Ordering and manipulation of MoS, platelets on differently charged micas by atomic force microscopy Suzanne Mulley," Angelo Sironi," Adriana De Stefanisb and Anthony A. G. Tomlinson*b "Dipartimento di Chimica Strutturale e Stereochimica Inorganica, Universitci degli Studi di Milano, Via G. Venezian, 21, 20133 Milan, Italy bIstituto di Chimica dei Materiali, Area della Ricerca di Roma del C.N.R., C.P. 10 Monterotondo Staz., 00016 Rome, Italy Atomic force microscopy (AFM) has been used to investigate how dispersed MoS, platelets (from colloidal suspensions) deposit onto three natural (Muscovite) mica (001) surfaces. Different platelet arrangements are observed, which are attributed to defects and charging effects of the mica (as well as the concentration of starting colloid).The provenance of the mica influences the self- organisation of the platelets into long tape-like assemblies (South Dakota mica) or individual flakes (Alps mica). Atomic scale imaging of the tapes reveals a distorted octahedral (Oh)-based local structure, different from the trigonal prismatic structure found in the 2H-polytype of annealed MoS,, in agreement with previous structural results on water-dispersed MoS, platelets. The buckling and susceptibility to stripping of the tapes is ascribed to the presence of a water layer between the substrate mica and MoS,, and after stripping by the tip, the platelets ultimately form small clusters. The ordering of these clusters depends not only on the defect and charge structure of the mica, but also on complex hydration reactions between the H,O layer associated with the MoS, and K+ ions of the mica.Relatively symmetrical squares may be lifted out of the tapes, supporting the presence of weak bonding between tape and mica. Conversely, Alps and Bihar micas give rise to separate platelets, which are resistent to tip manipulation, which is attributed to hydrophobic interaction between MoS, platelets and the mica surface. Much recent work on the interaction between AFM tips and surfaces has shown that it is possible not only to probe assembly, topology and atomic structure' but also to manipu- late a surface nanometrically (dig holes in it, remove specific areas, and even cut a line into a substrate to form a nano- gate)., During work on the assembly of MoS, platelets from colloidal suspensions on quartz, semiconductor and (001 ) mica surface^,^ it was found that they adopted tape-like assemblies on the mica, presumably directed by the charged surface of the substrate mica.We have now found that such tapes are sensitive to manipulation by the AFM tips (a phenomenon not previously observed) and also to the type of mica. In the interim, AFM investigation has revealed the atomic structure variations on mica surfaces with differing cation exchange and ~alcination.~We here describe the strong influence that charges on mica have on the ordering and morphology of small MoS, platelets and the resultant surface topographies produced by their interactions with AFM tips. Experimental As reported previ~usly,~ the platelets were obtained by simple casting of colloidal MoS,, prepared by 'blowing apart' Li,MoS, via forced hydration under argon.' A drop of colloid solution was placed on freshly cleaved mica and allowed to dry, and film purity (i.e.absence of Li') was assessed by optical spectro~copy.~ The samples of mica used as substrates originated from three different localities: South Dakota (ident- ical to that used previously3) the Alps, and a Ruby mica from Bihar, India. AFM measurements were carried out in air using a commer- cial microscope (Digital Instruments, Nanoscope 111).Images were made by scanning the samples with Si3N, cantilevers having integrated oxide sharpened pyramidal tips with spring constants of 0.12 and 0.58 N m-l.The microscope was oper- ated in contact mode and the contact forces between the tip and the surface were calculated to be no greater that 30nN (in many cases much less). A scanner with maximum oper- ational area 15 x 15 pm2 was used for both topological and zoomed scans. Samples were investigated both immediately after preparation and a few days later. Energy-dispersive analysis by X-rays (EDAX) studies were carried out on a Hitachi S2400 scanning electron microscope (SEM) fitted with a Kevex EDAX analyser. All three mica substrates were found to have identical chemical compositions, and no Mg2+ or other cations were detected. Results and Discussion South Dakota mica Of the three mica substrates investigated, only that from South Dakota produced the distinct tape-like features shown in Plate 1.The tapes, evidently formed from ordered MoS, platelets, were found to have a very uniform thickness (ca. 3.5-4.0 nm) with widths typically around 500-600 nm (although occasionally up to 2 or 3 pm) and lengths ranging from 3 to 10 pm. Some structures (reminiscent of the ribbon- like features recently observed for linear chalcogenides6) on the top of these tapes provided evidence that they are composite in nature. This was further