RAPID COMMUNICATIONS

PHYSICAL REVIEW B VOLUME 61, NUMBER 10 1 MARCH 2000-II

Finite-size effects in nickel nanowire arrays

L. Sun and P. C. Searson
Department of Materials Science and Engineering, The Johns Hopkins University, Baltimore, Maryland 21218

C. L. Chien
Department of Physics and Astronomy, The Johns Hopkins University, Baltimore, Maryland 21218
共Received 11 June 1999兲
Nickel nanowire arrays with diameters in the range 30–500 nm have been fabricated by electrochemical
deposition into nanoporous, single-crystal mica templates, which allow measurements of the magnetic prop-
erties of nickel nanowire arrays at high temperatures. The Curie temperature is found to be reduced by as much
as 51 K for the 30 nm diameter nanowires. The Curie temperature shift with wire diameter follows the
finite-size scaling relation with ␭⫽0.94 and ␰ 0 ⫽22 Å.

The influence of reduced physical dimensions on mag- the Curie temperature for bulk Ni 共⬃630 K兲. The porous
netic entities is of both fundamental and technological inter- templates were fabricated by nuclear track etching. Particle
est. In bulk magnetic systems, the correlation length ␰ in- tracks were formed in 5 ␮m thick mica wafers by exposure
creases with temperature and diverges at the bulk transition to ⬇6 meV ␣ particles from a 100 ␮CiCf-252 source in a
temperature T C (⬁). When one or more dimensions in the chamber at a pressure of about 10⫺3 Torr. The alignment of
system are small, the growth of ␰ will eventually be limited the particle tracks was within 5° through collimation of the
by the smallest dimension d and the system displays a re- ␣-particle beam. The particle tracks were etched by immers-
duced transition temperature T C (d) due to finite-size effects. ing the tracked wafers in 20 wt. % HF at room temperature.
Technologically, as the feature sizes of magnetic structures From electrical conductance measurements we determined
continue to decrease, the influence of dimensionality on their the etch rate of the particle tracks to be about 1200 Å s⫺1.26
magnetic properties has also become an important issue. To The lateral etch rate of the bulk mica was determined to be
date, reports of finite-size effect in magnetic systems have 0.35 Å s⫺1.26 From the ratio of the normal and lateral etch
been largely limited to ultrathin films and quasi-two- rates we determine that the taper on the pore walls due to
dimensional systems. Measurements on quasi-one- etching is 0.02°. Pore diameters in the range 30–500 nm
dimensional systems have been hampered by the difficulty in were obtained by varying the etching time. In order to mini-
fabricating structures that are sufficiently small and stable mize the number of overlapping pores, the volume fraction
over the required temperature range. of nickel in the films was set at about 0.02 for all samples by
In quasi-two-dimensional systems, thickness dependent adjusting the track density from 1⫻107 cm⫺2 for the 500 nm
phase transition temperatures have been measured in ultra- pores to 2⫻109 cm⫺2 for the 30 nm diameter pores. The
thin ferromagnetic films of Ni,1–6 Fe,7,8 Co,9 and Gd 共Refs. relevant parameters are shown in Table I. For a volume frac-
10–12兲 and CuMn spin glass films.13,14 Recently, finite-size tion of 0.02, the fraction of tracks resulting in overlapping
effects have also been reported in antiferromagnetic CoO pores is less than 4%.26
共Refs. 15 and 16兲 and Cr 共Ref. 17兲 thin films. In quasizero Figure 1 shows a scanning electron microscope image of
dimensional systems, such as granular thin films, phase tran- etched particle tracks in mica. The pores are perpendicular to
sitions are not observed due to the onset of superparamag- the film plane and the pore walls are smooth and free from
netism when the particle size is very small.18 etching residues. The pores are diamond shaped with the
Electrochemical deposition of metals into porous polymer pore walls defined by the 兵110其 planes corresponding to the
films has been used to fabricate quasi-one-dimensional oxygen terminated planes in the mica lattice.25 For conve-
nanowire arrays19–22 and the ferromagnetic properties of Ni nience the pore diameter is defined as the diameter of the
and Co have been studied both at low temperatures and at
room temperature.22–25 However, measurements of magnetic
TABLE I. Effective diameter 共d兲, number density of tracks 共n兲,
phase transitions are not possible in these structures since the and volume 共area兲 fraction 共f 兲 of pores.
Curie temperatures for elements such as Ni, Co, and Fe are
much higher than the temperature range of polymer tem- d 共nm兲 n 共cm⫺2兲 f
plates. In this paper we describe the fabrication of quasi-one-
dimensional nickel nanowire arrays in porous mica films and 500 1⫻107 0.0196
show that this approach can be used to determine the tem- 200 5⫻107 0.0157
perature dependence of the magnetic properties at elevated 150 1⫻108 0.0177
temperatures. 100 3⫻108 0.0236
Nickel nanowire arrays were fabricated by electrochemi- 50 1⫻109 0.0196
cal deposition into porous single-crystal muscovite mica 30 2⫻109 0.0141
templates. Mica is chemically stable up to 770 K, well above

