# Fileset

[NatComm_Proton.pdf](https://mdr.nims.go.jp/filesets/42d72bd2-3a6d-4e59-bb35-f6b8e1935512/download)

## Creator

Yu Ji, Guang-Ping Hao, Yong-Tao Tan, Wenqi Xiong, Yu Liu, Wenzhe Zhou, Dai-Ming Tang, [Renzhi Ma](https://orcid.org/0000-0001-7126-2006), Shengjun Yuan, [Takayoshi Sasaki](https://orcid.org/0000-0002-2872-0427), Marcelo Lozada-Hidalgo, Andre K. Geim, Pengzhan Sun

## Rights

[Creative Commons BY Attribution 4.0 International](https://creativecommons.org/licenses/by/4.0/)

## Other metadata

[High proton conductivity through angstrom-porous titania](https://mdr.nims.go.jp/datasets/b049af21-d3e2-421c-831a-338c3a00ac3c)

## Fulltext

High proton conductivity through angstrom-porous titaniaArticle https://doi.org/10.1038/s41467-024-54544-zHighprotonconductivity throughangstrom-porous titaniaYu Ji1, Guang-Ping Hao 2 , Yong-Tao Tan3,4, Wenqi Xiong5,6, Yu Liu1,Wenzhe Zhou1, Dai-Ming Tang 7, Renzhi Ma7, Shengjun Yuan 6,Takayoshi Sasaki 7, Marcelo Lozada-Hidalgo 3,4 , Andre K. Geim 3,4 &Pengzhan Sun 1Two dimensional (2D) crystals have attracted strong interest as a new class ofproton-conducting materials that can block atoms, molecules and ions whileallowing proton transport through the atomically thin basal planes. Although2Dmaterials exhibit this perfect selectivity, the reportedproton conductivitieshave been relatively low. Here we show that vacancy-rich titania monolayersare highly permeable to protons while remaining impermeable to helium withproton conductivity exceeding 100 S cm−2 at 200 °C and surpassing targets setby industry roadmaps. The fast and selective proton transport is attributed toan extremely high density of titanium-atom vacancies (one per square nm),which effectively turns titania monolayers into angstrom-scale sieves. Ourfindings highlight the potential of 2D oxides as membrane materials forhydrogen-based technologies.Proton-permeable two dimensional (2D) crystals, such as grapheneand hexagonal boron nitride (hBN), display high transparency tothermal protons while retaining complete impermeability to all ionsand gases1–5. Their excellent selectivity turns them into attractiveproton-conducting membrane materials6–9 for hydrogen-basedtechnologies. However, those applications require membranes withvery high proton (areal) conductivity10, typically exceeding11 5 S cm−2,which stimulates research into new 2D proton-conducting materialswith higher conductivities than that of graphene and hBN12,13. Thiscould be achieved by engineering atomic scale defects13,14, nanoscalecorrugations and strain5,15 in the existing 2D materials or by growingdesigner 2D crystals with intrinsic angstrom-scale pores16–19 (e.g.,various graphynes). However, it is challenging to control pores’shape, sizes and other characteristics via these routes, which wouldallow both high permeability and selectivity. Alternatively,increasing the operation temperature T could in principle lead to anexponential increase in conductivity because proton transporttypically involves a finite energy barrier E1,2. Besides higher con-ductivity, materials that can operate at elevated temperatures (forexample, 200—500 °C) are highly sought after because many che-mical engineering and energy conversion applications are moreefficient at such T. However, the above temperature range—com-monly referred to as the proton materials gap10,12— remains challen-ging for both 2D and 3D materials. In addition to their potential inhydrogen-based technologies, proton-permeable 2D materials arealso of interest for the use as atomically thin barrier layers for pro-tection and control in catalytic and electrochemical processes20–22.In this work, we explore proton transport through titaniamonolayers23–25. Thematerial consists of a 2D array of TiO6 octahedra(Fig. 1a) and inherits its 3D parent’s stability in aqueous, oxidizingand reducing environments at elevated T. We find an unexpectedlyhigh proton permeability of 2D titania, including at temperaturesabove 200 °C, and attribute this to an extremely high density ofangstrom-scale vacancies.Received: 14 June 2024Accepted: 13 November 2024Check for updates1Institute of Applied Physics and Materials Engineering, University of Macau, Macau, China. 2State Key Laboratory of Fine Chemicals, School of ChemicalEngineering, Dalian University of Technology, Dalian, Liaoning, China. 3Department of Physics and Astronomy, University of Manchester, Manchester, UK.4National Graphene Institute, University of Manchester, Manchester, UK. 5Institute of Quantum Materials and Physics, Henan Academy of Sciences,Zhengzhou, China. 6Key Laboratory of Artificial Micro- and Nano-Structures of Ministry of Education, School of Physics and Technology, Wuhan University,Wuhan, China. 7Research Center for Materials Nanoarchitectonics, National Institute for Materials Science, Ibaraki, Japan.e-mail: guangpinghao@dlut.edu.cn; marcelo.lozadahidalgo@manchester.ac.uk; geim@manchester.ac.uk; pengzhansun@um.edu.moNature Communications |        (2024) 15:10546 11234567890():,;1234567890():,;http://orcid.org/0000-0001-5849-9965http://orcid.org/0000-0001-5849-9965http://orcid.org/0000-0001-5849-9965http://orcid.org/0000-0001-5849-9965http://orcid.org/0000-0001-5849-9965http://orcid.org/0000-0001-7136-7481http://orcid.org/0000-0001-7136-7481http://orcid.org/0000-0001-7136-7481http://orcid.org/0000-0001-7136-7481http://orcid.org/0000-0001-7136-7481http://orcid.org/0000-0001-6208-1502http://orcid.org/0000-0001-6208-1502http://orcid.org/0000-0001-6208-1502http://orcid.org/0000-0001-6208-1502http://orcid.org/0000-0001-6208-1502http://orcid.org/0000-0002-2872-0427http://orcid.org/0000-0002-2872-0427http://orcid.org/0000-0002-2872-0427http://orcid.org/0000-0002-2872-0427http://orcid.org/0000-0002-2872-0427http://orcid.org/0000-0003-3216-7537http://orcid.org/0000-0003-3216-7537http://orcid.org/0000-0003-3216-7537http://orcid.org/0000-0003-3216-7537http://orcid.org/0000-0003-3216-7537http://orcid.org/0000-0003-2861-8331http://orcid.org/0000-0003-2861-8331http://orcid.org/0000-0003-2861-8331http://orcid.org/0000-0003-2861-8331http://orcid.org/0000-0003-2861-8331http://orcid.org/0000-0002-5838-0945http://orcid.org/0000-0002-5838-0945http://orcid.org/0000-0002-5838-0945http://orcid.org/0000-0002-5838-0945http://orcid.org/0000-0002-5838-0945http://crossmark.crossref.org/dialog/?doi=10.1038/s41467-024-54544-z&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-024-54544-z&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-024-54544-z&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-024-54544-z&domain=pdfmailto:guangpinghao@dlut.edu.cnmailto:marcelo.lozadahidalgo@manchester.ac.ukmailto:geim@manchester.ac.ukmailto:pengzhansun@um.edu.mowww.nature.com/naturecommunicationsResultsDevice