immobilization of apoferritin-templated seeds for si nanowire growth
TRANSCRIPT
DOI: 10.1002/cvde.201006896
Full Paper
Immobilization of Apoferritin-Templated Seeds for SiNanowire Growth**
By Zhang Zhang, Lianbing Zhang,* Stephan Senz, and Mato Knez
Small Au nanoparticles with a confined size are synthesized exclusively inside the protein cavity of apoferritin (Au-apo). A
modified immobilization process is used to decorate a large area of a Si substrate with monodisperse Au-apo. After removing
the protein, the immobilized Au nanoparticles act as catalysts for a subsequent sub-10 nm diameter silicon nanowire (SiNW)
growth by CVD. The immobilization of Au catalysts by linking the biotemplate to a Si substrate provides a possible route to
control the location of the semiconductor NW in advance of the growth.
Keywords: Apoferritin, Gold nanoparticle, Protein immobilization, Si nanowire
1. Introduction
Si nanowires (SiNWs) are considered to be promising
building blocks for applications in future post-CMOS
technologies approaching dimensions on the sub-10 nm
scale. Compared to other growth techniques, CVD has the
advantage of growing SiNWs with a high length-to-diameter
ratio and diminishing parasitic growth on the substrate
surface.[1] In a conventional vapor-liquid-solid (VLS)
growth mechanism,[2,3] smaller catalytic seeds lead to
smaller diameters. A crucial point for the catalyzed growth
by the VLS mechanism is the control of the size and lateral
distribution of the seeds. In free space, Au-catalyzed growth
of SiNWs with diameters below 10 nm in the <110> growth
direction was observed when the seeds were as small as
3 nm.[4] This is considered to be the thermodynamically
allowed minimum diameter for the VLS growth mode,[5]
however an integrated growth of sub-10 nm SiNWs on a Si
substrate with a high density and uniform size distribution is
not easily obtained from densely packed Au nanoclusters.
One specific reason is the Au/Si eutectic droplet formation,
in which larger Au/Si droplets grow at the expense of
smaller ones in their neighborhood (Ostwald ripening). To
avoid this obstacle, a low-temperature, plasma-enhanced
[*] Dr. Z. Zhang,[+] Dr. L. Zhang, Dr. S. Senz, Dr. M. KnezMax Planck Institute ofMicrostructure PhysicsWeinberg 2, 06120Halle(Germany)E-mail: [email protected]
[+] Present address: School of Physics & Telecommunication Engineering,South China Normal University, Guangzhou (P. R. China)
[**] The authors acknowledge the financial support of the German ministryof education and research (BMBF) under the contract number03X5507, the International Max Planck School for Science and Tech-nology of Nanostructures, and partial support by the European NODEproject 015783 and the DFG project GO 704/5-1.
Chem. Vap. Deposition 2011, 17, 149–154 � 2011 WILEY-VCH Verlag Gm
(PE)CVD process was applied to fabricate SiNWs with
small diameters on Si substrates.[6] Such techniques,
however, did not yield sub-10 nm SiNWs with high quality,
since the plasma also induced decomposition of the gaseous
precursors, leading to an enhanced growth rate and
structural defects. In this work, we show that the size of
Au particles is controlled in the interesting regime with
apoferritin as the template, and the catalytic growth of sub-
10 nm SiNWs on a Si substrate with a narrow size
distribution was successfully obtained after an immobiliza-
tion of the apoferritin.
The sizes and shape of nanoparticles are easily con-
trollable with bottom-up synthetic approaches comprising a
metal salt, a molecular stabilizer, and a reducing agent.
Biological templates play a special role for the template-
guided nanoparticle synthesis. Many of these templates are
easily available, cheap, and most importantly, identical. This
guarantees a perfect reproducibility of the synthetic
approaches. A commonly used nanometer-scaled biotem-
plate is apoferritin. It is a large globular protein composed of
24 subunits with an outer diameter of 12 nm and an inner
cavity of 7.6 nm. The junctions of the subunits provide
14 channels, 3 - 4 A in diameter, which perforate the protein
shell and serve as pathways between the exterior and
interior.[7] This feature enables the apoferritin molecules to
be used as highly accurate templates for a preparation of
metallic nanoparticles inside the cavity, which has already
been seen for the preparation of various functional
nanomaterials.[8–10] Besides the good solubility and the
upper limit for the particle size provided by the protein shell,
the protein can be easily modified and immobilized on
surfaces in several ways. This unique advantage, offered by
the protein, even enables patterning of large surface areas.
