TECHNICAL FIELD

[001] The present disclosure relates to a thermocouple and a thermopile and devices formed using such a thermocouple. More particularly, the present disclosure relates to thermocouples formed using black phosphorous.

BACKGROUND

[002] With the advent of internet of useful things, gas sensor networks are receiving attention in applications involving spatial control over air-quality both in built and natural environments and in early prevention of industrial hazards caused by fugitive emission, or early intervention. Various methods of gas sensing have been developed so far and each particular sensor needs to meet certain technical and business requirement to qualify as a commercially viable solution.

[003] Furthermore, the emerging applications are increasingly demanding multiple gas sensing capability in the fashion of a gas sensor array and high selectivity and reliability in a mixed and complex environment. On the one hand, commercially available gas sensors based on gas-selective materials (Electrical, Acoustic and Calorimetric) have limited selectivity and life-time against a mixed environment and suffer from large power consumption as the number of array element increases. On the other hand, techniques based on in-situ gas decomposition (Gas Chromatograph) are not on-chip solutions and thereby are limited for applications warranting massive parallel monitoring at low installation and maintenance cost.

[004] Monitoring harmful gases is imperative to the safety of the inhabitants and overall well-being of an organization. Multi-gas sensing can be regarded as a rather holistic approach for combating air-quality issues and industrial hazards pertinent to workplace safety by looking at multiple gases at the same time. In this regard, gas spectrometry is an optical method that can detect a particular gas in a mixed environment by resolving and measuring the spectral components with high reliability and selectivity. Spectrometers can be seamlessly modified to sense and provide feedback on the spatial variation of C0 ₂ , CO and VOC on a single platform. [005] It is desirable to provide a gas sensing device, and other forms of device, of increased effectiveness when compared with the prior art, and to develop new devices that leverage off that increased effectiveness and/or at least provide useful alternatives to existing devices.

SUMMARY OF THE PRESENT DISCLOSURE

[006] The presently disclosure provides an in-plane optical platform for multi-gas sensing at mid-IR. Embodiments therefor may be considered a first of their kind, fully integrated and portable optical solutions developed on Si/AIN Photonics and MEMS (Micro Electro Mechanical System) technology with multiple gas sensing capability. Other salient features of proposed embodiments include power consumption :<=lmW per channel; miniaturized foot-print: W(5mm)XL(lcm)XH(0.5mm) and overall sensitivity down to sub-ppm (C0 ₂ as an example) with significantly improved lifetime.

[007] In-plane photonics used in IR spectroscopy is a relatively new field. In-plane spectrometers operating over mid- or far- IR are yet to be realized. AIN photonics particularly offers working range which covers the desired IR wavelength from 5.5 - 13.5 pm which conventional Silicon photonics does not provide.

[008] In addition, embodiments of the present invention are compatible with CMOS scale fabrication.

[009] Embodiments of the present invention may be, or employ, 2-dimensional materials - sometimes referred to as single-layer materials - that have photovoltaic, semiconductor and other applications. Relevantly, 2-dimensional black phosphorous (BP) shows great thermoelectric properties with high Seebeck coefficient, high carrier mobility and low thermal conductance.

[010] In accordance with the present disclosure, there is provided a thermopile comprising:

one or more thermocouple pairs, each thermocouple of each thermocouple pair comprising:

an n-type conductor formed from doped black phosphorous; and a p-type conductor formed from doped black phosphorous; and an absorber layer disposed between opposite ends of each thermocouple. [Oil] At least one of the n-type conductor and p-type conductor may be chemically doped. The n-type conductor may be doped with Aluminium. The p-type conductor may be doped with one of Carbon, Lithium and Tellurium.

[012] The thermopile may comprise two or more thermocouple pairs connected in series.

[013] The thermopile may further comprise a biasing element for establishing a predetermined bias on at least one location of the thermopile.

[014] The thermopile may be suspended on a membrane. The membrane may be silicon nitride.

[015] The thermopile may be mounted on a substrate, wherein the absorber layer comprises a cold side and a hot side, and the n-type conductor and the p-type conductor form a junction on the cold side, the thermopile comprising a via between one or both of the n-type conductor and p-type conductor to the substrate at the cold side.

[016] Also disclosed is a thermocouple comprising a black phosphorous doped by a dopant such that the thermocouple comprises an n-type conductor and a p-type conductor. The dopant may be chemical dopant. The dopant may be at least one of Carbon, Lithium and Tellurium. The dopant may be Aluminium.

[017] Also disclosed is a method for forming a thermocouple comprising:

forming at least two pairs of black phosphorous bodies, each pair comprising a first black phosphorous body and a second black phosphorous body, on a substrate in series;

n-doping the first black phosphorous body;

p-doping the second black phosphorous body; and

depositing an absorber layer between the first black phosphorous body and the second black phosphorous body of each pair.

[018] The first black phosphorous body may be n-doped before p-doping the second black phosphorous body, wherein p-doping comprises doping over both the second black phosphorous body and the n-doped first black phosphorous body.

[019] Also disclosed is a thermal imaging device, an integrated sensor or a Pirani gauge comprising a thermopile as described above, or a thermocouple as described above.

[020] Disclosed herein is an in-plane, on-chip gas detector comprising:

a sensing block comprising: a path having an input and an output, configured for guiding light of a wavelength band (the desired light band) along the path from the input to the output, the wavelength band being selected for interaction with a gas; and

an external surface around the path, for contacting the gas;

a light source bonded to the input, for emitting light (the emitted light), including light of the desired light band into the input; and

a photodetector bonded to the output, for detecting the desired light band.

[021] The light source may be configured to emit one of mid-infrared and far-infra red light. The light source may comprise a micro-heater light source. The light source may comprise a light emitting diode (LED) light source configured to produce light of the desired light band.

[022] The photodetector may be a thermopile. The thermopile may be a thermopile as described above. The thermopile may be a black phosphorous photodetector. The thermopile may be flip chip bonded to the output.

[023] The photodetector may be an integrated plasmonic na noantenna detector.

[024] The sensing block may comprise a resonator. The resonator may be one of:

a ring resonator configured to maximise light-gas molecule interaction at the desired light band;

a slow light resonator; and

a Vernier resonator.

[025] The sensing block may comprise a non-resonant device. The non-resonant device may comprise a spiral waveguide.

[026] The input may comprise a filter for selectively admitting the desired wavelength band into the path. Additionally, or instead, the output may comprise a filter for selectively admitting the desired wavelength band from the path to the photodetector, or a filter for selectively admitting a wavelength band corresponding to the desired wavelength band when phase-shifted by the gas, from the path to the photodetector.

[027] A multi-gas detector comprising a plurality of on-chip gas detectors, each on-chip gas detector being a gas detector as described above.

[028] The light source of the on-chip gas detectors may be a single, shared light source.

[029] The on-chip gas detectors may be disposed in parallel and formed concurrently and on a common substrate. [030] Advantageously, embodiments of the on-chip gas detector and, particularly, the multi-gas detector comprise an integrated light source and photodetector for gas sensing across wideband spectrum. Thus the detector is portable.

[031] By utilising resonant optical sensing in some embodiments, the detector can be miniaturized to reduce chip footprint.

[032] By provided a multi-gas detector is a single device, a single platform can be provided for monitoring and controlling air quality.

[033] The detectors described herein are free of fibres and bulk optics for sensor read outs. The, unwanted signals that might otherwise effect sensitivity are eliminated, and the sensor or detector module is simplified. This can be useful in, for example, wearable devices.

[034] Embodiments of the present invention may comprise integrated electrical terminals for read-out or connection of a wireless transmission module to enable communication with cloud systems and administrative systems.