fabrication and characterizationThe monolayer titania crystals used in this work were prepared bydelamination of layered bulk compound K0.8[Ti1.73Li0.27]O4 via ionexchange, following the recipe described previously23–25. In brief, thebulk compound consists of titania monolayers with some Ti atomssubstituted by Li and the space between the layers is filled with K+ions, which balance the layers’ negative charge. During an ionexchange process, both Li+ and K+ ions were substituted by protons.Then, in the second ion exchange process, the interlayer protonswere replaced by large cations (C4H9)4N+ which led to the crystals’delamination and yielded titania monolayers with an effectivethickness of ~1.1 nm (Fig. 1b), in agreement with the previousreports24,25. Typical lateral dimensions of 2D titania were a few µmbutsome flakes could reach tens of µm (Fig. 1b and Fig. S1). We searchedfor the largest monolayers (Fig. S1) and transferred them overapertures of 2-3 µm in diameter etched in silicon-nitride membranes(inset of Fig. 1b and Fig. S2, Supplementary Information), as descri-bed previously1,2. The resulting freestanding titania monolayers werefirst examined using atomic force microscopy (AFM), and samplesshowing cracks, tears or folds were discarded. We then characterizedthe remaining monolayers under high-resolution transmissionelectron microscopy (HRTEM). In the HRTEM images (Fig. 1c andFig. S3), titanium atoms appear as dark spots within an orthorhombiclattice (see the schematic in Fig. 1a). Some of those dark spots weremissing, resulting in rectangles with blurred centers (Fig. 1c).According to the previous report26, this structural feature corre-sponds to a Ti-atom vacancy. 2D titania has a unit cell of 0.3 nm×0.38 nm in size (Fig. 1a, left), according to the X-ray diffractionanalysis23, and the space available for proton permeation through aTi-atom vacancy is only a fraction of the empty space in the ball-and-stick model shown in Fig. 1a because of the dense electronic cloudssurrounding the atomic nuclei (Fig. S5 in Supplementary Informa-tion). The frequency of these vacancies estimated from the elementalanalysis of the material25 was 13.5%. Our HRTEM images, obtainedfrom a combined area of a few hundred nm2 (Fig. S3), yielded asomewhat lower occurrence of ~7.5%, which translates into about onevacancy per nm2, or ~1014cm−2. This discrepancy between chemicaland TEM analyzes is the same as in the previous work26. We attributeit to the fact that our monolayers were selected for their large sizeand high quality, whereas the elemental analysis was done for mac-roscopic samples that probably contained flakes with a higher con-centration of vacancies. The latter flakes are expected to break moreeasily and did not survive our selection.~1.1 nmVacancy0.3 nm0.38 nmVacancyTi Ti Ti TiacbFig. 1 | Studied monolayer titania. a Schematics of single-Ti-atom vacancies inmonolayer titania. Left: top view with the shown lattice parameters; right: tilted 3Dview. The blue and red balls denote Ti and O atoms, respectively. The green ballshighlight vacancy positions. b AFM of a titania sheet placed on an oxidized siliconwafer. Scale bar, 10 µm. Yellow profile, height-trace along the yellow solid lineindicating the titania thickness. Inset: same crystal after being transferred over a3μm diameter aperture etched in a silicon-nitride substrate. c HRTEM image of atitania monolayer. Scale bar, 1 nm. Dark spots are Ti atoms. They form an orthor-hombic lattice that is often interrupted bybrighter blurred rectangles, twoofwhichare indicated by circles.Article https://doi.org/10.1038/s41467-024-54544-zNature Communications |        (2024) 15:10546 2www.nature.com/naturecommunicationsTo ensure the absence of nanoscale pinholes (occasionallyobserved in HRTEM) that could have been missed under the AFMcharacterization, all our suspended titaniamonolayer deviceswereHe-leak tested. The test provided a sensitivity down to ~108 atoms s−1(Supplementary Information, Fig. S4), which would be sufficient todiscern gas flows through an individual pore of 1 nm in diameter. Inthese measurements, one side of the membrane was exposed tohelium gas using a maximum pressure of 1 bar. The other side faced avacuum chamber connected to a He-leak detector. Only two deviceswere found to exhibit gas flow rates of the order of 1013 atoms s−1 at the1 bar feed pressure (Fig. S4). Their retrospective examination under ascanning electron microscope revealed a single pinhole of ~50 nm insize (Fig. S4), consistent with the observed Knudsen flow. The leakydevices were excluded from further measurements. All the othermembrane devices (20 in total) exhibited no helium leakage, showingthat they did not contain even a single one-nm pinhole. The heliumtests also demonstrated that numerous vacancies observed in 2Dtitania by HRTEM are practically impermeable to gases.Proton transportSome of the devices impermeable to helium were then tested forprotonpermeation. To that end, both sidesof the suspended2D titaniawere coated with a proton-conducting polymer (Nafion) and elec-trically connected to proton-injecting electrodes (Pt on carbon),as reported previously1,2 (inset of Fig. 2a). Typical