52,99 €
Mehr erfahren.
- Herausgeber: Wiley-VCH
- Kategorie: Wissenschaft und neue Technologien
- Sprache: Englisch
Long awaited new edition of this highly successful textbook, provides once more a unique introduction to the concepts, techniques and applications of nanoscale systems by covering its entire spectrum up to recent findings on graphene.
Sie lesen das E-Book in den Legimi-Apps auf:
Seitenzahl: 678
Veröffentlichungsjahr: 2015
Ähnliche
Table of Contents
Cover
Title Page
Related Titles
Copyright
Dedication
Preface
Glossary of Abbreviations
Chapter 1: Introduction
1.1 Nanometers, Micrometers, and Millimeters
1.2 Moore's Law
1.3 Esaki's Quantum Tunneling Diode
1.4 QDs of Many Colors
1.5 GMR and TMR 100–1000 Gb Hard Drive “Read Heads”
1.6 Accelerometers in Your Car
1.7 Nanopore Filters
1.8 Nanoscale Elements in Traditional Technologies
References
Chapter 2: Systematics of Making Things Smaller, Pre-quantum
2.1 Mechanical Frequencies Increase in Small Systems
2.2 Scaling Relations Illustrated by a Simple Harmonic Oscillator
2.3 Scaling Relations Illustrated by Simple Circuit Elements
2.4 Thermal Time Constants and Temperature Differences Decrease
2.5 Viscous Forces Become Dominant for Small Particles in Fluid Media
2.6 Frictional Forces Can Disappear in Symmetric Molecular Scale Systems
References
Chapter 3: What Are Limits to Smallness?
3.1 Particle (Quantum) Nature of Matter: Photons, Electrons, Atoms, and Molecules
3.2 Biological Examples of Nanomotors and Nanodevices
3.3 How Small Can You Make it?
References
Chapter 4: Quantum Nature of the Nanoworld
4.1 Bohr's Model of Nuclear Atom
4.2 Particle–Wave Nature of Light and Matter, DeBroglie Formulas
λ
=
h
/
p
,
E
=
hν
4.3 Wavefunction
Ψ
for Electron, Probability Density
Ψ
*
Ψ
, Traveling and Standing Waves
4.4 Maxwell's Equations;
E
and
B
as Wavefunctions for Photons, Optical Fiber Modes
4.5 The Heisenberg Uncertainty Principle
4.6 Schrodinger Equation, Quantum States and Energies, Barrier Tunneling
4.7 The Hydrogen Atom, One-Electron Atoms, Excitons
4.8 Fermions, Bosons, and Occupation Rules
References
Chapter 5: Quantum Consequences for the Macroworld
5.1 Chemical Table of the Elements
5.2 Nanosymmetry, Diatoms, and Ferromagnets
5.3 More Purely Nanophysical Forces: van der Waals, Casimir, and Hydrogen Bonding
5.4 Metals as Boxes of Free Electrons: Fermi Level, DOS, Dimensionality
5.5 Periodic Structures (e.g., Si, GaAs, InSb, Cu): Kronig–Penney Model for Electron Bands and Gaps
5.6 Electron Bands and Conduction in Semiconductors and Insulators; Localization versus Delocalization
5.7 Hydrogenic Donors and Acceptors
5.8 More about Ferromagnetism, the Nanophysical Basis of Disk Memory
5.9 Surfaces are Different; Schottky Barrier Thickness
W
= [2
εε
o
V
B
/
eN
D
]
1/2
5.10 Ferroelectrics, Piezoelectrics, and Pyroelectrics: Recent Applications to Advancing Nanotechnology
References
Chapter 6: Self-Assembled Nanostructures in Nature and Industry
6.1 Carbon Atom 1s
2
2p
4
(0.07 nm)
6.2 Methane (CH
4
), Ethane (C
2
H
6
), and Octane (C
8
H
18
)
6.3 Ethylene (C
2
H
4
), Benzene (C
6
H
6
), and Acetylene (C
2
H
2
)
6.4 C
60
Buckyball (∼0.5 nm)
6.5 C
∞
Nanotube (∼0.5 nm)
6.6 InAs Quantum Dot (∼5 nm)
6.7 AgBr Nanocrystal (0.1–2 µm)
6.8 Fe
3
O
4
Magnetite and Fe
3
S
4
Greigite Nanoparticles in Magnetotactic Bacteria
6.9 Self-Assembled Monolayers on Au and Other Smooth Surfaces
References
Chapter 7: Physics-Based Experimental Approaches to Nanofabrication and Nanotechnology