[035] Embodiments of the present invention enables mid-infrared (mid-IR), and potentially far-infrared (far-IR), optical sensing without requiring a gas sensitive membrane. This can result in high selectivity, e.g. of a wavelength band associated with a gas, low hysteresis and long longevity.

[036] Embodiments of the present invention employ aluminium nitride (AIN) photonics - e.g. mounting the gas detector to a AIN cantilever - to facilitate sensing at far-IR wavelengths for gases that do not interact, or only comparatively weakly interact, with near- and mid-IR wavelengths.

BRIEF DESCRIPTION OF THE DRAWINGS

[037] Some embodiments of thermocouples, thermopiles and other devices will be now be described by way of non-limiting example only, with reference to the accompanying drawings in which:

FIG. 1 provides schematic diagrams of: a cantilever-shaped thermopile - Fig. la; a bridge-shaped thermopile - Fig lb; a cross-bridge shaped thermopile - Fig. lc, in accordance with the present teachings; and a detailed image of a thermopile in which no metal filling is provided to link the thermopile to the substrate; FIG. 2, comprising FIGs. 2a to 2f, illustrates steps in a method for forming the thermopile of FIG. lc;

FIG. 3 illustrates the calculated responsivity and specific detectivities of thermopiles with: a 10pm x 10pm active area in a vacuum - FIG. 3a; a lOOprn x lOOprn active area in a vacuum; a 10pm x 10pm active area in ambient conditions - FIG. 3a; a lOOpm x lOOpm active area in ambient conditions;

FIG. 4 provides: a schematic illustration of a photo-thermoelectric detector for integrated sensing at mid-IR and the corresponding Seebeck effect in 2D material ribbon integrated with the nanoscale heat source, using a diabolo antenna as a test case - FIG. 4a; localized temperature distribution at the corresponding H-field distribution of the resonant diabolo nanostructure under infrared excitation - FIG. 4b; and (c) geometric scaling of diabolo antenna absorption cross-section by varying the length parameter - FIG. 4c;

FIG. 5 provides schematic representations of a suspended BP ribbon-based Pirani Gauge, particularly a cross-section of the device - FIG. 5a, and various heat loss mechanisms in Pirani Gauge - FIG. 5b;

FIG. 6 provides: in FIG. 6(a) a conceptual schematic of an in-plane optical configuration for multi-gas sensing; and, in FIG. 6(b), spectral fingerprints of various gases at mid-IR spectra (bl), high quality resonance assisted round-trip light-matter interaction in sensing (b2), and the effect of gas absorption on the energy received at the output of the waveguide (b3);

FIG. 7 illustrates: a detailed configuration and packaged version of the present integrated chip-scale spectrometer - FIG 7(a); an individual gas specific channel including a sensing block comprising a band rejection filter and a non-resonant spiral waveguide - FIG. 7(b); light being coupled in and out of the structure of Figure 7(b) using properly designed grating couplers - FIG. 7(c); and mid-IR spectral finger-prints of various gases and the quality factor and spectral similarity of closely spaced gas absorption peaks - FIG. 7(d);

FIG. 8 illustrates a grating coupler as used in the gas detectors disclosed herein, along with the computed sensitivity thereof;

FIG. 9 shows: a photo of a fabricated, high quality factor resonator - FIG. 9(a); experimentally obtained high quality factor line-width for sensing - FIG. 9(b); theoretically observed effect of gas concentration on quality factor - FIG. 9(c); FIG. 10 (a)Optical Image of daisy chain pattern on bottom chip (b) SEM image of Au bumps (c) Thickness of the Al wedge (d) Two bonded chips via Al interface;

FIG. 11 illustrates the process flow for integrating the thermopile and waveguide into a single chip, along with attachment of the light source;

FIG. 12 provides: a schematic view of in-plane photonic components - FIG. 12(a); a schematic view of a system level integrated in-plane optical sensor - FIG. 12(b); simulated surface temperature and radiation power of the microheater of FIG. 12(a); and simulated thermopile output voltage response versus incident IR power - FIG. 12(d);

FIG. 13 illustrates the simulated effect of a beam profile for an LED light source on grating performance, in which: FIG. 13(a) is the Gaussian approximation; FIG. 13(b) is the simulation setup for various angles of divergence; FIG. 13(c) shows 12% coupling obtained from the LED to the waveguide under optimized conditions; FIGs. 13(d) and 13(e) illustrates the electric field distribution of the coupled mode under beam divergence of q=40 deg (FIG. 13(D)) and q=10 deg (FIG. 13(e)); and FIG. 13(f) illustrates the effect of beam polarization on the grating coupling efficiency;

FIG. 14 are images of a microheater and associated measurements, including: a schematic cross-section view - FIG. 14(a); a scanning electron microscope (SEM) view - FIG. 14(b); and an optical image - FIG 14(c); the optical power and efficiency vs power input - FIG 14(d); and resistance vs electrical power input of the fabricated microheater - FIG. 14(e);

FIG. 15 shows a micro-heater laboratory setup - FIG 15(a); the bias dependent broadband emission - FIG. 15(b); and the obtained integrated emission power of 80 pW - FIG 15(c);

FIG. 16 shows: an LED chip - FIG. 16(a); dimensions of the LED chip of FIG. 16(a) - FIG. 16(b); outputs of various monochromatic LEDs - FIGs. 16(c) to 16(e); and the radiant flux v pulse forward current of the LED source - FIG. 16(f);

FIG. 17 illustrates various grating couplers and the corresponding coupling efficiency;

FIG. 18 shows a grating coupler area - FIG. 18(a) - and the coupling efficiency curve - FIG. 18(b);

FIG. 19 illustrates the calculated detection limits of the presently proposed multiple gas detector platform;

FIG. 20 illustrates flip chip bonded BP thermopiles; FIG. 21 shows the simulated response curves for various BP-based photodetector designs;

FIG. 22 shows a plasmonic nanoatenna detector used as an alternative detector device;

FIG. 23 illustrates the flip-chip bonded photonics chip, in which: FIG. 23(a) is an optical image of the fabricated output grating coupler after Au-pad deposition; FIG. 23(b) is a SEM image of the wedges formed on Au-pad; FIG. 23(c) is a SEM image of the bonding area with Au pads visible on the Si-photonic chip side and Al interface; and FIG. 23(d) is an optical image of fabricated spiral waveguides gas sensing applications;

FIG 24 illustrates: the optical spectra of the fabricated photonic chip (sensing block) - FIG. 24(a); outputs of the photonics chip integrated with a thermopile - FIG. 24(b); open circuit voltage and short circuit current output from thermopile - FIG. 24(c); and the absorption spectrum of thermopile active area - FIG. 24(d); and

FIG. 25 illustrates the photovoltage output from integrated thermopile when short FIG. 25(a and b) and long FIG. 25(c and d) spiral waveguides (sensing blocks) are exposed to N ₂ 0 (target) and C0 ₂ (reference) gasses.

DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS

[038] Proposed herein is a novel thermopile. The thermopile may be used in sensing, such as gas and temperature sensing applications. In general, the following description will describe devices using back phosphorous (BP) in thermocouples. Two-dimensional (2D) BP shows great thermoelectric properties with high Seebeck coefficient, high carrier mobility and low thermal conductance. Aluminium (Al) and carbon (C)/Lithium (Li)/ Tellurium (Te) doping processes can create n-type and p-type BP respectively and will allow the formation of one or more thermocouples, or one or more pairs of thermocouples, for thermopile operation. The BP may be formed in ribbons. The BP thermocouples may be used with an absorber layer to form the thermopile. The absorber layer may be a surface absorber layer or volume absorber layer, and may be formed from any appropriate material for a particular application, such as heteropolysiloxane with or without carbon doping. Other materials may be used if context allows. The absorber layer may also be patterned or otherwise adapted for a particular application. [039] As shown in Figure 1, comprising Figures 1(a), 1(b), 1(c) and 1(d), thermopiles 100, 200, 300 and 400 in accordance with the present invention broadly comprise:

102, 202, 302: a thermopile formed from a plurality of thermocouples, each thermocouple formed from doped black phosphorous;

an absorber layer disposed between an end of the n-type thermocouple and an end of the p-type thermocouple.

[040] The thermopile can be realized in various forms depending on the application requirements. With reference to Figure 1(d), a thermopile 400 is shown. The thermopile 400 is on a cantilever substrate similar to Figure 1(a) and comprises two thermocouple pairs connected in series. Each thermocouple comprises an n-doped black phosphorous (BP) conductor 402 and a p-doped BP conductor 404. The thermopile 400 is formed on or transferred onto a membrane - e.g. S N membrane 406 - and an absorber layer 408 formed on top of the junction of each thermocouple - i.e. the location at which the conductors 402 and 404 of a thermocouple meet. The absorber layer 408 may be formed at the hot end of the thermopile 400. At the junction between neighbouring thermocouples - which may be the hot end of the thermopile 400 - a via 410 may be provided to transfer heat to an underlying substrate - e.g. silicon substrate 412. The via 410 is not essential and is thus shown in broken lines. Notably, the thermocouples show in Figure Id include diagonal n-doped conductors 402. Flowever, the conductors may be arranged in parallel or any other orientation as needed.

[041] While thermopile 400 includes two pairs of thermocouples, thermopiles described herein may have one or more thermocouple pairs as needed, to achieve the desired voltage differential between the opposite terminals of the thermopile (i.e. between the input end of end of a first thermocouple, past any intervening thermocouples and from the far end of the last thermocouple). The absorber layer of Figure 1(d) is shown over the junction between conductors of each thermocouple. Flowever, the absorber layer may be positioned at another location between the opposite ends of each thermocouple - e.g. at a point between ends 414 and 416 of a thermocouple shown in Figure 1(d). The opposite ends 414, 418 of the thermopile, may then be connected to a load (e.g. a voltmeter for testing voltage differential across the thermopile).

[042] Figures 1(a), 1(b) and 1(c) illustrate three exemplary thermopile forms 100, 200, 300, each of the figures showing top and cross-section views of each thermopile 100, 200, 300. The thermocouples connected to produce thermopiles 100, 200, 200 are formed from BP. BP is anisotropic in terms of thermal and electrical conductivity and also in Seebeck coefficient. It has lower thermal conductance and higher carrier mobility in the armchair direction, when compared with other directions, which makes armchair direction a preferred direction for thermoelectric applications.

[043] While various alignments are possible, to improve performance of the present embodiments the armchair direction is aligned along the bridge 202, 302 or cantilever direction 102 for the best performance. This process may require the exfoliation of BP onto a substrate. Such a substrate may be a substrate such as

Polydimethy!siloxane {PDMS). Raman measurement can be used for determining the directions of the BP which is then transferred onto the S N substrate or membrane 104, 204, 304 in the desired orientation. Since for every side of the thermopile (1 for cantilever shaped, 2 for bridge shaped and 4 for cross-bridge shaped) the aforementioned steps may be necessary, one can choose a simpler structure in case one doesn't need large number of thermocouples, or can increase the complexity if additional thermocouples are required. The complexity of the fabrication process is thus adjusted to suit.

[044] The thermopile may be based on black phosphorous (BP) as the thermoelectric material. The BP may be two dimensional (2D) BP. With its high Seebeck coefficient, high carrier mobility and low thermal conductance, BP shows enhanced thermoelectric properties when compared with many other materials.

[045] Black Phosphorus ribbons may be chemically doped to produce one or both of the n-type and p-type conductors. For example, the BP may be doped with Al and C/Li/Te, serve as n-type and p-type thermoelectric materials respectively, to form a thermocouple for thermopile operation. There may be one or more thermocouple pairs in the thermopile. The one or more thermocouple pairs may be connected in series. The one or more thermocouple pairs may alternatively be connected in parallel.

[046] The thermocouple may include a biasing element 106, 206, 306 for establishing a predetermined bias on at least one location of the thermopile. The bias may be a gate bias. With reference to Figure 1(a), the gate bias may be applied at a cold side 102 of the absorber layer 104. The gate bias allows post-fabrication adjustment of the thermoelectric performance for further optimization. [047] Where the thermocouple has a gate bias, the gate bias may dynamically modulate the thermoelectric performance of the n-type and p-type thermocouples and thereby modify the device's overall performance.

[048] The BP may be chemically doped. The chemical doping of the BP may enhance the thermoelectric properties of BP. This may enable the BP to form the thermocouple. This is because BP is ambipolar. The doping may force the otherwise ambipolar BP to be n-type or p-type, but not both.

[049] A cold junction may be formed between thermocouples - i.e. on the cold side of the absorber layer. This may be at end or side 108, 208, 308, at the junctions between neighbouring thermocouples in the series forming the thermopile 100, 200, 300. A metal filling 110, 210, 310 may be added to the cold junction area 108, 208, 308 of the thermopile 100, 200, 300 thereby connecting the thermocouples to the substrate 112, 212, 312. This ensures the cold junction remains cold.

[050] BP readily oxidises. To prevent oxidization, a passivation layer must be deposited and the time the 2D flakes are exposed to ambient atmosphere, should be minimized. 2- dimensional (2D) BP flake sizes can usually be <50pm. This may limit the scalability of the device.

[051] The BP ribbons along with the absorber layer may be suspended on top of a low stress silicon nitride (Si3N ₄ ) membrane 104, 204, 304 discussed above, through isotropic etching of Si or another process. The Aluminium oxide (AI ₂ O ₃ ) that is formed during the doping process may be used as, or form, the passivation layer for the BP ribbons. Furthermore, AI ₂ O ₃ may also serve as a gate dielectric for a first - i.e. top - gate. The first gate may be formed through another metal deposition step. The further metal deposition step may further optimize the device and also do dynamic tuning if needed, as discussed above with regard to gate bias.

[052] The via to the substrate next to the cold junction side of the thermocouples 100, 200, 300 will allow the cold junction to thermally contact to the substrate, generating an effective heat sink. This via can be opened through S13N4 etching. Metal deposition (Au or some other metal with high thermal conductance) onto the resulting opening will result is metal formations 110, 210, 310 that create an effective cold junction and thus increase thermopile's performance.

[053] S1O2 has been used as the absorber layer 106, 206, 306 in the active area in the accompanying figures, since it has high absorption in far infrared region. It is also easy to deposit through plasma-enhanced chemical vapour deposition (PECVD), low pressure chemical vapour deposition (LPCVP), sputter or atomic layer deposition (ALD). The material and the deposition process can be changed or selected to meet high absorption requirements at any desired wavelength. Also, the active area of the thermopile can be patterned - e.g. using reactive ion etching - with Meta materia I or Nanoantenna to make achieve wavelength selectivity.

[054] Thermocouples formed using present teachings may be smaller than known thermocouples, thereby enhancing specific detectivity and enabling better thermoelectric performance. BP shows enhanced thermoelectric (TE) performance and hence the active area can be made small to increase the specific detectivity for sensing applications.

