formation of oriented nanostructures in diamond using...
TRANSCRIPT
Formation of oriented nanostructures in diamond using metallic nanoparticles
This article has been downloaded from IOPscience. Please scroll down to see the full text article.
2012 Nanotechnology 23 455302
(http://iopscience.iop.org/0957-4484/23/45/455302)
Download details:
IP Address: 147.173.65.96
The article was downloaded on 15/04/2013 at 11:01
Please note that terms and conditions apply.
View the table of contents for this issue, or go to the journal homepage for more
Home Search Collections Journals About Contact us My IOPscience
IOP PUBLISHING NANOTECHNOLOGY
Nanotechnology 23 (2012) 455302 (7pp) doi:10.1088/0957-4484/23/45/455302
Formation of oriented nanostructures indiamond using metallic nanoparticles
H-A Mehedi, C Hebert, S Ruffinatto, D Eon, F Omnes and E Gheeraert
Institut Neel, CNRS and Universite Joseph Fourier, F-38042 Grenoble Cedex 9, France
E-mail: [email protected]
Received 9 July 2012, in final form 11 September 2012Published 22 October 2012Online at stacks.iop.org/Nano/23/455302
AbstractA simple, fast and cost-effective etching technique to create oriented nanostructures such aspyramidal and cylindrical shaped nanopores in diamond membranes by self-assembledmetallic nanoparticles is proposed. In this process, a diamond film is annealed with thinmetallic layers in a hydrogen atmosphere. Carbon from the diamond surface is dissolved intonanoparticles generated from the metal film, then evacuated in the form of hydrocarbons and,consequently, the nanoparticles enter the crystal volume. In order to understand and optimizethe etching process, the role of different parameters such as type of catalyst (Ni, Co, Pt, andAu), hydrogen gas, temperature and time of annealing, and microstructure of diamond(polycrystalline and nanocrystalline) were investigated. With this technique, nanopores withlateral sizes in the range of 10–100 nm, and as deep as about 600 nm, in diamond membraneswere produced without any need for a lithography process, which opens the opportunities forfabricating porous diamond membranes for chemical sensing applications.
(Some figures may appear in colour only in the online journal)
1. Introduction
Diamond thin films are now easily produced on large areaswith nanocrystalline or polycrystalline microstructures, andvarious applications are under development that take intoaccount the exceptional properties of diamond [1–4]. Amongthem, chemical sensors are of particular interest, due to thebio-compatibility of diamond, its chemical inertness and widepotential window [5, 6]. In most cases, the device fabricationprocess involves an etching step which is performed almostexclusively by the reactive ion etching (RIE) technique withthe help of lithographic processes [7–9]. However, in thecase of sensors based on nanopores in a membrane [10–12],this etching technique fails to create narrow (10–50 nm indiameter) and long (typically longer than 300 nm) holes inthe diamond membrane due to the limitation of lithographicresolution. Based on reported works [13, 14] on catalyticetching of diamond, we have developed a fast and simpleetching method through the catalytic gasification of carbonby metallic nanoparticles in a hydrogen atmosphere. Inthis process, the etching of diamond proceeds with: (1)the formation of nanoparticles from the initially deposited
metal film, (2) the carbon dissolution into nanoparticlesthrough the breaking of crystallinity at the diamond–metalinterface [15–17] and, finally, (3) the desorption of carbonwith the assistance of a hydrogen atmosphere. Hydrogengets catalytically dissociated into hydrogen atoms by metalnanoparticles and gasifies the dissolved carbon in theparticles, and the amorphous carbon at the interface formsmethane [18] or some CHx complex.
In this study, the key parameters of the etching processwere investigated: type and thickness of the metallic layer; theannealing temperature; the hydrogen pressure; the diamondmicrostructure and the annealing time. We emphasizedthe etching of [100]-oriented diamond planes, since itis shown that metal-assisted diamond etching is slowestfor [111]-oriented planes and that they act as stoppingplanes [13].