7.1 Silicon Technology: The INTEL-IBM Approach to Nanotechnology
7.2 Lateral Resolution (Linewidths) Limited by Wavelength of Light, Now 65 nm
7.3 Sacrificial Layers, Suspended Bridges, Single-Electron Transistors
7.4 What Is the Future of Silicon Computer Technology?
7.5 Heat Dissipation and the RSFQ Technology
7.6 Scanning Probe (Machine) Methods: One Atom at a Time
7.7 STM as Prototype Molecular Assembler
7.8 Atomic Force Microscope Arrays
7.9 Fundamental Questions: Rates, Accuracy, and More
7.10 Nanophotonics and Nanoplasmonics
References
Chapter 8: Quantum Technologies Based on Magnetism, Electron and Nuclear Spin, and Superconductivity
8.1 Spin as an Element of “Quantum Computing”
8.2 The Stern–Gerlach Experiment: Observation of Spin-½ Angular Momentum of the Electron
8.3 Two Nuclear Spin Effects: MRI (Magnetic Resonance Imaging) and the “21.1 cm Line”
8.4 Electron Spin ½ as a Qubit for a Quantum Computer: Quantum Superposition, Coherence
8.5 Hard and Soft Ferromagnets
8.6 The Origins of GMR (Giant Magnetoresistance): Spin-Dependent Scattering of Electrons
8.7 The GMR Spin Valve, a Nanophysical Magnetoresistance Sensor
8.8 The Tunnel Valve, a Better (TMR) Nanophysical Magnetic Field Sensor
8.9 Magnetic Random Access Memory
8.10 Spin Injection: The Johnson–Silsbee Effect
8.11 Magnetic Logic Devices: A Majority Universal Logic Gate
8.12 Superconductors and the Superconducting (Magnetic) Flux Quantum
8.13 Josephson Effect and the Superconducting Quantum Interference Device (SQUID)
8.14 Superconducting (RSFQ) Logic/Memory Computer Elements
References
Chapter 9: Silicon Nanoelectronics and Beyond
9.1 Electron Interference Devices with Coherent Electrons
9.2 Carbon Nanotube Sensors and Dense Nonvolatile Random Access Memories
9.3 Resonant Tunneling Diodes, Tunneling Hot Electron Transistors
9.4 Double-Well Potential Charge Qubits
9.5 Single Electron Transistors
9.6 Experimental Approaches to the Double-Well Charge Qubit
9.7 Ion Trap on a GaAs Chip, Pointing to a New Qubit
9.8 Quantum Computing by Quantum Annealing with Artificial Spins
References
Chapter 10: Nanophysics and Nanotechnology of Graphene
10.1 Graphene: Record-Breaking Physical and Electrical Properties
10.2 Consequences of One-Atom Thickness: Softness and Adherence
10.3 Impermeability of Single-Layer Graphene
10.4 Synthesis by Chemical Vapor Deposition and Direct Reaction
10.5 Application to Flexible, Conducting, and Transparent Electrodes
10.6 Potential Application to Computer Logic Devices, Extending Moore's Law
10.7 Applications of Graphene within Silicon Technology
References
Chapter 11: Looking into the Future
11.1 Drexler's Mechanical (Molecular) Axle and Bearing
11.2 The Concept of the Molecular Assembler is Flawed
11.3 Could Molecular Machines Revolutionize Technology or Even Self-Replicate to Threaten Terrestrial Life?
11.4 The Prospect of Radical Abundance by a Breakthrough in Nanoengineering
11.5 What about Genetic Engineering and Robotics?
11.6 Possible Social and Ethical Implications of Biotechnology and Synthetic Biology
11.7 Is there a Posthuman Future as Envisioned by Fukuyama?
References
Some Useful Constants
Exercises
Index
End User License Agreement
Pages
xv
xvi
xvii
xviii
xix
xx
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
Guide
Cover
Table of Contents
Preface
Begin Reading
List of Illustrations
Chapter 1: Introduction
Figure 1.1 Silicon nanowires in a harp-like array. Due to the clamping of the single-crystal silicon bars at each end and the lack of applied tension, the situation is more like an array of xylophone keys. The resonant frequency of the wire of 2 μm length is about 400 MHz. After reprinted with permission from Ref. [5], Copyright 1999. American Institute of Physics.
Figure 1.2 Moore's law [6]. The number of transistors in successive generations of computer chips has risen exponentially, doubling every 1.5 years or so. Gordon Moore, cofounder of Intel, Inc., predicted this growth pattern in 1965, when a silicon chip contained only 30 transistors. The number of dynamic random access memory (DRAM) cells follows a similar growth pattern. The growth is largely due to continuing reduction in the size of key elements in the devices, to less than 45 nm, with improvements in optical photolithography. Clock speeds have similarly increased, presently around 2 GHz. For a summary, see [7]. Reprinted with permission from Ref. [6].