I–V characteristicsfor 2D titania at small biases V≲ 100mV are shown in Fig. 2a. Thecurrent I increased linearly with V, which allowed us to extract theareal conductivity σ. Analysis of several titania devices yieldedσ = 2.0 ± 0.8 S cm−2 (Fig. 2b). This proton conductivity is over 100 timeshigher than that ofmonolayer graphene, andmore than 10 times largerthan for monolayer hBN (Fig. 2a, b). Note that despite the large con-ductivity, the devices’ resistance was still ~2 orders of magnitudehigher than that of our reference devices with a bare aperture (nocrystal). This confirmed that the measured σ was intrinsic to titaniamonolayers and the series resistance from Nafion was negligible. Togain further insight, we measured how the areal conductivity evolvedwith temperature. Figure 2c shows that σ increased with T, followingroughly the Arrhenius behavior, σ∝ exp(-E/kBT). The fitting yields theactivation energy E =0.34 ±0.06 eV. We attribute the observed highconductivity to the high density of Ti vacancies identified underHRTEM, a conclusion supported by our theoretical simulations(Fig. S5, Supplementary Information).The temperaturedependence inFig. 2c suggests thatmuchhigherσ can be achieved for T inside the proton materials gap10. However,Nafion can be used as an electrical contact to titania only over a limitedT range because of its dehydration at higher temperatures11. For thisreason and because of a mechanical strain inflicted on suspended 2Dcrystals within heated Nafion, we had to limit our T to ~60 °C to avoidtheir damage, as reported previously1. This constraint is, however, notfundamental. Our suspended devices without Nafion could sustain Tup to ~260 °C (at higher T they cracked, presumably because of dif-ferent thermal expansion with respect to the silicon-nitride substrate).Furthermore, ourX-rayphotoelectron spectroscopy andAFManalyzesrevealed that 2D titania retained its crystallographic and chemicalstructure after beingexposed for several hours to 300 °C invarious gasatmospheres including argon, air, and hydrogen (Fig. S6).To measure the proton conductivity at temperatures higher thanthat allowed by Nafion, we coated the suspended 2D titania on bothsides with porous Pt films (~10 nm thick) and placed the devices into achamber containing humid hydrogen atmosphere (upper inset ofFig. 3). In this configuration, Pt absorbs H2 gas and provides protonsfor transport through titania12. Our measurements using the lattersetup are shown in Fig. 3. At room T, σwas ~3 times lower than that forthe Nafion-based devices. We attribute this to either some vacanciesbeing blocked by Pt atoms or a lower proton density on the titaniasurface covered with Pt (Fig. S7, Supplementary Information). Fromthe measurements at higher T, we extracted the activation energyE ≈0.36 eV (Fig. S7, Supplementary Information). Within our accuracy,this is the same E as observed using Nafion devices (Fig. 3 and S7),suggesting the same mechanism governing proton transport in bothsetups.The areal conductivity reached 100 S cm−2 at 200 °C (Fig. 3) and200 S cm−2 at 260 °C (Fig. S7). This is an order of magnitude higherthan for the industry standard, Nafion 117, that is, 200 µm thick Nafionfilms measured at 80 °C (the finite thickness is essential to minimizewater and hydrogen permeation)11 and, also, surpasses the USDepartment of Energy target (50S cm−2) for proton-conductingmaterials in hydrogen and fuel cell technologies27.We verified that the observed high σ did not involve electrontunneling through titania monolayers. To that end, we measured thesame devices in vacuum (Fig. 3). At room T, no current could be-0.1 0.10-0.2-0.10.10.20hBNmcA(ytisnedtnerruC-2)Bias (V)TitaniaGraphene3.0 3.1 3.2 3.3 3.41510201000/T (K-1)mcS(ytivitc udn ocl ae rA-2)55 45 35 25T (oC)10-310-210-1110hBNmcS(ytivitcudnoclaerA-2)Titania Graphenea bH+NafionProton-injecting electrodeProton-injecting electrodeNafionA-0.1 0.10-10161 oC43 oCI (A cm-2)V (V)23 oCFig. 2 | Proton transport through 2D titania. a Examples of I–V characteristics fortitania, hBN, and graphene monolayers (color coded). Inset, schematic of themeasurement setup usingNafionas the conductingmedia. Dashedblack linemarkszero voltage.b Proton areal conductivity of titania devices is comparedwith that ofgraphene andhBNmonolayersmeasured using the same setup. For 2D titania, eachbar represents a different device. The solid horizontal line marks the averageconductivity for all 8 devices with the shaded area indicating SD. For hBN andgraphene, the error bars show the average conductivity and SD found from mea-surements using at least 3 devices. The graphene and hBN data are in quantitativeagreement with the previous reports1,2,5. c Temperature dependence for one of ourtitania devices. Symbols: experimental data. Solid line: best exponential fit, yieldingE =0.34 ± 0.06 eV. Inset, examples of I–V curves fromwhich σ in themain panelwasextracted (same color coding as in the main figure).Article https://doi.org/10.1038/s41467-024-54544-zNature Communications |        (2024) 15:10546 3www.nature.com/naturecommunicationsdetected within our accuracy of ~10 pA. At higher T, the backgroundcurrent started to increase, but σ was still three orders of magnitudelower than that in the hydrogen atmosphere for all T. This unam-biguously corroborates that the high σ observed in hydrogen was dueto proton conduction and electron tunneling provided a negligiblecontribution to the overall conductivity, consistent with titania’s largebandgap (3.8 eV)28. Note also that the slope of the T dependence invacuum (Fig. 3) was somewhat close to that in the hydrogen atmo-sphere. This is perhaps unsurprising as not-ultrahigh vacuum systemsinevitably contain remnant water adsorbed on surfaces whereas, asshown below, water on 2D titania can serve as a source of protons.Ion selectivityThe above experiments show thatmonolayer titania blocks helium butis highly permeable to thermal protons. In principle, this leaves