Fabrication of Graphene Nanomesh FET Terahertz Detector

micromachines

Article Fabrication of Graphene Nanomesh FET Terahertz Detector

Yuan Zhai 1, Yi Xiang 1, Weiqing Yuan 2, Gang Chen 2, Jinliang Shi 1, Gaofeng Liang 2, Zhongquan Wen 2 and Ying Wu 1,*

1 Intelligent Technology and Engineering, Chongqing University of Science and Technology, Chongqing 401331, China; [email protected] (Y.Z.); [email protected] (Y.X.); [email protected] (J.S.) 2 College of Optoelectronic Engineering, Chongqing University, Chongqing 400044, China; [email protected] (W.Y.); [email protected] (G.C.); [email protected] (G.L.); [email protected] (Z.W.) * Correspondence: [email protected]

Abstract: High sensitivity detection of terahertz waves can be achieved with a graphene nanomesh as grating to improve the coupling efﬁciency of the incident terahertz waves and using a graphene nanostructure energy gap to enhance the excitation of plasmon. Herein, the fabrication process of the FET THz detector based on the rectangular GNM (r-GNM) is designed, and the THz detector is developed, including the CVD growth and the wet-process transfer of high quality monolayer graphene ﬁlms, preparation of r-GNM by electron-beam lithography and oxygen plasma etching, and

the fabrication of the gate electrodes on the Si3N4 dielectric layer. The problem that the conductive metal is easy to peel off during the fabrication process of the GNM THz device is mainly discussed. The photoelectric performance of the detector was tested at room temperature. The experimental results show that the sensitivity of the detector is 2.5 A/W (@ 3 THz) at room temperature.

Keywords: graphene nanomesh; Si3N4 dielectric layer; FET; terahertz detector

Citation: Zhai, Y.; Xiang, Y.; Yuan, W.; Chen, G.; Shi, J.; Liang, G.; Wen, 1. Introduction Z.; Wu, Y. Fabrication of Graphene Nanomesh FET Terahertz Detector. Graphene, a conjugated carbon sheet arranged in a 2D hexagonal lattice [1] and an Micromachines 2021, 12, 641. https:// important alternative to extend the validity of Moore’s law of electrons in semiconduc- doi.org/10.3390/mi12060641 tors [2], has good transmission performance, a larger volume miniaturization space and a lower cost by virtue of ultrahigh electron mobility and ultrathin material thickness [3–5]. Academic Editor: Francesco La Via Due to the ultrathin planar structure of graphene, the performances of graphene-based field-effect transistor (GFET) devices are not obviously reduced when they shrink in size. Received: 25 February 2021 The fabrication of the device is compatible with current CMOS technology, making GFET a Accepted: 17 May 2021 highly competitive choice for high-performance, high-integration chips in the future [6–8]. Published: 31 May 2021 At present, terahertz technology is widely used in many areas, such as defense, medical diagnosis, security monitoring, communication technology and space exploration. Publisher’s Note: MDPI stays neutral The development of terahertz technology has put forward higher requirements for terahertz with regard to jurisdictional claims in detectors. However, due to the inherent limitations of electron velocity, the performance of published maps and institutional affil- traditional microwave electronic transistors decreases rapidly as the frequency approaches iations. to the terahertz (THz) band (>0.1 THz). It is also difficult for the infrared optical devices to have good applications at frequencies below 20 THz [9]. The special position of THz in the electromagnetic spectrum (both electronic and optical devices are involved) poses a severe challenge to the modern solid-state devices [10]. With high carrier mobility, tunable Copyright: © 2021 by the authors. electronic properties and unique photoelectric properties, graphene provides a new idea for Licensee MDPI, Basel, Switzerland. the research of terahertz direct detectors. Terahertz detectors based on the GFET structure This article is an open access article are developed and reported [6,7,11–13]. distributed under the terms and Although some methods can open the band gap of graphene, such as using graphene conditions of the Creative Commons nanoribbons and double-layer graphene or applying stress to graphene, due to a series of Attribution (CC BY) license (https:// challenges it has not been able to achieve mass production in the existing semiconductor creativecommons.org/licenses/by/ process [11–14]. So far, no mature solid-state devices have been widely used in the terahertz 4.0/).

Micromachines 2021, 12, 641. https://doi.org/10.3390/mi12060641 https://www.mdpi.com/journal/micromachines Micromachines 2021, 12, x FOR PEER REVIEW 2 of 12

Micromachines 2021, 12, 641 2 of 11 process [11–14]. So far, no mature solid-state devices have been widely used in the terahertz band. This study aims to solve the problem encountered in the process of graphene band.transfer This and study lamination. aims to Insolve the paper, the problem high quality encountered graphene in materials the process were of prepared graphene by transferchemical and vapor lamination. deposition In (CVD). the paper, The l higharge- qualityarea and graphene uniform graphene materials nanomesh were prepared struc- byture chemical was fabricated vapor deposition by electron (CVD).-beam Thelithography large-area (EBL). and uniformThe fabri graphenecation process nanomesh of the structureterahertz was detector fabricated based by on electron-beam the graphene lithographynanomesh was (EBL). designed The fabrication and the terahertz process de- of thetector terahertz was developed. detector based on the graphene nanomesh was designed and the terahertz detector was developed. 2. Fabrication Process of Graphene Terahertz Detector 2. FabricationBased on Process the principles of Graphene of the TerahertzgrapheneDetector FET terahertz detector, the pitch size and dimensionsBased on of the the principles graphene of nanomesh the graphene and FETstructural terahertz parameters detector, are the decided pitch size and and de- dimensionsscribed in Table of the 1 graphene. nanomesh and structural parameters are decided and described in Table1. Table 1. Parameters of the graphene FET terahertz detector. Table 1. Parameters of the graphene FET terahertz detector. Parameters of Graphene Value MinimumParameters width of Graphene of nanostructures 30 nm~6 Value0 nm PeriodMinimum of structural width of nanostructures array 1 μm 30 nm~60 nm µ ChannelPeriod of length structural array14 μm 1 m Channel length 14 µm ChannelChannel widthwidth 60 μm 60 µm ThicknessThickness ofof dielectric dielectric layer layer 60 nm 60 nm

AccordingAccording to to the the device device structure structure we we designed, designed, and and the the micro–nano micro–nano processing processing plat- plat- formform of of the the National National Center Center for for Nanoscience Nanoscience and and Technology Technology of China, of China, the overall the overall fabrication fabri- processcation process of the device of the is device divided is divided into the followinginto the following five steps five as shownsteps as in shown Figure in1. Figure 1.

