Ambient-temperature fabrication of microporous carbon terminated with graphene walls by sputtering...

This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/authorsrights

http://www.elsevier.com/authorsrights

Author's personal copy

Ambient-temperature fabrication of microporous carbon terminated with graphenewalls by sputtering process for hydrogen storage applications

Arghya Narayan Banerjee a,⁎, Sang Woo Joo a,⁎⁎, Bong-Ki Min b

a School of Mechanical Engineering, Yeungnam University, Gyeongsan-Si 712-749, South Koreab Center for Research Facilities, Yeungnam University, Gyeongsan-Si 712-749, South Korea

a b s t r a c ta r t i c l e i n f o

Article history:Received 31 August 2012Received in revised form 6 April 2013Accepted 16 April 2013Available online 2 May 2013

Keywords:Graphitic carbonMicroporous carbonMultilayer grapheneAmbient substrate temperatureMetal nanoparticlesSputteringDisplacement cascadeHydrogen storage

A very thin amorphous carbon film (10–30 nm), has been bombarded with sputtered Cr nanoparticles,resulting in inelastic collision between the nanoparticles and the nuclei of the C-atoms causing atom displace-ment and re-arrangement into graphene layers. The process occurs at ambient temperature. Fabrication ofgraphitic microporous carbon terminated with few-to-multilayer graphene walls has been verified byRaman spectroscopy and scanning transmission electron microscopy. High resolution transmission electronmicrographs reveal that the formation of graphene layers is highly sensitive to the sputtering parameters.With a gradual increase in the sputtering voltage/current density/time from 3.5 kV/40 mA–cm−2/1.0 minto 5.0 kV/70 mA–cm−2/3.0 min the graphitic domains are found to transform from semi-graphitized layersto well-defined, highly ordered, larger-area graphene walls within the microporous network. The mechanismof this graphitic microporous carbon formation is assumed to be due to two simultaneous processes: in onehand, the sputtering plasma, containing energetic ions and sub-atomic particles, act as dry-etchant to activatethe a:C film to transform it into microporous carbon, whereas on the other hand, the charged metal nanoparticle/ion bombardment under sputtering resulted in the inelastic collision between the nanoparticles/ions and the nucleiof the C atoms followed by atom displacement (and displacement cascade) and re-arrangement into orderedstructure to formgraphitic domainswithin themicroporous carbon network. H2 storage experiment of the samplesdepicts excellent hydrogen storage properties. This simple, cost-effective, complementary-metal-oxide-semicon-ductor-compatible, single-step process of metal-graphene hybrid nanomaterial formationmay find interesting ap-plications in the field of optoelectronics and biotechnology. Additionally, this method can be adopted easily for theincorporation of transitionmetals into graphene and similar graphitic carbon nanostructures andmay enhance thehydrogen storage capacity for energy-related applications.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

Hydrogen storage is one of the most important applications in thefield of clean energy generation because H2 is considered to be anideal source of energy for efficient and environmental friendly applica-tions [1]. Because of its tremendous importance, the US Department ofEnergy (DOE) has set a gravimetric goal of 8–9 wt.% hydrogen storageby 2017, and various groups around the globe are actively engaged inthe research of safe and compact storage of hydrogen in solid-statematerials to achieve this target. Among various solid sources used forhydrogen storage medium, carbon-based nanomaterials are found tobe promising because of the possibility of efficient reversibility, fastkinetics, and high capacity owing to large surface area [2]. Among differ-ent carbon-based nanostructures, graphitic micro/mesoporous carbon

andmetal-incorporated/corrugated graphene nanostructures are wide-ly explored for improved hydrogen storage [3–5]. Especially, metal-incorporated/decorated graphene is reported to show very high hydro-gen uptake capacity due to its hybridization/polarization with metalatoms [2,5,6]. On the other hand, micro/meso-porous carbon, containingpartially graphitic structures, also shows enhanced hydrogen uptakecapacity because of the high surface area into the porous morphologyand favorable hydrogen binding energy (BE) at the graphene walls ofthe localized graphitic network [3,4,7]. Therefore, metal-incorporated,graphitic micro-/meso-porous carbon, containing few-to-multilayergraphene walls, may potentially enhance the hydrogen BE to the desiredrange of ~0.2–0.6 eV to fulfill the elusive DOE goal [2,8]. But the mainissue in this regard is the lack of proper fabrication techniques to incorpo-ratemetal atoms into the porous carbon/graphene network to satisfy thedesired H2 BE, and hence, warrants much attention.

Fabrication of graphene has been realized by various physical andchemical-based techniques [9–12], whereas metal incorporation/decoration is performed by a solution-based process [13]. On the otherhand, partially graphitic micro/meso-porous carbon formation as well

Thin Solid Films 537 (2013) 49–57

⁎ Corresponding author. Tel.: +82 53 810 2453; fax: +82 53 810 2062.⁎⁎ Corresponding author. Tel.: +82 53 810 1456; fax: +82 53 810 2062.

E-mail addresses: [email protected] (A.N. Banerjee), [email protected](S.W. Joo).

0040-6090/$ – see front matter © 2013 Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.tsf.2013.04.070