verified by the existence of one or two particularly flat ribbons, such as that shown in Plate l(b), in which holes corresponding to 'missing' platelets and bumps corresponding to stacked platelets, are clearly visible.The average thickness of these flat tapes was found to be ca. 0.65 nm which is in keeping with the thickness of a single layer of MoS, (0.615 nm). Despite zooming in on a large number of wide-scale images, joins between platelets were not observed, although smaller groups of platelets, such as those shown in Plate 2(a), did give a hint of their agglomerating nature. When the same colloidal suspension was diluted tenfold, a different dispersion on this mica was found; platelets grouped together to form flat discs up to 1.0 pm across [see Plate 2(b)]. We originally attributed this type of dispersion to individual drops of colloid solution breaking up into many tiny micro- droplets during sample preparation caused by the charged surface of the mica ~ubstrate.~ Another, perhaps more plausible, J.Mater. Chem., 1996, 6(4), 661-666 661 Plate 1 (a) Large scale micrograph of MoS, colloidal platelets on (001) mica (image 5.0 x 5.0 pm). Note the tape-like development. (b)Thin ribbon (average height ca. 0.65nm, i.e. corresponding to single platelet thickness) running diagonally from top left to bottom right of image showing holes and bumps corresponding to missing and extra platelets, respectively (image 2.5 x 2.5 pm) Average height of surrounding ribbons is 3.5-4.0 nm. Plate 2 (a) Small scan of MoS, tape showing joins between agglomerated platelets (image 820 x 820 nm). (b)Dilute MoS, platelet dispersion aggregating at point defects on the mica surface (image 7.0 x 7.0 pm).The roughly circular features have uniform thickness of ca. 1.3 nm and are surrounded by a thinner halo; the centres of the circular features rise into small points ca. 10 nm in height. (c) Profile showing the ‘flatness’ and height of a typical flat disc of MoS, platelets. 662 J. Muter. Chem., 1996, 6(4), 661-666 Plate 3 Atomic scale images of MoS, tape. (a) Raw data (10 x 10 nm); (b)close-up (2.56 x 2.56 nm) of the same, filtered image. The image was filtered in two different ways; one by choosing appropriate masks to cover all the points in the power spectrum, the second by choosing two vectors in reciprocal space and fitting them via an iterative method to all the points in the power spectrum (36 significant values). The two methods produced virtually identical images, the apparent extra structure was therefore deemed to be real.explanation could be that the mica surface contains many point defects which act as seeding centres for agglomerating platelets. Nevertheless, the platelet thickness was again found to be very uniform even in this dispersion: ca. 1.3-1.5 nm in general, although when small groups of only two or three platelets were observed, they have a minimum thickness of ca. 0.6-0.7 nm. These changes imply directing of the platelets by the mica su~face,~and evidence to support this suggestion comes from atomic scale images of very thin parts of the tape (taken at low contact forces before tape break-up occurred). Plate 3 shows a Fourier filtered image of the surface, which appears to be corrugated with an almost 2a0 x 2a0 unit cell (ao is the lattice spacing of the 2H-MoS2 structure found on annealed plat$ets3).Analysis gives S-S bond lengths of ca. 2.5 and 3.7 A, in keeping with a distorted pseudo-octahedral based local structure and in good agreement with those found in XRPD and EXAFS studies for the MoS, platelets dispersed in water.’ Such a distorted structure is thought to arise from the presence of water bilayers associated with the MoS, layer, although the corrugation found previously is smaller than that found here, and a 2u, x 1 unit cell was deduced for the metastable MoS, layer.6 We suggest that both effects are caused by coordination of the layer-associated water to K + ions exchanged on the mica surface.The MoS, tapes on this mica were found to be exceedingly sensitive to the AFM tip, persistent scanning of a particular area generally resulting in the complete break-up of the material. Plate 4 shows a consecutive sequence of images of a single tape repeatedly swept by the tip (at an applied force of ca. 30 nN). The tape structure breaks up, initially into strips, most of the stripped material being swept to the sides of the area scanned [see Plate4(f)]. Close inspection of the final swept area shows that small, almost circular, ‘nanohenges’ are present Plate 4(d),(e)].The diameter of a henge is typically ca. 15 nm with each component ‘stone’ having an average height of 0.7 nm and comparable width. Although atomic-level analy- sis was not feasible, each ‘stone’ feature is suggestive of the presence of a cluster (There are no recorded instances of tips giving rise to such features on mica itself and removal of MoS, from a 2H-MoS2 crystal surface via STM rastering gives initially triangular features and then removal of a layer.)8 We speculate that