achance that small ions like Li+ can also permeate through 2D titania. Toassess the latter possibility, we used another experimental setup inwhich the titania devices separated two reservoirs filled with liquidelectrolytes (Fig. 4a, top inset). As a reference, we first filled bothreservoirs with HCl solutions and measured the membranes’ arealconductivity σ using Ag/AgCl electrodes3. The conductivity extractedfrom the linear I–V responsewas ~1.8 S cm−2 for 0.1MHCl (Fig. 4a). Thisagrees with σ measured for our Nafion-coated devices in Fig. 2, inwhich Nafion provided a similar proton concentration1. In contrast, ifwe used 0.1M solutions of KCl or LiCl, σ was ~180 times smaller. Thisclearly indicates that monolayer titania exhibits high selectivitybetween protons (H+) and other small cations.We corroborated the high proton selectivity using drift-diffusionmeasurements (inset of Fig. 4b) which provided information aboutrelative contributions of different ions into the total conductance3. Tothis end, one of the reservoirs was filled with HCl at a relatively high-concentration (Ch = 1M) and the other one at Cl = 0.1M, which pro-vided the concentration gradient ΔC =Ch/Cl = 10. The measured I–Vcharacteristics using this setup included the well-known contributiondue to redox reactions at electrodes3, which was subtracted from themeasure voltage, allowing us to extract the membrane potential Vm(Fig. S8). Figure 4b shows an example of typical I–Vm characteristicsfound in our drift-diffusion experiments. At zero Vm, the current waspositive. The concentration gradient drives both H+ and Cl− from highto low concentration reservoirs, and the positive current unambigu-ously shows that protons contributed most. The potential drop Vm*,that is required to stop the diffusion current across the membrane, isgiven by3 Vm* = −(tH − tCl) kBT/e ln(ΔC) where tH, tCl are the transportnumbers for protons and Cl− (both numbers are positive andtH + tCl ≡ 1). Our measurements yielded Vm* ≈ −59mV (Fig. 4b). Thistranslates into tH ≈ 1, that is, practically all the current was carried byprotons and the Cl− contributionwas small. Our accuracy inmeasuringVm* was ~1mV as found from several replicated measurements. Weperformed similar drift-diffusion experiments using LiCl and KClsolutions and, as expected, the observed Vm* was close to zero withinthe same accuracy.This left the question of where the small but clearly discernableareal conductivity σ0 observed for KCl and LiCl solutions came from(Fig. 4a). Indeed, σ0 was close to 10−2 S cm−2, nearly an order of mag-nitude larger than our detection limit, whereas the finite accuracy ofthe drift-diffusion experiments might still allow minute flows of ionsthrough 2D titania membranes. To address the above question, weperformed additional experiments and found that σ0 did not dependonKCl and LiCl concentrations (Fig. 4c) and only slightly changed if weutilized other chlorine solutions (MgCl2, CaCl2 and Ru(bipy)3Cl2;Fig. 4d). Moreover, the same areal conductivity σ0 was found fordeionized water (Fig. 4c). This shows that the observed σ0 cannot beattributed to ions. Their translocation through single-atom vacanciesshould be blocked because of the relatively large diameter of hydratedions. However, we also cannot rule out a role of hydrocarbon con-tamination that is practically unavoidable for surfaces prepared in air,not under ultra-high vacuum conditions. Hydrocarbon moleculeswould then be expected to reduce the space available for ion passagebut to be less detrimental for proton transport. In either case, theobserved σ0 can be attributed to residual protons that are alwayspresent in water. Indeed, although the bulk conductivity of deionizedwater is insufficient to account for the observed value of σ0, note thatthe titania surface is well known for its water dissociative properties29and our titania monolayers carried a large negative charge (Fig. S9,Supplementary Information). Accordingly, this should have resulted ina high density of protons adsorbed on titaniamembranes, which couldthen diffuse along the surface and transfer through vacancies, givingrise to the finite σ0 observed for all aqueous solutions. Further work isrequired to understand the reason for the finite conductivity observedfor salt solutions.To providemore information about the proton transport throughmonolayer titania, we studied the isotope effect. To this end, deuter-ium chloride (DCl) dissolved in heavy water (D2O) was used andcompared with HCl in H2O for the same range of concentrations(Fig. 4c). We observed qualitatively the same dependence of 2D tita-nia’s areal conductivity as a function of H+ and D+ concentrations.However, σ for deuterons D+ was lower than that for protons H+ by afactor of 1.6 ± 0.16 (using the linear fits for the high-concentrationregime in Fig. 4c). The isotope effect clearly corroborates that theobserved conductance was indeed due to protons. Our theoreticalanalysis for the observed E and the D+/H+ separation factor suggeststhat protons first attach to broken bonds of Ti-atom vacancies andthen translocate through the 2D crystal. The broken bonds at thevacancy edges make the titania membrane highly negatively charged,as evidenced by the zeta potential measurements (Fig. S9), whichconsequently attracts a high density of protons. When a voltage bias is2.0 2.4 2.8 3.210-310-210-1102110In vacuum1000/T (K-1)mcS(ytivitcudnoclaerA-2)In H2200 140 80 20T (oC)-0.1 0.10-110In H2138 oCI (A cm-2)V (V)39 oCH2H2H+APtFig. 