[055] The technology can be further extended for integrated sensing and thermal imaging with the integration of wavelength selective optical metamaterial structures allowing multispectral sensing and imaging. In addition, the developed fabrication process is directly relevant for the realization of an ultra-sensitive pirani gauge for pressure sensing applications.

[056] The BP thermocouple may be used on a metamaterial and/or in a nanoantenna absorber. Metamaterials and nanoantenna patterns can be incorporated onto the active area of the thermopile to realize a wavelength selective infrared (IR) detector.

[057] The thermocouple may be able to be disposed on a flexible substrate. The low temperature fabrication process described with reference to Figure 2, may allow realization of a BP thermopile able to be fabricated or transferred onto a flexible substrate.

[058] FIG. 2 shows a proposed fabrication method using mechanical exfoliation to extract BP flakes. The release of the structure can be done using silicon (Si) as the sacrificial layer and xenon difluoride (XeF2) or sulfur hexafloride (SFs) gasses to isotropically etch the Si and release the structure. Since the proposed active area is small, the suspended part of the structure will be small and hence no backside release will be necessary, which enables deep Si etching for backside release to be avoided.

[059] 1. Chemical Doping of BP for Thermocouple Formation

[060] The thermoelectric properties of an ambipolar 2D material including BP can be modulated using gate bias. Thus, thermocouple pairs can be generated using intrinsic 2D flakes. Described herein is the chemical doping of the 2D flake into n-type and p-type thermopiles using BP flakes. This is to be contrasted with electrical-bias tuning methods, i.e., fine adjustment of the carrier concentration of the thermocouples using gate bias. Proposed herein is the use of Al contacts to n-type BP to realize low contact resistance and stronger n-characteristics. Al-doping of BP can be realized with low thermal budget at 120C. The doping process, after forming the metal contacts, will also work as an annealing step to further decrease the contact resistance.

[061] This way, better thermoelectric performance and, hence, better responsivity and specific detectivity can be expected from the fabricated device. BP can be tuned into n- type thermopile by Al doping and into enhanced p-type by carbon (C), lithium (Li), arsenic (As), sulfur (S) or tellurium (Te) doping, in accordance with present teachings.

[062] 2. Enhanced Specific Detectivity (D*) and better Thermoelectric Performance

[063] High thermoelectric power (Seebeck Coefficient > 300) of BP along with high carrier mobility and low thermal conductance results in enhanced thermoelectric performance. Black Phosphorus is thus a good thermoelectric two-dimensional material that can be used to build high performance thermopiles and other thermoelectric devices such as energy harvesters. However, a fabrication constraint on BP exists that with mechanical exfoliation methods, wafer scale flakes cannot be extracted. However, flakes of size 5-50 pm are common and easy to extract. This enables the use of a small active area along with the BP flakes in thermopiles. This increases the performance of the thermopiles in ambient conditions. Moreover, the smaller active area allows simpler fabrication methods to be used, especially in configurations having a suspended structure with the thermocouples and active area.

[064] FIG. 3 shows calculated photo-thermal responsivity and specific detectivities of small (10pm x 10pm active area) and large (100pm x 100pm active area) thermopiles in vacuum and ambient conditions. The calculations take ±300pV/K for the Seebeck coefficients of BP, 20 and 30 W/mK for the thermal conductivities of n-type and p-type BP respectively with 1000 & 2000 cm ² /Vs for electron and hole conductivities with 1019 cm ³ carrier concentration for both types. They also consider 100 nm S N with 3W/mK thermal conductivity for the membrane as well as for the passivation layer. The calculations support the idea mentioned previously that a smaller thermopile will increase the performance of the device in ambient conditions FIG. 3c versus FIG. 3d.

[065] 3. Thermopile on a flexible Substrate [066] Since BP thermopile fabrication processes described herein may be performed at low temperatures, the structure can be transferred or fabricated on top of a flexible substrate to, for example, synthesise a BP transistor on a poly-ethylene terephthalate (PET) substrate, form a BP-based inverter on a Polyimide (PI) substrate using a transfer method, or produce a WS ₂ (n-type) and NbSe ₂ (p-type) based thermoelectric nanogenerator on a polydimethylsiloxane (PDMS) substrate.

[067] To date there has been no thermopile formation on a flexible substrate for, for example, radiation detection. In view of present teachings, it may be possible to fabricate thermopiles on, or transfer thermopiles onto, flexible substrates due to the use of a 2D-material based thermopile and low-temperature doping process.

[068] 4. BP Thermopile Integrated with Wavelength-Selective Metamaterial

Absorber

[069] For ultra-sensitive response at mid-IR range, two dimensional materials such as black phosphorous are considered. FIG. 4a shows a nanostructure-based photo thermoelectric detector 500 for integrated sensing. Sensing may be performed at mid- IR. The photo-thermoelectric detection originates from the Seebeck effect of material. The temperature difference between the hot side 502 and cold side 504 of the material 506 can be converted to an electrical signal as shown. The hot side 502 is heated by optical absorption resonance allowed by the nanostructure or nanopattern designed for mid-IR. FIG. 4b shows a temperature increase of 4.5°C (background temperature set at 300K) of the hot end 502 relative to the cold end 504 for an input optical power of lmW. Such a temperature increase is highly wavelength selective and can be harnessed for band-gap independent photodetection at mid-IR.

[070] The wideband scalability of the underlying concept can be supported by the length study of the geometry in FIG. 4c. The magnitude of the absorption cross-section corresponds to an overall antenna length varying from 280 nm to 360 nm at a fixed gap of 100 nm. The signal output is expected to be increased further by defining the n- and p- doping concentration of the 2D material nanoribbon and forming thermocouples. The photovoltage to be generated will be the outcome of the product (ASAT) instead of SAT in a single nanoribbon structure, where S is the Seebeck coefficient and T is the temperature.

[071] 5. BP based Pirani Gauge [072] Also described herein is a BP based Pirani Gauge 600. A doped BP thermocouple, or a thermopile comprising two or more doped BP thermocouples as taught herein, may be used in the Pirani Gauge. The Pirani Gauge 600 may comprise a gap d between the Pirani bridge 602 formed using the BP doped thermocouple and a substrate 604 of the Pirani Gauge - i.e. between the heat sink and heating element. The gap d may be a nanoscale gap. The size of the gap d in the Pirani Gauge 600, between the heat sink and the heating element, modifies the sensitivity of the Pirani Gauge 600. Reducing the gap d may achieve sensitivity at low pressure ranges - e.g. milli-torr and beyond.

[073] The Pirani gauge 600 design is based on the present BP based thermopile where the thermopile structure is provided without the absorber material. Instead of the absorber material, a heated wire 606 has been placed on top of the S N membrane (FIG. 1(b)). The metal wire is heated with electrical bias and the temperature at the membrane is sensed with BP thermocouples. The steady state temperature of the membrane depends on the applied bias, the geometric design and materials used in thermopile and the surrounding air pressure. With the constant bias, materials and geometry, the steady state temperature of the suspended membrane only depends on the air pressure. By sensing this temperature through BP-based thermopile, the air pressure can be calculated.