The immobilized nanoparticles in the cavity of the
apoferritin can subsequently act as catalysts, for example,
for the growth of 1D nanostructures.[11,12]
bH & Co. KGaA, Weinheim wileyonlinelibrary.com 149
Full Paper
2. Results and Discussion
Figure 1a shows a transmission electron microscopy
(TEM) image of Au-apo. It confirms that small Au
nanoparticles, which appear as black spots (ca. 4 nm in
diameter), form inside the cavity. The surrounding protein
appears as a bright ring around the dark spots. It is worth
noting that a treatment of the solution with spin columns
prevents a formation of Au nanoparticles outside the
apoferritin. This procedure was previously reported by
Zhang et al.,[9] however the filling factor, even with optimized
synthetic conditions, is still below 100%. Protein particles
without metal loading usually do not show the typical intact
spherical structure (Fig. 1a). Presumably, the reason for their
presence is a destruction of the protein during the preparation
process. A further purification step, for example with gel
filtration, may be beneficial to isolate the Au-apo in higher
purity. In Figure 1a, the apoferritins marked with black
arrows lack Au cores. White arrows mark apoferritin
containing Au nanoparticles which are smaller than expected
(�1 - 2nm). A TEM image of unstained apoferritin (Fig. 1b)
was used to determine the size distribution of the prepared
Au nanoparticles. The inset histogram shows the size
distribution with a mean diameter of the prepared Au
nanoparticles of 4.4� 1.1 nm, indicating that the cavities are
only partially filled with Au. This value appears reasonable.
In nature, the iron(III) oxyhydroxide nucleated inside ferritin
was found to be hydrophilic.[13] Therefore, within the
confinement of the ferritin cavity, the nucleation of
HAuCl4 � xH2O (x � 2) was assumed to occupy the whole
space. The unit volumes of KAuCl4 and Au are 674.96
(PCPDFWIN PDF70-1363) and 67.85 (PDF04-0784), respec-
tively. The volume ratio between the dihydrate Au salt and
the reduced Au is approximately 10:1. Considering the cavity
of apoferritin having an inner diameter of 7.6 nm and being
saturated with the hydrated gold complex, after reduction an
Fig. 1. a) TEM image of Au nanoparticles synthesized within apoferritin cavities stain
arrows mark apoferritins in which Au nanoparticles with sizes of 1 - 2 nm were detect
TEM image of an unstained sample.
150 www.cvd-journal.de � 2011 WILEY-VCH Verlag GmbH &
approximate particle size of around 4nm is expected
(7.6/101/3). The process demarks a natural size limitation.
The immobilization of the Au catalysts onto a Si substrate
was performed in a modified form of the procedure reported
by Oliveira et al.[14] With this strategy, Au-apo molecules
were covalently bound to a pretreated Si surface. The
individual steps for this immobilization process are
illustrated in Figure 2. In order to increase the packing of
Au-apo molecules on the surface of a Si substrate, the
concentration of Au-apo in the solution requires adjustment
to a certain value resulting in a monolayer of Au-apo
attached to the Si surface, however a direct use of Au
nanoparticles with such a high density would form bigger Au
droplets during heating (Ostwald ripening), which would
result in a serious increase in the diameter of the NWs. Thus,
the Au-apo in the solution was diluted to avoid further Au
aggregation during the CVD process. The same procedure
can also be used for patterning the Si substrate with Au-apo,
if, for example, microcontact printing (mCp) is applied.
Figure 3 shows the immobilization behavior of Au-apo on
patterned substrates. The patterns were produced with a
PDMS stamp. The regions with a bright contrast show
passivated Si surfaces without modification by APTES and
L-ascorbic acid, whereAu-apo is scarcely found. The density
of the bright spots, which correspond to Au cores, is strongly
enhanced within the patterned (functionalized) lines (inset
of Fig. 3), whereas only a few cores were dispersed on the
passivated Si surface. This unspecific binding of the proteins
to the silicon dioxide surface is caused by electrostatic
interactions and may be avoided by adjusting the concen-
tration or the pH of the protein solution. The image
demonstrates that Au-apo molecules can be, to a certain
extent, selectively immobilized on a Si substrate after a
surface modification.