3 | High-temperature proton transport using Pt-coated devices. Arrheniusplots for the device measured using porous Pt electrodes in H2 gas (1 bar) andvacuum (~10−2mbar), respectively (color coded). Standards deviations (SD) usingdifferent devices (not shown) increasedwith increasingTbut didnot exceed 50%ofthe measured values for all temperatures. Solid red line, best exponential fit,yielding E ≈0.36 eV. The gray shaded area marks our lower detection limit. Upperinset, schematic of the experimental setup; lower inset, examples of I–V curves atdifferent T (color coded).Article https://doi.org/10.1038/s41467-024-54544-zNature Communications |        (2024) 15:10546 4www.nature.com/naturecommunicationsapplied across the membrane, protons from the surrounding media(electrolytes or Pt films) are injected into the crystal. Our calculations(Fig. S5) show that protons can then hop between oxygen atoms alongthe vacancy, leading to a proton current. This resembles proton hop-ping along water chains in the Grotthuss mechanism known for bulkwater, except that in our case water molecules are replaced withoxygen bonds in titania pores (Supplementary Information providescomparison of the inferred processwith those in known biological andsolid-state 1D channels).OutlookOur experiments show that protons can permeate throughmonolayertitania crystals whereas helium atoms are excluded. At roomtemperature, the observed areal conductivity of protons inmonolayertitania is orders of magnitude higher than that of graphene andhBN monolayers. The titania conductivity exceeds 100 S cm−2at 200 °C, making it an attractive proton-conductive material withinthe infamous proton materials gap10. In principle, titania monolayerscan be prepared via scalable routes involving soft-chemistryprocedures23–25 and assembled over large areas to form quality mem-branes via techniques such as layer-by-layer electrostatic assembly andLangmuir–Blodgett deposition30,31. Furthermore, the density ofmonovacancies in 2D titania canbe changed if required from ~9%up to~18% using different compositions of the original bulk compound usedfor expoliation32,33. Not only titania but also other 2D oxides canpotentially be used as membranes, separators and protective coatings-0.1 0.10-0.50.51.00HClMembrane potential (V)mcA(ytisnedtn err uC-2)-59 mV-0.1 0.10-0.10.10Bias (V)mcA(ytisnedtnerruC-2)HClLiClKClHClH+AHClAg/AgCl Ag/AgCla bc d-0.05 0.050-0.0010.001010-6 10-4 10-2 110-310-210-1110mcS(ytivitcudnoclaerA-2)LiCl KClH2ODCl in D2OConcentration of H(D)Cl (M)HCl in H2O10-310-210-1110[Ru(bipy) 3]Cl 2CaCl 2MgCl 2KClLiClHClmcS(ytivitcudnoclaerA-2)1 M0.1 MAFig. 4 | Ion selectivity measurements. a Examples of I–V characteristics for HCl,KCl and LiCl (color coded). Bottom inset, zoom-in. Top inset, schematic of theexperimental setup. b Example of I–V characteristics in the drift-diffusion experi-ments usingHCl in concentrations0.1 and 1Mas illustrated in thebottom inset. Theblue arrow indicates the membrane potential of ~59mV that corresponds to theperfect proton selectivity with respect to Cl ions. c Concentration dependencesusing HCl dissolved in H2O and DCl in D2O (color coded). Symbols, data taken in 5different measurements with the error bars indicating SD (shown if larger than thesymbols). Solid lines, best linear fits at high concentrations. Also shown are σ forLiCl and KCl solutions at different concentrations and for deionized water (colorcoded). d Conductivities for various 0.1M salt solutions (color coded; SDwere lessthan 40% of all the solutions). The gray areas in (c, d) mark our detection limitbecause of leakage currents. Blue areas, σ measured using deionized water.Article https://doi.org/10.1038/s41467-024-54544-zNature Communications |        (2024) 15:10546 5www.nature.com/naturecommunicationsin renewable energy applications such as fuel cells, electrolyzers andcatalytic systemswhere rapid proton transport combinedwith gas andion impermeability is essential.MethodsDevice fabricationWe followed the well-established soft-chemistry procedures23–25 to pre-pare monolayer titania crystals. In brief, layered titanate compoundsK0.8[Ti1.73Li0.27]O4 were obtained by mixing potassium carbonate(K2CO3, Sigma–Aldrich), lithium carbonate (Li2CO3, Sigma–Aldrich) andtitanium dioxide (TiO2, rutile form) according to a molar ratio2.4:0.8:10.4, followed by decarbonating at 800 °C for 0.5 h and a furthercalcination at 1100 °C for 20h. The products were stirred vigorously in1M of HCl solution for a few days so that the interlayer potassium ionsand intralayer lithium ions were fully extracted and exchangedfor protons, resulting in protonic compounds H1.07Ti1.73O4 as deter-mined by chemical analysis. To delaminate for monolayers, the materialwas dispersed in a tetrabutylammonium hydroxide [(C4H9)4NOH]aqueous solution and mildly shaken for 10 days, resulting in 2Dtitania Ti0.87O20.52−.The 2Dcrystalswere castedonto a freshly cleanedoxidized siliconwafer and then checkedunder anopticalmicroscope. FigureS1a showsthat the typical lateral dimensionof the crystalswas a few µmand somereached a few tens of µm. These large flakes were carefully examinedunder dark field and differential interference contrast (DIC) modes(Fig. S1b, c) to ensure they were in high quality. Only those free fromany contaminations, wrinkles, cracks and other imperfections werechosen for device fabrication. They were transferred over an aperture2-3 µm in diameter that was microfabricated in a silicon-nitride (SiNx)chip (500nm thick) using the technique standard for van der Waalsassembly34,35. Details for making the SiNx microapertures were well-documented previously1,2 and are schematically illustrated in Fig. S2.To make the Nafion-coated device for proton transport mea-surements, we carefully drop-casted Nafion (Sigma–Aldrich, 5wt%1100EW) solution on both sides of the fabricated device (step 5 ofFig. S2), followed by electrically connecting to a pair of proton-injecting electrodes (Pt on carbon) (inset of Fig. 2a). The assembly wasbaked at 130 °C under 100% relative humidity to crosslink the Nafionmonomers so that the resulting polymer layers were highly proton-conductive but remained electron-insulating. For measurements atelevated temperatures T, instead of casting Nafion, we sputteredporous Pt films (a few to tens of nm thick) on both sides of the devicefollowing the procedures developed previously12 (step 6, Fig. S2).These Pt films served as both electrodes