2. Methods and materials
The etching technique consists of two steps, (i) the depositionof thin metallic layers on a diamond substrate and (ii) thesubsequent annealing in a hydrogen atmosphere. Metals (Ni,
10957-4484/12/455302+07$33.00 c© 2012 IOP Publishing Ltd Printed in the UK & the USA
Nanotechnology 23 (2012) 455302 H-A Mehedi et al
Figure 1. SEM images of the (100) surface of polycrystalline diamond after 10 min of annealing with a 3 nm film of (a) Ni, (b) Co, (c) Ptand (d) Au, (800 ◦C, 60 Torr H2); the corresponding enlarged images show etch pits with flat sidewall faces and sharply defined edges forNi, Co, Pt and finely dispersed nanoparticles for Au.
Co, Pt and Au) with thickness in the range of 1–10 nm weredeposited on diamond substrates by e-beam evaporation, thethickness being measured by a quartz crystal oscillator. Theannealing step was carried out in a hot-filament-assisted CVDdiamond growth reactor at 750–900 ◦C in 60 Torr flowinghydrogen with 100 sccm (sccm denotes cubic centimeterper minute at STP) for between a few minutes to severalhours. The base pressure of the furnace before annealingwas <10−6 Torr. The temperature was increased at a rateof ∼200 ◦C min−1 and measured with an infrared pyrometerwith a reading precision of ±10 ◦C. After the annealing step,the samples were cooled down for 30 min under a continuoushydrogen flow.
Two types of diamond films were used for theseexperiments, namely polycrystalline diamond with a grainsize of typically 50 µm and nanocrystalline diamond (grainsize 100 nm). Polycrystalline diamond was grown by ElementSix (electrochemical grade and boron doped CVD diamond).Nanocrystalline diamond (undoped) with a thickness of about1.5 µm was grown on silicon substrates in a microwaveplasma CVD reactor (5200 Seki Technotron MPCVDreactor) using the bias-enhanced nucleation method [19]with following parameters: 900 ◦C temperature, 1000 Wmicrowave power, 0.3% methane in hydrogen, and pressureof 30 Torr.
Annealed samples were characterized using a fieldemission gun scanning electron microscope (FEG-SEM;ZEISS Ultraplus) and Raman spectroscopy (HORIBAScientific, Raman spectroscopy system T64000, 514 nmexcitation).
3. Results and discussion
3.1. Effect of type and thickness of metallic layer
In order to study the effect of metal type on the etchingprocess, Ni, Co, Pt, and Au with a thickness of 3 nm wereevaporated on polycrystalline diamond and annealed at 800 ◦Cfor 10 min. Figure 1 shows the SEM images of the (100)surface of annealed polycrystalline diamond.
During annealing, nanoparticles were generated from themetallic layers, prior to etching, in order to minimize thesurface energy. This effect is well known and relates to themelting point lowering effect of thin metal films and hydrogendiffusion into the metal [20–23]. From the SEM observation,we estimated the diameter of the resulting nanoparticles (Ni,Co and Pt) to be approximately fifteen times larger than theinitially evaporated metal layer thickness.
Etching was observed for Ni, Co, and Pt, figures 1(a)–(c),forming etch pits with flat sidewall faces and sharplydefined edges. In contrast, finely dispersed Au nanoparticleswere formed, figure 1(d), and neither etch pits nor nano-channels were apparently found on the diamond surface. Ninanoparticles showed higher etching activity than Co and Pt.
The observed difference in etching behavior between themetals may be caused by the combined effect of the differencein the size and melting point of metallic nanoparticles as wellas carbon solubility into the nanoparticles at the temperaturesin our experiment. The solubilities of carbon into Ni, Coand Pt at 900 ◦C estimated from M–C, metal–carbon, phasediagrams are 1 at.%, 0.8 at.%, and 0.04 at.% respectively [24].On the other hand, unlike the iron-group elements (Ni, Co),the carbon solubility of Au is extremely low (0.018 at.%at 800 ◦C in bulk phase) [25]. It has recently been reportedthat when the Au particle size becomes several tens ofnanometers, it has carbon solubility depending on the natureof the substrate and source of carbon [26]. However, thismodel of carbon solubility in nanosized Au particles wouldnot be similar to the uptake of carbon from diamond surface.