(a) (b) (c)

(d) (e)

Figure 1. The fabrication process of the graphene terahertz field-effect transistor (FET) detector. (a) Chemical vapor deposi- Figure 1. The fabrication process of the graphene terahertz field-effect transistor (FET) detector. (a) Chemical vapor dep- tion (CVD) preparation and substrate transfer of graphene. (b) Fabrication of source and drain electrodes. (c) Fabrication of osition (CVD) preparation and substrate transfer of graphene. (b) Fabrication of source and drain electrodes. (c) Fabrica- graphene nanogrid by electron-beam lithography (EBL). (d) CVD deposition of the Si3N4 dielectric layer. (e) Fabrication of tion of graphene nanogrid by electron-beam lithography (EBL). (d) CVD deposition of the Si3N4 dielectric layer. (e) Fabri- thecation gate of electrode. the gate electrode. 2.1. Chemical Vapor Deposition (CVD) and Transfer of Graphene 2.1. Chemical Vapor Deposition (CVD) and Transfer of Graphene Graphene film is prepared on copper foil (2 cm × 2 cm × 25 µm in volume) by Graphene film is prepared on copper foil (2 cm × 2 cm × 25 μm in volume) by CVD CVD [4,15,16], and then transferred to the prepared Si/SiO2 substrate. Firstly, the desired copper[4,15,16] foil, and substrate then transferred was obtained to the prepared with the Si/SiO method2 substrate. of chemical Firstly, immersion the desired cleaning copper combinedfoil substrate with was electrochemical obtained with polishing. the method Then, of graphene chemical crystals immersion were cleaning grown in combined a single- temperature-zonewith electrochemical CVD polishing. tube furnace. Then, g Theraphene temperature crystals were was grown raised in in a asingle low-pressure-temperature-zone CVD tube furnace. The temperature was raised in a low-pressure environment environment (the pressure was set at 600–800 mTorr), and H2 was introduced when the(the temperature pressure was was set raised. at 600– Under800 mT theorr), condition and H2 was of 1050 introduced◦C, the when copper the foil temperature substrate was raised. Under the condition of 1050 °C , the copper foil substrate was annealed for 1.5 was annealed for 1.5 h and maintained at this temperature, and H2 (catalyst) and CH4 (carbonh and maintained source) were at introducedthis temperature, for chemical and H2 reaction (catalyst) preparation. and CH4 (carbon The reaction source) timewere wasintroduced controlled forwithin chemical half reaction an hour. preparation. After the graphene The reaction film time was was formed, controlled the PMMA within was half coatedan hour. as theAfter protective the graphene layer. film The was thickness formed, of the the PMMA PMMA was is about coated 200 as nmthe whenprotective the spin speed is 4000 RPM and the concentration of the PMMA is 6%. Ferric chloride and

Micromachines 2021, 12, x FOR PEER REVIEW 3 of 12

Micromachines 2021, 12, 641 3 of 11 layer. The thickness of the PMMA is about 200 nm when the spin speed is 4000 RPM and the concentration of the PMMA is 6%. Ferric chloride and hydrochloric acid were used as hydrochloricetching solution acids wereto soak used the as copper etching foil solutions substrate to soak and thesome copper inorganic foil substrate impurities. and someAfter inorganiccopper foil impurities. etching, the After PMMA/graphene copper foil etching, was cleaned the PMMA/graphene with deionized water. was cleaned The selected with deionizedsubstrate is water. a highly The doped selected p-type substrate silicon is with a highly a thickness doped p-typeof 525 μm. silicon The with thickness a thickness of the ofoxide 525 µ layerm. The above thickness the silicon of the wafer oxide layeris 285 above nm. Before the silicon the wafer transfer, is 285 the nm. substrate Before was the transfer,cleaned with the substrate deionized was water cleaned and then with dried deionized with waternitrogen. and The then PMMA/graphene dried with nitrogen. was Thedirectly PMMA/graphene transferred to the was substrate. directly transferredAt last, the PMMA to the substrate. was dissolved At last, and the removed PMMA in was an dissolvedacetone solution. and removed In order in anto acetoneprevent solution.graphene In damage order to and prevent avoid graphene using ultrasound, damage and the avoidPMMA using was ultrasound,removed by the 80 PMMA°C water was bath removed heating. by 80 ◦C water bath heating.

2.2.2.2. Fabrication of thethe SourceSource andand DrainDrain ElectrodesElectrodes GrapheneGraphene ﬁlmfilm isis essentiallyessentially aa singlesingle layerlayer ofof carboncarbon atoms.atoms. DuringDuring thethe processprocess designdesign ofof the device device,, graphene graphene damage damage should should be be avoided avoided as far as faras possible. as possible. Therefore, Therefore, the strip- the strippingping process process is chosen is chosen for electrode for electrode evaporation evaporation in order in to order avoid to damage avoid to damage the graphene to the graphenecaused by caused strong byacid strong etching acid solution etchings solutionsin the electrode in the electrodemanufacturing manufacturing process. process. However,However, thethe metal metal pattern pattern formed formed on on graphene graphene by by the the stripping stripping process process is easy is easy to fall to off.fall Theoff. The main main reason reason is that is that the bondingthe bonding between between graphene graphene and substrateand substrate is very is very low duelow todue the to van the dervan Waals der Waals force. force. When When the metal the metal ﬁlm isfilm evaporated is evaporated on graphene, on graphene, the metal the metal does notdoes directly not directly make make contact contact with with the substrate, the substrate, and grapheneand graphene may may also also act as act a “strippingas a “strip- adhesive”ping adhesive in the” in stripping the stripping process process to strip tothe strip metal the directly.metal directly. The phenomenon The phenomenon of metal of peelingmetal peeling is observed is observed obviously, obviously, especially especially in the in location the location of the of large-areathe large-area metal metal pad pad as shownas shown in Figurein Figure2. 2.

FigureFigure 2.2. MetalMetal sheddingshedding occursoccurs whenwhen thethe graphenegraphene actsacts asas aa “stripping“stripping adhesive”.adhesive”.

InIn orderorder toto solvesolve thisthis problem,problem, aa processprocess ofof etchingetching graphenegraphene isis addedadded onon thethe basisbasis ofof thethe conventionalconventional strippingstripping process,process, andand thenthen thethe metalmetal evaporationevaporation strippingstripping processprocess isis carriedcarried outout twice.twice. TheThe optimizedoptimized processprocess isis shownshown inin FigureFigure3 3.. Before spin-coating, the graphene substrate is pretreated and soaked in acetone at room temperature for half an hour. After that, the graphene substrate is cleaned with isopropanol and deionized water, and dried with nitrogen successively. Next, the pre- baking temperature for the substrate is set to 120 ◦C and the heating time is 3 min. Then, a 4% concentration of PMMA (950 k) is selected as the positive glue to spin onto the substrate with a spin-coating speed of 5000 RPM. After spin-coating, the substrate is post-baked at 150 ◦C for 2 min, as shown in Figure3a. Figure3b shows that the positive photoresist stripping process is used to expose the pattern of the metal wire and pad, in which the graphene would be removed, excluding the part of the source and drain electrodes located in the central part framed in the red dotted line. Figure4 shows the layout pattern for electron-beam direct writing with an exposure dose of 850 µC/cm2. In order to improve the exposure efﬁciency, the electron-beam step is set at 50 nm.