the ‘stones’ may be Mo-S clusters (which are legion)’ and, significantly their heights are close to those of the most ubiquitous and stable Mo-S building block, [Mo~S,,]~-,which in bulk, complex inorganic chemistry form readily in the presence of K+ ions in aqueous solution.” Similar topologies were found after repeated ribbon removal over the sample surface, and in no case were other types of features (curled up, nanotubules, etc.) observed.This suggests that some platelet-surface interactions remain throughout the process. Also visible in Plate 4(e)is what appears to be a step with a height of 0.25-0.3 nm. This is much less than that of a single mica layer (ca. 1.0 nm) so it cannot be attributed to a growth plane of the substrate material. It also cannot be attributed to cations other than K+ (e.g. Mg2+) which would give rise to a step generated by the difference in cationic radii. The step appears exactly where one of the ribbon edges was, indicating that it plays an important role in the original tape growth in that particular area, which also provides clues about the underlying surface chemistry in operation. We recall that: (i) Hartmann et a!.” could not locate interlayer cations in 2 :1 clay minerals (e.g.montmorillonite, illite, having a basal struc- ture analogous to micas); (ii) according to Nishimura et ~l.,~ a similar observation for K+-exchanged Ruby mica under water can be ascribed to hydration, ‘keying-out’ of the K+ from the hexagonal hole and removal by tip (unlike Lif and Mg2+, K’ does not irreversibly ‘fix’ into the hexagonal 0-cavity of mica’,); (iii) in water, K+ ions readily react with the S atoms of MoS, (giving MoS2).l3 This suggests a simple model for the interaction between MoS, platelets and South Dakota mica. Hydrated platelets dock with hydration of the surface mica K +,the tip then ‘keys out’ (now larger) hydrated Kf ions, at the same time breaking J.Muter. Chem., 1996, 6(4),661-666 663 Plate 4 (a),(b),(c): Consecutive scans over a section of tape illustrating progressive destruction (image 2.7 x 2.7 pm). Note the area in the top right hand corner of the images; initially this appears to be one of the thin ribbons, whilst progressive sweeping of the area leaves a number of clusters (‘henges’). (d) Area left after sweeping clean with AFM tip (image 520 x 520 nm). The clusters appear to be sitting in a shallow dip in the mica (step height measured as ca. 0.3 nm). (e) Close-up showing step in mica surface and henges (image 200 x 220 nm). (f)Large scale scan after sweeping operation (image 15.0x 15.0 pm). The central part of the image clearly shows tape material has been deposited at the sides of the smaller scan area during the sweeping.-b Scheme 1 Suggested mechanism of removal of material by the tip, to leave an area of keyed-out hydrated K+ ions on the mica basal plane. weak axial Mo...S bonds of the metastable Oh-MoS2, as shown in Scheme 1. Reaction then occurs between hydrated K+ and [ MoS,]” fragments. Although speculative (AFM provides no chemical analytical information), the model fits the results (and is also in agreement with the completely flat 2H-MoS2 observed after ~alcination,~ which does not undergo any of these processes). In addition, and more convincingly, this tip/platelet/mica reaction can be used to systematically lift out pieces, as shown in Plate 5. This operation was accomplished by zooming in to 20 nm scale from 200 nm scale, applying a force to the tip and then zooming back.The square left, of dimensions ca. 50 x 50 nm, is presumably owing to a platelet having been removed, facilitated by the only weak attachment to the mica. There is a pronounced ‘tail’ to the hole left by platelet removal shown in Plate 5. This is suggestive of the tip point of attachment pressing down the lower platelet, making it adhere more strongly to the surface, and in turn being left behind after lifting*Note that an AFM tip taking Plate 5 Small Scan ca. 50 x 50 nm produces a hole in the tape (image Part in a chemical reaction, (hydrogen-bond breakage) has 150 x 150 nm). Note that the hole is not cubic, indicating that the hole recently been suggested.14 is formed uia removal of more than one platelet.664 J. Mater. Chem., 1996, 6(4), 661-666 Alps and Bihar micas Although substrate-directing effects were observed on the other two mica samples, compact tapes were not generated. On Alps mica, the platelets agglomerated to form flakes up to several microns in diameter (Plate 6) a morphology comparable to that described previ~usly,~ except that here the single platelets combining to form each flake are clearly distinguishable. Minimum platelet thicknesses measured lie in the range 1.2-1.5 nm. Platelets in this type of assembly were much more stable to tip manipulation than the tapes, indeed, repeated scanning over small areas and increasing the loading on the tip caused no visible damage to the material.In contrast, Bihar mica yielded a platelet distribution some- where between the other two micas. Here the platelets