The observation also suggests that the size of the etchedstructures depends on the size of the initially formed metalnanoparticles. Moreover, the size can be adjusted to a certainextent by the metal layer thickness, as shown in figure 2.Nanoholes with a lateral size of 20 ± 5 nm, figure 2(a),and 150 ± 10 nm, figure 2(b), were achieved by etchingnanocrystalline diamond with Ni particles that originatedfrom 1.5 nm to 10 nm Ni films respectively.
3.2. Effect of annealing temperature
To understand the influence of annealing temperature on theetching process, polycrystalline diamond with a 3 nm film
2
Nanotechnology 23 (2012) 455302 H-A Mehedi et al
Figure 2. SEM images of etched nanocrystalline diamond after 30 min of annealing with (a) 1.5 and (b) 10 nm Ni films, (800 ◦C, 60 TorrH2); the inset showing nanoholes with a lateral size of 20 ± 5 nm.
Figure 3. SEM images of a (100) surface of polycrystalline diamond after 30 min annealing at (a) 750 ◦C, (b) 800 ◦C, (c) 850 ◦C and (d)900 ◦C, (3 nm Ni, 60 Torr H2).
of Ni was annealed at 750, 800, 850 and 900 ◦C in 60 TorrH2 atmosphere. Figure 3 shows the SEM images of a (100)surface of polycrystalline diamond after 30 min of annealing.
The observation suggests that etching occurs typicallyin the range of 800–850 ◦C. The absence of etching at750 ◦C, figure 3(a), could be due to the very low solubilityof carbon into Ni at this temperature. On the other hand,the density of etched nanostructures decreased dramaticallyat 850 ◦C, figure 3(c). It seems that, at this temperature,some of the particles are not active, in particular the smallestparticles. The effect is even stronger at 900 ◦C, where noetching was observed at all, instead particles started to formclusters, figure 3(d). We attribute this to the random nucleationand growth of carbon overlayers on the metal particles,poisoning them at higher temperatures, where the rate ofcarbon diffusion to the metal–gas interface was the highest. Ahigh surface concentration of carbon enhances the probabilityof nucleation and the subsequent growth of an inactive carbonlayer.
3.3. Effect of hydrogen pressure
Figure 4 shows SEM micrographs of a (100) surface ofpolycrystalline diamond annealed with a 3 nm Ni film for10 min at 800 ◦C in vacuum, 60 Torr Ar, 60 Torr H2, and 1 atmH2.
Neither etch pits nor nano-channels were apparentlyfound on the diamond surface when there was no hydrogenin the furnace, i.e. under vacuum, figure 4(a), and Aratmosphere, figure 4(b). Moreover, in both cases Ninanoparticles were covered by overlayers, as apparentin the SEM images observed by an in-lens energy andangle selective backscatter detector (EsB), right panels offigures 4(a) and (b). The overlayers were confirmed ascarbon by energy dispersive x-ray analysis (EDX), and mostlikely could be amorphous carbon, since nickel carbidesbecome unstable at temperatures as low as 350 ◦C [27]. Thisobservation indicates that when there is no hydrogen, andthereby no desorption, carbon absorbed by the Ni can leadto the nucleation and growth of inactive carbon layers that
3
Nanotechnology 23 (2012) 455302 H-A Mehedi et al
Figure 4. SEM images of a (100) surface of polycrystalline diamond after 10 min annealing in (a) vacuum, (b) 60 Torr Ar, (c) 60 Torr H2and (d) 1 atm H2, (3 nm Ni, 800 ◦C); the right panels of images (a) and (b) are SEM images observed by an in-lens energy and angleselective backscatter detector (EsB), showing the presence of carbon overlayers.
eventually decrease the catalytic activity of the metal. So,hydrogen is instrumental in the etching process. Moreover,the higher the pressure of hydrogen gas in the furnace, thefaster the diamond is etched, figure 4(d). We attributed thisto the increased concentration of atomic hydrogen, generatedcatalytically, that, indeed, enhances the desorption rate.