Figure 1.3 Transmission electron micrograph (TEM) image of one 5 nm CdSe quantum dot particle. Courtesy Andreas Kornowski, University of Hamburg, Germany.
Figure 1.4 Schematic of quantum dot with coatings suitable to assure water solubility, for application in biological tissue. This ZnS-capped CdSe quantum dot is covalently coupled to a protein by mercaptoacetic acid. The typical QD core size is 4.2 nm [8]. Reprinted with permission from [8], Copyright 1998 AAAS.
Figure 1.5 Schematic diagram of the GMR read head, showing two current leads connected by the sensing element, which itself is a conducting copper sheet sandwiched between a pair of hard and soft magnets. Courtesy Prinz, G.A., U.S. Naval Research Laboratory, Washington, DC.
Chapter 2: Systematics of Making Things Smaller, Pre-quantum
Figure 2.1 Nested carbon nanotubes. This is a computer-generated image. Zettl has experimentally demonstrated relative rotation and translation of nested nanotubes, a situation very similar to that in this image. The carbon–carbon bonding is similar to that in graphite. The space between the tubes is simply vacuum. (Image courtesy of Zettl Research Group, University of California at Berkeley, and Lawrence Berkeley National Laboratory.)
Figure 2.2 TEM image of partially nested nanotubes, after relative translation [5]. It was found that the inner tube could repeatedly be slid and rotated within the outer tube, with no evidence of wear or friction. Attractive forces very rapidly pulled a freed tube back into its original full nesting. (Reprinted with permission from Ref. [5] Copyright 2000 AAAS.)
Chapter 3: What Are Limits to Smallness?
Figure 3.1 Spasmoneme spring. (a) The spasmoneme in
Vorticella
in its fully extended (left), fully contracted (middle), and partly extended (right) states [6]. (b) The extended spring state (left) consists of aligned filaments held apart by negative charges (dots). On the right, with plus and minus charges (dots) equally likely, the stalk collapses to a rubber-like state. (Reprinted with permission from Ref. [6], Copyright 2000 AAAS.)
Figure 3.2 Models of the motion of muscle myosin and conventional kinesin [8]. (a) Frame 1: Muscle myosin is a dimer of two identical motor heads, anchored to a thick filament (top). Frames 2–4 show docking of heads on actin filament (lower), which serves to move the actin. The motion is about 10 nm per ATP hydrolyzed. (b) The two heads of the kinesin dimer move along a tubulin filament as indicated in Frames 1–4. In these frames, the coiled coil extends to the top and to the attached cargo. The step length is about 8 nm. The head motion is associated with hydrolysis of ATP to ADP. The scale bar in A is 6 nm and that in B is 4 nm. (Taken from Reprinted with permission from Ref. [8]), Copyright 2000 AAAS.)
Figure 3.3 Fluorescently labeled actin filament permits observation of rotation of the
c
subunit in ATP synthase (F
0
F
1
) [20]. A
c
subunit of Glu
2
was replaced by cysteine and then biotinylated to bind streptavidin and the actin filament. The
γ
,
ε
, and
c
units are thus shown to be a rotor, while the
α
,
β
,
δ
,
a
, and
b
complex is the stator. The rate of rotation of the actin filament in the viscous medium was found to depend on its length. Rotational rates in the range 0.5–10 Hz were measured, consistent with a torque
τ
of 40 pN nm. (Reprinted with permission Ref. [20], Copyright 1999 AAAS.)
Figure 3.4 Biomolecular rotary motor-powered propellers [21]. (a) 80 nm Ni post from array. (b) Schematic view of F
1
-ATPase molecular motor. (c) Array of Ni propellers, 750–1400 nm in length, 150 nm in diameter. (d) Schematic view of one assembled device from array. Rotation in fluid of propeller (0.8–8.3 rps) fueled with ATP is 50% efficient. (Reprinted with permission from Ref. [21], Copyright 2000 AAAS.)
Figure 3.5 Model for Ca
++
-gated K channel. Here (a) and (b) respectively depict the channel as closed and open. (After Reprinted with permission from Nature: [22], Copyright 2002, Macmillan Publishers Ltd.)
Figure 3.6 Models for a voltage-gated potassium ion channel [26]. Here (a) and (b), respectively , depict the conventional and new models. (Reprinted with permission from Nature [26], Copyright 2003, Macmillan Publishers Ltd.)
