The formation of dimensionally ordered germanium nanowires within mesoporous silica

The formation of dimensionally ordered germanium nanowires within mesoporous silica

27 July 2001 Chemical Physics Letters 343 (2001) 1±6 www.elsevier.com/locate/cplett The formation of dimensionally ordered germanium nanowires with...

171KB Sizes 4 Downloads 49 Views

27 July 2001

Chemical Physics Letters 343 (2001) 1±6

www.elsevier.com/locate/cplett

The formation of dimensionally ordered germanium nanowires within mesoporous silica N.R.B. Coleman a, K.M. Ryan a, T.R. Spalding a, J.D. Holmes b, M.A. Morris a,* b

a Dimensional Solids Research Group, Department of Chemistry, University College Cork, Cork, Ireland Supercritical Fluid Centre, Material Chemistry Section, Department of Chemistry, University College Cork, Cork, Ireland

Received 9 April 2001; in ®nal form 29 May 2001

Abstract Herein is described a supercritical ¯uid solution-phase method for the synthesis of high aspect ratio nanowires of true nanometre diameter within the pores of an ordered mesoporous material. Using powder X-ray di€raction (PXRD), 29 Si MAS±NMR and transmission electron microscopy (TEM), it is possible to show that the quantum con®ned nanowires formed partially ®ll the pores of the mesoporous solid and are orientated so that the h1 0 0i plane of the wires runs parallel to the pore direction. We also show that the wires have an expected diamond-like structure although there is a measurable lattice expansion compared to bulk forms of germanium. Ó 2001 Elsevier Science B.V. All rights reserved.

Very recently, we described a supercritical ¯uid solution-phase method whereby ordered nanowires of silicon could be formed within the pores of a silica mesoporous material [1,2]. In order to show that the method used in the previous work is generally applicable to the formation of other materials, we have extended our preparative method to grow germanium nanowires within a mesoporous silica matrix. The greater scattering power of germanium atoms, compared to silicon atoms, allowed details of the wire crystallography and in particular the orientation of the wires within the porous material to be readily determined by powder X-ray di€raction (PXRD). Brie¯y, hexagonal mesoporous silica was prepared using a polyethylene oxide (PEO)±poly-

*

Corresponding author. Fax: +353-21-4274097. E-mail addresses: [email protected] (J.D. Holmes), [email protected] (M.A. Morris).

propylene (PPO) triblock copolymer surfactant …PEO26 PPO39 PEO26 † available from Uniquema, Belgium and referred to as Synperonic PE/P85 [1,2]. The resultant solid, indicated here as P85 mesopore, was shown by SANS, PXRD and transmission electron microscopy (TEM) to have a pore-to-pore distance of 8 nm and a pore size of around 5 nm. Germanium nanowires were grown by the decomposition of diphenylgermane (0.011 mol) in hexane at 773 K and 375 bar in the presence of 0.5 g of the P85 mesopore for 15 min. The white mesoporous material changed colour from white to silvery-black during the course of the pore-®lling reaction. No colour change was observed in the absence of diphenylgermane. After the reaction had ®nished, the contents of the cell were washed with hexane and the relatively large (1 mm dimensions) particles of the mesoporous silica incorporating germanium were manually extracted and dried for analysis. 29 Si MAS±NMR data were collected on a Chemagnetics CMX-lite

0009-2614/01/$ - see front matter Ó 2001 Elsevier Science B.V. All rights reserved. PII: S 0 0 0 9 - 2 6 1 4 ( 0 1 ) 0 0 6 4 7 - 9