Contents lists available at SciVerse ScienceDirect

Thin Solid Films

j ourna l homepage: www.e lsev ie r .com/ locate / ts f


as metal incorporation are performed via template-based or chemicaletching techniques followed by elevated temperature annealing foractivation/carbonization/graphitization [3,4,7,14–17]. As most of theseprocesses are solution-based and involve high temperature treatments,they are not compatible to modern solid state fabrication techniquesand also not suitable for low-cost high-throughput processes. Thus,the adoption of alternative vacuum-based, low/ambient-temperaturemethods is needed for diverse device applications. And in that respect,partial graphitization of solid carbon sources via ion/electron irradiationtechnique is found to gain renewed interest for possible controlledgraphitization through clean processes [18,19]. Among various solidcarbon sources, amorphous carbon (a:C) is found to be a useful sourcematerial for this type of irradiation method, because it consists ofmostly sp2-bonded carbon atoms, and therefore, the derived graphenelayers will be extremely pure. Also it is very easy to synthesize undermost of the vacuum-based techniques (which are complementarymetal-oxide semiconductor {CMOS} compatible) under ambient tem-perature, and therefore, can easily be integrated with modern solidstate device fabrication technologies for future applications. But theformation of stable graphitic nanostructures under ion/electron beam(e-beam) irradiation from a:C is found to be challenging because of thepresence of dangling bonds in the unordered structure of a:C, which ap-pears to influence the dissolution of the ordered domain into theunordered species [20] once the irradiation is withdrawn. To overcomethis, catalytic graphitization of a:C is performed by incorporating metalparticles (Fe, Co Ni etc.) into the a:C layers followed by e-beamirradiation, causing the dissolution of carbon atoms into metal particlesto form eutectic followed by the supersaturation and orderedprecipitation of graphitic layers at the metal-film interface [21,22]. Bute-beam lithography is a highly time-consuming process for large areagraphitization of a:C film, and therefore only suitable for short-range gra-phitic network formation, and impractical for volume production as thelonger time exposure to electron irradiation of the sample is reported tocause the re-amorphization of the already created graphitic layers [23].

In this current study, we have shown the graphitization of a:C thinfilm by irradiation of energetic metal nanoparticles using sputteringprocess. At an optimum sputtering voltage, current (density) andtime, the thin a:C layer transforms into metal-incorporated micropo-rous carbon network embedded with short-range few-to-multilayergraphene walls due to the energy transfer of the bombarded sputteredmetal (Cr) nanoparticles/ions into the carbon matrix at ambienttemperature. The number of graphene layer formation and the degreeof crystallization are found to be highly sensitive to the sputteringparameters. The microporosity formation is facilitated by a dry-etchingprocess caused by the sputtering plasma. Previously, we have reportedthe preliminary results of the site-specific formation of partially graphiticmicroporous carbon on some pre-selected areas (using etchedmask), viametal sputtering process for potential hydrogen storage applications [24].In this report we are presenting the detailed study of the effect ofsputtering current/voltage and time on the number of graphene layerformation and the degree of ordering of the graphene walls within themicroporous network at various sputtering conditions. Since the resultssuggest that the degree of graphitization and number of graphene layerformation are highly dependent on the sputtering parameters, largearea, single-to few-layer graphene can be fabricated by this simplemethod by a proper optimization of the deposition conditions. Therefore,this ambient temperature, CMOS-compatible process can find interestingapplications in device fabrication industry for the large-area synthesis ofmetal-incorporated graphitic microporous carbon, terminated withhighly ordered few-to-multilayer graphene walls, for hydrogen storageand related clean energy applications.

2. Experimental details

Fabrication of graphitic microporous carbon is performed by treatingthermally evaporated ultra-thin a:C film under sputtering plasma.

Initially, a:C thin film (thickness ~10–30 nm) is deposited on commer-cially available Si wafer (Sigma Aldrich, Korea) and formvar-stabilized200-mesh Cu-grid (Ted Pella Inc., US) via a thermal evaporation system(JEOL, JEE-400) using standard carbon rod. The thickness of the film ismonitored by in-situ quartz crystal monitor. Next, these a:C-coatedsubstrates are introduced into a direct-current (dc) sputtering chamber(Quorum Technologies/Emitech, UK) which consists of a Cr sputter-target as upper electrode with a base pressure of 10−3 Pa. The layout ofthe experimental process and the sputtering parameters is furnished inFig. 1. The substrate temperature is kept at ambient condition while thesputtering voltage, current (density) and time are varied to observetheir effect on the degree of graphitization of the samples. The substratetemperature is measured before and after each sputter-deposition viathermocouple arrangements. Since the samples are placed on a metallicsubstrate holder, the local heat generated from the waste kinetic energyof the incident particles is rapidly dissipated through it, keeping thesubstrate at relatively lesser temperature. The details of the experimentalprocess are reported elsewhere [24]. The samples are then characterizedunder confocal Raman microscope (alpha300#, WITec), at a laser exci-tationwavelength of 532 nm, to verify the formation of graphiticmicro-porous carbon. To observe the degree of graphitization at differentsputtering conditions, microstructural and morphological characteriza-tions are performed by scanning transmission electron microscope(S-TEM, Tecnai G2 F20 with 200 and 80 kV e-beam energy). Graphiticmicroporous carbon, fabricated on the a:C-coated Cu-grid, is directlyintroduced into the S-TEM unit for high-resolution (HR)-TEM analysis.To verify that the formation of graphitic microporous carbons on twodifferent substrates (Si wafer, Cu-grid) under identical sputteringparameters shows similar features (i.e. negligible substrate effect),Raman characterizations are done on samples prepared on both typesof substrates before and after sputtering treatment. Both sets of thesamples show similar Raman curves (shown and described later), indi-cating minimal substrate effect. The only difference is that the Ramanpeaks for graphitic microporous carbon prepared on Cu-grid via sputtertreatment show a very strong background noise due to fluorescencegenerated by the Raman laser, and mask the signals considerably(shown later). Hence the samples are prepared on Si substrate for prop-er Raman characterizations. Additionally, the negligible substrate effectis further justified by the fact that the entire fabrication is done atambient substrate temperature. The necessary activation energy forgraphitization of the a:C film is provided by the energetic sputterednanoparticles/ions by a top-down approach through the film surface(explained later). In our experimental condition, we believe that thesubstrate effect can only come into play for graphitization, if the activa-tion energy is provided by a bottom-up approach (probably throughsubstrate heating). With the current operating conditions, the localtemperature enhancement at the a:C film surface due to sputteringplasma is found to be at most 20 °C, which is much less than therequired temperature (>500 °C) for graphitization through thermaltreatment [4,7,15].