For the investigation of the influence of the protein shell
on NW growth, three procedures have been performed; the
ed with 1.5% uranyl acetate. Black arrows mark empty apoferritins, and white
ed. b) The size distribution of the obtained Au nanoparticles calculated from a
Co. KGaA, Weinheim Chem. Vap. Deposition 2011, 17, 149–154
Full Paper
Fig. 2. Schematic flow chart of the immobilization ofAu-apo on a Si substrate: 1. oxidative activation of the silicon wafer surface with Piranha solution; 2. silanization
with 3-aminopropyltriethoxysilane; 3. binding dehydroascorbic acid; 4. immobilization of Au-apo.
NW growth with as-prepared Au-apo catalysts, a pretreat-
ment of the Au-apo with HF steam, and a pretreatment of
the Au-apo with O2-plasma. Figure 4a is a top-view scanning
electron microscopy (SEM) image of Au-apo immobilized
on a Si substrate. Each of the bright dots shows an Au core
inside an apoferritin molecule. Although this bionanopro-
cess has not yet been optimized for a homogeneous,
hexagonally close-packed alignment of Au nanoparticles
on a Si surface, the assembly shows a distribution without
aggregation on a large scale. Figure 4b shows a top-view
SEM image after the SiNW growth, which was obtained
without pretreatment of the Au-apo. The observed SiNWs
show a worm-like morphology with a low growth density,
which indicates that the protein shell strongly inhibits the
effective Si incorporation at the liquid-solid interface and
introduces many defects into the Si structure. The reason is
presumably a carbon contamination after pyrolysis of the
protein shell. In order to verify this, the Au-apo,
immobilized on Si, was exposed to HF-containing steam
in a sealed Teflon beaker for 30min. Figure 4c shows Au
nanoparticles after the HF treatment, which appear better
resolved due to the removal of the non-conductive coating.
Note that the HF also attacks the native SiO2 on the wafer.
Fig. 3. Top-view SEM images of Au-apo molecules adsorbed on a linearly
pre-patterned Si substrate, with a polydimethylsiloxane (PDMS) stamp. Scale
bar in the inset is 100 nm.
Chem. Vap. Deposition 2011, 17, 149–154 � 2011 WILEY-VCH Verlag G
Under identical CVD growth conditions, as shown in
Figure 4d, the SiNWs grew on the Si(111) substrate in a
straight shape with a high density, however without the
protein shells and the SiO2 layer, the Au diffusion was
enhanced and initially smaller particles coalesced to eutectic
Au/Si droplets with larger diameters. Most of the measured
SiNWs show diameters larger than 20 nm, and grew in the
<111> direction.
For removal of the protein shells, the sample was
alternatively treated with O2-plasma. As a result, the neat
Au cores remained immobile on the Si substrate which was
still covered with a SiO2 layer. No aggregation of the Au
nanoparticles was observed after the treatment. Finally, the
ultra-high vacuum (UHV)-CVD process was applied for a
SiNW growth with the Au particles as catalysts. The CVD
parameters were kept constant. Figure 5a shows a high-
magnification, side-view SEM image of SiNWs grown on the
Si(111) substrate after O2-plasma treatment. No obvious
aggregation of the Au nanoparticles was observed, which
can be derived from the persisting sub-10 nm diameter of the
straight SiNWs grown. Figure 5b shows a low-magnification,
top-view SEM image, fromwhich the average diameter of 25
randomly selected SiNWs was 9.4 nm, with a small size
distribution. We approximated the size distribution with a
Gaussian fit and obtained a full width at half maximum
(FWHM) of 20%. An interesting observation is that
although straight, sub-10 nm SiNWs with a small size
distribution were grown, the growth density is much lower
than expected from the Au nanoparticle distribution, and no
obvious epitaxial relationship between the growth direction
and the Si(111) substrate could be distinguished from the
{110} breaking edge. By the VLS mechanism, a higher
supersaturation is required for smaller Au catalysts.[15]
Considering the Si supersaturation under our growth
conditions, it appears that Au nanoparticles larger than
4 nm could more easily catalyze the growth of SiNWs. The
bright spots with a weak contrast on the Si surface are
presumably remaining Au nanoparticles smaller than the
critical size, which did not induce SiNW growth. In order to
increase the growth density of SiNWs, a higher partial
pressure of silane may be required.