Micromachines 2021, 12, x FOR PEER REVIEW 4 of 12 Micromachines 2021, 12, 641 4 of 11

(a) (b)

(e) (f)

(g) (h)

(i) (j)

Si SiO2 Graphene PMMA Positive lithography Metal

FigureFigure 3. A 3. schematicA schematic diagram diagram of the ofsource thesource and drain and electrode drain electrode production production process. ( process.a) Spin-coat- (a) Spin- ingcoating positive positive photoresist photoresist (b) Electron (b)- Electron-beambeam lithography lithography (c) Positive (c photoresist) Positive photoresist development development (d) Micromachines 2021, 12, x FOR PEER REVIEW 5 of 12 ICP(d etching) ICP etching (e) First (e evaporation) First evaporation of metal of (f metal) Stripping (f) Stripping (g) Spin- (coatingg) Spin-coating positive photoresist positive photoresist (h) (h) ElectronElectron-beam-beam lithography lithography (i) Second (i) Second evaporation evaporation of metal of metal (j) Stripping (j) Stripping.

Before spin-coating, the graphene substrate is pretreated and soaked in acetone at room temperature for half an hour. After that, the graphene substrate is cleaned with isopropanol and deionized water, and dried with nitrogen successively. Next, the pre-baking temperature for the substrate is set to 120 °C and the heating time is 3 min. Then, a 4% concentration of PMMA (950 k) is selected as the positive glue to spin onto the substrate with a spin-coating speed of 5000 RPM. After spin-coating, the substrate is post-baked at 150 °C for 2 min, as shown in Figure 3a. Figure 3b shows that the positive photoresist stripping process is used to expose the pattern of the metal wire and pad, in which the graphene would be removed, excluding the part of the source and drain electrodes located in the central part framed in the red dotted line. Figure 4 shows the layout pattern for electron-beam direct writing with an exposure dose of 850 μC/cm2. In order to improve the exposure efficiency, the electron- Figure 4. The layout of etching graphene under metal. beamFigure step 4. isThe set l ayoutat 50 nmof etching. graphene under metal. Figure3c shows the result of the positive photoresist development. The developer was aFigure mixture 3c ofshows MIBK: the IPA result = 1:3. of Afterthe positive 100 s of photoresist developing development time, the substrate. The developer was then wassoaked a mixture in isopropanol of MIBK: for IPA 50 = s.1:3. In After order 100 to avoids of developing the stripping time, effect the substrate of graphene was uponthen soakedthe metal, in isopropanol oxygen plasma for 50 was s. In used order to etchto avoid the graphenethe stripping with effect an etching of graphene time ofupon 8 s andthe metal,an etching oxygen power plasma of 150 was W used under to aetch pressure the graphene of 5 Pa. with The developmentan etching time result of 8 iss and shown an etchingin Figure power3d. Then, of 150 the W ﬁrst under evaporated a pressure metal of of5 Pa. 8 nm/40 The development nm Ti/Au, shown result inis Figureshown3 e,in Figurecould make3d. Then, contact the with first the evaporated SiO2/Si substrate metal of to 8 enhancenm/40 nm the Ti/Au, adhesion shown force. in Meanwhile, Figure 3e, couldthe useless make graphene contact with easily the fell SiO off.2/Si Therefore, substrate waterto enhance bath heatingthe adhesion and longer force. soaking Meanwhile, time thecan useless be used graphene to enhance easily the strippingfell off. Therefore effect without, water using bath heating ultrasound. and longer The stripping soaking result time canis shown be used in to Figure enhance3f. the stripping effect without using ultrasound. The stripping result is shownThe samein Figure PMMA 3f. positive photoresist was used for spin-coating again. The peeling effect ofThe the same subsequent PMMA process positive could photoresist be enhanced was used by appropriately for spin-coating increasing again. the The thickness peeling effectof the of spin-coating, the subsequent as shown process in c Figureould be3g. enhanced The source by andappropriately drain electrodes increasing are designedthe thickness of the spin-coating, as shown in Figure 3g. The source and drain electrodes are designed and shown in the enlarged picture in Figure 5. This scheme causes the metal to directly make contact with the substrate to enhance the adhesion of the metal, and it can also reduce the contact resistance by making the source and drain electrodes directly make contact with graphene, as shown in Figure 3h. The parameters of electron-beam exposure are the same as those of the first exposure.

Figure 5. The layout of evaporating metal for the source and drain.

The evaporated metal is Au with a thickness of 100 nm as shown in Figure 3i, including the patterns of wire and pad, and source and drain electrodes. In order to ensure the continuity of the whole metal pattern, the evaporation thickness is required to exceed the first overall thickness. Figure 3j shows the source and drain electrodes and metal pads obtained by the same stripping method presented above.

2.3. Fabrication of Graphene Nanomesh by EBL and OPE In order to obtain the large-area and uniform graphene nanogrid, the micro–nano fabrication method of electron-beam lithography (EBL) and oxygen plasma etching (OPE) was used to obtain the corresponding size of graphene nanogrid structure materials.

Micromachines 2021, 12, x FOR PEER REVIEW 5 of 12

Figure 4. The layout of etching graphene under metal.

Figure 3c shows the result of the positive photoresist development. The developer was a mixture of MIBK: IPA = 1:3. After 100 s of developing time, the substrate was then soaked in isopropanol for 50 s. In order to avoid the stripping effect of graphene upon the metal, oxygen plasma was used to etch the graphene with an etching time of 8 s and an etching power of 150 W under a pressure of 5 Pa. The development result is shown in Figure 3d. Then, the first evaporated metal of 8 nm/40 nm Ti/Au, shown in Figure 3e, could make contact with the SiO2/Si substrate to enhance the adhesion force. Meanwhile, the useless graphene easily fell off. Therefore, water bath heating and longer soaking time can be used to enhance the stripping effect without using ultrasound. The stripping result is shown in Figure 3f. Micromachines 2021, 12, 641 5 of 11 The same PMMA positive photoresist was used for spin-coating again. The peeling effect of the subsequent process could be enhanced by appropriately increasing the thickness of the spin-coating, as shown in Figure 3g. The source and drain electrodes are de- sandigned shown and inshown the enlarged in the enlarge pictured inpicture Figure in5 .Figure This scheme 5. This causesscheme the causes metal the to directlymetal to directlymake contact make with contact the substratewith the substrate to enhance to the enhance adhesion the of adhesion the metal, of andthe itmetal, can also and reduce it can alsothe contactreduce the resistance contact byresistance making by the mak sourceing the and source drain and electrodes drain electrode directlys makedirectly contact make withcontact graphene, with graphene as shown, as inshown Figure in3 h.Figure The parameters3h. The parameters of electron-beam of electron exposure-beam exposure are the aresame the as same those as of those the ﬁrst of the exposure. first exposure.