3. Results and discussion

Fig. 2 represents the TEM micrographs of untreated and sputter-treated samples, indicating the formation of micropores within a:Cfilms. Fig. 2(a) shows the untreated a:C films indicating smooth sur-face structure, whereas Fig. 2(b) depicts the formation of micro-pores (pore diameter b2 nm) under sputtering treatment of3.5 kV/40 mA cm−2/1.0 min of sputtering voltage/current (densi-ty)/time, respectively. Fig. 3 represents the HRTEM micrographsof graphitic microporous carbon fabricated on a:C-coated Cu-grid atdifferent sputtering conditions. Fig. 3(a-i) and (a-ii) shows thesemi-graphitized wall formation within the microporous carbonnetwork at sputtering voltage/current-density/time (V/J/T) of3.5 kV/40 mA–cm−2/1.0 min and 3.5 kV/40 mA–cm−2/3.0 min,respectively. With increase in the deposition time, the semi-graphitized

50 A.N. Banerjee et al. / Thin Solid Films 537 (2013) 49–57


domains are found to increase, although the graphitic walls are notwell-defined in both cases. When the V/J/T values are increased to4.0 kV/50 mA–cm−2/1.0 min (cf. Fig. 3b-i), ordered graphitic domainsstarted to form within the microporous network with better degree ofcrystallization with respect to previous conditions (Fig. 3a). The forma-tion of a mixture of 5- to 10-layers of graphene walls is observed in thiscase. When the deposition time is increased to 3.0 min (keeping otherparameters identical to 3b-i), graphitic domains started to formon larg-er area (cf. Fig. 3b-ii). In both cases, well-defined graphene wall forma-tion is observed. Inset of Fig. 3(b-i) represents the line intensity profileacross a 5-layer graphenewalls, indicating the proper lattice spacing be-tween the graphene layers. Similarly, the inset of Fig. 3(b-ii) representsa portion of the sample having ordered graphene layers indicating theformation of proper lattice spacing. With further increase in the valuesof V/J/T to 4.5 kV/60 mA–cm−2/1.0 min, larger-order graphitic domainformation is observed havingmore than 15-layers of graphenewalls (cf.Fig. 3c-i). When the deposition time is increased to 3.0 min, well-aligned, long-order multilayer graphene walls are formed with higherdegree of graphitization (cf. Fig. 3c-ii). With further increase in the V/J/Tvalues to 5.0 kV/70 mA–cm−2/1.0 min and 3.0 min, highly-aligned,very long-order and larger-domain multilayer graphene formation isobserved (cf. Fig. 3d-i/ii). Also the formation of proper lattice spacingwithin the graphene layers is indicated within the micrographs. All ofthe above HRTEM micrographs are imaged with a 200 kV e-beam ofthe STEM unit, whereas the inset of Fig. 3(d-ii) represents the HRTEM

image of the sample prepared with identical conditions as that usedfor Fig. 3(d-ii) and imaged with an 80 kV e-beam under STEM. Basi-cally, this low-energetic-beam-imaging is performed to verify thatthe observed graphitization process in our samples is purely due tothe bombardment of energetic nanoparticles/ions, rather thanin-situ graphitization of carbon matrix via high-energetic e-beamduring TEM imaging, because sometimes, carbon nanostructuresare found to be unstable under highly accelerating electron beam[19]. The graphene layer formation, with proper lattice spacing, is clear-ly observed in this micrograph, verifying that the graphitization processis purely the result of charged nanoparticles/ion–carbon atom collisionprocess (explained in details later). Due to the lower resolution of theimages under 80 kV e-beam, higher energetic e-beam (@ 200 kV) isused to obtain good images for representation. Also, it has beenshown later in the Raman characterization, that the samples preparedon Si substrates show the signature of graphitic microporous carbon, al-though those samples have never been exposed to e-beam, thus prov-ing the collision mechanism to be correct. Also it is noteworthy thatthe commercially available formvar-stabilized carbon-coated coppergrids are routinely used for HRTEM imaging as they are designed for200 kV e-beam, and hence, highly stable.

The mechanism of the graphitic microporous carbon formation isbelieved to be due to two simultaneous effectsmanifested by sputteringprocess. In one hand, the bombardment of a shower of energeticnanoparticles/atoms/ions (present within the sputtering plasma) onto

Fig. 1. Flow-chart of the fabrication process of graphitic microporous carbon.

5 nm5 nm

(a) (b)

Fig. 2. TEM micrographs of (a) untreated, and (b) sputter-treated samples indicating the conversion of a:C films into microporous carbon under sputtering treatment of 3.5 kV/40 mA cm−2/1.0 min of sputtering voltage/current (density)/time, respectively.

51A.N. Banerjee et al. / Thin Solid Films 537 (2013) 49–57


2 nm

2 nm2 nm

2 nm 2 nm

2 nm

0.34 nm

0.34 nm

0.34 nm0.34 nm

0.34 nm

2 nm 2 nm

(a-i) (a-ii)

(b-i) (b-ii)

(c-i) (c-ii)

(d-i) (d-ii)

0 0.5 1.0 1.5

0.34nm

2 nm

0.34 nm

3.5 kV, 40 mA-cm-2,

1.0 min

3.5 kV, 40 mA-cm-2,

3.0 min

4.0 kV, 50 mA-cm-2,

1.0 min

4.0 kV, 50 mA-cm-2,

3.0 min

4.5 kV, 60 mA-cm-2,

1.0 min

4.5 kV, 60 mA-cm-2,

3.0 min

5.0 kV, 70 mA-cm-2,

3.0 min

5.0 kV, 70 mA-cm-2,

1.0 min

Fig. 3. HRTEM micrographs of graphitic microporous carbon fabricated at different sputtering voltage, current (density) and time of (a) 3.5 kV, 40 mA–cm−2 and (i) 1.0 min,(ii) 3.0 min; (b) 4.0 kV, 50 mA–cm−2 and (i) 1.0 min, (ii) 3.0 min; (c) 4.5 kV, 60 mA–cm−2 and (i) 1.0 min, (ii) 3.0 min; (d) 5.0 kV, 70 mA–cm−2 and (i) 1.0 min, (ii) 3.0 min,respectively. Inset of 3(b-i): Line intensity profile across a 5-layer graphene (X-axis is scaled from 0 to 1.75 nm). Inset 3(b-ii): Selected portion of the sample showing the properlattice spacing between graphene layers. Inset 3(d-ii): HRTEM micrograph of graphitic microporous carbon at 80 keV e-beam.