Figure 3.7 Turbine wheel produced on silicon wafer with deep reactive ion etching. Courtesy Schmidt, M.A. Microsystems Technology Laboratories at Massachusetts Institute of Technology.
Figure 3.8 Assembling a ring of 48 Fe atoms on a (111) Cu surface with an STM. The diameter of the ring is 14.3 nm. (Courtesy IBM Research Almaden Research Center., Unauthorized use not permitted.)
Figure 3.9 Cesium and iodine on Cu (111). This pattern represents a molecule on the copper surface which contains eight cesium and eight iodine atoms. This image illustrates that the assembled atoms may choose their own structure, beyond control of the tip, in this case driven by the strong ionic bond. (Courtesy IBM Research Almaden Research Center, Unauthorized use not permitted.)
Figure 3.10 Scheme [30] for fabricating a chain using a flexible photomask and electrochemical welding. (a) Metalized glass capillaries coated with photoresist, by pulling them slowly from bulk solution. (b) Capillaries hard-baked at 105 °C for 3 min. Exposure (8s) of coated capillary to UV light through a flexible mask (design shown in (c)) wrapped around its surface. (d) Under optical microscope, align two patterned capillaries to be in close proximity with their patterns matched to form a chain. (e) and (f) (Links correspond to openings in the photoresist. Dotted lines represent links on the undersides of the capillaries that are not visible from the top.) Electroplating nickel for 30 min at density 20 mA cm
−2
welded together the ends of the chain links, in those areas defined by the photoresist. Finally, release the nickel chain by dissolving photoresist in acetone, dissolving silver metallization in aqueous ferricyanide bath, and dissolving the titanium and glass in concentrated HF [30]. (Reprinted with permission from Ref. [30], Copyright 1998 AAAS.)
Figure 3.11 Optical micrograph of a free-jointed nickel chain formed by the process shown in Figure 3.10 [30]. The final thickness of the nickel was about 50 µm. (Reprinted with permission from Ref. [30], Copyright 1998 AAAS.)
Figure 3.12 Use of optical tweezers (a) to observe single RNAP molecule pulling DNA (b) [31]. (Reprinted with permission from Ref. [31], Copyright 1998 AAAS.)
Figure 3.13 Fourfold rotating optical trap (a) stabilizes 3D arrays (b) [32]. (Reprinted with permission from Ref. [32], Copyright 2002 AAAS.)
Figure 3.14 Assembly of a DNA-templated FET and Au contacts [33]. (i) Short single-strand DNA (ssDNA) is reacted with RecA protein derived from
E
.
coli
bacteria to form a nucleoprotein filament, which is 500 bases (250 nm) in length. (ii) Long double-stranded DNA (dsDNA) molecule serves as scaffold. The assembly process is guided by the information encoded in ssDNA and dsDNA molecules: the dsDNA was synthesized so that its “sequence” (see Figure 5.7) is identical to the dsDNA at the designated location of the FET. “Homologous recombination” locates the short dsDNA (RecA-coated) segment at the proper location on the longer dsDNA, ending step (ii). (The RecA coating later helps locate a (streptavidin-coated) SWNT, and protects the covered DNA segment against metallization.) (iii) A streptavidin-coated SWNT is bound to the scaffold using a primary antibody to RecA and a biotin-conjugated secondary antibody. (iv) Incubation in an AgNO
3
solution leads to Ag clusters on sections not protected by RecA. (v) “Electroless” gold deposition, using the Ag clusters as nucleation centers, results in the formation of two DNA-templated Au wires contacting the SWNT bound at the gap. (Reprinted with permission from Science, copyright AAAS.)
Figure 3.15 Localization of an SWNT at a specific address on the scaffold dsDNA using RecA [33]. (a) Atomic force microscope (AFM) topograph of 250-base (250 nm) RecA nucleoprotein filament (black arrow) located at matching segment on DNA scaffold molecule. Bar, 200 nm. (b) AFM image of streptavidin-coated SWNT (white arrow) bound to 500-base-long nucleoprotein filament localized on DNA scaffold molecule. Bar, 300 nm. (c) Scanning conductance image of the same region as in (b). The conductive SWNT (white arrow) yields a signal (black indicates conductivity), while the insulating DNA is hardly resolved. Bar, 300 nm. (Reprinted with permission from Science, copyright AAAS.)
Chapter 4: Quantum Nature of the Nanoworld
Figure 4.1 Sketch of transverse electric mode TE10 in a rectangular metallic waveguide.
