating sensors: a) fiber Bragg grating, b) titled fiber Bragg grating, and c) long period grating.

Figure 4 .
Figure 4. Schematic of evanescent wave-based optical fiber gas sensor.

Figure 5 .
Figure 5. Schematic of surface plasmon wave-based optical fiber gas sensor.

Figure 6 .
Figure 6.The structure of photoacoustic cell.Reproduced according to the terms of the CC BY license.[150]Copyright 2022, the authors, published by MDPI.

Figure 7 .
Figure 7. a) Schematic diagram of gas sensing system and optical fiber sensor integrated with ZIP-8 film.b) Dynamic response at ≈242 nm of ZIP-8 coated optical fiber after exposure to various gases.Reproduced with permission.[80]Copyright 2019, American Chemical Society.

Figure 8 .
Figure 8. a) Schematic of LPG-SPR sensor with graphene film, and b) longitudinal section of fiber structure.Reproduced according to the terms of the CC BY license.[91]Copyright 2017, the authors, published by MDPI.

Figure 9 .
Figure 9. a) Structure of SPR H 2 sensor with TFBG in the core.b)Transmission spectrum of the sensor after exposure to H 2 concentrations of 0% and 1.02%.Reproduced with permission.[107]Copyright 2022, Elsevier.

Figure 10 .
Figure 10.a) Schematic representation of structure and sensing mechanism of optical fiber sensor coated with Au and IRMOF-20.b) Plasmon absorption band in different H 2 concentrations.c) Sensor response to 4 and 20% hydrogen concentrations at different ambient temperatures.d) Sensor response to 4 and 20% hydrogen concentrations at different ambient humidity.Reproduced with permission.[114]Copyright 2019, American Chemical Society.

Figure 11 .
Figure 11.Four electrode stacking structures.a) Single sheet stacking; b) Z-stacking; c) cylindrical winding; and d) Prismatic winding.Reproduced according to the terms of the CC BY license.[119]Copyright 2019, the authors, published by IOP Publishing.

Figure 12 .
Figure12.a) FBG sensor (D1) was inserted in the central void of jellyroll from the negative pole.Reproduced according to the terms of the CC BY license.[121]Copyright 2018, the authors, published by Elsevier.b) FBG sensor was inserted into the central void of jellyroll from the positive pole to monitor temperature.c) The complete sensing system.Reproduced according to the terms of the CC BY license.[122]Copyright 2018, the authors, published by Elsevier.

Figure 13 .
Figure13.a) Pouch cell with Au-coated optical fiber inserted.The fiber was integrated between anode and separator.Reproduced according to the terms of the CC BY license.[128]Copyright 2022, the authors, published by MDPI.b,c) FBG sensor inserted on the surface of anode and implanted inside anode.Reproduced with permission.[129]Copyright 2016, Wiley.d) Schematic of FBG integrated with pouch cell.e,f) SEM images of pouch cells without and with heat-sealing film surrounded respectively.Reproduced with permission.[130]Copyright 2017, Elsevier.

Lin
Gan is an associate professor at the School of MSE, HUST.He received his bachelor's degree from Wuhan University (2001-2005, Chemistry) and doctoral degree from Peking University (2007-2012, Physical Chemistry), and then continued his research at Hong Kong University of Science and Technology (HKUST, 2012-2014, CBME).After that, he joined HUST in July 2014.His research interest is focused on controllable synthesis and phase modulation of optoelectronic 2D materials in the initial several years, which further developed into integrated sensor applications based on 2D materials in recent years.Xin Guo is a distinguished professor at the School of MSE, HUST, the Executive Committee Member of the Chinese Society for Solid State Ionics, and the Editorial Board Member of journals "Solid State Ionics" and "National Science Open."He was a senior scientist until 2012, at the Research Center Juelich, Germany; from 1998 to 2002, he worked at the Max Planck Institute, Stuttgart, Germany.In 2005, he received the Ross Coffin Purdy Award.His current research focuses are 1) gas sensors and intelligent sensing systems, 2) neuromorphic devices and systems, and 3) solid electrolytes and solidstate batteries.

Table 1 .
Amount of gas generation.

Table 2 .
Summary of the optical fiber gas sensors.