the ultra-thin a:C film facilitates the graphitization of the films via‘knock-on’ effect [19,25,26], and on the other hand, the Ar+-plasmaprovides the necessary activation (by acting as a dry-etchant) for micro-porosity formation. The threshold energy (Eth) for the above-mentionedstructural ordering of carbon nanostructures via a high energetice-beam is reported to be typically around 10 eV [19]. Therefore,for energetic nanoparticles/ions, having much larger size andmass (than that of e-beam), this threshold would be much lesserbecause of higher collision cross-section (and hence a larger energytransfer). It is well-known that sputtering plasma consists largelyof charged nanoparticles/atomic-clusters/ions with wide varia-tions in the size/mass/charge, the mean of which is the functionsof the sputtering parameters used [27–29]. Therefore, sputteredCr atoms/charged nanoparticles/atomic clusters can acquireenough kinetic energy through the applied dc bias within oursputtering system (4–5 kV) and inelastically collide with the car-bon nuclei to transfer necessary energy (over the threshold) forthe structural ordering via atomic displacements, resulting in aphase transformation from amorphous to graphitic carbon. Addi-tionally, the ‘knock-on’ atomic displacement in the sub-thresholdenergy regime can also be possible in our case as we have observeda small amount of hydrogenated carbon in our samples (shownlater in Raman micrographs). The process may basically follow atwo-step collision method, where lighter sputtered particles cantransfer sufficient energy to the surface hydrogen atoms, which inturn, can displace lattice carbon atoms by a large momentum trans-fer process [19].

It is well-known that in a sputtering system, total energy transferredto the substrate consists of kinetic energy (KE) of the sputtered atoms,KE of ions and electrons originating from sputtering plasma, KE of neutralparticles reflected towards the substrate and plasma radiation energy[30]. Among these, the KE of sputtered atoms will contribute maximumto the total transferred energy against the other components as themean free path of the ions/electrons will be very low under the currentoperating pressure, and hence, would experience very high scattering[30]. The maximum KE of the sputtered atoms near the substrate surface(ξf) will be expressed as [31,32]:

ξf ¼ δξion ξBE

ξth

� �cosθ; ð1Þ

where, ξion, ξBE and ξth are the sputtering ion energy (which is related tothe cathode potential), surface binding energy and sputtering thresholdof the target material, respectively. δ is the degradation factor (becauseof the loss of energy of sputtered atoms due to collision with thesputtering gas during transport towards the substrate [32]) and θ is thescattering angle in the lab frame (with respect to the normal to the sub-strate and target in a vertical diode configuration) of the sputtered atomsduring collision, the averages of which can be obtained from the follow-ing relations [33]:

ln δð ÞjAvg ¼1−Mg

.Ms

� �2

2 Mg�Ms

� �264

375 ln

1þ Mg

.Ms

� �

1−Mg�Ms

� �24

35−1 ð2Þ

cosθjAvg ¼ 1−Mg

.Ms

� �2

3

264

375 ð3Þ

and,Ms,Mg aremolecular weights of sputtering target and sputteringgas, respectively. These relations are modeled by taking into accountthe pressure, temperature, electrode distance of the sputtering sys-tem, mean free path of the scattered atoms and the scatteringcross-section of the interacting particles/ions [33]. According to ourexperimental conditions of chromium sputtering at 1.0 Pa argonpressure at ambient temperature (300 K), we have approximated

ξion|max = 5 keV (assuming that majority of the sputtering ions suf-fer negligible collision with the neutral gas atoms before fallingthrough the cathode potential to the target, as the negative glow re-gion of a discharge tube, from which these ions are generated, isformed very close to the cathode of a sputtering system operating at ab-normal glow regime [34,35]), ξBE = 3.37 eV [36], ξth = 20.52 eV [37],δ = 0.4 and cosθ|avg = 0.8 (calculated from Eqs. (2) and (3), respective-ly). With these values, the maximum KE (ξf) of the sputtered atoms(moving perpendicular to the substrate) at the substrate surface isapproximately obtained around 250 eV. Now the energy spectrum ofthe sputtered atoms, according to the Thompson–Sigmund model [38],will vary from 0 to ξf. As the threshold energy for structural evolution ofa:C is around 10 eV [19], therefore it becomes apparent that a consider-able amount of energetic atoms exists near the substrate surface, whoseenergies are higher than the threshold energy to manifest the knock-ongraphitization. Apart from these, the calculations of the plasma radiationenergy depict that the corresponding contribution is much less than thatof the threshold energy for a:C to induce graphitization. Therefore, it isbelieved that this manifests mainly the micropore formation of the a:Cfilm [39].

It is well-known that at the onset of ‘abnormal glow’ discharge ofsputtering process (which is the standard working regime for most ofthe sputtering systems) the sputtering current (density) is an increasingfunction of the sputtering voltage as in [35].

V ¼ A1 þ A2

ffiffiJ

p 1p

� �ð4Þ

where V is the applied sputtering voltage, J is the sputtering currentdensity, p is the sputtering pressure during deposition and A1/A2 areconstants, which depend on the electrode material/geometry andsputtering gas composition. Therefore, Eq. (4) shows that when thesputtering voltage is increased (other independent parametersremain the same within the experimental limits), more energy is trans-ferred into the system, resulting in the higher degree of crystallization.This clearly explains the higher degree of ordering in the multilayergraphenewalls with the increase in the sputtering voltage/current densi-ty (cf. Fig. 3). Also, an increase in the sputtering time allows for the trans-lation of higher energy into the carbon matrix to manifest larger atomicdisplacement and displacement cascade for crystallization into longer-order graphitic domain. Several models have been proposed for thethree-dimensional graphitization of carbon nanomaterials. Some ofthese models infer that the thermally activated displacement and rota-tion of partially graphitic layers manifest the three-dimensional graphiti-zation process, whereas some other models speculated that the removalof defects during activation process might have constituted the graphiti-zation or assumed the presence of curled and (or) twisted grapheneribbons within the amorphous layer that, in order to reduce their surfaceenergy, tend to lieflat during high-temperature carbonization and activa-tion treatments to form closely packed graphitic walls [40–45]. A quanti-tative analysiswas carried out to determine the specific binding energy oftwo graphene layers by summing up the binary interaction energies ofeach atom of one layer with all the atoms of the other layer by theatom–atom potential method, followed by the optimization of the inter-action potential of the two neighboring graphene layers as a function oftheir relative arrangements according to the expression [40,42,43]:

ζ inij ¼ −α1 ρ

→ij

�� −n þ α2 exp −α3 ρ→

ij

�� ð5Þ

where ζijin is the interaction energy between ith carbon atomof one layerand the jth carbon atom of the other layer separated by a distance ρijandα1,α2, andα3 are constants and n is an exponent. Thisminimizationprocess seems to be reasonable as the layerswithin the graphitic carbonconsist of hexagonally coordinated and covalently bonded carbonatoms, and therefore, favorable for three-dimensional graphitizationthrough a change in the relative orientation and energy of the partially



graphitic layers. The minimization process depicts that at a certaindimension of the layers (which may correspond to two- to threegraphene layers), a set of layer-displacement points emerges wherethe potential energy becomes minimum. This leads to near-negligiblepotential barrier between these points, resulting in the displacementsof neighboring layers with respect to each other without any activation.This manifests a displacement cascade for the longer order graphitiza-tion of the samples [40]. In our case, we speculate that initially, theenergetic nanoparticle and (or) ion irradiation process provide the nec-essary energy into the a:C to initiate the structural ordering to formshort-range few-layer graphitic walls and then the interaction betweenC-atoms of different layers produces the potential energy minimum toundergo displacement cascade with very little activation, resulting inthe formation of long-order graphene walls. With further increase inthe number of graphene layers, the geometry of the region ofdisplacement-points changes and a new potential energy minimum isattained. The process continues until it is thermodynamically limitedby the final overall dimension of the graphene layers [40].

As far as the microporosity formation is concerned, several models[45–48] assumed the presence of non-hexagonal rings within the hex-agonal network of carbon, which are distributed randomly throughoutthe carbonmatrix and create localized curvature andnanoscale porositybecause of the presence of fullerene-like elements. Traditionally, themicroporosity formation is manifested by some chemical activationprocess, where various reagents are used which act as dehydratingagents or chemical etchants to remove surface atoms to createmicropo-rosity [3,49]. In our case, we assume that during activation process of a:C film under sputtering plasma, the energetic ions and sub-atomicparticles create the defect within the hexagonal network of carbon togenerate non-hexagonal (pentagon and/or heptagon) rings to formcurved and closed structures of microporous carbon. The entire mecha-nismof the formation of graphiticmicroporous carbon, terminatedwithfew-to-multilayer graphene walls, is schematically shown in Fig. 4. Itdepicts that when the hexagonally arranged carbon network (cf.Fig. 4a) is treated under sputtering plasma, some of the loosely boundsurface atoms are etched out by scattered Ar+ and other energetic

ions/sub-atomic particles to form nonhexagonal rings and creates local-ized curvature and nanoscale porosity (cf. Fig. 4b). Simultaneously, thebombardment of higher energetic sputtered-nanoparticles/ions ontothe carbon network manifests the atomic displacement and displace-ment cascade for crystallization into ordered graphitic domains withinmicroporous carbon terminated with few-to-multilayer graphenewalls (cf. Fig. 4c). With an increase in the sputtering voltage and cur-rent density, higher rate of energy translation into the carbon matrixoccurred, resulting in higher degree of crystallization. Hence, wehave observed a gradual increase in the order of graphitizationfrom semi-graphitized domain formation at lower sputtering voltage/current density (@ 3.5 kV/40 mA–cm−2, cf. Fig. 3a) to highly-orderedand well-aligned multilayer graphene wall formation at highersputtering conditions (@ 5.0 kV/70 mA–cm−2, cf. Fig. 3d). The proposedmechanism of graphitic microporous carbon formation is further corrob-orated by the HRTEM image shown in Fig. 5(a), where a nanoparticleof ~ 5 nm diameter is observed to constitute graphitic layers within mi-croporous carbon. Low magnification TEM (Fig. 5b and inset) imagesalso confirm the formation of nanoparticle-graphitic microporous carbonnanocomposite. The presence ofmetal (Cr) nanoparticle is verified by en-ergy dispersive X-ray analysis reported earlier [24]. Fig. 5(c) and (d)shows the TEM images of 3-to-5-layers of graphitic microporous carbonsheets, the edges of which are speculated to be forming the graphenewalls.

To further verify the formation of graphitic domain within themi-croporous carbon network, as well as to qualitatively determine thequality of the graphene layers, Raman characterizations are performed.Firstly, Raman characterizations are done on the starting a:C films depos-ited on Si-wafer andCu-grid. Raman curves of a:Cfilms deposited onboththe substrates show similar nature (cf. Fig. 6a). Both the curves resemblewith a typical disordered amorphous carbon, indicating minimal sub-strate effect. The broad band around 1500 cm−1 and the small peak at~975 cm−1 originate from 1st-order Raman scattering within a:C andT-peak of hydrogenated a:C (sp3-bonded carbon), respectively [50].For the a:C film deposited on Si wafer (red curve), the sharp peakat ~520 cm−1 corresponds to the fundamental Si Raman scattering,

(b)

(c)

Sputtering plasma

Graphene walls

Closed graphitic structures

(a)

Fig. 4. Schematic representation of the formation mechanism of graphitic microporous carbon.



originated from the substrate [51]. Fig. 6(b) represents the Ramanspectrum of graphitic microporous carbon prepared on Cu-gridunder the sputtering voltage, current (density) and time of 5.0 kV,70.0 mA–cm−1 and 3.0 min, respectively. The existence of RamanD and G peaks clearly indicates the formation of graphitic domainswithin the microporous carbon. As mentioned above, the graphalso depicts that due to very high background noise (generated dueto the fluorescence at Cu-grid under Raman laser), the signal resolu-tion becomes very low. Hence, the 2D Raman mode of graphitic car-bon (speculated to be around 2600–2700 cm−1, as shown in theinset of Fig. 6b) is found to be completely masked due to the poorsignal-to-noise ratio. Since the 2D band is derived from the doubleresonance Raman process involving phonon-charge carrier interac-tion within the first Brillouin zone of graphite, therefore proper mea-surement of this band as well as its relative intensity with respect toD and G bands are very important to understand the degree of graph-itization and the quality of the graphene layers formed within thegraphitic domains of the microporous carbon. Hence we have fabricatedgraphitic microporous carbon on Si substrate and Raman analysisdepicts the existence of Raman D (~1347 cm−1), G (~1579 cm−1)and 2D (~ 2600–2700 cm−1) bands within the samples fabricatedat sputtering parameters (V/J/T) of 5.0 kV, 70.0 mA–cm−1 and3.0 min, respectively (cf. Fig. 6c and d). Especially, the Raman Gand D bands are typical of partially graphitic micro-/meso-porouscarbon, reported earlier [4,7,16], indicating the formation of graphiticdomains within our samples. The presence of G-band, which is theRaman active E2g mode of two-dimensional graphite (i.e. graphenesheets) and originates from the interplanar sp2-bonded C\C stretchingvibrations, clearly indicates the formation ofwell-defined graphitic layers