aggre- gated on the surface of the substrate in long bands similar to the tape dispersion, but the joins between single platelets could still be distinguished (Plate 7). Moreover, distinct layers of platelets (minimum thickness ca. 0.7 nm) are easily discernible, confirming our previous suggestion that MoS, platelets deposit onto a mica surface via a side-on parking mechanism. Again, the platelet dispersion was much less sensitive to tip manipulations. Plate6 Flake morphology of MoS, platelets on Alps mica (image 90Ox900nm). Each single plate is easily identified. Brighter areas indicate at least two platelets stacked one uuon the other.Plate7 MoSz platelets deposited in layers on Bihar mica (image 2.0 x 2.0 pm). The difference between platelet distributions on the different micas is not obvious; -all three show the same chemical composition (from EDAX measurements) and have the authi- genic 1M polytype structure.14 Hence, invoking differences in the ordering of K+ ions between the top and bottom surfaces of the cleave owing to inherent non-commensurability is unconvincing. Further, although having a directional effect on platelet deposition [Plates 2(b) and 61, surface defect phen- omena alone do not appear to be responsible. Also, the relative closeness of platelet approach seems to be determined by the particular mica substrate rather than the colloid preparation; platelets from the same MoS, preparation aggregate closer on Bihar mica than on Alps mica.(Preliminary results on a saponite clay, having platelets of dimensions similar to those of MoS,, show that on South Dakota mica they form the same compact ribbon-type films as found with MoS,.”). Extreme tip sensitivity of the tapes on South Dakota mica implies less strongly attached platelets (both to each other and to mica). In turn, reduced tip sensitivity to damage in Alps and Bihar mica implies that MoS, platelets dock without an associated water layer. Presumably, differences in water organ- isation (mono-, bilayer, etc.), which in turn depend on changes in interstratification immediately before MoS, docking, are responsible for the more ‘hydrophobic’ tip response.This would explain the different minimum measured platelet heights of ca. 0.7 and 1.4 nm on these two micas. It is notoriously difficult to follow the water involved in interstrati-fication ordering,16 and further work is underway on modified platelets. Conclusions We have shown that the provenance of mica used as a ‘flat support’ for AFM may, in fact, condition a complex micro aqueous chemistry, which for MoS, may be utilised to cut and manipulate platelet pieces at the nm-scale. However, the platelet ordering immediately before deposition has such a large influence on surface docking that it is unlikely that tip manipulation on water-containing species can lead to rational nano-engineering.we thank Dr. Moret for his s* is grateful to the HCM programme for a Fellowship and A.A.G.T. the BRITE-Euram programme of the E. C. (contract No. BRE2-CT93-0450) for their continuing financial support. References 1 G. Binnig, Ultramicroscopy, 1992, 42-44, 7; S. N. Magonov, Appl. Spectrosc. Rev., 1993,28, 1. 2 T. Schimmel, B. Winzer, R. Kemnitzer, T. Koch, J. Kuppers, M. Schoerer, C. M. Lieber and Y. Kim, Adv. Muter., 1993, 5, 392; E. Gardner, Science, 1994,206,543. 3 S. Foglia, A. A. G. Tomlinson, A. Sironi and S. Mulley, J. Muter. Chem., 1995,5,1191. 4 S. Nishimura, S. Biggs, P. J. Scales, T. W. Healy, K. Tsuematsu and T. Tateyama, Lungmuir, 1994, 10,4554. 5 P. Joensen, R. F. Frindt and S. R. Morrison, Muter.Res. Bull., 1986, 13, 487. 6 W. Liang, M. H. Whangbo, M. Evain, L. Monconduit, R. Brec, H. Bengel, H-J. Cantow and S. N. Magonov, Chem. Muter., 1994, 6, 678. 7 L. D. Yang, S. Jimenez-Sandoval, W. M. R. Divigalpitaya, J. C. Irwin and R. F. Frindt, Phys. Reu. B, 1991,43,1200. 8 B. Parkinson, in Suprumoleculur Architecture, ed. T. M. Bein, ACS Symp. Ser., vol. 499, Washington, 1992, p. 76; B. Parkinson, J. Am. Chem. SOC., 1990,112,7498. J. Muter. Chem., 1996, 6(4), 661-666 665 9 Eg D Coucouvanis, A Tompakis, S-M Koo and 13 J Rouxel, in Supramolecular Architecture, ed T M Bein, ACS A Hadjikyriacou, Polyhedron, 1989, 8, 1705 and refs therein, G-X Lin, A Muller and M Penk, 2 Naturforsch, B Chem Sci , 14 Symp Ser ,vol 499, Washington, 1992, p 88 G Bar, B Scott, S R Johnson, B I Swanson, J Ren and 1991,46,25 and refs therein M-H Whangbo, Chem Muter, 1995,7,391 10 W Clegg, N Mohan, A Muller, W Rittner and G M Sheldrick, 15 A De Stefanis, S Mulley and A A G Tomlinson, Clays Clay 11 Inorg Chem , 1980,19,2066 H Hartmann, G Sposito, A Young, S Manne, S A C Gould and P K Hansma, Clay Clay Miner, 1990,38,337 16 Miner, 1996, submitted Eg P H Nadeau, M J Wilson, W J Hardy and J M Tait, Science, 1984,225,923 12 G W Brindley and G M Brown, Crystal Structures of Clay Minerals and their X-ray IdentlJication, The Mineralogical Society, London, 1980, N Guven, 2 Kristallogr ,1971, 134, 196 Paper 5/04313A, Received 4th July, 1995 666 J Muter Chem, 1996, 6(4),661-666