Figure 5. The l layoutayout of evaporating metal for the source and drain.drain.

The evaporated metalmetal isis AuAu with with a a thickness thickness of of 100 100 nm nm as as shown shown in in Figure Figure3i, including3i, includ- ingthe the patterns patterns of wireof wire and and pad, pad, and and source source and and drain drain electrodes. electrodes. In In order order toto ensureensure thethe continuity of the whole metal pattern, the evaporation thickness is required to exceed the firstfirst overall thickness. Figure Figure 33jj showsshows thethe sourcesource andand draindrain electrodeselectrodes andand metalmetal padspads obtained by t thehe same stripping method presented above.above. 2.3. Fabrication of Graphene Nanomesh by EBL and OPE 2.3. Fabrication of Graphene Nanomesh by EBL and OPE In order to obtain the large-area and uniform graphene nanogrid, the micro–nano In order to obtain the large-area and uniform graphene nanogrid, the micro–nano fabrication method of electron-beam lithography (EBL) and oxygen plasma etching (OPE) fabrication method of electron-beam lithography (EBL) and oxygen plasma etching (OPE) was used to obtain the corresponding size of graphene nanogrid structure materials. Firstly, was used to obtain the corresponding size of graphene nanogrid structure materials. a large-area single-layer graphene was grown by chemical vapor deposition on a copper substrate. It was then transferred onto heavily doped p-type Si substrates with a 285 nm SiO2 layer using polymethyl methacrylate (PMMA)-assisted wet-transfer techniques. The silicon wafer (substrate size 1.2 cm × 1.2 cm) with the transferred graphene film was soaked in an acetone solution for 12 h, then soaked in isopropanol, slightly washed with deionized water, and finally dried (<80 ◦C). Then, the PMMA with a concentration of 4% was used as a resist, and the spin-coating speed was set at 5000–6000 RPM (the corresponding coating thickness was about 200 nm). For different structures and sizes of GNM, the corresponding exposure dose was determined and selected for electron-beam exposure. At last, a graphene nanomesh structure was fabricated by oxygen plasma etching. The layout of the graphene channel between the source and drain electrodes is shown in Figure6a, which was processed by EBL [17]. The exploration of the graphene nanomesh is shown in the right image of Figure6b. After the electron-beam direct writing exposure process is completed, the channel image of the device is obtained and shown on the left of Figure6b.

Figure 6. (a) Layout of rectangular graphene nanomesh (r-GNM) channel (b) SEM of rectangular graphene nanomesh (r-GNM) channel.

Micromachines 2021, 12, x FOR PEER REVIEW 6 of 12

Firstly, a large-area single-layer graphene was grown by chemical vapor deposition on a copper substrate. It was then transferred onto heavily doped p-type Si substrates with a 285 nm SiO2 layer using polymethyl methacrylate (PMMA)-assisted wet-transfer techniques. The silicon wafer (substrate size 1.2 cm × 1.2 cm) with the transferred graphene film was soaked in an acetone solution for 12 h, then soaked in isopropanol, slightly washed with deionized water, and finally dried (<80 °C). Then, the PMMA with a concentration of 4% was used as a resist, and the spin-coating speed was set at 5000–6000 RPM (the corresponding coating thickness was about 200 nm). For different structures and sizes of GNM, the corresponding exposure dose was determined and selected for electron- beam exposure. At last, a graphene nanomesh structure was fabricated by oxygen plasma etching. The layout of the graphene channel between the source and drain electrodes is shown in Figure 6a, which was processed by EBL [17]. The exploration of the graphene nanomesh is shown in the right image of Figure 6b. After the electron-beam direct writing exposure process is completed, the channel image of the device is obtained and shown on the left of Figure 6b.

(a) (b)

20μm 1μ m Micromachines 2021, 12, 641 6 of 11 Figure 6. (a) Layout of rectangular graphene nanomesh (r-GNM) channel (b) SEM of rectangular graphene nanomesh (r-GNM) channel.

SEMSEM imagesimages ofof thethe graphenegraphene nanomeshnanomesh structurestructure afterafter oxygenoxygen plasmaplasma etchingetching areare shownshown inin FigureFigure7 7.. It It can can be be seen seen that that the the fabricated fabricated graphene graphene nanogrid nanogrid can can maintain maintain the the integrityintegrity ofof thethe periodicperiodic structurestructure inin thethe large-area.large-area.

FigureFigure 7.7. SEMSEM imagesimages ofof thethe graphenegraphene nanomeshnanomesh structurestructure afterafter oxygenoxygen plasmaplasma etching.etching.

2.4.2.4. CVDCVD Deposition of thethe DielectricDielectric LayerLayer AsAs thethe devicedevice willwill workwork inin THzTHz frequencyfrequency bands,bands, thethe dielectricdielectric constantconstant ofof thethe gategate dielectricdielectric layerlayer underunder thethe gategate electrodeelectrode shouldshould bebe asas largelarge asas possible.possible. ForFor graphenegraphene thatthat hashas beenbeen processedprocessed intointo nanostructure,nanostructure, thethe bondingbonding abilityability withwith thethe dielectricdielectric layerlayer onon thethe surfacesurface ofof graphenegraphene and and the the question question of of easy easy damage damage should should be consideredbe considered carefully carefully in the in subsequentthe subsequen processes.t processes. Therefore, plasma-enhanced chemical vapor deposition (PECVD) was chosen to de- posit Si3N4 on the surface of the graphene nanomesh as the dielectric layer. Because the combination of Si3N4 and graphene is much better than that of a silicon oxide dielectric layer created by the thermal oxidation process, the dielectric constant of Si3N4 (ε = 6.6) is

much higher than that of silicon oxide (ε = 3.9), and Si3N4 has a higher polarized photo– phonon frequency, which can reduce the phonon scattering of the graphene conductive channel and is conducive to the photoelectric application of graphene [18]. Simultaneously, the PECVD process worked at a low-temperature environment, which could effectively prevent the damage of graphene. The gas source of Si3N4 CVD was the inert gas of N2 and NH3. In addition, the power source was a low-density plasma with a power of only 40 W. These factors could ensure that the damage to graphene is minimized in the manufacturing process of the dielectric layer. The growth process of the Si3N4 dielectric layer is shown in Figure8. The result of spin-coating positive photoresist is shown in Figure8a. The rotation speed was set at 4000–5000 RPM to spin-coat with the PMMA 950 k positive photoresist and control the photoresist thickness above 200 nm. The pre-baking and post-baking time are 1–2 and 2 min, respectively. The pre-baking and post-baking temperatures are 120 and 150 ◦C, respectively. As the pattern of the dielectric layer is simple and the requirement of dimensional accuracy is low, only accurate alignment is required for electron-beam direct writing. The exposure dose of electron beam was 850 µC/cm2, and the step setting was 50 nm. The fabrication of EBL is shown in Figure8b. The mixture of MIBK and IPA (1:3) was used as developer. The positive photoresist development time was 100 s, and isopropanol was used for ﬁxing for 30 s. After the development was obtained, as shown in Figure8c, the Si 3N4 was grown by using the Si 500 D PECVD equipment (SENTECH). SiH4/N2 and NH3 were used as the gas source, and the reaction equation is:

3SiH4 + 4NH3 → Si3N4 + 12H2↑ Micromachines 2021, 12, x FOR PEER REVIEW 7 of 12

Therefore, plasma-enhanced chemical vapor deposition (PECVD) was chosen to de- posit Si3N4 on the surface of the graphene nanomesh as the dielectric layer. Because the combination of Si3N4 and graphene is much better than that of a silicon oxide dielectric layer created by the thermal oxidation process, the dielectric constant of Si3N4 (ε = 6.6) is much higher than that of silicon oxide (ε = 3.9), and Si3N4 has a higher polarized photo– phonon frequency, which can reduce the phonon scattering of the graphene conductive channel and is conducive to the photoelectric application of graphene [18]. Simultane- ously, the PECVD process worked at a low-temperature environment, which could effectively prevent the damage of graphene. The gas source of Si3N4 CVD was the inert gas of N2 and NH3. In addition, the power source was a low-density plasma with a power of only 40 W. These factors could ensure that the damage to graphene is minimized in the manufacturing process of the dielectric layer. The growth process of the Si3N4 dielectric layer is shown in Figure 8. The result of spin-coating positive photoresist is shown in Figure 8a. The rotation speed was set at 4000–5000 RPM to spin-coat with the PMMA 950 k positive photoresist and control the photoresist thickness above 200 nm. The pre-baking and post-baking time are 1–2 and 2 min, respectively. The pre-baking and post-baking temperatures are 120 and 150 °C, respectively. As the pattern of the dielectric layer is simple and the requirement of dimensional accuracy is low, only accurate alignment is required for electron-beam direct writing. The exposure dose of electron beam was 850 μC/cm2, and the step setting was 50 nm. The fabrication of EBL is shown in Figure 8b. The mixture of MIBK and IPA (1:3) was used as developer. The positive photoresist development time was 100 s, and isopropanol was used for fixing for 30 s. After the development was obtained, as shown in Figure 8c, the Si3N4 was grown by using the Si 500 D PECVD equipment (SENTECH). SiH4/N2 and NH3 were used as the gas source, and the reaction equation is: Micromachines 2021, 12, 641 7 of 11 3SiH4 + 4NH3 → Si3N4 + 12H2↑

(a) (b)

(e)

Si SiO2 Graphene PMMA Positive lithography Metal Si3N4

Figure 8. The preparation process of the dielectric layer. (a) Spin-coating positive photoresist (b) Electron-beam lithography Figure 8. The preparation process of the dielectric layer. (a) Spin-coating positive photoresist (b) Electron-beam lithogra- (c) Positive photoresist development (d) CVD Si3N4 (e) Stripping. phy (c) Positive photoresist development (d) CVD Si3N4 (e) Stripping. While the RF power, temperature, pressure and deposition time are set to 40 W, 55 ◦C, While the RF power, temperature, pressure and deposition time are set to 40 W, 55 50 Pa and 1 h, respectively, the thickness of the Si3N4 ﬁlm would be about 60 nm, as shown Micromachines 2021, 12, x FOR PEER REVIEW°C, 50 Pa and 1 h, respectively, the thickness of the Si3N4 film would be about 60 nm,8 of as12 in Figure8d. The last step of the dielectric layer deposition is stripping, which used an acetoneshown in solution Figure to8d. remove The last the step PMMA, of the dielectric and the result layer is deposition shown in is Figure stripping,8e. which used an acetone solution to remove the PMMA, and the result is shown in Figure 8e. 2.5.2.5. FabricationFabrication ofof thethe GateGate ElectrodeElectrode

AfterAfter thethefabrication fabrication of of the the dielectric dielectric layer layer Si3 SiN43N, the4, the graphene graphene nanomesh nanomesh was was covered cov- byered silicon by silicon nitride, nitride, and the and possibility the possibility of damage of dama wasge reduced, was reduced as shown, as shown in Figure in 9Figure. There- 9. fore,Therefore, the conventional the conventional metal evaporationmetal evaporation stripping stripping process process can be usedcan be for used the fabricationfor the fab- ofrication different of different gate electrodes. gate electrodes. The speciﬁc The processspecific ﬂowprocess is shown flow is in shown Figure in 10 Figure. 10.

FigureFigure 9.9. AnAn opticaloptical photographphotograph ofof thethe devicedevice afterafter thethe fabricationfabrication ofof thethe dielectricdielectric layer.layer.

Figure 10a shows the spin-coating result, among which PMMA (950 k) stock solution was used for spin-coating, the spin-coating speed was 4500 RPM, and the thickness of the adhesive was controlled above 200 nm. The pre-baking and post-baking time both were ◦ 100 s,(a and) the temperatures were set to 80 and 150 C, respectively.(b) While the electron-beam exposure dose was set at 900 µC/cm2 and the step was set at 50 nm, the overlay precision could be ensured for the electron-beam direct writing exposure of the gate electrode pattern. The EBL result is shown in Figure 10b. The positive development still used the mixture of MIBK and IPA (1:3) as developer in Figure 10c. The development time was 100 s and then isopropanol was used for ﬁxing. The thickness of the evaporated gate metal should be as thin as(c) possible, as shown in Figure 10d. Therefore, we( evaporatedd) 10/50 nm thick Ti/Au metal, which is conducive to complete stripping. The PMMA adhesive was removed by immersing the substrate in an acetone solution to complete the stripping process, as shown

(e)

Si SiO2 Graphene PMMA Positive lithography Metal Si3N4

Figure 10. The process flow of the gate electrode. (a) Spin-coating positive photoresist (b) Electron-beam lithography (c) Positive photoresist development (d) Evaporating metal (e) Stripping.

Figure 10a shows the spin-coating result, among which PMMA (950 k) stock solution was used for spin-coating, the spin-coating speed was 4500 RPM, and the thickness of the adhesive was controlled above 200 nm. The pre-baking and post-baking time both were 100 s, and the temperatures were set to 80 and 150 °C, respectively. While the electron- beam exposure dose was set at 900 μC/cm2 and the step was set at 50 nm, the overlay precision could be ensured for the electron-beam direct writing exposure of the gate electrode pattern. The EBL result is shown in Figure 10b. The positive development still used the mixture of MIBK and IPA (1:3) as developer in Figure 10c. The development time was 100 s and then isopropanol was used for fixing. The thickness of the evaporated gate metal should be as thin as possible, as shown in Figure 10d. Therefore, we evaporated 10/50 nm thick Ti/Au metal, which is conducive to complete stripping. The PMMA adhesive was removed by immersing the substrate in an acetone solution to complete the stripping process, as shown in Figure 10e. A water bath with long-time soaking was used to ensure

Micromachines 2021, 12, x FOR PEER REVIEW 8 of 12

2.5. Fabrication of the Gate Electrode

After the fabrication of the dielectric layer Si3N4, the graphene nanomesh was covered by silicon nitride, and the possibility of damage was reduced, as shown in Figure 9. Therefore, the conventional metal evaporation stripping process can be used for the fabrication of different gate electrodes. The specific process flow is shown in Figure 10.