The pretreatments with both HF steam and O2-plasma
can remove the protein shell effectively, however HF also
attacks the SiO2 layer on the surface of the Si substrate,
which facilitates the diffusion and aggregation of seeds and
mbH & Co. KGaA, Weinheim www.cvd-journal.de 151
Full Paper
Fig. 4. Top-view SEM images of a) Au-apo immobilized on a pretreated Si(111) substrate, and b) SiNWs grown after UHV-CVD, whereas c) shows the Au
nanoparticle array on Si(111) after HF-steam treatment, and d) after subsequent CVD SiNW growth.
makes the desired size-control of NWs impossible. In
comparison, treatment with O2-plasma is a better way to
remove the protein shell without negative effects on the seed
size. Since prolonged treatment with plasma can cause an
oxidation and deformation of metal nanoparticles,[16] the
treatment was kept short. Figure 6a shows a top-view SEM
Fig. 5. a) High-magnification, side-view SEM image of SiNWs catalyzed by Au-ap
magnification, top-view SEM image and the size distribution of 25 randomly select
152 www.cvd-journal.de � 2011 WILEY-VCH Verlag GmbH &
image of Au seeds on a Si substrate after the O2-plasma
treatment. Due to the deformation of the Au/Si alloy during
the process, the SiNWs had a larger diameter than the
synthesized Au seeds.[17] In addition, no obvious epitaxial
relationship between the growth direction and the Si(111)
substrate was observed. Figure 6b shows a magnified view of
o, immobilized on a Si(111) substrate after O2-plasma treatment. b) Low-
ed SiNWs with unspecific growth directions.
Co. KGaA, Weinheim Chem. Vap. Deposition 2011, 17, 149–154
Full Paper
Fig. 6. Top-view SEM images of a) Au-apo immobilized on a pretreated Si(111) substrate after O2-plasma treatment, scale bar: 100 nm; and b) SiNW grown from
a) after UHV-CVD, scale bar: 100 nm. c) HRTEM image of the top part of a SiNW including both Si and Au catalyst lattice planes. Insets are FFT patterns fromAu
and Si, respectively. Both FFT patterns are not to scale.
a single SiNW. The Au catalyst was found on top of the NW
as a bright dot. To avoid further parasitic growth on the NW
surface, the typical lengths of SiNWs grown by this process
were kept below 500 nm. Longer NWs can also be obtained,
but on the expense of a diameter increase. The wire has a
diameter below 10 nm, uniform along its length, and a disc-
shaped base, which is a typical feature observed for
semiconductor NWs grown on an oxide surface.[18]
As already known, SiNWs with larger diameters grow in
the<111> direction, while the<112> and<110> directions
are dominant for smaller diameters.[19] To determine the
crystallographic structure of both Si and Au, the SiNWs
were placed on a carbon-coated Cu grid and investigated by
TEM. Figure 6c shows a high-resolution (HR)TEM image of
the top part of a SiNW. The electron beam was adjusted
close to the [1-10] zone axis. The diameter of the SiNW was
about 7 nm, and a possible thin SiO2 layer can not be
distinguished due to the amorphous carbon support
beneath. The dark contrast in the upper part of the NW
corresponds to Au with a cubic lattice. The Au tip is around
6 nm in diameter, slightly smaller than the Si beneath.