[074] For the design of Pirani gauge, minimizing the gap distance (d) between the heated wire and the heat sink (the substrate 603) down to nanoscale is considered for extending the dynamic range of pressure sensing at high vacuum conditions (FIG. 5a). The suspended membrane in the embodiment shown has the thermopile and wire on top of the S N membrane, with an insulating layer between the thermopile and wire. The heat sink maintains a substantially constant temperature (usually around room temperature) and the heated wire generates a temperature gradient from the middle of the structure - mid-bridge - to the substrate - heat sink 603. The temperature generated depends on the air pressure as the air molecules take away heat and decrease the temperature. Thus, if the pressure decreases, the heat taken away decreases so that the temperature in the middle of the structure increases, and this temperature can be measured by the thermopile. It is expected that the heat loss by air convection can be increased in this manner to achieve better sensitivity at the low- pressure regime of the Pirani gauge (FIG. 5b). Aggressive scaling (to reduce dimensions) of the Pirani gauge or element shown will be approached for improving the ohmic heating at a relatively reduced electrical power consumption (of the order of milliwatt). Reducing the Pirani gap (d) will increase the sensitivity of the sensor and decrease the operating power. The scaling will further reduce the solid heat conduction of the Pirani element for ultra-sensitive performance. In the present design, the gap (d) is scaled down and power consumption is reduced by scaling down the lateral dimensions of the membrane (i.e. active area). With the proposed suspension techniques mentioned in section 2 and small lateral dimensions, small gaps (d) and low power consumption will be possible.

[075] The resultant heating may then be described by the formula:

in which w, L, and t are the channel width, channel length and channel thickness, respectively, and lb, Ro and K _b are the electric current, room temperature channel resistance and channel thermal conductivity, respectively.

[076] The above thermopile can, along with other and alternative instruments, be integrated into light-sources and detectors with Si/aluminium nitride (AIN) photonic technology. In the following discussion, parallel waveguide resonators will be implemented on-chip to perform the spectrometer operation. Advantageously, the number of waveguide channels can be increased as needed - this can enable the sensing of multiple gases and other substances on a single chip. Advantageously, the resonators can significantly miniaturize the sensor area by mimicking large travelling distances of optical beams, to build up information from the sensing gas molecules. Proposed herein are various types of resonant devices (e.g., ring cavity, slow-light and Vernier) and non resonant device (e.g., waveguide and spiral geometries) with high performance at the technologically relevant mid-IR spectrum, used in applications such as gas detection.

[077] Figure 6(a) illustrates an in-plane, on-chip gas detector 700. The detector 700 broadly comprises:

a sensing block 702;

a light source 704; and

a photodetector 706.

[078] The sensing block 702 comprises a path 708 having an input 710 and an output 712. The path 708 is configured for guiding light of a wavelength band (the desired light band) along the path from the input 710 to the output 712. The wavelength band is selected for interaction with a gas sought to be detected by the detector 700.

[079] An external surface 714 around the path 708 contacts the gas, and absorbs energy from light passing along the path.

[080] The light source 704 is bonded to the input 710. This enables the light source 704 to be positioned close to the input 710, thereby reducing power consumption. The light source 704 emits light (the emitted light), including light of the desired light band into the input 710. In cases such as that shown in Figure 6(a), a single light source can be used for multiple gas sensing channels (i.e. gas detectors). Thus, whether the light source emits light for a single gas detector 700, or for all detectors in the system 716, it may be configured to emit one or both of mid-IR and far-IR light as required to sense a desired gas. This reduces electrical power consumption in the gas sensor array format when compared with micro-hotplate ceramic based gas sensors.

[081] The photodetector 706 is, similarly, bonded to the output 712, for detecting the desired light band from the light escaping the output 712.

[082] In integrating the light source and detector, particularly for gas sensing across wideband spectra, the gas detectors disclosed herein are highly portable. Moreover, resonant optical sensing facilitates miniaturization of the device. Consequently, the system level implementation can be done on a CMOS compatible platform, realizing low-cost, high-volume production for implementing gas sensor networks.

[083] While the gas detector 700 is for sensing a single gas. As shown in Figure 6(a), any number of detectors, which may be disposed in parallel as shown, may be provided on a single device. The device can therefore be built up to provide the ability to sense multiple different gases. A single platform or device may therefore be used for air- quality monitoring or control. Moreover, the system can replace in-situ gas decomposition methods such as gas chromatography.

[084] Notably, the path 708 may comprise a waveguide. The waveguide may extend all or part of the distance from the input 710 of the path 708 to the output 712. Therefore, light emitted from the light source 704 travels through the device, is manipulated by surrounding gas, and is subsequently detected at the photo detector 706, without the need for wires. Replacement of bulk-optics and fibres in typical optical sensors while providing electrical read-out enables seamless integration with wearable and smart phone devices and interfacing with cloud servers for machine-learning and administrative processing applications. This eliminates unwanted signals which might otherwise affect sensitivity, and enables simplification of the sensor module.

[085] Due to the variability of the path length on a compact, miniature footprint, embodiments of the gas detector may perform mid-IR or far-IR optical sensing without requiring a gas sensitive membrane. Since different gases are sensed through, for example, path length variation, this can remove the need for using gas selective materials. These advantages lead to high-sensitivity, low hysteresis and significant longevity. Moreover, the platform is particularly wavelength scalable for aluminium nitride (AIN) photonics, being suitable for far-IR wavelength detection for identifying the presence of a greater range of gases.

[086] Detectors taught herein are also readily provided with electrical read out, for integrating with a wired or wireless transmission module and the like, for remote data transmission, facilitating machine learning and administrative process or air control measures.

[087] Figure 6(b) illustrates, at 6(b)(bl) the spectral fingerprints of various gases on the mid-IR spectrum. It shows some delineation between the mid-IR spectra for each of a variety of gases. The system 716 of Figure 6(a) may therefore, given a sufficient number of single or individual gas detectors 700, detect some or all of the gases indicated. Figure 6(b)(b2) illustrates the high quality resonance resulting from light passing around the ring sensor of the sensor module 702 of Figure 6(a), as indicated by external surface reference 714. The circumference of the ring may be selected to be a whole number multiple of the wavelength, to enable constructive interference and thus resonance within the ring. Figure 6(b)(b3) illustrates absorption of energy by gas impacting on the ring sensor, read from the output 712 of the path 708 - the path may alternatively be referred to as a waveguide.

[088] Figure 7(a) illustrates the configuration and packaged version of the present integrated chip-scale spectrometer 800. The light source 802 of the present embodiment comprises an on-the-shelf emitter for producing a broadband infrared signal. That signal may be coupled into a waveguide coupler 804 via a focusing lens 806. The lens 806 can be shaped to ensure maximum collection of the incoming energy by the grating coupler structure.

[089] Per system 716, a spectrometer formed using the device 800 will generally consist of multiple gas specific channels. Each channel may be as shown in Fig. 2 (b), being a non-resonant spiral device. In alternative embodiments, combinations of devices may be used such as ring resonators for some gases and spiral waveguides for longer other gases such as those that absorb energy from longer wavelengths - e.g. far-IR.

[090] While various band rejection filters may be used, a one dimensional (ID) photonic crystal cavity can be readily employed at the chip-scale, as a band rejection filter for overcoming the spectral overlap among the channels. Geometric scalability of the filters can be observed in the inset. The photonic crystal cavity will only admit the desired wavelength or wavelengths through the waveguide.

[091] Broad-band mid-IR absorption peaks of various gases are presented in Fig. 19.

The emission spectrum of the infrared emitter in consideration in the inset shows maximal coverage of the peaks on platform consisting of a single light-source. This evidences that various gasses can be sensed in the wavelength spectrum shown and can thus, if necessary, be identified using a common light source.

[092] The seamless configuration of the band-rejection filter with desired quality factor enables precise matching and scanning across the spectral bandwidth for a particular gas. Such approach is especially suitable for discerning two closely spaced gases, e.g. C0 ₂ and CO, on a single platform as illustrated in Figure 2(d) - i.e. the band of the filter can be adjusted to exclude wavelengths that are common to both gases. The major spectral features of CO2 and CO are located at 4.26 pm and 4.4 pm, respectively requiring a spectral separation of 140 nm, to be detected selectively on the same platform.