Micromachines 2021, 12, 641 8 of 11

in Figure 10e. A water bath with long-time soaking was used to ensure complete stripping, Figurebecause 9. An due optical to the photograph narrow metal of the spacing device after it was the easy fabrication to cause of the incomplete dielectric stripping.layer.

(a) (b)

(e)

Micromachines 2021, 12, x FOR PEER REVIEW 9 of 12

Si SiO2 Graphene PMMA Positive lithography Metal Si3N4

Figure 10. The process ﬂow of the gate electrode. (a) Spin-coating positive photoresist (b) Electron-beam lithography (c) Figure 10. The process flow of the gate electrode. (a) Spin-coating positive photoresist (b) Electron-beam lithography (c) Positive photoresist developmentcomplete (d) Evaporatingstripping, because metal (e )due Stripping. to the narrow metal spacing it was easy to cause incom- Positive photoresist developmentplete (stripping.d) Evaporating metal (e) Stripping. UsingUsing the the above above process, process, we we fabricated fabricated a ann r r-GNM-GNM FET FET THz THz device device with with a graphene Figure 10a shows the spin-coating result, among which PMMA (950 k) stock solution nanomeshnanomesh size size of of 14 14 nm nm ×× 6060 nm, nm, which which is is shown shown in in Figure Figure 1 111. . was used for spin-coating, the spin-coating speed was 4500 RPM, and the thickness of the adhesive was controlled above 200 nm. The pre-baking and post-baking time both were 100 s, and the temperatures were set to 80 and 150 °C, respectively. While the electron- beam exposure dose was set at 900 μC/cm2 and the step was set at 50 nm, the overlay precision could be ensured for the electron-beam direct writing exposure of the gate electrode pattern. The EBL result is shown in Figure 10b. The positive development still used the mixture of MIBK and IPA (1:3) as developer in Figure 10c. The development time was 100 s and then isopropanol was used for fixing. The thickness of the evaporated gate metal should be as thin as possible, as shown in Figure 10d. Therefore, we evaporated 10/50 nm thick Ti/Au metal, which is conducive to complete stripping. The PMMA adhesive was removed by immersing the substrate in an acetone solution to complete the stripping process, as shown in Figure 10e. A water bath with long-time soaking was used to ensure

FFigureigure 11. 11. OpticalOptical photographs photographs of of the the top top gate gate FET FET d device.evice.

3.3. Measurement Measurement Experiments Experiments AAccordingccording to to the the size size of of the the silicon silicon substrate, substrate, the thePCB PCB board board is designed is designed for the for con- the venienceconvenience of testing of testing the electrical the electrical properties properties and terahertz and terahertz detection detection of the of FET the device FET device.. Fig- ureFigure 12 shows 12 shows that thatthe metal the metal pads pads of the of device the device are connected are connected to the topins the of pins the PCB of the board PCB byboard gold by wires gold. wires.

Figure 12. Physical drawings of devices on PCB.

3.1. Electrical Transfer Characteristics The transfer characteristics of the Gr layer are shown in Figures 13. Figure 13a shows the transfer characteristics at Vds = 2 V for the devices based on r-GNMs with different neck widths of 30, 40, 50, and 60 nm, from which we could determine the corresponding Ion/Ioff ratios of ~40, ~25, ~5, and ~4, respectively. The transfer characteristics for the devices based on c-GNMs with different neck widths of 30, 40, 50, and 60 nm are presented in Figure 13b. Comparing Figure 13a with Figure 13b, we can see that the conduction current of c-GNMs is much larger than that of r-GNMs (about two times). As a result of GNM being able to be viewed as an interconnected network structure of graphene, the actual area of c-GNM delivering current is greater than that of r-GNM, which leads to the current of c-GNM being greater than r-GNM under the same conditions. Additionally, the Ion/Ioff ratios of r-GNMs with different neck widths of 30, 40, 50, and 60 nm obtained were ~100, ~25, ~8, and ~3, respectively, indicating that the Ion/Ioff ratio of the GNM-based devices

Micromachines 2021, 12, x FOR PEER REVIEW 9 of 12

complete stripping, because due to the narrow metal spacing it was easy to cause incomplete stripping. Using the above process, we fabricated an r-GNM FET THz device with a graphene nanomesh size of 14 nm × 60 nm, which is shown in Figure 11.

Figure 11. Optical photographs of the top gate FET device.

3. Measurement Experiments According to the size of the silicon substrate, the PCB board is designed for the con- Micromachines 2021, 12, 641 venience of testing the electrical properties and terahertz detection of the FET device9. ofFig- 11 ure 12 shows that the metal pads of the device are connected to the pins of the PCB board by gold wires.

FigureFigure 12.12. PhysicalPhysical drawingsdrawings ofof devicesdevices onon PCB.PCB.

3.1.3.1. ElectricalElectrical TransferTransfer CharacteristicsCharacteristics TheThe transfertransfer characteristicscharacteristics of the Gr layer are shown in Figures Figure 13 13.. FigureFigure 13 13aa showsshows thethe transfertransfer characteristicscharacteristics atat VdsVds == 2 VV forfor thethe devicesdevices basedbased onon r-GNMsr-GNMs withwith differentdifferent neckneck widthswidths ofof 30,30, 40,40, 50,50, andand 6060 nm,nm, fromfrom whichwhich wewe couldcould determinedetermine thethe correspondingcorresponding Ion/IoffIon/Ioff ratios ratios of of ~ ~40,40, ~ ~25,25, ~5, ~5, and and ~4, ~4, respectively. respectively. The The transfer transfer characteristics characteristics for forthe thede- devicesvices based based on on c- c-GNMsGNMs with with different different neck neck widths widths of of 30, 30, 40, 40, 50, 50, and and 60 nm are presentedpresented inin FigureFigure 13b.13b. Comparing Comparing Figure Figure 13a 13 awith with Figure Figure 13b, 13 b,we we can can see seethat that the theconduction conduction cur- currentrent of c of-GNMs c-GNMs is much is much larger larger than than that thatof r-GNMs of r-GNMs (about (about two times). two times). As a result As a resultof GNM of GNMbeing beingable to able be viewed to be viewed as an interconnected as an interconnected network network structure structure of graphene, of graphene, the actual the actual area of c-GNM delivering current is greater than that of r-GNM, which leads to the Micromachines 2021, 12, x FOR PEER REVIEWarea of c-GNM delivering current is greater than that of r-GNM, which leads to the current10 of 12 current of c-GNM being greater than r-GNM under the same conditions. Additionally, the of c-GNM being greater than r-GNM under the same conditions. Additionally, the Ion/Ioff Ion/Ioff ratios of r-GNMs with different neck widths of 30, 40, 50, and 60 nm obtained were ratios of r-GNMs with different neck widths of 30, 40, 50, and 60 nm obtained were ~100, ~100, ~25, ~8, and ~3, respectively, indicating that the Ion/Ioff ratio of the GNM-based ~25, ~8, and ~3, respectively, indicating that the Ion/Ioff ratio of the GNM-based devices devicescan be readily can be tuned readily by tuned varying by varying the neck the width, neck which width, plays which an plays important an important role in charge role in chargeof transport of transport properties. properties.