2

N.R.B. Coleman et al. / Chemical Physics Letters 343 (2001) 1±6

300 MHz machine. Cross-polarisation was not used because of very long relaxation times. Pulse widths of 4 ls were used corresponding to ¯ip angles of about 70 deg and pulse delay times of at least 60 s were used. Spin speeds of 5 kHz were used routinely. PXRD data were collected on a Philips MPD system using 1/8 mm Stoller slits. CuKa radiation from an anode run at 40 kV and 40 mA was used in all experiments. Peak positions (2-theta values) are referenced to data from a powdered silicon single crystal sample with a  TEM data were colh1 1 1i re¯ection at 3.137 A. lected on a JEOL 100EX electron microscope with an acceleration voltage of 80 kV. 29 Si MAS±NMR provides evidence for ®lling of the pores. The P85 mesoporous matrix displays spectra typi®ed in Fig. 1a which are similar to data previously described for mesoporous silicon nanowire formation [2]. Two resonances can be resolved by curve ®tting as shown in the ®gure. The ®rst of these at )110.9 ppm is due to Q4 …Si…OSi†4 † sites and the second at )103.0 ppm due to Q3 …Si…OSi†3 …OH†† sites [1]. It is, thus, relatively simple to assign the Q4 site to the silica species within the cell wall between pores (refer to bulk silica) and Q3 to sites at the wall (surface silica) which have a terminal ±OH linkage. Curve®tting allowed the Si wall …Q3 † to Si bulk …Q4 † ratio to be measured as 0.40 which is consistent with a pore volume to total volume of 30% [1]. After the deposition of germanium in the pores the 29 Si MAS±NMR data suggest that complete pore ®lling is not achieved as a Q3 contribution can still be resolved (Fig. 1b). Curve ®tting does reveal that the Si wall …Q3 † to Si bulk …Q4 † ratio has decreased from the original value of 0.40 to 0.24. Since PXRD suggests that no change in mesopore structure has occurred (see below) it can be concluded that around 40% of the pores have been ®lled. PXRD analysis of the product reveals the presence of metallic germanium. Data are shown in Fig. 2. A background and Ka2 subtracted pro®le is also shown for convenience. Strong peaks are observed at 27.08, 45.13, 53.49 and 68.23 deg (2theta) which are equivalent to d-spacings of 3.2822 (3.2663, h1 1 1i re¯ection), 2.009 (2.000, h2 2 0i),  (1.4143 A,  1.7113 (1.7057 h3 1 1i) and 1.4214 A

Fig. 1. (a) Curve-®tted 29 Si MAS±NMR spectra of P85 mesopore. Curve-®tting was based on least squares minimisation of residuals and peak shapes were 70% Gaussian and 30% Lorenzian. (b) Curve-®tted 29 Si MAS±NMR spectra of the P85 mesopore with germanium nanowires. The peak shapes and widths were transferred from those found in Fig. 1a.

h4 0 0i) typical of metallic germanium and the values in brackets indicate expected peak positions and re¯ection assignments from previous work [3]. It can be seen that there is a slight increase in the dspacings of the con®ned germanium nanowires, relative to the accepted bulk values for germanium, corresponding to a lattice expansion from  [4]. As well as these germanium 5.6574 to 5.6856 A (diamond structure) derived re¯ections, weaker re¯ections can also be resolved particularly at the low angle side of the main h1 1 1i re¯ection and  can be depeak positions at 3.477 and 3.386 A termined. These have been assigned to the presence of germanium oxide (t-quartz structure d ˆ 3:430  [5]) and graphite (3.376 A  [6]). Although other A re¯ections are dicult to observe the expected

N.R.B. Coleman et al. / Chemical Physics Letters 343 (2001) 1±6

3

Fig. 2. PXRD data showing (a) experimental PXRD data from the mesoporous germanium nanowires. The marked lines are peak positions from reference ®les indicated in the text. (b) A baseline-subtracted pro®le showing the main re¯ections of carbon as graphite and GeO2 (t-quartz structure). Also shown are insets of expanded plots of the h3 1 1i and h4 0 0i re¯ections (left and right of the plot, respectively). The (3 1 1) and (4 0 0) data have been smoothed using a seven-point polynomial ®t to the data.

positions of all three materials are shown as solid lines in Fig. 2 and the agreement with each material is excellent across the 2-theta range. The extremely broad peaks (>20 deg) shown in Fig. 2a are simple due to amorphous type scattering from the host matrix. This scattering has been stripped out of Fig. 2b for convenience. Estimates of the

amount of graphite and GeO2 present are at about 3% and 5%, respectively, as estimated by Rietveld ®tting of the data. Because the GeO2 features observed are relatively sharp they must be associated with a small amount of bulk oxidation process occurring on air exposure rather than as a result of direct bonding of germanium atoms to the wall