Figure 4.2 Calculated modes for cylindrical dielectric waveguide, assumed to extend out of the page. (By courtesy of RP Photonics Consulting GmbH, from: Encyclopedia of Laser Physics and Technology. http://www.rp-photonics.com/fibers.html.)
Figure 4.3 Adding waves creates regions of localization that move at the group velocity.
Figure 4.4 Finite potential step at
x
= 0.
Figure 4.5 Electrons trapped in a small 2D box on the (111) surface of copper. A rectangular array of iron atoms serves as a barrier, reflecting electron waves (Courtesy IBM Research and Almaden Research Center. Unauthorized use not permitted.)
Figure 4.6 Geometry of a (111) plane, shown shaded. Copper is a face-centered cubic crystal, but only if the surface is cut to consist of the indicated (111) plane will the 2D electron effects be present.
Figure 4.7 Indium phosphide nanowires [7]. InP nanowires grown by laser-assisted catalytic growth, in 10, 20, 30, and 50 nm diameters, were studied by atomic force microscope image (a) and also by observation of photoluminescence (b) and (c) under illumination by light with energy
hc
/
λ
>
E
g
. The bandgap is about 1.4 eV. In (a), the white scale bar is 5 µm, so the wires are up to 10 µm in length. (b) and (c) Gray-scale representation of light emitted from 20 nm diameter InP nanowire excitation in (b) with bandgap light linearly polarized along the axis of the wire produces a large photoluminescence, but (c) no light is emitted when excited by bandgap light linearly polarized
perpendicular
to wire axis. Inset shows dependence of emission intensity on polarization angle between light and wire axis. Reprinted with permission from Ref. [7], Copyright 2001 AAAS. Reproduced from Ref. [8].
Figure 4.8 Probability
P
n
(
x
) density for
n
= 13 state of simple harmonic oscillator. Dashed line is classical
P
(
x
), with classical turning points
x
′ and
x
″ indicated. Reproduced from Ref. [9].
Figure 4.9 2p wavefunctions in schematic form. (a) Complex forms carry angular momentum. (b) Linear combinations have the same energy, now assume aspect of bonds. (Reproduced from Ref. [9])
Figure 4.10 Five allowed orientations of angular momentum
l
= 2, length of vector and
z
-projections in units of . Azimuthal angle is free to take any value.
Figure 4.11 Radial wavefunctions of one-electron atoms exhibit
n
−
ℓ
− 1 nodes, as illustrated in sketch for (bottom to top) 1s, 2s, 3s, 2p, and 3p wavefunctions. Reproduced from Ref. [9].
Chapter 5: Quantum Consequences for the Macroworld
Figure 5.1 Energy curves for bonding and antibonding states of the hydrogen molecule. The bonding state requires antiparallel spins. The equilibrium separation is 0.074 nm. Ref. [2].
Figure 5.2 Melting points of molecular solids versus number of electrons per molecule. The melting point is a measure of the cohesive energy, presumably here largely originating in van der Waals interactions. (After Ref. [5].)
Figure 5.3 van der Waals interaction energies for several extended geometries: (a) atom-surface; (b) sphere-surface; (c) two spheres; and (d) two surfaces. Here,
n
v
represents the number of atoms per unit volume, is the Hamaker constant, a tabulated material parameter, and
S
represents the area of two facing surfaces. (After Ref. [7].)
Figure 5.4 Schematic of measurement of the Casimir force between sphere and planar surface, measured with
V
= 0 [9]. Planar capacitor plates on paddle rotate against torsion fiber (center black dot) whose restoring force is calibrated using
V
and known Coulomb force between sphere and plane. Rotation angle
θ
is measured by capacitance difference between paddle plates and lower fixed plates, whose spacing
t
is 2 µm. Piezo stage lifts paddle assembly toward sphere, adjusting
z
down to 75 nm. Reprinted with permission from Ref. [9]. Copyright 2001 AAAS.
Figure 5.5 Micromachined planar capacitor plates freely suspended to rotate above silicon crystal on twin silicon torsion fibers [9]. (a) Rectangular paddle 500 µm on a side is freely suspended 2 µm above Si chip by twin Si fibers, front and back. (b) Detail of paddle (upper rotating capacitor plate) showing its separation from Si chip and showing the front Si suspending torsion fiber. A 200 µm radius metalized sphere (not shown) is placed a variable distance
z
above center of the right portion of the paddle (see Figure 5.4). Reprinted with permission from Ref. [9]. Copyright 2001 AAAS.
Figure 5.6 Measured Casimir force between sphere and plane versus spacing
z
, shown by points fitted to theoretical curve. (After [9].) (a) Data points and Casimir theory curve (lower trace) versus spacing
z.