3.4. Effect of diamond microstructure
Polycrystalline diamond and nanocrystalline diamond wereannealed with a 3 nm Ni film at 800 ◦C under 60 Torrhydrogen atmosphere. After 30 min of annealing, etching wasobserved on both microstructures of diamond. No particulareffect related to the grain boundaries was observed. Etchpits with an inverted four-sided pyramid were formed on[100]-oriented planes of polycrystalline diamond, figure 5(a),whereas a mixture of pyramidal, hexagonal, and trigonalshaped etch pits were observed on nanocrystalline diamond,figure 5(b). Note that the nanocrystalline diamond sampleexhibits a mixture of [111]-oriented and [100]-orientedplanes, and the etching behavior is different depending on thecrystallographic orientation.
The orientation dependence of diamond etching has beenstudied before using molten Ce [28], Fe particles [29], Co
nanoparticles [30], and molten Ni [13], as catalysts. It hasbeen shown that Ni-assisted diamond etching is slowestfor the [111]-oriented diamond planes and that they act asstopping planes [13]. Since the uptake of carbon atoms fromdiamond by the metal nanoparticles occurs from the weaklybonded atoms, the carbon atoms of the closest packing plane,the {111} planes, are the most stable against the dissolutionof carbon in nanoparticles, as compared to the other crystalplanes [30]. As a result, etch pits walled with four {111} planesare formed on the {100} planes, whereas {111} show etchingpits with equilateral triangles and hexagons [13].
3.5. Change in microstructure during etching
Two diamond samples, [100]-oriented single crystalline(CVD diamond plate, Element Six) and nanocrystallinediamond (1.5 µm thickness, 100 nm grain size), wereannealed with 3 nm Ni for 10 min at 800 ◦C. In order todetect an eventual change in microstructure, such as theappearance of graphitic phases on the surface, both sampleswere then characterized by Raman spectroscopy using 514 nmexcitation, and the spectra were compared with those recordedbefore annealing.
4
Nanotechnology 23 (2012) 455302 H-A Mehedi et al
Figure 5. SEM images of (a) (100) surface of polycrystalline diamond and (b) nanocrystalline diamond after 30 min annealing with 3 nmNi film, (800 ◦C, 60 Torr H2), showing the formation of nanopores with no obvious dependence on the diamond microstructure.
Figure 6. Raman spectra of (a) single crystal diamond ([100]-oriented, CVD diamond, Element Six), (b) nanocrystalline diamond (1.5 µmthickness, 100 nm grain size), showing no change in the diamond structure after the catalytic etching by Ni nanoparticles, (3 nm Ni, 800 ◦C,10 min, 60 Torr H2).
Figure 6(a) shows Raman spectra for a [100]-orienteddiamond crystal recorded before (red curve) and after (bluecurve) annealing. Both spectra show a similar characteristicdiamond line at 1332.4 cm−1, and no graphitic component isobserved before or after treatment. On the other hand, D andG peaks, which lie at 1337 and 1590 cm−1 respectively, wereobserved for the nanocrystalline diamond sample, figure 6(b).The band at 1337 cm−1 is assigned to the sp3-bonded carbonin diamond (D band) and the broad peak around 1590 cm−1
is attributed to the sp2-bonded carbon (G band). The peakintensity ratio of the D band to the G band (ID/IG) is ameasure of the amount of graphitic impurities. The measuredID/IG values of the treated and untreated samples were1.02 and 1.01 respectively. The observation suggests that thediamond structure was maintained for both samples (singlecrystalline and nanocrystalline diamond) after the catalyticetching by Ni nanoparticles.
3.6. Effect of annealing time
Nanocrystalline diamond with a thickness of about 1.5 µmwas annealed with a 3 nm Ni film for 6 h at 800 ◦C. Afterevery 10 min, the cross section of the sample was observedin SEM and the depth of etch pits was roughly estimatedfrom the observation, figure 7(a). Figure 7(b) shows the SEMmicrograph, of the cross section of the etched nanocrystallinediamond after 90 min of annealing, observed by an EsBdetector. The white spots correspond to Ni nanoparticles.