and proton reservoirs in ahumid hydrogen atmosphere.HRTEM imagingThe exfoliated titania monolayers were characterized using a trans-mission electron microscope (JEOL JEM-ARM200F) which was equip-ped with an image corrector of corrected electron optical systems(CEOS). To remove hydrocarbon and other contaminations on thecrystals’ surface, they were first UV-treated for 2 h and then baked foranother 2 h at 150 °C. The acceleration voltagewas set to 80 kV and thecurrent density was ~1.7 pA/cm2. To image the atomic structure, amaximum magnification of 800,000 times was used. Figure 1c showsone of our obtained HRTEM images and a larger view is provided inFig. S3. They were captured using a Gatan OneView camera under anexposure time of 6.5 s with drift-correction.Helium leak testsIn addition to the extensive AFM imaging of the transferred titaniamonolayers, we also performed helium leak tests for all our devices tocheck for any nm-scale pinholes and other imperfections. In thosemeasurements, the tested device separated two vacuum chambers(inset of Fig. S4). One of them (feed chamber) was connected to ahelium-gas reservoir. The injection of helium gas was electricallycontrolled by a dosing valve and its pressure was recorded by a pres-sure gauge. The other (permeate chamber) was connected to a leakdetector (Leybold L300i). Its sensitivity limit with respect to heliumflows was of the order of ~108 atoms s−1, as determined by controlmeasurements using a piece of bare silicon wafer. This sensitivityallows discerning Knudsen flows through a single pinhole down to1 nm in size. Prior to real tests, the setup was sealed and each chamberwas leak-checked to ensure that the only possible gas pathwaybetween the two chambers was through the device.Electrical measurements for the transport of protons and ionsTomeasure the transport of protons through 2D titania, Nafion-coateddevices were placed inside a chamber filled with 1 bar of H2 at 100%relative humidity. The I–V characteristics were measured using thesource meter Keithley 2636B and collected using software LabVIEW.Normally, we limited the applied voltage to ≲100mV to ensure linearresponse and at a sweep rate of 5mV s−1. The measurement tempera-tureTwas limited to 60 °C to avoid dehydration in theNafionfilms.Wenote that according to the previous reports32,33, the porosity of 2Dtitania can be tuned from ~9% to ~18% if required by using layeredprecursors of different compositions and in principle, we wouldexpect a porosity dependence of the areal conductivity using sampleshaving different porosities. However, our experimental sensitivity inproton transport experiments (2.0 ± 0.8 S cm−2, Fig. 2b) was about40%, which is greater than the described porosity range. Such varia-tions are therefore not expected to result in ameasurable difference inthe areal conductivity of protons. For higher Tmeasurements such asthose in Fig. 3 and S7, Pt-coated devices were used instead and all theother conditions remained the same.To find out whether or not ions could permeate through thetitania monolayers, we employed a customized setup (inset of Fig. 4a)which consisted of two reservoirs separated by the fabricated device.Prior to measurements using different salt solutions, the reservoirswere first flushed with an isopropanol/water mixture (1:1 in volume)and then deionized water to ensure proper wetting of themembrane’ssurface. The following solutions were tested: HCl, LiCl, KCl, MgCl2,CaCl2, and [Ru(bipy)3]Cl2. Each of them was carefully introduced intothe two reservoirs simultaneously and Ag/AgCl electrodes wereinserted for electricalmeasurements. All these experiments were doneat room T (297 ± 3 K).Density functional theory calculationsTo provide theoretical insights for the observed proton transport, weperformeddensity functional theory (DFT) calculations using theVASPpackage36. The exchange-correlation potential and ion-electron inter-actions were described using the generalized gradient approximation(GGA) and projected augmented wave (PAW) methods37,38. A kineticenergy cutoff of 500 eV was employed. The van derWaals interactionswere addressed by the semi-empirical DFT-D2 method39,40. All atomswere allowed to fully relax to the ground state by taking into accountthe spin-polarization. The relaxation resulted in the optimized latticeconstants of 3.77 Å and 3.03 Å, respectively, for the 2D titania crystal.Then a single-Ti-atom vacancy was created in the 2 × 3 supercell. Thisresulted in eight under-coordinated oxygen atomsbonded to the edgeof the vacancy: two of them coordinated to a single-Ti atom; anothertwo coordinated to two Ti atoms and the rest to three Ti atoms. As perthe previous study26, the vacancy model of removing one Ti atomtogether with two single-bondedO atomswell reproduced the HRTEMimaging results because those O atoms are most reactive and tend todesorb from the vacancy. This model was adopted in our simulations.Due to the other unsaturated O atoms, the vacancy was negativelycharged. This is in qualitative agreement with our zeta potentialmeasurements (Fig. S9, details see below), showing a large negativesurface charge under low proton concentrations. Comparing with theArticle https://doi.org/10.1038/s41467-024-54544-zNature Communications |        (2024) 15:10546 6www.nature.com/naturecommunicationslattice constants for an intact titania crystal, the vacancy modelexhibited slightly larger interatomic separations of 3.97 Å × 3.38Å.Nonetheless, the space available for proton permeation should besmaller, due to the electronic cloud surrounding the atomic nuclei, asshownby the electron density calculations (Fig. S5a) and in theHRTEMimage (Fig. 1c). To simulate the transport process of protons throughthe vacancy, we first put a proton at infinity and allowed it to movetoward the vacancy under the electrostatic attraction from the nega-tively charged surface. The pathway was fixed perpendicularly to thecrystal’s basal plane. After crossing