FigureFigure 13.13. GNMGNM transfer transfer characteristic characteristic curves curves with with dimensions dimensions of 30,of 40,30, 5040, and 50 and 60 nm, 60 respectively.nm, respectively (a) r-GNM. (a) r-GNM (b) c-GNM. (b) c- GNM. 3.2. Terahertz Photocurrent Measurement 3.2. TerahertzThe IR-30 Photocurrent steady-state Measurement infrared light source (HawkEye Technologies) was used as the light sourceThe IR- for30 thesteady THz-state response infrared test. light The source excitation (Hawk sourceEye of Technologies) the light source wa wass used powered as the bylight 2.5 source V DC andfor the could THz be response regarded test. as a The blackbody excitation light source source of at the a stable light source working was state. pow- A 3ered THz by band-pass 2.5 V DC filterand could is used be behind regarded the as light a blackbody source to obtainlight source a 3 THz at a electromagnetic stable working radiationstate. A 3 wave.THz band The- BPF3.0pass filter filter is (TYDEX)used behind was the used light to obtain source the to transmissionobtain a 3 TH spectrum,z electro- whichmagnetic is shown radiation in Figurewave. 14Thea. B InPF order3.0 filter to obtain (TYDEX the) influencewas used ofto obtain the gate the voltage transmission on the terahertzspectrum, photocurrent, which is shown the in gate Figure voltage 14a. In was order changed to obtain to test the the influence photocurrent of the gate for r-GNMvoltage withon the a 60terahertz nm width. photocurrent, The gate voltage the gate had voltage a decisive wa effects changed on the to dispersion test the photocurrent of the graphene for r-GNM with a 60 nm width. The gate voltage had a decisive effect on the dispersion of the graphene terahertz plasmon by changing the Fermi level, but the effect on the terahertz absorption of the graphene nanomesh was not obvious.

(a) (b)

Figure 14. (a)The terahertz transmission of BPF3.0 (b) the curves of the photocurrent of the FET terahertz detector versus the gate voltage.

Figure 14b shows that the increase in gate voltage increased the photocurrent, but the increase after the neutral point tends to be saturated. The reason is that the increase in gate voltage will also increase the Fermi energy level, and then affect the carrier concentration, so that the excitation of graphene plasmon is weak when the Fermi energy level is relatively low, resulting in a slightly small photocurrent. When the Fermi level increases to a certain extent, the saturation of the terahertz absorption makes the photocurrent no longer increase. In order to further determine the response rate of the graphene nanomesh FET terahertz detector, the standard Golay cell GC-1P terahertz detector (TYDEX) was used to

Micromachines 2021, 12, x FOR PEER REVIEW 10 of 12

can be readily tuned by varying the neck width, which plays an important role in charge of transport properties.

Figure 13. GNM transfer characteristic curves with dimensions of 30, 40, 50 and 60 nm, respectively. (a) r-GNM (b) c- GNM.

3.2. Terahertz Photocurrent Measurement The IR-30 steady-state infrared light source (HawkEye Technologies) was used as the light source for the THz response test. The excitation source of the light source was powered by 2.5 V DC and could be regarded as a blackbody light source at a stable working state. A 3 THz band-pass filter is used behind the light source to obtain a 3 THz electro- Micromachines 2021,magnetic12, 641 radiation wave. The BPF3.0 filter (TYDEX) was used to obtain the transmission 10 of 11 spectrum, which is shown in Figure 14a. In order to obtain the influence of the gate voltage on the terahertz photocurrent, the gate voltage was changed to test the photocurrent for r-GNM with a 60 nm width. The gate voltage had a decisive effect on the dispersion of the graphene terahertzterahertz plasmon plasmon by changing by changing the theFermi Fermi level level,, but but the the effect effect on onthe the terahertz terahertz absorption absorption of ofthe the graphene graphene nano nanomeshmesh was was not not obvious. obvious.

(a) (b)

Figure 14. (aFigure)The terahertz 14. (a)The transmission terahertz transmission of BPF3.0 (b of) the BPF3.0 curves (b) of the the curves photocurrent of the photocurrent of the FET terahertz of the FET detector versus the gate voltage.terahertz detector versus the gate voltage.

Figure 14b showsFigure that 14 bthe shows increase that thein gate increase voltage in gate increase voltaged the increased photocurrent, the photocurrent, but but the the increase afterincrease the neutral after the point neutral tends point to be tends saturated. to be saturated. The reason The is that reason the is increase that the in increase in gate gate voltage willvoltage also willincreas alsoe increasethe Fermi the energy Fermi level, energy and level, then and affect then the affect carrier the concen- carrier concentration, tration, so thatso the that excitation the excitation of graphene of graphene plasmon plasmon is weak is weak when when the theFermi Fermi energy energy level level is relatively is relatively low,low, resulting resulting in in a slightly a slightly small small photocurrent. photocurrent. When When the the Fermi Fermi level level increases increases to a certain to a certain extent,extent, the the saturation saturation of of the the terahertz terahertz absorption makesmakes thethe photocurrent photocurrent no no longer increase. longer increase. In order to further determine the response rate of the graphene nanomesh FET ter- In order ahertzto further detector, determine the standard the response Golay rate cell of GC-1P the graphene terahertz nano detectormesh(TYDEX) FET te- was used to rahertz detector,calibrate the standard the power Golay of the cell terahertz GC-1P terahertz radiation detector source by (TYDEX replacing) was the used graphene to nanomesh FET terahertz detector in the same position. According to the voltage sensitivity formula:

V RV = (1) Po

where RV is the hardness of high Leghorn. When the reference frequency was 10 Hz, the photovoltage signal intensity measured by the Golay GC-1P was 6.47 µW. Under this condition, the sensitivity of the Golay was 81.3 KV/W, and the output optical power Po was about 0.08 µW. The current sensitivity formula of the detector is as follows:

i Ri = (2) Pin

where Ri is the current sensitivity of the detector, i is the detection current, and Pin is the actual incident light power. The current sensitivity of the graphene terahertz FET detector is 12 mA/W under 3 THz radiation.