within the microporous network in the form of multilayer graphenewalls. The Raman D band, which is the Raman active A1g mode, corre-sponds to the structural disorders within the as-synthesized multilayergraphene walls within the microporous network and mostly found inthe wall edges. This band actually originates from the vibrations of Catoms with dangling bonds in planer termination of disordered graphiticstructures [52]. Generally, the ratio between the intensities of G-band (IG)andD-band (ID) indicates the degree of graphitizationwithin the graphit-ic network, and in our case, we found this value (IG/ID) to be around 1.6,which is comparable to the chemically and physically grown few-to-multilayer graphene reported earlier [11,53,54]. This clearly indicatesthe formation of well-ordered graphitized domain within our micropo-rous network, containing few-to-multilayer graphene walls. Also, a com-parison of the IG/ID-value of our samples with respect to previouslyreported partially graphitic porous carbon samples depicts that thisratio is considerably less than 1.0 (~0.48 to 0.56) in previous cases, indi-cating disordered semi-graphitized wall formation within those porouscarbon networks [4,7,16] in contrast to our samples, which showwell-defined graphene layer formation. But still the D-band intensity inour case is relatively higher than that of single-to-few-layer graphenefabricated by exfoliation technique [9,10]. We speculate that the forma-tion of metal-graphene nanocomposite in our case probably increasesthis D-band intensity, because several groups have already reported ear-lier that the decoration of graphene by metal-nanoparticle or intercala-tion of foreign molecules into graphene creates higher disorder, leadingto an increment in the D-band intensity [11,13]. Additionally, we pre-sume that because of our dynamic fabrication process of metal nanopar-ticle sputtering, disorders are induced by some corrugated structureswithin the graphitic domain, which tends to decrease this IG/ID ratio

50 nm 5 nm

5 nm 0.5 µm 50 nm

(c) (d)

(a) (b)

Fig. 5. TEM images showing the (a) formation of graphitic domain due to the bombardment of a 5 nm metal nanoparticle, (b and inset) incorporation of metal-nanoparticles intographitic microporous network, and (c and d) 5-to-3-layer graphene sheets, the edges of which are forming the graphene walls.



with respect to previously reported single-to-few-layer graphene.However, it may also be noted in this connection that, the Raman Dpeak alone cannot be used to accurately assess the degree of disorderas edges of graphene are always present even in perfect graphene andwill appear in Raman spectra if the laser beam includes these edges[52]. As far as the Raman 2D band in the graphitic porous carbon net-work is concerned, we have reported the presence of this band in ourgraphitic microporous network against previously reported partiallygraphitic porous carbon structures [4,7,16]. Fig. 6(d) represents theRaman 2D band in our sample, the y-axis of which is magnified toshow the two deconvoluted (2D1, and 2D2) peaks, which are typicalof graphitic carbon [50]. This clearly indicates the formation of well-ordered, larger domainmultilayer graphenewalls in our sample. The in-tensity ratio between the G-band and 2D-band (IG/I2D) in our sample isfound to be around ~10–12, which is typical of physically and chemical-ly grown few-to-multilayer (>10 layers) graphene reported earlierunder similar Raman excitation wavelength [11]. However, this valueis found to be relatively higher against exfoliated single-to-few-layergraphene reported earlier [9,10]. We speculate that the metal nanopar-ticle incorporation in our samples via sputtering treatment affects sig-nificantly the 2D band intensity because of the Coulombic chargetransfer, which alters the phonon-charge carrier scattering probability,and hence decreases the 2D peak intensity [11,13]. But in any case,since the 2D-band intensity is highly sensitive to the electronic bandstructure of graphene, the evolution of 2D band in our sample with re-spect to previously reported partially graphitic porous carbon networks,clearly indicates that our process is highly effective to fabricate largerarea graphitic domains within metal-incorporated microporous carbonnetwork, terminated by highly ordered multilayer graphene walls.

We have also performed the H2 storage experiment via the electricalresistance measurements during molecular hydrogen adsorption on theas-prepared samples. Initial electrical resistance (R0) of the sample isfirst measured under vacuum (0.7 Pa) until the resistance is stabilized.Thereafter, hydrogen gas is introduced into the chamber @ 20 sccm(with a gas pressure of 16.0 Pa) and the corresponding changes in the

resistance (R − R0) of the samples as a function of time are measured.Fig. 7 represents the proportional changes in the electrical resistance ofa:C film (reference, untreated sample, cf. Fig. 7a) and the graphiticmicro-porous carbon films (for two different sputtering conditions, cf. Fig. 7band c) during molecular hydrogen absorption. In the later two cases,the resistance is observed to decrease nearly 30% to 50%with time againstthe reference sample, indicating excellent hydrogen adsorption proper-ties of the as-synthesized graphitic microporous carbon films. It hasbeen observed in Fig. 3(b) that under similar sputtering conditions aspresented in Fig. 7(b) and (c), the average number of graphene layers in-creases with sputtering time within the microporous network, thus en-hancing the hydrogen adsorption properties. Similar hydrogen storagebehavior is reported previously by Im et al. [55] for expanded graphitenetwork. The decrease in the resistance of the graphitic microporous car-bon is due to the destruction of the π-bonds in graphitic structures anddecrease in the electron hopping effects by virtue of hydrogenmolecularadsorption at the micropores [55,56].