The fast Fourier transform (FFT) pattern (lower inset)
shows both Si(111) and (220) planes, corresponding to a NW
with a [112] growth direction. The (111) planes are visible,
being parallel to the sidewalls marked with dashed lines. The
Si(111) plane has a low interface energy to the catalyst
droplet, and some models of the growth discuss a layer-by-
layer growth at (111) planes.[4] The interface between
catalyst and Si was usually flat, and a Si(111) plane
Chem. Vap. Deposition 2011, 17, 149–154 � 2011 WILEY-VCH Verlag G
terminated the SiNW.[20,21] In our case, however, the
interface between the SiNW and the catalyst particle was
not flat, as observed for the [112] wire, but showed a
considerable roughness. If also for [112] wires the {111}
planes would be planes of rapid growth, the idea of a
continuous stacking by adding material to the front of {111}
planes to build a wire can be expected.
3. Conclusions
In summary, small Au nanoparticles with a confined size
were synthesized exclusively inside the protein cavity of
apoferritin. On a large area, the Si substrate was decorated
and even patterned with Au-apo nanoparticles after a
modified bionanoprocess, which is difficult to achieve with
other biotemplates.[22] We demonstrate that, after the
protein immobilization, the Au nanoparticles can act as
heterogeneous catalysts, such as for crystalline SiNWs
growth via the VLS mechanism. The resulting NWs show
uniform diameters below 10 nm and a narrow size distribu-
tion. Patterning is an important issue to control the location
of NWs. The line size of the patterns by mCp is much bigger
than the apoferritin. Therefore, the apoferritin can only
randomly distribute on such patterns. For perfect ordering,
much smaller feature sizes for the silanization would be
needed, which can only be obtained with special methods,
such as dip-pen lithography, etc. In such approaches, real
chains of ferritins could be obtained. Together with various
mbH & Co. KGaA, Weinheim www.cvd-journal.de 153
Full Paper
protein immobilization methods, this bio-assisted approach
provides a possible route to control the location of NWs in
the sub-10 nm range, which can be a promising attribute for
further device engineering.
4. Experimental
For the synthesis of the Au nanoparticles inside apoferritin, a modifiedsynthesis procedure compared to that described in the literature [10] was usedto prevent formation of metal particles outside the protein shell and to avoidpost-synthesis purification. 5mgmL�1 apoferritin was incubated withHAuCl4 in a buffer solution at pH 7.4. The solution was mixed in darknessat 30 8C for 1 h to transport Au ions into the cavity. Excessive salt wasremoved with a Desalt Spin Column (Thermo Scientific Pierce) to preventformation of metal particles outside the protein shell. The desaltedapoferritin was immediately mixed with NaBH4 and stirred for 30min at30 8C. TheAuIII ions inside the cavity were reduced to yield Au nanoparticles.The excessive reducing agent was removed with spin columns after thereduction.
For the immobilization, first, a p-type Si(111) substrate was activated byimmersion in Piranha solution (1:3 H2O2/H2SO4) for 2 h. After rinsing withdeionized water and drying with nitrogen, the surface was silanized with 10%3-aminopropyltriethoxysilane (APTES) in toluene for 12 h at roomtemperature. For binding dehydroascorbic acid, the silanized Si substratewas placed into a saturated solution of L-ascorbic acid (VitaminC) in 90%ethanol and 10% methanol. After 30min. the sample was washed withethanol and dried in air. For the immobilization of Au-apo, the dried waferwas incubated with an aqueous solution of Au-apo over night at 4 8C. Thesample was then washed with deionized water to remove unbound Au-apo.
For the patterning of apoferritin-encapsulated Au nanoparticles (Au-apo), microcontact printing (mCp) was used. Briefly, patterned polydi-methylsiloxane (PDMS) (Sylgard 184, Dow Corning) stamps were fabricatedby casting an elastomeric polymer against silicon masters, which wererendered hydrophobic with fluoroalkyl-trichlorosilane vapor before use. ThePDMS stamps were immersed for 2min. in ethanol containing 10% APTES,dried with a nitrogen stream, and contacted for 20 - 30 s with activated Sisubstrates. Before the following protein immobilization steps, the Sisubstrates were dried at 80 8C to allow the polymerization of patternedAPTES.
The protein shells were selectively removed with O2-plasma treatment for2min (0.9 Torr, 100W, technics plasma 100-E, TePla). As a result, bare Aucores were produced on the Si substrate atop the SiO2 layer.