[093] As a result of the integrated filter design, along with the tortuous path waveguide (e.g. spiral) enabling a full wavelength path to be contained in a small area, the device foot-print for sensing 5 specific gases can be kept within 500 pm (width) x 1 cm (length). Including the integrated light source and photodetector enables the complete packaging to have a footprint of not more than 5 cm (length) x 3 cm (width) x 3 cm (height). The dimensions of the completed package will be primarily determined by the size of the light-source.

[094] The densely arranged waveguide geometry - i.e. the geometry of path 708 of figure 6(a) - used for the sensing element enables realization of a long interaction length on a relatively small foot print. Figure 8 schematically illustrates the interaction of the surrounding gas molecules 808 with a sensing waveguide or path, presently embodied by a spiral waveguide 810. In an embodiment, the length of the waveguide is selected for excitation by CO2. A length of 3.4cm can be obtained from the dense spiral waveguide geometry with a diameter of 100 pm implying a total device foot-print of 0.0079 mm ² . The electrical power consumption is around 250W for a device footprint of 30mm x 30mm x 50mm including the integrated light source and detector, and the resolution is adjustable up to 2.3 nm. The calculated sensitivity is found to be 20 ppm for CO2 gas. Selecting the interaction length to enhance coupling between the waveguide mode and surrounding gas - e.g. by selecting the waveguide length to be a whole number multiple of the wavelength of light the energy of which can be absorbed by the gas - can yield or approach the limit of detection offered by the conventional systems, but on a drastically reduced foot-print.

[095] In alternative embodiments, the waveguide may have the form of a rounded rectangle or square, or an irregular tortuous path of the desired length. Accordingly, the waveguide shape may be adapted to suit surrounding electrical or electronic components on the CMOS wafer or chip. In the embodiment of Figure 9, the sensing block 850 comprises a ring resonator 852 similar to that disclosed by Figure 6(a). The shape and fabricated structure of the resonator 852 shown in Figure 9(a) is matched to maximize the light-gas molecule interaction on a small footprint, for a specific gas. Figure 9(b) shows the effect on the output power of the presence of the gas sought to be detected on the ring resonator 852. There is a clear drop in output power 858 for wavelengths around the wavelength the energy of which is expected to be absorbed by the gas.

[096] Arm 854 may comprise part of the input, or may comprise part of the path between the input and output, depending on context. Moreover, an optical grating may be incorporated into arm 854 to selectively reject wavelengths outside the desired light band - i.e. the wavelengths the energy of which can be absorbed by gas molecules incident on the ring resonator 852. Similarly, arm 856 may comprise part of the output and may also include an optical grating for the same purpose.

[097] Preliminary gas sensing results using the ring resonator 852 as part of sensing block 850 are shown in Figure 9(c). Clearly, the quality factor of the resonance keeps decreasing as the concentration of the surrounding gas increases. This implies change of the energy received at the output 856 of the waveguide 850. Unlike typical refractive index based optical sensors, mid-IR optical sensors disclosed herein may directly probe into the absorption fingerprint of a gas molecule. This can achieve high sensitivity and selectivity even in a mixed environment - e.g. multiple different gases concurrently present. Moreover, by selecting an appropriate diameter to achieve resonance, the ring resonator 852 can realize a large effective length on a significantly reduced foot-print when compared with other gas sensor systems - e.g. table top sized commercial systems.

[098] The proposed chip-scale spectrometer can be realized on suspended a beam or cantilever - e.g. AIN beam. This facilitates low-loss operation at the far-IR wavelength. Moreover, it facilitates use of the thermopile described with reference to Figure 1(a) to be used as the photodetector where, for example, the hot end or side of the thermopile has been patterned to match the wavelength of incident light of the desired light band and will thus heat up in the presence of light of the desired light band.

[099] Such an arrangement is expected to cover wavelengths up to 13.6 pm. This allows a broader range of gases to be explored than was previously achievable on chip- scale devices. Various traces gases, e.g., Ozone possess unique fingerprints in the far-IR wavelength range which could not be otherwise sensed by traditional nondispersive infrared (NDIR) systems.

[100] The embodiment shown in Figure 9 enables highly efficient, on-chip transmission to be achieved. Moreover, the quality factor is adjustable to optimize the sensing performance in terms of selectivity and signal to noise ratio. This provides a viable alternative to thin film based devices, and manually assembled bulk filters. The present approach is highly scalable and can be realized by a single monolithic process. This reduces the cost of manufacturing for volume production.

[101] With the exception of some components - e.g. the light source - the present device may be mostly realized by wafer scale, massively parallel processes. Moreover, for energy efficiency, a single light source can be integrated with (e.g. attached to) the inputs of all parallel sensor blocks. The core sensor can be fabricated on 6"/8" wafer with a sub-dollar die level cost at the manufacturing phase while the integration of detector and light-source can be realized by high speed, parallelly operated flip-chip bonding tool bringing down the overall cost, and/or by forming thermopiles such as BP thermopiles discussed above into the fabrication process - e.g. using flip chip bonding.

[102] The sensor block may be fabricated using known methods. This process may include creation of contacts 860 according to a predefined pattern 862 as shown in Figure 10 on the chip 866 with the sensor block, at opposite ends of the sensor block. These contacts facilitate flip chip bonding. Solder beads or balls 864 (e.g. aluminium wedges or gold balls) are then deposited on the contacts 860. The chip 866 is then flipped over onto the light source and photodetector. The solder balls 864 are then reflowed to attached the light source and photodetector to the chip 866.

[103] FIG. 11 illustrates the chip integration process for bonding the photodetector and light source onto the chip. After formation of the chip 868 - FIG. 11(a) - and placement of the solder balls on the contacts, the chip 868 is flipped over and the solder balls at the output end 870 are reflowed to attach the detector. The detector presently comprises a separately formed thermopile 872 with electrical read out contact block 874.

[104] Once the photodetector 872 has been attached, or concurrently with its attachment, the light source 876 may be attached using the same flip chip and reflow bonding process.

[105] Flip-chip bonders are used to align (in some cases with 2 pm accuracy) and vertically bond the light-source 876 and detector 872 chips on the prefabricated sensor device 868. The various components can be precisely aligned as shown, so that the active area 878 of the light source 876 and the active area 880 of the detector 872 are positioned between respective solder contact regions 882. As can be seen, the solder contact regions 882 also serve to connect electrical readout terminals 883 to take a reading from the photodetector.

[106] Necessary space between components - e.g. 1cm between adjacent groups of sensor blocks - will be provided to accommodate the individual chips. The spacing between the light-source 876 and the waveguide chip 868 can be maintained at couple of micrometers depending on the flip-chip bonding process and materials. This ensures high transferral of light from the light source 868 to the input of the sensing block.

[107] FIG. 12 illustrates a schematic of the arrangement of FIG. 11. In this arrangement, a light source 884 is provided for a single sensor block 884. The sensor block 884 employs a ring resonator 886 such as that described above. The light source 884 is a microheater that provides IR light.

[108] The IR detector 888 is a thermopile. The thermopile 888 of the present embodiment is a cross-bridge thermopile as also shown in Figure 1(c). The light source 882 and thermopile 888 are coupled through respective grating couplers 890, 892 in the normal direction relative to the in-plane photonic devices 884. Integration of the light source 890 and detector 892 is achieved through flip chip bonding onto contacts 894 formed on the photonic device 884. Figure 12(a) illustrates the formed sensor block or photonic device 884, with the light source 884 and detector 888 awaiting attachment. Figure 12(b) shows the integrated product on attachment of the light source 884 and detector 888 - e.g. by flip chip bonding.