4

N.R.B. Coleman et al. / Chemical Physics Letters 343 (2001) 1±6

±OSi units as this very thin interfacial structure would not be observable using PXRD. The observation of a small lattice expansion of the nanowire germanium should be noted. Lattice size changes are expected for small-dimensioned particles. It is well accepted that contraction of lattice dimensions can occur because of hydrostatic pressures and surface tension that result from their ®nite size [7]. However, opposed to what might be expected, lattice expansion in Si and SiGe quantum dots has been observed and is thought to be due to surface strains and lattice mismatch [8]. Whilst it is clear that more work is required before de®nitive discussion can be made, it can be argued on the basis of the results reported here that the contraction of unit cell parameters expected on the basis of surface tension and similar e€ects can be ignored. This is because the particles essentially ®ll the pores and in the direction orthogonal to the wire direction (i.e., the long wire direction) contact is made with the pore wall and negligible hydrostatic pressures are developed. Thus, only lattice expansions might be expected for this type of nanowires and these probably result from the presence of longer bond distances due to incomplete band formation [9]. The background subtracted PXRD data also provide quanti®able data related to the shape of the peaks for each individual re¯ection. It is seen that the h4 0 0i re¯ection peak is very sharp whilst the h3 1 1i feature, for example, seems to show a sharper feature above a broader base. The existence of either asymmetric or multiple peak structures existing within a single peak envelope for a re¯ection is conventionally explained as being due to lattice strains [10]. However, this is

considered unlikely for the materials prepared here as curve-®tting shows the individual peaks to be highly symmetrical about almost common axis. Further, the shape and width of the broader component have a form closely related to those expected from nanoparticles [11]. Instead the data are interpreted as being related to the fact that two e€ective particle sizes exist within a wire; those in the long direction or parallel to the wire and those perpendicular to the wire direction. Using the peak ®t derived widths the Scherrer formula can be used to estimate these e€ective particle sizes. These results are shown in Table 1. It can be seen that the wires are orientated in a h1 0 0i direction which is the largest particle size that can be determined. The data suggest that the wires have lengths of the order of 500 nm. It should be noted that Scherrer formulism cannot be used to accurately measure dimensions of this size and the number is best viewed as the best approximation. The derived particle sizes for the other re¯ections are what might be expected from simple crystallographic consideration of the data; the h2 2 0i directions would be the shortest and the h3 1 1i direction the longest in terms of planes running across the width of the wire (obviously there are orthogonal directions which are longer and give rise to the narrow component of the peak envelope for each of these re¯ections). If the h2 2 0i data are converted to a wire width the thickness of the wire can be estimated as 6.29 nm which is close to the 5 nm pore size of the mesoporous material estimated by TEM [1,2]. This analysis supports the argument that wires have been formed within the pore structure of the host matrix.

Table 1 Summary of peak ®t information and Scherrer calculations of e€ective thickness from the data shown in Fig. 2 2-Theta angle (deg)

Corrected width (deg)

Derived size (nm)

Crystal plane

27.08 45.13 53.49 68.23

0.855 1.016 0.907 0.020

10.2 8.9 10.4 479.7

h1 1 1i h2 2 0i h3 1 1i h4 0 0i

The corrected widths of peak re¯ections are experimental measurements of the broad component of a re¯ection the curves being ®tted by a least square-®t method using Gaussian peak shapes. A broad component cannot be seen for the h4 0 0i probably due to poor background signal to noise for this low-intensity feature re¯ection and is consequently observed as an extremely narrow feature. Instrumental resolution has been estimated from a powdered sample of single crystal silicon.