According to figure 7(a), etching proceeded rapidlywithin first 60 min, then slowed down, and eventually wassaturated after 90 min of annealing. After 6 h of annealing, themean depth was estimated to be about 520 nm. The observedreduction and saturation of the etching rate for long processingtimes could be explained by the increase in the distance fromthe top surface down to the metal nanoparticle. As the Ninanoparticle moves deeper into the diamond, away from thesource of molecular hydrogen, the carbon desorption from thenanoparticle becomes less effective, eventually leading to thesaturation of the catalyst with carbon.
4. Conclusion
The careful study of the process parameters shows thatnanopores can be formed in diamond, whatever itsmicrostructure. The process requires a catalyst with highcarbon solubility, such as nickel, in a hydrogen atmosphere ata temperature of 800–850 ◦C. The etching process is stoppedfor catalyst particles located deep inside the crystal volume,probably because of a limited gas exchange between thesurface and the particle, leading to carbon saturation. Afteroptimization of the process parameters, nanopores as longas 600 nm were formed in nanocrystalline diamond, whichwill allow the fabrication of porous diamond membranes forchemical sensor applications.
5
Nanotechnology 23 (2012) 455302 H-A Mehedi et al
Figure 7. (a) Variation of etch depth with annealing time measured for nanocrystalline diamond after 6 h of annealing with 3 nm Ni film,(800 ◦C, 60 Torr H2). Bottom and top line of the box represent the 25th and 75th percentile of each data set. The band and square insiderefer to the median and mean respectively; minimum and maximum values are indicated by the cross. Whiskers refer to mean ± standarddeviation. After every 10 min the cross section of the etched sample was observed by SEM and pore depth was estimated roughly from theSEM image. Note that the Ni particles that were present on the diamond surface after annealing were excluded from the depth measurement.(b) Corresponding SEM image recorded by using an EsB detector after 90 min, revealing Ni nanoparticles embedded into the crystalvolume.
Acknowledgments
The authors would like to thank D Carole and A Vo-Ha foratmospheric pressure experiments. The work was supportedby the ANR 2010-BLAN-812-03 (VLOC) program.
References
[1] Berman R 1979 The Properties of Diamonded J E Field (London: Academic) pp 4–22
[2] El-Dasher B S, Gray J J, Tringe J W, Biener J, Hamza A V,Wild C, Worner E and Koidl P 2006 Crystallographicanisotropy of wear on a polycrystalline diamond surfaceAppl. Phys. Lett. 88 241915
[3] Auciello O, Birrell J, Carlisle J A, Gerbi J E, Xiao X C,Peng B and Espinsoa H D 2004 Materials science andfabrication processes for a new MEMS technology based onultrananocrystalline diamond thin films J. Phys.: Condens.Matter 16 R539
[4] Kriele A, Williams O A, Wolfer M, Brink D,Muller-Serbert W and Nebel C E 2009 Tuneable opticallenses from diamond thin films Appl. Phys. Lett. 95 031905
[5] Lud Q, Steenackers M, Jordan R, Bruno P, Gruen D M,Feulner P, Garrido J A and Stutzmann M 2006 Chemicalgrafting of biphenyl self-assembled monolayers onultrananocrystalline diamond J. Am. Chem. Soc. 128 16884
[6] Nebel C E, Rezek B, Shin D, Uetsuka H and Yang N 2007Diamond for bio-sensor applications J. Phys. D: Appl.Phys. 40 6443
[7] Hausmann B M, Khan M, Zhang Y, Babinec T M,Martinick K, McCutcheon M, Hemmer P R andLonear M 2010 Fabrication of diamond nanowires forquantum information processing applications DiamondRelat. Mater. 19 621–9