4. Conclusions

Well-defined, highly-ordered, larger-domain graphene layer forma-tion is observed within a microporous carbon network via a sputter-deposition technique. Initially, ultra-thin films of thermally-evaporated a:C are deposited on commercially available Si-waferand Cu-grid followed by metal (Cr) nanoparticle sputtering ontothe a:C layer. The Ar-plasma facilitates for the microporosity for-mation by creating defects in terms of non-hexagonal rings withinthe hexagonal network of carbon to form localized curvature andnanoscale porosity. On the other hand, high-energetic chargednanoparticles/ions/atomic clusters, present within the sputteringplasma, are bombarded onto the a:C film via a ‘knock-on’ inelasticcollision process to manifest the atomic displace and displacementcascade for local graphitization of the a:C film to transform it fromamorphous to crystalline structure. According to the HRTEM imag-ing, the sputtering parameters are observed to be highly sensitive

GD

0 1000 2000 3000

a:C on Cu grid a:C on Si substrate

2600 2650 2700

2D2

2D1

Raman Shift (cm-)

Inte

nsit

y (a

rb. u

nits

)

1250 1500 1750

G

D

2560 2640 2720 2800

2D

Raman Shift (cm -1)

(a) (b)

(c) (d)

0 1000 2000 3000

Fig. 6. Raman spectra of (a) a:C film deposited on Si-wafer and Cu-grid, (b) graphitic microporous carbon on Cu-grid (inset: deconvolution of Raman 2D band), (c) RamanD andG bands, and(d) Raman 2D band of graphitic microporous carbon fabricated on Si wafer under a sputtering voltage, current (density) and time of 5.0 kV, 70.0 mA–cm−2 and 3.0 min, respectively.



to the degree of graphitization of the samples. At a lower sputtering volt-age and current (density), semi-graphitized layers are found to formwithin the microporous carbon network and with a gradual increase inthe sputtering voltage and current (density), well-defined, highly or-dered, multilayer graphene wall formation is observed because of thehigher rate of energy translation into the carbon matrix during collisionprocess. Also at higher deposition time, larger-domain graphitic layer for-mation is observed and attributed to the larger amount of energy transla-tion at higher time. Raman spectroscopy confirms the formation offew-to-multilayer graphene walls within the microporous network.This fabrication ofmetal-incorporated, graphiticmicroporous carbonnet-work, terminated with multilayer graphene walls from highly thin a:Cfilms at ambient temperature using a sputter deposition process, canfound interesting applications in energy-related devices, because of itsseveral advantages. First of all, the technique is cost-effective, CMOS-compatible and low/ambient-temperature process, and therefore, highlysuitable for high-throughput, volume production. Secondly, the metalsputtering technique constitutes a single step process for metal-graphene-microporous carbon hybrid nanomaterial formation, whichcan have very interesting applications in hydrogen fuel cells. Also othermetals with comparable energy threshold and sputtering yield to Cr areexpected to produce similar effect, thus expanding its versatility. Prelim-inary results with Ni and other transitional metals tend to support thisargument and will be reported later. Thirdly, this dynamic depositionprocess with corrugated graphene structures can decrease the metalclustering and thus increase the hydrogen binding energy at local curva-ture for better H2 storage. Finally, other vacuum-based deposition tech-niques, such as pulsed dc sputtering can easily be adopted to create ashower of energetic nanoparticles (via a very short and/or optimalpulse duration) having enough energy for graphitization to create fullyrelaxed, highly controlled graphene layers within microporous networkfor diverse device applications.

Acknowledgments

We acknowledge the financial support of the World Class UniversityGrant # R32-2008-000-20082-0 of the National Research Foundation ofKorea. We also acknowledge Dr. H.S Jang, Korea Research Institute ofStandards and Science, S. Korea, for hydrogen storage measurements.

References

[1] L. Schlapbach, A. Züttel, Nature (London) 414 (2001) 353.[2] H. Lee, J. Ihm, M.L. Cohen, S.G. Louie, Nano Lett. 10 (2010) 793.[3] G.P. Meisner, Q. Hu, Nanotechnology 20 (2009) 204023.[4] Z. Wu, Y. Yang, D. Gu, Y. Zhai, D. Feng, Q. Li, B. Tu, P.A. Webley, D.Y. Zhao, Top.

Catal. 52 (2009) 12.[5] X. Yang, R.Q. Zhang, J. Nim, Phys. Rev. B 79 (2009) 075431.[6] D.W. Boukhvalov, M.I. Katsnelson, J. Phys. Chem. C 113 (2009) 14176.[7] W. Gao, Y. Wan, Y. Dou, D. Zhao, Adv. Energy Mater. 1 (2011) 115.[8] Y.H. Kim, Y. Zhao, A. Williamson, M.J. Heben, S.B. Zhang, Phys. Rev. Lett. 96 (2006)

16102.[9] K.S. Novoselov, A.K. Geim, S.V. Mozorov, D. Jiang, Y. Zhang, S.V. Dubonos, I.V.

Grigorieva, A.A. Firsov, Science 306 (2004) 666.[10] K.S. Novoselov, D. Jiang, F. Schedin, T.J. Booth, V.V. Khotkevich, S.V. Morozov, A.K.

Geim, PNAS 102 (2005) 10451.[11] C.N.R. Rao, K. Biswas, K.S. Subrahmanyam, A. Govindaraj, J. Mater. Chem. 19 (2009)

2457.[12] W. Choi, I. Lahiri, R. Seelaboyina, Y.S. Kang, Crit. Rev. Solid State Mater. Sci. 35

(2010) 52.[13] R. Muszynsky, B. Seger, P.V. Kamat, J. Phys. Chem. C 112 (2008) 5263.[14] C. Liang, Z. Li, S. Dai, Angew. Chem. Int. Ed. 47 (2008) 3696.[15] Y. Zhai, Y. Dou, X. Liu, S.S. Park, C.S. Ha, D. Zhao, Carbon 49 (2011) 545.[16] Z. Lei, Y. Xiao, L. Dang, S. Bai, L. An, Microporous Mesoporous Mater. 109 (2008) 109.[17] A. Wang, J. Ren, B. Shi, G. Lu, Y. Wang, Microporous Mesoporous Mater. 151