SiNWs were grown by the UHV-CVD method with SiH4 as the gaseousprecursor. Preheating at 110 8C in the UHV environment with a backgroundpressure about 1� 10�9 mbar was performed for 2 h to remove the remainingwater and anchor the seeds to avoid aggregation during growth. Subse-quently, the temperature was raised to 450 8C, and Ar gas with 5% SiH4 was
154 www.cvd-journal.de � 2011 WILEY-VCH Verlag GmbH &
flowed into the chamber for 20min under a total pressure of 5Torr for theNW growth.
Received: November 18, 2010
Revised: March 23, 2011
[1] H. J. Fan, P. Werner, M. Zacharias, Small 2006, 2, 700.
[2] R. S. Wagner, W. C. Ellis, Appl. Phys. Lett. 1964, 4, 89.
[3] E. I. Givargizov, J. Cryst. Growth 1975, 31, 20.
[4] Y. Wu, Y. Cui, L. Huynh, C. J. Barrelet, D. C. Bell, C. M. Lieber, NanoLett. 2004, 4, 433.
[5] T. Y. Tan, N. Li, U. Gosele, Appl. Phys. A 2004, 78, 519.
[6] S. Hofmann, C. Ducati, R. J. Neill, S. Piscanec, A. C. Ferrari, J. Geng,R. E. Duni-Borkowski, J. Robertson, J. Appl. Phys. 2003, 94, 6005.
[7] G. C. Ford, P. M. Harrison, D. W. Rice, J. M. A. Smith, A. Treffry, J. L.White, J. Yariv, Phil. Trans. Roy. Soc. Lond. B 1984, 304, 551.
[8] K. K. W. Wong, S. Mann, Adv. Mater. 1996, 8, 928.
[9] L. B. Zhang, L. Laug, W. Munchgesang, E. Pippel, U. Gosele,M. Brandsch, M. Knez, Nano Lett. 2010, 10, 219.
[10] L. Zhang, J. Swift, C. A. Butts, V. Yerubandi, I. J. Dmochowski, J. Inorg.Biochem. 2007, 101, 1719.
[11] D. Takagi, A. Yamazaki, Y. Otsuka, H. Yoshimura, Y. Kobayashi,Y. Homma, Chem. Phys. Lett. 2007, 445, 213.
[12] S. Kumagai, T. Ono, S. Yoshii, A. Kadotani, R. Tsukamoto, K. Nishio,M. Okuda, I. Yamashita, Appl. Phys. Express 2010, 3, 015101.
[13] T. G. St. Pierre, N. T. Gorham, P. D. Allen, J. L. Costa-Kramer, K. V.Rao, Phys. Rev. B 2001, 65, 024436.
[14] E. M. Oliveira, S. Beyer, J. Heinze, Bioelectrochem. 2007, 71, 186.
[15] V. Schmidt, J. Wittemann, S. Senz, U. Gosele, Adv. Mater. 2009, 21,2681.
[16] B. Gehl, A. Fromsdorf, V. Aleksandrovic, T. Schmidt, A. Pretorius,J. Flege, S. Bernstorff, A. Rosenauer, J. Falta, H. Weller, M. Baumer,Adv. Func. Mater. 2008, 18, 2398.
[17] A. I. Hochbaum, R. Fan, P. Yang, Nano Lett. 2005, 5, 457.
[18] M. Lorenz, E. M. Kaidashev, A. Rahm, Th. Nobis, J. Lenzner,G. Wagner, D. Spemann, H. Hochmuth, M. Grundmann, Appl. Phys.Lett. 2005, 86, 143113.
[19] V. Schmidt, S. Senz, U. Gosele, Nano Lett. 2005, 5, 931.
[20] S. Hofmann, R. Sharma, C. T. Wirth, F. Cervantes-Sodi, C. Ducati,T. Kasama, R. E. Dunin-Borkowski, J. Drucker, P. Bennett, J. Robert-son, Nature Mater. 2008, 7, 372.
[21] T. Xie, V. Schmidt, E. Pippel, S. Senz, U. Gosele, Small 2008, 4, 64.
[22] Y. Sierra-Sastre, S. Choi, S. T. Picraux, C. A. Batt, J. Am. Chem. Soc.2008, 130, 10488.
Co. KGaA, Weinheim Chem. Vap. Deposition 2011, 17, 149–154