through the vacancy, the protonwas forced to move away from the crystal by overcoming the latter’selectrostatic resistance. For comparison, we also simulated the sameproton transport process but through an intact titania lattice. Thetransition states were searched using the climbing-image nudgedelastic band (CINEB) method41,42.Thermal stabilityTo assess the thermal stability of 2D titania, we employed X-ray pho-toelectron spectroscopy (XPS) and AFM to characterize its chemicaland crystallographic structures after thermal cycling to a higher T (upto 300 °C). To this end, the material was deposited on an oxidizedsilicon wafer and then annealed at 300 °C for 3 h. Different gas atmo-spheres were used for the annealing: air, 1 bar of H2 and 1 bar of Ar,respectively. The treated material was carefully characterized usingXPS (ESCALAB 250Xi, Thermo Fisher Scientific) equipped with Al KαX-rays (hν = 1486.7 eV). To quantify the composition and relativeamount of Ti- and O-containing groups/bonds, the XPS Ti 2p and O 1 sspectra were analyzed and fitted by Gaussian–Lorentzian functions.Zeta potentialTo provide information about the surface charging state of 2D titania,we measured zeta potential (ζ) for an aqueous colloidal suspensioncontaining delaminated titania monolayers. Zeta potential is estab-lished in all solid-electrolyte systems and characterizes the potential atthe slipping surface outside the stationary Helmholtz layer. It isdetermined by the surface charge density ρs and the concentration Cof electrolyte. It is well known that ρs is also sensitive to the con-centration of protons (that is, solution pH) because the present pro-tons easily adsorb on the surface and tune its ρs. For this reason, weshould expect a strongdependenceof ζon the solutionpH. To seek forthis effect, we added HCl into the titania suspension to adjust its pH.This setup ensures the same chemical environment as in the ionselectivity measurements (Fig. 4) where HCl solutions of different Cwere also measured.Data availabilityAll data supporting the key findings of this study are available withinthe article and the Supplementary Information file. All raw data gen-erated during the current study are provided in the Source Datafile. Source data are provided with this paper.References1. Hu, S. et al. Proton transport through one-atom-thick crystals.Nature 516, 227–230 (2014).2. Lozada-Hidalgo, M. et al. Sieving hydrogen isotopes through two-dimensional crystals. Science 351, 68–70 (2016).3. Mogg, L. et al. Perfect proton selectivity in ion transport throughtwo-dimensional crystals. Nat. Commun. 10, 4243 (2019).4. Sun, P. Z. et al. Limits on gas impermeability of graphene. Nature579, 229–232 (2020).5. Wahab, O. J. et al. Proton transport through nanoscale corrugationsin two-dimensional crystals. Nature 620, 782–786 (2023).6. Kidambi, P. R., Chaturvedi, P. & Moehring, N. K. Subatomic speciestransport through atomically thin membranes: Present and futureapplications. Science 374, eabd7687 (2021).7. Bukola, S., Beard, K., Korzeniewski, C., Harris, J. M. & Creager, S. E.Single-layer graphene sandwiched between proton-exchangemembranes for selective proton transmission. ACS Appl. NanoMater. 2, 964–974 (2019).8. Zeng, Z. et al. Biomimetic N-doped graphenemembrane for protonexchange membranes. Nano Lett. 21, 4314–4319 (2021).9. Chaturvedi, P. et al. Deconstructing proton transport throughatomically thin monolayer CVD graphene membranes. J. Mater.Chem. A 10, 19797–19810 (2022).10. Norby, T. Solid-state protonic conductors: principles, properties,progress and prospects. Solid State Ionics 125, 1–11 (1999).11. Casciola, M., Alberti, G., Sganappa, M. & Narducci, R. On the decayof Nafion proton conductivity at high temperature and relativehumidity. J. Power Sources 162, 141–145 (2006).12. Mogg, L. et al. Atomically thin micas as proton-conducting mem-branes. Nat. Nanotechnol. 14, 962–966 (2019).13. Griffin, E. et al. Proton and Li-ion permeation throughgraphenewitheight-atom-ring defects. ACS Nano 14, 7280–7286 (2020).14. Sun, P. Z. et al. Exponentially selective molecular sieving throughangstrom pores. Nat. Commun. 12, 7170 (2021).15. Wu, Z. F. et al. Proton and molecular permeation through thebasal plane of monolayer graphene oxide. Nat. Commun. 14,7756 (2023).16. Qiu, H., Xue, M. M., Shen, C., Zhang, Z. H. & Guo, W. L. Graphynesfor water desalination and gas separation. Adv. Mater. 31, 1803772(2019).17. Gao, X., Liu, H. B., Wang, D. & Zhang, J. Graphdiyne: synthesis,properties, and applications. Chem. Soc. Rev. 48, 908–936 (2019).18. Neumann, C. et al. Bottom-up synthesis of graphene monolayerswith tunable crystallinity and porosity. ACS Nano 13, 7310–7322(2019).19. Toh, C. T. et al. Synthesis andproperties of free-standingmonolayeramorphous carbon. Nature 577, 199–203 (2020).20. Hu, K. et al. Catalytic activity of graphene-covered non-noblemetals governed by proton penetration in electrochemical hydro-gen evolution reaction. Nat. Commun. 12, 203 (2021).21. Kosmala, T. et al. Operando visualization of the hydrogen evolutionreaction with atomic-scale precision at different metal–grapheneinterfaces. Nat. Catal. 4, 850–859 (2021).22. Tong, J. et al. Control of proton transport and hydrogenation indouble-gated graphene. Nature 630, 619–624 (2024).23. Sasaki, T.,Watanabe,M., Hashizume,H., Yamada, H. &Nakazawa,H.Macromolecule-like aspects for a colloidal suspension of an exfo-liated titanate. Pairwise association of nanosheets and dynamicreassembling process initiated from it. J. Am. Chem. Soc. 118,8329–8335 (1996).24. Sasaki, T., Ebina, Y., Kitami, Y., Watanabe, M. & Oikawa, T. Two-dimensional diffraction of molecular nanosheet crystallites of tita-nium oxide. J. Phys. Chem. B 105, 6116–6121 (2001).25. Tanaka, T., Ebina, Y., Takada, K., Kurashima, K. & Sasaki, T. Oversizedtitania nanosheet crystallites derived from flux-grown layered tita-nate single crystals. Chem. Mater. 