4. Conclusions Monolayer graphene was grown on copper foil by CVD, and the high-quality graphene ﬁlm was obtained by wet transfer on the SiO2/Si substrate. Then, aiming at the problems of electrode stripping, the method of metal evaporation and photoresist stripping twice was used to realize the source and drain electrode lamination of graphene, and enhance the adhesion of the metal wire and electrodes with the substance. Next, the graphene nanomesh with a large-area and uniform size was fabricated by EBL and oxygen plasma etching. Lastly, the conventional metal evaporation stripping process was used to fabricate the gate electrode on the Si3N4 dielectric layer. In sum, the fabrication process of the terahertz detector based on the graphene nanomesh was designed and the terahertz detector was developed. The electrical characteristics of the graphene nanomesh were tested, which veriﬁed the regulating effect of the graphene nanomesh on the electronic energy gap. A Micromachines 2021, 12, 641 11 of 11

test system was built to test the developed detector, suggesting that the THz detection sensitivity of the detector can reach 2.5 A/W at room temperature.

Author Contributions: Y.W., G.C. and Z.W. conceived and designed the study; Y.X. and G.C. performed the theory; Y.Z., W.Y. and G.L. performed the experiments; W.Y., Y.X. and J.S. analyzed the experimental data; Y.Z., W.Y. and Y.W. wrote the paper; Y.W. and Z.W. provided guidance and modi- fication of the paper. All authors have read and agreed to the published version of the manuscript. Funding: This research work was financially supported by the Chongqing Research Program of Basic Research and Frontier Technology (NO: cstc2018jcyjAX0519). Conflicts of Interest: The authors declare no conflict of interest.

References 1. Wang, R.; Ren, X.; Yan, Z.; Jiang, L.; Sha, W.; Shan, G. Graphene based functional devices: A short review. Front. Phys. 2019, 14, 13603. [CrossRef] 2. Jacob, A.; Xie, R.; Sung, M.; Liebmann, L.; Lee, R.; Taylor, B. Scaling Challenges for Advanced CMOS Devices. Int. J. High Speed Electron. Syst. 2017, 26, 1740001. [CrossRef] 3. Buron, J.; Pizzocchero, F.; Jepsen, P.; Ptersen, D.; Caridad, J.; Jessen, B.; Booth, T.; Begglid, P. Graphene mobility mapping. Sci. Rep. 2015, 5, 12305. [CrossRef][PubMed] 4. Banszerus, L.; Schmitz, M.; Engels, S.; Dauber, J.; Oellers, M.; Haupt, F.; Watanabe, K.; Taniguchi, T.; Beschoten, B.; Stampfer, C. Ultrahigh-mobility graphene devices from chemical vapor deposition on reusable copper. Sci. Adv. 2015, 1, e1500222. [CrossRef] [PubMed] 5. Ren, W.; Cheng, H. The global growth of graphene. Nat. Nanotechnol. 2014, 9, 726–730. [CrossRef][PubMed] 6. Bandurin, D.; Gayduchencko, I.; Cao, Y.; Moskotin, M.; Principi, A.; Grigorieva, I.; Goltsman, G.; Fedorov, G.; Svintsov, D. Dual Origin of Room Temperature Sub-Terahertz Photoresponse in Graphene Field Effect Transistors. Appl. Phys. Lett. 2018, 112, 141101. [CrossRef] 7. Vicarelli, L.; Vitiello, M.; Coquillat, D.; Lombardo, A.; Derrari, A.; Knap, W.; Polini, M.; Pellegrini, V.; Tredicucci, A. Graphene field-effect transistors as room-temperature terahertz detectors. Nat. Mater. 2012, 11, 865–871. [CrossRef][PubMed] 8. Feng, W.; Zhang, R.; Cao, J. Progress of terahertz devices based on graphene. Acta Phys. Sin. 2015, 64, 229501. 9. Docherty, C.; Johnston, M. Terahertz Properties of Graphene. J. Infrared Millim. Terahertz Waves 2012, 33, 797–815. [CrossRef] 10. Hesler, J.; Prasankumar, R.; Tignon, J. Advances in terahertz solid-state physics and devices. J. Appl. Phys. 2019, 126, 110401. [CrossRef] 11. Spirito, D.; Coquillat, D.; Bonis, S.; Lombardo, A.; Bruna, M.; Ferrari, A.; Pellgegrini, V.; Tredicucci, A.; Knap, W.; Viteillo, M. High performance bilayer-graphene Terahertz detectors. Appl. Phys. Lett. 2014, 104, 97–105. [CrossRef] 12. Otsuji, T.; Tombet, S.; Satou, A.; Fukidome, H.; Suemitsu, M.; Sano, E.; Popov, V.; Ryzhii, M.; Ryzhii, V. Graphene materials and devices in terahertz science and technology. MRS Bull. 2012, 37, 1235–1243. [CrossRef] 13. Qin, H.; Sun, J.; Liang, S.; Li, X.; Yang, X.; He, Z.; Yu, C.; Feng, Z. Room-temperature, low-impedance and high-sensitivity terahertz direct detector based on bilayer graphene field-effect transistor. Carbon 2017, 116, 760–765. [CrossRef] 14. Yuan, W.; Wen, Z.; Li, M.; Chen, L.; Chen, G.; Ruan, D.; Gao, Y. Energy gap of novel edge-defected graphene nanoribbons. Jpn. J. Appl. Phys. 2016, 55, 085101. [CrossRef] 15. Li, X.; Cai, W.; An, J.; Kim, S.; Nah, J.; Yang, D.; Piner, R.; Velamakanni, A.; Jung, I.; Tutuc, E.; et al. Large-area synthesis of high-quality and uniform graphene films on copper foils. Science 2009, 324, 1312. [CrossRef][PubMed] 16. Li, X.; Cai, W.; Colombo, L.; Ruoff, R. Evolution of graphene growth on Cu and Ni studied by carbon isotope labeling. Nano Lett. 2009, 9, 4268–4272. [CrossRef][PubMed] 17. Yuan, W.; Li, M.; Wen, Z.; Sun, Y.; Ruan, D.; Zhang, Z.; Chen, G.; Gao, Y. The Fabrication of Large-Area, Uniform Graphene Nanomeshes for High-Speed, Room-Temperature Direct Terahertz Detection. Nanoscale Res. Lett. 2018, 13, 190. [CrossRef] [PubMed] 18. Zhu, W.; Neumayer, D.; Perebeinos, V.; Avouris, P. Silicon nitride gate dielectrics and bandgap engineering in graphene layers. Nano Lett. 2010, 10, 3572–3576. [CrossRef][PubMed]