Upper curve is force versus
z
for
V
= 136 mV, where Coulomb force matches Casimir force at closest approach (76 nm). (b) Deviation between measured points and fitted Casimir theory curve, on expanded scale. Reprinted with permission from Ref. [9]. Copyright 2001 AAAS.
Figure 5.7 Models of DNA replication fork, showing breaking of hydrogen bonds between upper and lower bases (center of figure) as DNAP engine (left) pulls double-helical DNA from right to left. (After [10].) Hydrogen bonds are depicted here as smaller diameter cylinders connecting the larger cylinders representing the four different bases A, C, G, and T, which, however, bond only as complementary pairs AT and CG. The AT pairs form double hydrogen bonds, while the CG pairs form triple hydrogen bonds.
Figure 5.8 Metal as a 3D box filled with noninteracting electrons up to the Fermi energy
E
F
, following the Pauli exclusion principle. The total depth of the potential well is
V
o
, the sum of
E
F
and the work function, here labeled
w
o
. Ref. [2].
Figure 5.9 Constant energy surfaces in quantum number space for particle in a 3D potential well, a starting representation for electrons in a metal. There are two electron states per lattice point. The number of states
N
out to radius
r
is therefore (2/8)(4π/3)
r
3
, where
r
=
E
/
E
o
and
E
o
=
h
2
/8
mL
2
(see Eq. (5.29)). Ref. [2].
Figure 5.10 Density of states
g
(
E
) and occupation
f
(
E
) at
T
= 0 (a) and
T
nonzero (b) in the 3D case. Ref. [2].
Figure 5.11 A simplified model potential, assumed extended periodically.
Figure 5.12 Kronig–Penney model, a schematic plot of ordinate
R
(
E
) versus abscissa
qa
. Allowed solutions (unshaded) occur only when ordinate
R
(
E
) has magnitude unity or less.
Figure 5.13 Schematic of bands
E
versus
k
in a periodic potential, based on Kronig–Penney model. The bands are restricted in
k
to values less than π/
a
. Energy gaps occurring at
k
= ±(π/
a
) are also physically understood on the basis of Bragg reflections at
k
= ±(π/
a
). Physical arguments easily show that each band accommodates exactly
N
/2 electrons, so that one electron per atom gives a half-filled band and a metal, while two electrons per atom gives a filled band, and an insulator.
Figure 5.14 Diamond and zincblende crystal structures. Each atom is covalently bonded to four nearest neighbors in tetrahedral directions. The directed bonds are linear combinations of s and p orbitals (see Table 4.1), and analogous to directed orbitals depicted in Figure 4.9. Specifically, 2s and 2p
3
for diamond (as in CH
4
) and 3s and 3p
3
for Si. There are four valence electrons per atom, leaving a band structure with filled bands.
Figure 5.15 Energy band structures for silicon (a) and GaAs (b). Energy is shown vertically, and
k
horizontally. The horizontal line marks the top of the filled “valence” bands; in pure samples the upper bands are empty except for thermal excitations (indicated by ++ and −−symbols). The zero of momentum is indicated as “
Γ
,” and separate sketches are given for
E
versus
k
in (111) left and (100) right directions.
Figure 5.16 Current voltage measurement of germanium tunnel diode (Esaki diode), in forward bias, emphasizing anomalous current peak and negative resistance region (
b–c
). For an interpretation, see Figure 5.17.
Figure 5.17 (a) Band shifts in forward bias of heavily doped PN junction, showing tunneling as the origin of the anomalous current peak in the Esaki tunnel diode (see Figure 5.16). (b) Forward bias in which electrons can tunnel directly into filled hole states on the left. (c) At larger forward bias, no current can flow because electrons face the energy gap of the P-type region, which does not allow current flow. (d) The thermal activation of electrons over the barrier leads to the usual exponential growth of current with forward positive bias
V
.
Figure 5.18 (a) A CdS nanowire (NW), coated above with thin alumina, and a Ti/Au electrode, lying on a heavily doped conductive P
+
Si substrate, acts as an injection laser as described in the text. Electrons flow into the nanowire from the Ti/Au electrode, and holes flow into the device from the substrate. (b) The device is seen in a top view in the upper panel, and is seen to be emitting light from its end in the lower panel. The scale bar in (b) is 5 µm. In this device, the reflective cavity is provided by the discontinuity in refractive index at the accurately cleaved ends of the single-crystalline (001) CdS nanowire [13]. (Reprinted from Nature with permission from Macmillan Publishers, Ltd.)