0163-1829/2000/61共10兲/6463共4兲/$15.00 PRB 61 R6463 ©2000 The American Physical Society

RAPID COMMUNICATIONS

R6464 L. SUN, P. C. SEARSON, AND C. L. CHIEN PRB 61

FIG. 1. Scanning electron microscope image of 2 ␮m pores in

single-crystal mica.

circle having the same cross-sectional area as the diamond-

shaped pores.
Nickel nanowire arrays were fabricated by electrochemi-
cal deposition into the mica templates. The electrode contact
was provided by a sputter deposited gold layer. Nickel was
deposited from a solution of 20 g l⫺1 NiCl2•6H2O, FIG. 3. 共a兲 Normalized magnetization and 共b兲 normalized de-
rivatives of the M-T curves vs temperature for nickel nanowire ar-
515 g l⫺1 Ni共H2NSO3兲2•4H2O, 20 g l⫺1 H3BO3 buffered to
rays with diameters 共from left to right兲 of 30, 50, 100, 150, 200, and
pH 3.4 at a potential of ⫺1.0 V 共Ag/AgCl兲.25 Scanning elec-
500 nm. The solid curves correspond to bulk nickel. All measure-
tron microscopy and atomic force microscopy were used to
ments were obtained at H⫽2000 Oe.
verify that ⬃100% of the pores were filled during the depo-
sition process and that the nickel nanowires grew at a similar
squareness 共the ratio of the remanence and saturation mag-
rate. An advantage of this geometry is that the nickel nanow-
netization兲 is strongly dependent on the physical properties
ires are embedded in the mica films, reducing oxidation and
of the porous template, particularly the alignment of the
structural damage at elevated temperatures.27
pores. Recently, we have shown25 that values of the square-
After dissolving the mica matrix in concentrated HF共⬃48
ness as high as 0.96 can be obtained with mica templates due
wt. %兲, free standing Ni nanowires were obtained on the gold
to the improved collimation of the pores, the uniform pore
electrode. Figure 2 shows an SEM image of free standing
cross section, and the low density of overlapping pores.
150 nm nanowires. In the cross section the wires have a
Figure 3共a兲 shows M-T curves for nickel nanowire arrays
diamond shape due to conformal filling of the diamond
as a function of the wire diameter from 30 to 500 nm. The
shaped pores in the mica.
temperature was calibrated prior to each measurement using
The magnetization measurements were conducted on a
a bulk nickel sample. Figure 3共b兲 shows the derivative of the
vibrating sample magnetometer. Due to the strong shape an-
M-T curves from which we determined the Curie tempera-
isotropy of the nickel nanowires, the applied external mag-
ture; identical results were obtained from examination of the
netic field was parallel to the wire axis. The room-
hysteresis loops. These results clearly show that the Curie
temperature magnetic coercivity of the nickel nanowire
temperature progressively decreases with decreasing wire di-
arrays increases with decreasing wire diameter, reaching val-
ameter. The Curie temperatures obtained by this method
ues of 800–900 Oe for 30 nm diameter wires.22–25 The
were independent of the applied magnetic field over the mea-
sured range 2 000 to 12 000 Oe. Figure 4共a兲 shows the de-
pendence of the measured Curie temperature on the wire
diameter.
The reduction of the Curie temperature for nanowires
with decreasing diameter is due to finite-size effects. The
correlation length ␰ (T) of a magnetic system increases with
temperature and at temperatures close to the bulk transition
temperature T C (⬁), the correlation length ␰ (T) has an
asymptotic behavior described by

冏
␰ 共 T 兲 ⫽ ␰ 0 1⫺
T
T C共 ⬁ 兲
冏 ⫺␯
, 共1兲

where ␰ 0 is the correlation length extrapolated to T⫽0, and ␯

is the critical exponent for correlation.28 For a magnetic sys-
FIG. 2. Scanning electron microscope image of 150 nm diam- tem with a small dimension of size d 共e.g., the diameter of
eter Ni nanowires after etching of the mica template. the nanowires兲, the growth of ␰ (T) with increasing tempera-
 RAPID COMMUNICATIONS