N.R.B. Coleman et al. / Chemical Physics Letters 343 (2001) 1±6

Final con®rmation of wire formation can be seen by the TEM image shown in Fig. 3. Here the much greater scattering power of germanium compared to the silica matrix allows germanium wires to be imaged directly. Fig. 3a shows germanium nanowires extending from the edge of a mesoporous silica particle and Fig. 3b shows nanowires and nanorods of germanium within the pores. These wires are parallel and of uniform thickness as constrained by the mesoporous solid. The average width of the wires can be estimated at 6.4 nm in close agreement with that measured by PXRD. Further work continues in relation to the properties of these constrained wires in particular if reaction conditions can be used to de®ne the wire orientation. The question of lattice expansion in these wires must also be addressed. In a previous work on silicon nanowires embedded in mesoporous matrix arrays it was found that the unit cell dimensions were almost exactly as that expected of bulk silicon [1,2]. It might be suggested that the silicon (of the wire) to pore wall silica interaction should result in only low lattice strains since this process is similar to native oxidation whilst in the germanium nanowire case there might be strains induced due to size mismatch of GeO2 and SiO2 . However, this is a very tentative proposal, which requires extensive investigation, and as discussed there may be important electronic e€ects [9] and/or surface relaxation or reconstruction [12]. It is also interesting to note that both PXRD and TEM suggest that inclusion of germanium into the pore structure results in an expansion of the mesoporous material pore diameter. This was not observed for the silicon wire inclusion [1,2]. PXRD would suggest that the pore-to-pore distance is unchanged by the inclusion of germanium (analysis of low angle PXRD peaks which indicate the pore-to-pore distance suggests only a 0.5±1% change in this parameter) and suggest that the silicon walls are being decreased in thickness when germanium is included. Work in these laboratories suggests that the decomposition of some precursor materials can grossly e€ect the silicate structure and evolution of ordered silicates has been observed [10]. Clearly the technique is as yet

5

Fig. 3. TEM data showing (a) mesoporous silica showing growth of germanium nanowires extending from the edge (scale bar ˆ 50 nm) and (b) germanium nanowires/nanorods incorporated inside the pores of the same mesoporous silica sample (scale bar ˆ 50 nm).

only partially understood and a more extensive study is required if the technique is to be considered generally applicable.

6

N.R.B. Coleman et al. / Chemical Physics Letters 343 (2001) 1±6

Acknowledgements The authors gratefully acknowledge the expertise and work of the sta€ of the Electron Microscopy Unit at UCC. Enterprise Ireland is thanked for the ®nancial support through the Strategic Science Scheme and K.R. wishes to thank Intel for the award of a postgraduate scholarship. References [1] N.R.B. Coleman, M.A. Morris, T.R. Spalding, J.D. Holmes, J. Am. Chem. Soc. 123 (2001) 187. [2] N.R.B. Coleman, N. O'Sullivan, K.M. Ryan, T.A. Crowley, M.A. Morris, T.R. Spalding, D.C. Steytler, J.D. Holmes, J. Am. Chem. Soc., in press.

[3] E.T. Lippmaa, M. Magi, A. Samosan, G. Engelhardt, A.R. Grimmer, J. Am. Chem. Soc. 102 (1980) 4889. [4] R.W.G. Wycko€, Crystal Structures, second ed., vol. 1, 1964, 26. [5] G. Smith, P. Isaacs, Acta Cryst. 17 (1964) 842. [6] R.W.G. Wycko€, Crystal Structures, ®rst ed., vol. 1, 1963, 27. [7] M. Dubiel, H. Hofmeister, E. Schurig, E. Wendler, W. Wesch, Nucl. Instr. and Meth. B 166 (2000) 871. [8] W.-X. Ni, J. Burch, Y.S. Tang, K.B. Joelsson, C. Sotomayor-Torres, A. Kvick, G.V. Hansson, Thin Solid Films 294 (1997) 300. [9] S. Furukawa, T. Miyasato, Phys. Rev. B 38 (1988) 5726. [10] L.H. Schwarz, J.B. Cohen, Di€raction from Materials, second ed., Springer-Verlag, New York, 1987. [11] N.R.B. Coleman, M.A. Morris, T.R. Spalding, J.D. Holmes, submitted. [12] See for instance discussions in H. Luth, Surfaces and Interfaces of Solid Materials, third ed., Springer-Verlag, Berlin, 1995.