Figure 5.19 Light emission spectrum [13] of the CdS nanowire at 8 K and injection currents 200 μA (lower) and 280 μA (upper). The sharp peak in the upper trace at 494 nm indicates that the laser threshold has been reached, although it was not concluded by the authors that only a single mode was involved. (Reprinted from Nature with permission from Macmillan Publishers, Ltd.)
Figure 5.20 Bulk moment
M
versus applied magnetic field
H
, for typical ferromagnet. Hysteretic behavior arises from motion of domain walls. A large remanent
M
at
H
= 0 is typical of a permanent magnet. Ref. [14].
Figure 5.21 The saturation magnetic moment
M
S
(
T
) of a typical ferromagnet versus temperature, normalized to critical temperature
T
c
. Circles are data for ferromagnet nickel. Solid line comes from solution to Eq. (5.59). Ref. [14].
Figure 5.22 Schematic band fillings for normal metal (a) and ferromagnet (b). Normal metal has equal numbers of spin-up and spin-down electrons, here depicted in a partially filled s-band. A ferromagnet is depicted with relatively narrow d-bands, shifted by 2
μB
E
, where
B
E
is the
internal exchange field
, representing the action of the exchange interaction
J
E
. A portion of the spin-down band is empty, leaving a net excess of spin-up electrons.
Figure 5.23 Light emission [19]: single-bubble sonoluminescence spectra for argon gas above 85% H
2
SO
4
in water at ultrasound pressure amplitudes ranging from 2.3 to 5.5 bar. The curves are smooth at small wavelength, and are fit by blackbody radiation curves (dashed) with temperatures
T
ranging from 9000 to 12 300 K. The
λ
= 700–900 nm spectra reveal identifiable emission lines of the argon atom. (These lines are known to arise from transitions among Ar atomic levels, which lie 13 eV above the ground state.) Fits to these line spectra indicate that
T
rises from 8000 K to more than 15 000 K as the pressure amplitude
P
0
rises from 2.3 to 5.5 bar. The emitted power densities here are relatively high: it was noted that by changing from H
2
O to H
2
SO
4
(higher density and lower vapor pressure), the power density increased by a factor of 2700 [19]. At the highest
P
0
, the Ar line spectra broaden and disappear, indicating short lifetimes of atomic levels, due to rapid ionization events by hot electrons from an inner plasma core of much higher temperature. The Ar atom levels identified in these spectra are too high in energy (>13 eV or 156 000 K) to be consistent with the emission temperatures (12 300–15 000 K). The authors [19] suggest that there is an inner (opaque) plasma core (at perhaps 156 000 K) which showers the outer Ar with high-energy electrons, exciting the observed Ar levels, known to lie more than 13 eV (156 000 K) above the ground state. This analysis suggests a temperature gradient within the collapsing bubble, so the plasma core is much hotter than 15 000 K at the outer region. The
T
of the outer region, near the liquid interface, is known to fall as the thermal conductivity of the gas is increased, by admixing Ne into Ar. (Reprinted from Nature with permission from Macmillan Publishers, Ltd.)
Figure 5.24 (a) Compact pyroelectric-based field-ionizing neutron generator [18]. The pyroelectric crystal (center left) produces a strong dipole-like set of electric field lines, which appear as a function of temperature. The strongly (positive) biased (up to 80 kV) tip of diameter 100 nm (b) is attached to a copper electrode encasing the LiTaO
3
pyroelectric crystal, producing a distinct set of radial electric field lines which trace toward the (grounded) target, at the right-hand side of the apparatus. The right surface of the pyroelectric crystal reaches a potential of +80 kV or so relative to its left surface, which is grounded. The ambient gas is D
2
, dilute such that the mean free path exceeds the dimension of the chamber. The electric field near the tip (at voltage) locally ionizes the deuterium gas to produce D
+
ions, which then fall down the potential gradient and impact the (deuterated) target at 80 keV energy. This energy is enough to drive the well-documented D + D =
3
He + n reaction, verified by direct observation of the neutrons. (Reprinted from Nature with permission from Macmillan Publishers, Ltd.)