[109] Figure 12(c) shows a simulated surface temperature v radiation power curve for the microheater light source 884. The curve shows the change in surface temperature for varying input electrical power. Figure 12(d) shows a simulated output voltage response of the thermopile 888 for given, varying incident IR power at the input of the sensor block 884.

[110] The in-plane photonic device shown in Figure 12 can be scaled up in dimensions to confine the far-IR wavelengths into the sensor block. This can be achieved through increased dimensions of, for example, a ring resonator or by increasing the path length of, for example, the spiral waveguide. To increase the wavelength of permissible incident light, the sensor block may be a AIN device with the microheater flip chip bonded thereonto.

[111] The grating couplers 890, 892 couple the normally incident IR waves emitted by the light source 884 into the in-plane AIN waveguide 884 in a known manner. While the proposed BP based mid IR detection has response only up to 5.5 pm, presently proposed are two approaches to overcome this upper response limit. Firstly, the photodetector is a thermopile detectors bonded at output end of the waveguide 884. Thermopile based broadband IR detectors responsive at far IR region can be flip chip bonded at the output port of far-IR photonic devices (i.e. far-IR sensor blocks such as one mounted on AIN) via grating couplers. After the resonator filter, the high-Q factor IR signal will be delivered to the thermopile detectors of each channel via a respective coupler - e.g. grating couplers. The received IR radiation power from a thermopile can be calibrated with respect to the concentration of an unknown molecule in the environment, at a specific wavelength. The fabricated array of gas sensor devices, in which each gas sensor is configured for detecting one particular molecule at a defined wavelength, the system of gas detectors can cover a 5.5 pm to 13.5 pm sensing range.

[112] By bonding the light source onto the chip as illustrated in Figure 12, simulation results show at least 12% coupling of light into the waveguide structure under a Gaussian approximation of the incoming beam. Notably, polarization dependence of the grating coupler can cause the 50% loss of the incoming power.

[113] This simulated effect is illustrated in Figure 13. The present low-form factor platform may be seamlessly integrated to produce a gas sensor network for visualization with a high degree of spatial granularity. Furthermore, the simultaneous detection of multiple gases provides more holistic information to acquire situational awareness as per environmental regulations. Conventional wall-plugged gas detector solutions are often limited to detecting a single gas at a time whereas the present system, with multiple gas detectors (e.g. multiple detectors 700 of Figure 6(a)), can be extended to detect any number of gases without significantly increasing the overall system footprint. Furthermore, with the development of technology for monolithically integrating LED right onto the waveguide system, the coupling efficiency may be increased significantly and thereby, the electrical power required by the system can be provided by battery circuits on a truly handheld platform. This will eventually open up the possibility of deploying the multi-channel gas sensor for participatory sensing in industry 4.0 applications.

[114] The simulated results suggest the present gas detector, when compared with a common Fabry Perot device, may achieve line widths set out in the table below. It is clear that the waveguide-based chip-scale spectrometer (i.e. gas detector) presently taught can achieve narrower linewidth. This results in increased resolution and less spectral overlap for gas analysis in a highly mixed conditions - i.e. where multiple gases are present.

Table 1: comparison of proposed technology with other technology under development [115] While various light sources may be used as desired, two particular forms have been described herein - a microheater form and a light emitting diode (LED) form. A microheater embodiment 896 in shown in FIG. 14. The microheater 896 is an energy- efficient broadband micro-heater light source. The space 898 between the microheater element 900 and the base 902 may be to ensure the microheater heat output does not affect the grating of the sensor block or the path itself, or to provide a cavity for insertion of a lens for focusing the emitted light into the grating. Figures 14(b) and 14(c) also show the pads 904 for attaching to solder balls to facilitate flip chip bonding.

[116] FIG. 14(d) shows the power output 899 vs electrical power 901 of the microheater. There is no failure of the microheater response over a wide range. Moreover, there is only a small change in resistance of the microheater over a wide variation in input power as shown in FIG. 14(e). The microheater may therefore be efficiently employed over a wide wavelength range.

[117] FIG 15 shows a laboratory setup for testing the microheater - FIG. 15(a). The radiation measurement reflected in FIG. 15(b) shows an integrated output power of 80 uW over a broadband spectrum. In Figures 15(b) and 15(c) the gap at 4.26um wavelength indicates absorption of energy by atmospheric CO2. Thus, from the microheater spectrum, the presence of CO2 is confirmed. Figure 15(b) shows the spectra of the microheater at different biasing conditions and Figure 15(c) shows the resulting spectrum for microheater biasing at 5V.

[118] In addition to microheaters, LED light sources are being explored with high output power at a particular wavelength for an individual gas sensing. A monochromatic output power can be as high as lmW which can serve the need for a minimum power of 1 uW received at the end of the output of the integrated sensor block. FIG. 16 shows the a LED chip - FIG. 16(a) - with dimensions shown in FIG. 16(b). The outputs of various monochromatic LEDs are provided in FIGs. 16 (c) to (e), showing sharp peaks around the desired wavelength, and radiant flux v pulse forward current is shown in FIG. 16(f).

[119] Various devices can be used to couple the emitted light from the LED, microheater or other device, into the waveguide. Grating coupler devices are used in some of the present embodiments, for coupling light from the free space into the sensor structure - i.e. the space between the light source and the waveguide. FIG. 17 shows a fabricated grating coupler 906 along with couplers 908, 910, 912 having different shaped apertures. A maximum coupling efficiency of -8dB is demonstrated. The grating coupler area can be increased to increase or maximise power received from the flip-chip bonded light sources. The three lines shown are the resulting coupling efficiency for different wavelengths, using the grating coupler designs shown in Figures 17(d), 17€ and 17(f).

[120] FIG. 18(a) shows a large grating coupler area and, in FIG. 18(b) the corresponding coupling efficiency curve. Notably, a peak coupling efficiency of around 15% is achieved.

[121] Once the light source, detector and grating or other coupler have been designed, the chip (i.e. sensor block) can be fabricated. Once fabricated, the Photodetectors are to be flip-chip bonded to complete the integrated, multi-gas sensor. Based on preliminary sensing results, a detection limit of sub-ppm can be achieved for C02 for given performance characteristics of the photodetectors, i.e., with a noise margin of 1.5 pW. The details of the calculation can be referred by Fig. 15.

[122] For use as the photodetector, semiconductor BP based detectors such as those described above may be used. Mid-IR photodetection is demonstrated for low bandgap semiconductor platforms using BP. The material enables detection up to 4.2 urn wavelengths. This range covers important gases such as some volatile organic compounds (VOCs), nitrous oxide (N ₂ 0) and carbon monoxide (CO). To achieve this, BP may be used as a photodetector integrated into the waveguide rather than forming a separate thermopile photodetector. For larger wavelengths - e.g. 4.0um or 4.2um upwards - the thermopile embodiment may be preferred.

[123] The developed photodetector response curves are identified in FIG. 20.

[124] Complete thermopile devices may be provided using BP thermocouples as described above. Simulation results for thermopiles of FIGs. 1(a) to 1(c) are shown in FIG. 21. These simulation results show the potential of achieving the required responsivity in ambient conditions. The prefabricated thermopile chip will be flip chip bonded on the waveguide chip later on.