[8] Janssen W and Gheeraert E 2011 Dry etching of diamondnanowires using self-organized metal droplet masksDiamond Relat. Mater. 20 389–94
[9] Janssen W, Faby S and Gheeraert E 2011 Bottom–upfabrication of diamond nanowire arrays Diamond Relat.Mater. 20 779–81
[10] Li J L, Gershow M, Stein D, Brandin E andGolovchenko J A 2003 DNA molecules and configurationsin a solid-state nanopore microscope Nature Mater. 2 611
[11] Storm A J, Chen J H, Ling X S, Zandbergen H W andDekker C 2003 Fabrication of solid-state nanopores withsingle-nanometre precision Nature Mater. 2 537
[12] Yamaguchi A, Uejo F, Uchida T, Tanamura Y,Yamashita T and Teramae N 2004 Self-assembly of asilica–surfactant nanocomposite in a porous aluminamembrane Nature Mater. 3 337
[13] Smirnov W, Hees J J, Brink D, Muller-Sebert W, Kriele A,Williams O A and Nebel C E 2010 Anisotropic etching ofdiamond by molten Ni particles Appl. Phys. Lett. 97 073117
[14] Ralchenko V G, Kononenko T V, Pimenov S M,Chernenko N V, Loubnin E N, Armeyev V Y andZlobin A Y 1993 Catalytic interaction of Fe, Ni and Pt withdiamond films: patterning applications Diamond Relat.Mater. 2 904–9
[15] Anton R 2008 On the reaction kinetics of Ni with amorphouscarbon Carbon 46 656
[16] Clifton P H and Evans S 1995 XPS studies ofdiamond/transition–metal interfaces Ind. Diamond Rev.55 26–31
[17] Lander J J, Kern H E and Beach A L 1952 Solubility anddiffusion coefficient of carbon in nickel: reaction rates ofnickel–carbon alloys with barium oxide J. Appl. Phys.23 1305
[18] Takasu Y, Konishi S, Sugimoto W and Murakami Y 2006Catalytic formation of nanochannels in the surface layers ofdiamonds by metal nanoparticles Electrochem. Solid-StateLett. 9 C114
[19] Barrat S, Saada S, Dieguez I and Bauer-Grosse E 1998Diamond deposition by chemical vapor deposition process:study of the bias enhanced nucleation step J. Appl. Phys84 1870
[20] Gladkich N T, Niedermayer R and Spiegel K 1966 Nachweisgroßer Schmelzpunktserniedrigungen bei dunnenMetallschichten Phys. Status Solidi b 15 181
[21] Buffat P and Borel J P 1976 Size effect on the meltingtemperature of gold particles Phys. Rev. A 13 2287
[22] Yang P C, Liu W, Schlesser R, Wolden C A, Davis R F,Prater J T and Sitar Z J 1998 Surface melting in the
6
Nanotechnology 23 (2012) 455302 H-A Mehedi et al
heteroepitaxial nucleation of diamond on Ni J. Cryst.Growth 187 81
[23] Pundt A 2004 Hydrogen in nano-sized metals Adv. Eng. Mater.6 11
[24] Massalski T B, Okamoto H, Subramanian P R andKacprzak L 1990 Binary Alloy Phase Diagrams vol1 (Ohio: ASM International) p 835
[25] Okamoto H and Massalski T B 1984 The Au–C (gold–carbon)system J. Phase. Equil. 5 378–9
[26] Takagi D, Kobayashi Y, Hibino H, Suzuki S andHomma Y 2008 Mechanism of gold–catalyzed carbonmaterial growth Nano Lett. 8 832–5
[27] Schouten F C, Gijzeman O L J and Bootsma G A 1979Interaction of methane with Ni(111) and Ni(100): diffusion
of carbon into nickel through the (100) surface: anAES-LEED study Surf. Sci. 87 1–12
[28] Jin S, Zhu W, Siegrist T, Tiefel T H, Kammlott G W,Graebner J E and McCormack M 1994 Anisotropy indiamond etching with molten cerium Appl. Phys. Lett.65 2675
[29] Sonin V M, Chepurov A I and Fedorov I I 2003 The action ofiron particles at catalyzed hydrogenation of {100} and {110}faces of synthetic diamond Diamond Relat. Mater.12 1559–62
[30] Konishi S, Ohashi T, Sugimoto W and Takasu Y 2006 Effectof crystal plane on the catalytic etching behaviour ofdiamond crystallites by cobalt nanoparticles Chem. Lett.35 1216–7
7