(2012) 287.[18] H. Duan, E. Xie, L. Han, Z. Xu, Adv. Mater. 20 (2008) 3284.[19] F. Banhart, Rep. Prog. Phys. 62 (1999) 1181.[20] M.S. Zwanger, F. Banhart, A. Seeger, J. Cryst. Growth 163 (1996) 445.[21] E. Iwamura, Rev. Adv. Mater. Sci. 5 (2003) 166.[22] S. Aikawa, T. Kizu, E. Nishikawa, Carbon 48 (2010) 2989.[23] K. Nakai, C. Kinoshita, A. Matsunaga, Ultramicroscopy 39 (1991) 361.[24] A.N. Banerjee, S.W. Joo, B.K. Min, A. Sharma, Phys. Status Solidi (RRL) 6 (2012)

315.[25] J.Y. Huang, F. Ding, B.I. Yakobson, P. Lu, L. Qi, J. Li, PNAS 106 (2009) 10103.[26] C. Jin, H. Lan, L. Peng, K. Suenaga, S. Iijima, Phys. Rev. Lett. 102 (2009) 205501.[27] A.N. Banerjee, B. Das, Rev. Sci. Instrum. 79 (2008) 033910.[28] A.N. Banerjee, R. Krishna, B. Das, Appl. Phys. A 90 (2008) 299.[29] B. Das, A.N. Banerjee, Nanotechnology 18 (2007) 445202.[30] C. Paturaud, G. Farges, M.C. Sainte Catherine, J. Machet, Surf. Coat. Technol. 98

(1998) 1257.[31] M. Stepanova, S.K. Dew, Vac. Sci. Technol. A 19 (2001) 2805.[32] S.D. Ekpe, S.K. Dew, in: D. Depla, S. Mahieu (Eds.), Reactive Sputter Deposition,

Springer, NY, USA, 2008, p. 230.[33] W.D. Westwood, J. Vac. Sci. Technol 15 (1978) 1.[34] W.D. Davis, T.A. Vanderslice, Phys. Rev. 131 (1963) 219.[35] L.I. Maissel, R. Glang, Handbook of Thin Film Technology, McGraw-Hill, New York,

1970.[36] Y. Kudriavtsev, A. Villegas, A. Godines, R. Asomoza, Appl. Surf. Sci. 239 (2005) 273.[37] Y. Matsunami, Y. Yamamura, Y. Itikawa, N. Itoh, Y. Kazumata, S. Miyagawa, Energy

Dependence of the Yields of Ion-induced Sputtering of Monatomic Solids, Insti-tute of Plasma Physics, Nagoya University, Japan, 1983.

[38] G. Falcone, Riv. Nuovo Cimento 13 (1990) 1.[39] A.N. Banerjee, B.K. Min, S.W. Joo, Appl. Surf. Sci. 268 (2013) 588.[40] E.A. Belenkov, A.I. Sheinkman, E.A. Belenkov, A.I. Sheinkman, Modeling the

Graphitization of Amorphous Carbon, Chelyabinsk State University, IzvestiyaVysshikh Uchebnykh Zavendenii, Fizika, 10, Plenum Publishing Co., 1991, p. 67.

[41] V.S. Ostrovskii, Artificial Graphite, Metallurgiya, Moscow, 1986.[42] A.I. Kitaigorodskii, Molecular Crystals, Nauka, Moscow, 1971.[43] V.G. Nagornyi, Carbon Structural Materials, Metallurgiya, Moscow, 1985. 68.[44] J. Maire, J. Mering, Chemistry and Physics of Carbon, vol. 6, Dekker, New York,

1970, p. 125.[45] P.J.F. Harris, A. Burian, S. Duber, Philos. Mag. Lett. 80 (2000) 381.[46] M. Acharya, M.S. Strano, J.P. Mathews, S.J.L. Billinge, V. Petkov, S. Subramoney,

H.C. Foley, Philos. Mag. B 79 (1999) 1499.[47] V. Petkov, R.G. Difrancesco, S.J.L. Billinge, M. Acharya, F.C. Foley, Philos. Mag. B 79

(1999) 1519.[48] S.J. Townsend, T.J. Lenosky, D.A. Muller, C.S. Nichols, V. Elser, Phys. Rev. Lett. 69

(1992) 921.[49] M. Molina-Sabio, F. Rodrıguez-Reinoso, Colloids Surf. A Physicochem. Eng. Asp.

241 (2004) 15.[50] A.C. Ferrari, J. Robertson, Phys. Rev. B 64 (2001) 075414.[51] A.N. Obraztov, E.A. Obraztova, A.V. Tyurnina, A.A. Zolotukhin, Carbon 45 (2007)

2017.[52] A.C. Ferrari, J.C. Meyer, V. Scardaci, C. Casiraghi, M. Lazzeri, F. Mauri, S. Piscanec, D.

Jiang, K.S. Novoselov, S. Roth, A.K. Geim, Phys. Rev. Lett. 97 (2006) 187401.[53] P.R. Somani, P.S. Somani, M. Umeno, Chem. Phys. Lett. 430 (2006) 56.[54] Z. Sun, Z. Yan, J. Yao, E. Beitler, Y. Zhu, J.M. Tour, Nature 468 (2010) 549.[55] J.S. Im, S.S. Jang, Y.-S. Lee, Bull. Korean Chem. Soc. 33 (2012) 3033.[56] J.S. Im, S.C. Kang, B.C. Bai, T.S. Bae, S.J. In, E. Jeong, Y.S. Lee, Carbon 49 (2011) 2235.

0.0 4.0x105

8.0x105

-0.6

-0.4

-0.2

0.0

0.2

(c)

(b)

(a)

(R-R

0)/R

0

Time (ms)

(a) Reference (untreated) sample

(b) 4.0 kV/50 mA cm-2/1.0 min

(c) 4.0 kV/50 mA cm -2/3.0 min

Fig. 7. Electrical resistance measurements of the graphitic microporous carbonfilms during hydrogen adsorption: (a) reference (untreated) sample, (b) preparedat sputtering voltage/current (density)/time of (b) 4.0 kV/50 mA cm−2/1.0 min,and (c) 4.0 kV/50 mA cm−2/3.0 min.


Ambient-temperature fabrication of microporous carbon terminated with graphene walls by sputtering...

Documents

Transcript of Ambient-temperature fabrication of microporous carbon terminated with graphene walls by sputtering...