[125] As an alternative, a novel plasmonic nanoantenna detector 920 may be provided as shown in FIG. 22. The detector 920 may be monolithically integrated with the optical sensor. The antenna 920 can work as a nano-scale heat source using an appropriately patterned nanoantenna - i.e. patterned to preferentially absorb energy from specific wavelengths of light - or, where all but the desired wavelength range is filtered from the light inputted into the sensing block, energy from all light exiting the sensing block.

[126] The nanoantenna 920 produces a temperature difference on the thermoelectric material 922 for the electronic read-out of the gas concentration induced signal change at the output. For example, the nanoantenna 920 will heat up, thereby heating thermoelectric material 922, if light of the desired wavelength range exits the sensing block 924, 926, 928. Notably, sensing block 928 is a linear waveguide of a predetermined length which may be selected for resonance with the desired light wavelength for interaction with gas molecules. Preliminary results show around a 4.5°C temperature increase of the hot end of the thermoelectric material 922 (i.e. the end contacting the nanoantenna 920) for a received power of 2.5 uW.

[127] Various different nanoantena configurations may be used. Presently, a bowtie nanoantenna 930, and a diablo nanoantenna 932 arrangement are proposed, though other arrangements may be suitable for other applications.

[128] The detector foot-print can be significantly miniaturized and monolithically fabricated on the waveguide device as shown in Figure 1(a) to 1(c) for the thermopile embodiment. For example, the nanoantenna may be integrated into the hot end of the thermopile arrangement by patterning the thermopile - hot end - or by providing the nanoantenna on the membrane to heat the membrane and thereby heat the hot end of the thermocouples in the thermopile. The nanoribbon, or thermopile structure may be any desired structure or may be as shown in Figure 1(d). The minimum detectable power is calculated to be 72pW. The plasmonic nanoantenna is excited by the waveguide modes and the electronic read-out is extracted by thermoelectric nanoribbons with high performance metrics.

[129] Preliminary results of the flip-chip bonding embodiment is shown in Fig. 23. After flip chip bonding, the gap between the sensing block (photonic chip) and thermopile was measured using SEM. The gap was 3-4pm when Al-wedge was used on one side only and 7pm when Al-wedges were formed on both the sensing block and detector (thermopile). The spacer gap can be controlled by varying the number of Al-wedge layers and applying different pressures and temperatures during the thermocompression bonding. The bonding interface was strong enough to survive multiple drop tests from heights >5 cm.

[130] FIG. 23 shows SEM images of the bonded photonics chip. FIG 23(a) is an optical image of an output end of the sensing block, including an inset close-up, showing the fabricated output grating coupler 934. This image is taken after Au-pad deposition to facilitate flip chip bonding. FIG. 23(b) is a SEM image of the wedges 936 formed on Au- pad 938. FIG. 23(c) is a SEM image of the bonding area with Au pads 938 visible on the Si-photonic chip side 940 and an Al interface 936 to the photodetector 940. FIG. 23(d) is an optical image of two fabricated spiral waveguides 942, 944 integrated with a chamber, presently PDMS gas chamber 946, for gas sensing applications.

[131] Following the integration of thermopile device, the responsivity of the overall system was assessed using continuous illumination (i.e. no chopping). Illumination was 65 mW of power at 3.72 pm. This corresponds to peak responsivity for the gas detector. The open-circuit voltage and short-circuit current from the thermopile were then measured, with a parameter analyser, to be 4.5 mV a nd 65 nA. This translates to 69 mV/W or 1 pA/W respectively (Fig. 22(c)). Later, the IR radiation was chopped at three different frequencies and photo-voltage spectrum was measured in each case (Fig. 22(b). The spectral difference between Fig. 22(a) and 22(b) is mainly due to the fact that in Fig. 22(a) two grating couplers contribute to the spectrum whereas in Fig. 22(b) only the input grating coupler has a significant effect on the spectrum whereas output coupling efficiency is constant (at ~50%) regardless of the wavelength [13] . Thermopile absorption spectrum is measured using Fourier-transform IR spectroscopy (FTIR) from three different regions (Fig. 22(d)) of thermopile whereas region one is the middle portion of thermopile which corresponds the area above the grating coupler and the others are slightly off from the middle part which still may receive some signal. In all cases, the spectra are relatively flat implying thermopile has little to no effect on the measured photovoltage spectrum.

[132] With reference to FIG. 24, the optical spectra of the fabricated photonic chip are shown. This shows a clear peak around the 3.72 pm wavelength, corresponding to the gas sought to be detected. FIG. 24(b) shows various photovoltage outputs of the photonics chip integrated with thermopile, at different chopping frequencies. The lower the chopping frequency, the higher the output photovoltage. It is foreseeable that the highest output photovoltage will result from continuous illumination. FIG. 24(c) shows the open circuit voltage and short circuit current output from thermopile flip chip bonded to the in-plane sensing block when input grating is illuminated with 65mW power at 3.72 pm - gaps in the output result from the light source being switched on and off. FIG. 24(d) shows the absorption spectrum of the active area of the thermopile. The different absorption spectra 945, 947, 949 result from different locations of the thermopile at which the readings are taken.

[133] Frequency spectrum of the bonded thermopile was measured by fixing the wavelength and varying the chopper frequency. The resulting frequency spectrum (Fig. 23a) can be fitted with a low-pass filter behavior (red curve) with a thermal time constants of 28.88ms (for bare thermopile), 1.87ms (for thermopile integrated with 3.72um waveguide), 4.99ms and 4.80ms (for the same thermopile with short and long spiral waveguides at 3.9um). The reduction of the time constant implies that there's heat loss to the substrate of the photonics device causing the device to operate at a larger bandwidth.

[134] An important application of an in-plane spectrometers (i.e. gas detector) as taught herein is in molecular sensing. To demonstrate the sensing capability of such a device, a thermopile can be integrated with a Mid-I R waveguide designed at 3.90um. This waveguide is made in spiral form to increase the interaction area with the target gas (e.g. N2O). Two such spiral waveguides are fabricated on a single chip and they are integrated with the same thermopile - i.e. a single photodetector may be used across multiple sensing blocks. The resulting spectrum with and without N2O are plotted in Fig. 24(c) for short spiral spectrum and 24(d) for long spiral spectrum. At 3.89um where N2O has its absorption peak, we see an intensity drop of the thermopile signal output caused by the absorption of that wavelength by the target molecule. This result suggests that the device can be operated at this wavelength to sense N2O gas in the environment. Time-domain signals are collected using 3.89um laser wavelength with the exposure of target gas (N2O) and reference gas (CO2) at different chopper frequencies using both short and long spiral waveguides and the results are presented in Fig. 25.

[135] The signal drops when the spirals are exposed to N2O gas and recover back when they are exposed to CO2 gas. In these figures CO2 and N2O are pumped alternatingly onto the waveguide. When the N2O is pumped, it absorbs the IR radiation from the light source that travels through the waveguide. Thus, the measured signal drops. When CO2 is pumped, there is no absorption at the detected wavelength so the signal returns to its original position.

[136] These results suggest that the integrated platform consisting of waveguide and thermopile IR detector can be used to detect target gasses in a label-free and real-time fashion. Such integrated systems are feasible for miniaturized spectrometers and molecular sensors.

[137] Many modifications will be apparent to those skilled in the art without departing from the scope of the present invention. [138] Throughout this specification, unless the context requires otherwise, the word "comprise", and variations such as "comprises" and "comprising", will be understood to imply the inclusion of a stated integer or step or group of integers or steps but not the exclusion of any other integer or step or group of integers or steps.

[139] The reference to any prior art in this specification is not, and should not be taken as, an acknowledgment or any form of suggestion that the prior art forms part of the common general knowledge in Australia.