PRB 61 FINITE-SIZE EFFECTS IN NICKEL NANOWIRE ARRAYS R6465

3D Heisenberg model (␭⫽1.4) and the 3D Ising model (␭

⫽1.58), 27,28 however both models assume nearest neighbor
interactions whereas nickel is a ferromagnet that exhibits
longer range interactions. Spin-wave studies of Ni indicate
that the magnetic interactions extend beyond the fourth
neighbors.30 For the same reason the critical exponent ␤ for
the magnetization of Ni is 0.4, much larger than 0.36 for a
3D Heisenberg system with nearest-neighbor interactions.31
It may be noted that Eq. 共1兲, from which Eq. 共3兲 is de-
rived, is valid for T close to T C (⬁), i.e., in the critical region
where T C (d) is close to T C (⬁) for systems with relatively
large d. Thus for the relatively large diameters of the nickel
nanowires, as required by the scaling law, the extrapolated
correlation length ␰ 0 carries a larger uncertainty. However,
the extrapolated correlation length ␰ 0 ⫽22 Å is close to the
value of 20 Å reported for polycrystalline nickel thin films,6
although it is somewhat larger than the values of 4–10 Å
obtained for epitaxial single-crystal nickel films.4,5
Finally, we comment on the possible effects on T C due to
FIG. 4. 共a兲 Curie temperature T C (d) of nickel nanowire arrays strain. Large strains are known to exist in thin films, how-
vs wire diameter d. 共b兲 The log-log plot showing the reduced tem- ever, the effects on T C from mismatch in lattice and thermal
perature 关 T C (⬁)⫺T C (d) 兴 /T C (⬁) normalized to the Curie tempera- expansion are rarely incorporated. This is because of the dif-
ture for bulk Ni 关 T C (⬁)⫽631 K兴 vs wire diameter. The solid line ficulty in quantifying strain and the lack of simple relation
in 共b兲 corresponds to ␭⫽0.94 and ␰ 0 ⫽22 Å.
between strain and the resultant T C . In thin films, the prob-
lem is further compounded by the variation of strain with
ture will be constrained by the wire diameter d, resulting in a
film thickness. As a result, practically all finite-size scaling
reduced Curie temperature defined by
studies in thin films, including all the results mentioned

冋 冉 冊册
T C 共 d 兲 ⫽T C 共 ⬁ 兲 1⫺
␰0
d
␭
共2兲
above, neglect any effects due to strain. In the nanowire ge-
ometry, as long as the volume fraction of nickel is constant
and sufficiently small that all of the strain occurs in the nano-
or wires, the lateral strain is expected to be independent of wire

冉 冊
diameter. Thus, an upper limit for such an effect is 4 K,
␭
T C 共 ⬁ 兲 ⫺T C 共 d 兲 ␰0 corresponding to the shift in T C for the largest wire diameter
⫽ , 共3兲
T C共 ⬁ 兲 d of 500 nm. However, as can be seen from Figure 4共b兲, since
all the data, including that of d⫽500 nm, follow the finite-
where T C (d) is the Curie temperature for nanowires with size scaling relation, the actual contribution due to strain is
diameter d, and ␭⫽1/␯ is the shift exponent. likely to be negligible.
Due to the relatively large diameters, the nanowires are In summary, we have measured the reduction of Curie
expected to behave as a constrained 3D system 关Eq. 共3兲兴,26 temperature of Ni nanowires as a function of the nanowire
without the complexity of 3D to 1D crossover.29 Figure 4共b兲 diameter from 30 nm to 500 nm. The mica templates in
shows a log-log plot of the reduced temperature 关 T C (⬁) which nickel nanowires are imbedded enable such measure-
⫺T C (d) 兴 /T C (⬁) versus wire diameter d, illustrating that the ments for nanowires. The T C reduction follows a finite-size
measured values for T C (d) follow the finite-size scaling re- scaling relationship.
lation of 关Eq. 共3兲兴. From this figure we obtain ␭⫽0.94 and an
extrapolated value of ␰ 0 ⫽22 Å. The observed exponent of This work was supported by NSF Grant No. DMR-96-
␭⫽0.94 is lower than the theoretical values predicted by the 32526.