15, 3564–3568 (2003).26. Ohwada, M., Kimoto, K., Mizoguchi, T., Ebina, Y. & Sasaki, T. Atomicstructure of titania nanosheet with vacancies. Sci. Rep. 3, 2801(2013).27. USDepartment of Energy, HFTO.DOE Technical Targets for PolymerElectrolyte Membrane Fuel Cell Components, (EERE Publica-tions, 2015).28. Sakai, N., Ebina, Y., Takada, K. & Sasaki, T. Electronic band structure oftitania semiconductor nanosheets revealed by electrochemical andphotoelectrochemical studies. J. Am. Chem. Soc. 126, 5851–5858(2004).29. Sun,C.H., Liu, L.M., Selloni, A., Lu, G.Q. &Smith, S. C. Titania-waterinteractions: a review of theoretical studies. J. Mater. Chem. 20,10319–10334 (2010).Article https://doi.org/10.1038/s41467-024-54544-zNature Communications |        (2024) 15:10546 7www.nature.com/naturecommunications30. Osada, M. et al. High-κ dielectric nanofilms fabricated from titaniananosheets. Adv. Mater. 18, 1023–1027 (2006).31. Matsuba, K. et al. Neat monolayer tiling of molecularly thin two-dimensional materials in 1 min. Sci. Adv. 3, e1700414 (2017).32. Gao, T., Fjellvåg, H. & Norby, P. Protonic titanate derived fromCsxTi2−x/2Mgx/2O4 (x = 0.7) with lepidocrocite-type layered struc-ture. J. Mater. Chem. 19, 787–794 (2009).33. Wang, L. & Sasaki, T. Titanium oxide nanosheets: graphene analo-gues with versatile functionalities. Chem. Rev. 114, 9455–9486(2014).34. Geim, A. K. & Grigorieva, I. V. Van der Waals heterostructures.Nature 499, 419–425 (2013).35. Wang, L. et al. One-dimensional electrical contact to a two-dimensional material. Science 342, 614–617 (2013).36. Kresse, G. & Furthmüller, J. Efficiency of Ab-initio total energy cal-culations for metals and semiconductors using a plane-wave basisset. Comput. Mater. Sci. 6, 15 (1996).37. Blöchl, P. E. Projector augmented-wave method. Phys. Rev. B 50,17953 (1994).38. Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradientapproximation made simple. Phys. Rev. Lett. 77, 3865 (1996).39. Kerber, T., Sierka, M. & Sauer, J. Application of semiempirical long-range dispersion corrections to periodic systems in density func-tional theory. J. Comput. Chem. 29, 2088 (2008).40. Grimme, S. Semiempirical GGA-type density functional con-structedwith a long-range dispersion correction. J. Comput. Chem.27, 1787 (2006).41. Sheppard, D., Terrell, R. & Henkelman, G. Optimizationmethods forfinding minimum energy paths. J. Chem. Phys. 128, 134106 (2008).42. Sheppard, D., Xiao, P., Chemelewski, W., Johnson, D. D. & Henkel-man, G. A generalized solid-state nudged elastic band method. J.Chem. Phys. 136, 074103 (2012).AcknowledgementsThis work was supported by the Science and Technology DevelopmentFund (FDCT), Macao SAR (0063/2023/RIA1), the Natural Science Foun-dation of China (NSFC, 52322319), UM research grant (SRG2022-00053-IAPME), UM and UMDF research grant (MYRG-GRG2023-00014-IAPME-UMDF), the European Research Council (grant VANDER), the Lloyd’sRegister Foundation (grant Designer Nanomaterials), UKRI (EP/X017745:M.L.-H), theRoyal Society (URF\R1\201515:M.L.-H.) andDirectedResearchProjects Program of the Research and Innovation Center for Grapheneand 2D Materials at Khalifa University (RIC2D-D001: M.L.-H. and A.K.G.).Author contributionsP.Z.S., A.K.G., M.L.-H. and G.-P. H. designed, directed the projectand analyzed the results. Y.J., P.Z.S., Y.L. and W.Z.Z. fabricated thetitania devices. Y.J. and G.-P. H. performed proton and ion transportmeasurements. Y.-T.T. performed gas permeation measurements.W.Q.X. and S.J.Y. performed DFT calculations. R.Z.M. and T.S.synthesized titania sheets. D.-M. T. performed HRTEM character-izations for the titania samples. P.Z.S., A.K.G. and M.L.-H. wrote themanuscript with input from all authors.Competing interestsThe authors declare no competing interests.Additional informationSupplementary information The online version containssupplementary material available athttps://doi.org/10.1038/s41467-024-54544-z.Correspondence and requests for materials should be addressed toGuang-Ping Hao, Marcelo Lozada-Hidalgo, Andre K. Geim orPengzhan Sun.Peer review information Nature Communications thanks StephenCreager, ChuanhuaDuanand theother, anonymous, reviewer(s) for theircontribution to the peer review of this work. A peer review file is avail-able.Reprints and permissions information is available athttp://www.nature.com/reprintsPublisher’s note Springer Nature remains neutral with regard to jur-isdictional claims in published maps and institutional affiliations.Open Access This article is licensed under a Creative CommonsAttribution 4.0 International License, which permits use, sharing,adaptation, distribution and reproduction in any medium or format, aslong as you give appropriate credit to the original author(s) and thesource, provide a link to the Creative Commons licence, and indicate ifchanges were made. The images or other third party material in thisarticle are included in the article’s Creative Commons licence, unlessindicated otherwise in a credit line to the material. If material is notincluded in the article’s Creative Commons licence and your intendeduse is not permitted by statutory regulation or exceeds the permitteduse, you will need to obtain permission directly from the copyrightholder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.© The Author(s) 2024Article https://doi.org/10.1038/s41467-024-54544-zNature Communications |        (2024) 15:10546 8https://doi.org/10.1038/s41467-024-54544-zhttp://www.nature.com/reprintshttp://creativecommons.org/licenses/by/4.0/http://creativecommons.org/licenses/by/4.0/www.nature.com/naturecommunications High proton conductivity through angstrom-porous titania Results Device fabrication and characterization Proton transport Ion selectivity Outlook Methods Device fabrication HRTEM imaging Helium leak tests Electrical measurements for the transport of protons and ions Density functional theory calculations Thermal stability Zeta potential Data availability References Acknowledgements Author contributions Competing interests Additional information