1
Y. Li and K. Baberschke, Phys. Rev. Lett. 68, 1208 共1992兲. 7
N. C. Koon, B. T. Jonker, F. A. Volkening, J. J. Krebs, and G. A.
2
C. A. Ballentine, R. L. Fink, J. Araya-Pochet, and J. L. Erskine, Prinz, Phys. Rev. Lett. 59, 2463 共1987兲.
Phys. Rev. B 41, 2631 共1990兲. 8
H. J. Elmers, J. Hauschild, H. Hoche, U. Gradmann, H. Bethge,
3
J. R. Childress, C. L. Chien, and A. F. Jankowski, Phys. Rev. B D. Heuer, and U. Kohler, Phys. Rev. Lett. 73, 898 共1994兲.
45, 2855 共1992兲. 9
C. M. Schneider, P. Bressler, P. Schuster, J. Kirschner, J. J. de
4
R. Bergholz and V. Gradmann, J. Magn. Magn. Mater. 45, 389 Miguel, and R. Miranda, Phys. Rev. Lett. 64, 1059 共1990兲.
共1984兲. 10
M. Farle and K. Baberschke, Phys. Rev. Lett. 58, 511 共1987兲.
5 11
F. Huang, G. J. Mankey, M. T. Kief, and R. F. Willis, J. Appl. M. Farle, K. Baberschke, U. Stetter, A. Aspelmeier, and F. Ger-
Phys. 73, 6760 共1993兲. hardter, Phys. Rev. B 47, 11 571 共1993兲.
6 12
H. Lutz, J. D. Gunton, H. J. K. Schurmann, J. E. Crow, and T. E. D. Tober, F. J. Palomares, R. X. Ynzunza, R. Denecke, J.
Mihalisin, Solid State Commun. 14, 1075 共1974兲. Morais, Z. Wang, G. Bino, J. Liesegang, Z. Hussain, and C. S.
RAPID COMMUNICATIONS

R6466 L. SUN, P. C. SEARSON, AND C. L. CHIEN PRB 61

Fadley, Phys. Rev. Lett. 81, 2360 共1998兲. 共1996兲.

13 24
G. G. Kenning, J. M. Slaughter, and J. A. Cown, Phys. Rev. Lett. R. Ferre, K. Ounadjela, J. M. George, L. Piraux, and S. Dubois,
59, 2596 共1987兲. Phys. Rev. B 56, 14 066 共1997兲.
14 25
A. Gavrin, J. R. Childress, C. L. Chien, B. Martinez, and M. B. L. Sun, P. C. Searson, and C. L. Chien, Appl. Phys. Lett. 74, 2803
Salamon, Phys. Rev. Lett. 64, 2438 共1990兲. 共1999兲.
15
T. Ambrose and C. L. Chien, Phys. Rev. Lett. 76, 1743 共1995兲. 26
L. Sun, C. L. Chien, and P. C. Searson, J. Mater. Sci. 35, 1079
16
T. Ambrose and C. L. Chien, J. Appl. Phys. 79, 5920 共1996兲. 共2000兲.
17 27
E. E. Fullerton, K. T. Riggs, C. H. Sowers, S. D. Bader, and A. C. M. Schneider, P. Bressler, P. Schuster, J. Kirschner, J. J. de
Berger, Phys. Rev. Lett. 75, 330 共1995兲. Miguel, and R. Miranda, Phys. Rev. Lett. 64, 1059 共1990兲.
18
A. Aharoni, Introduction to the Theory of Ferromagnetism 共Ox- 28
See e.g., H. E. Stanley, Introduction to Phase Transitions and
ford University Press, New York, 1996兲. Critical Phenomena 共Oxford University Press, New York,
19
W. D. Williams and N. Giordano, Rev. Sci. Instrum. 55, 410 1971兲, p. 46.
共1984兲. 29
K. Binder, in Finite Size Scaling and Numerical Simulation of
20
J. T. Madsen and N. Giordano, Phys. Rev. B 31, 6395 共1985兲. Statistical Systems, edited by V. Privman 共World Scientific
21
C. R. Martin, Science 266, 1961 共1994兲. Press, New York, 1990兲, p. 174.
22
T. M. Whitney, J. S. Jiang, P. C. Searson, and C. L. Chien, Sci- 30
S. N. Kaul, Phys. Rev. B 27, 5761 共1983兲.
ence 261, 1316 共1993兲. 31
M. Seeger, S. N. Kaul, H. Kronmuller, and R. Reisser, Phys. Rev.
23
J. Meier, B. Doudin, and J. Ph. Ansermet, J. Appl. Phys. 79, 6010 B 51, 12 585 共1995兲.