Figure 5.25 Documentation of one cycle of neutron generation [18]. (a) At time
t
= 0, a ramp of temperature from 240 to 280 K leads to crystal potential rising from 0 to about 80 kV near
t
= 230 s. (b) The onset of X-rays which come as electrons released from the ionization events fall onto the positively charged copper plate encasing the tantalate pyroelectric crystal. The X-ray energies, observed up to about 100 keV, can only come from a tip potential of the same order, which produces a local electric field sufficient to strip electrons from the deuterium gas. (c) The ionic current, presumably the sum of electron current into the copper and positive ion current into the right-hand side electrode, adds to 4 nA maximum. This number can be checked by elementary considerations involving the radius and surface area around the tip leading to certain ionization, the density of the gas at the stated pressure, and the number of deuterium molecules in random gas diffusion that would cross that surface per unit time. (d) The measured number of neutrons per second. The satisfactory coincidence of the peaking of the several indicators at about 230 s puts these results really beyond question. (Reprinted from Nature with permission from Macmillan Publishers, Ltd.)
Chapter 6: Self-Assembled Nanostructures in Nature and Industry
Figure 6.1 Modeled thermal conformation changes of octane (C
8
H
18
) at 400 K. . (From Ref. [1] Copyright © 1992 by Wiley Publishing, Inc. All rights reserved. Reproduced here by permission of the publisher.)
Figure 6.2 Carbon 60 molecule, in perspective model view. Each carbon atom participates in one double bond and in two single bonds. All valence electrons in each carbon are bonded. Each hexagonal benzene ring shares sides with three hexagonal and three pentagonal rings. Courtesy Joseph W. Lauher, Chemistry Department, SUNY Stony Brook.
Figure 6.3 Images of carbon nanotubes [2]. (a), (b), and (c) Armchair (metallic), zigzag (small bandgap), and chiral (semiconducting) nanotubes, respectively. The twist of the chiral nanotube is clearly evident in (c), a perspective view along the tube axis. (d) An STM image of a 1.3-nm-diameter chiral nanotube. (e) A TEM image of a nine-walled nanotube, a concentric cylindrically nested assembly, in which the binding between the adjoining nested tubes is very weak. (Panels (a), (b), (c), and (e) reprinted with permission from Ref. [2a], Copyright 2002 AAAS. Panel D reprinted with permission from Nature: Ref. [2b]. Copyright 1998, Macmillan Publishers Ltd. and with permission from C. Dekker, Delft University of Technology.)
Figure 6.4 Incorporation of 1.4-nm-diameter single-wall nanotube (SWNT) as active element of field effect transistor [3]. (a) 1.4-nm-diameter SWNT placed 150 nm above Gate (P
+
Si), spaced by 150 nm thermally grown SiO
2
dielectric barrier layer. Source and drain contacts to the nanotube are thermally annealed Ti, which establish covalent bonding to the C nanotube by the formation (during annealing) of titanium carbide. Electrically, the Ti–TiC–C contacts are highly transmissive Schottky barriers. The whole device is covered with 10 nm of SiO
2
. (b) Current–gate voltage characteristics at fixed source–drain voltage. Current is small except in positive and negative bias ranges when the nanotube becomes metallic in hole or in electron conduction. Saturation current of 600 nA corresponds to current density of >10
8
A cm
−2
. Reprinted with permission from Ref. [3], Copyright 2003 AAAS.
Figure 6.5 A laser ablation method for growing free-standing semiconductor nanowires of controlled diameters by metal nanocluster-catalyzed growth [4]. (a) Laser ablation with photons of energy
hv
of the Si
1−
x
Fe
x
target creates a dense, hot vapor of Si and Fe species. (b) The hot vapor condenses into small clusters as the Si and Fe species cool through collisions with the buffer gas. The furnace temperature is controlled to maintain the Si–Fe nanoclusters in a liquid state. (c) Nanowire growth begins once the liquid becomes supersaturated in Si and continues as long as the Si–Fe nanoclusters remain in a liquid state and the Si reactant is available. (d) Growth terminates when the nanowire passes out of the hot reaction zone in the carrier gas flow onto a cold finger and the Si–Fe nanoclusters solidify. (Reprinted with permission from Science, copyright AAAS.)
Figure 6.6 Crystals contained in photographic emulsion are typically silver bromide. (Image courtesy © Eastman Kodak Company. KODAK is a trademark.) (Eastman Kodak Research)
Figure 6.7 Image of magnetotactic bacteria, showing linear array of magnetic crystals. The nanocrystals are greigite, an iron sulfide, with dimensions of the order of 40 nm, and are yet each large enough to be antiferromagnets. Arrow in box locates a bicrystal defect in the magnetic chain (see Figure 6.8) [6]. (Reprinted with permission from Ref. [6], Copyright 1998 AAAS.)
Figure 6.8 Magnetic field mapping [6] of central boxed region of magnetic bacterium image, shown in Figure 6.7 (see box and arrow). More precisely [6], this image is an off-axis electron hologram obtained in field-free conditions. The apparent magnetic field lines thread along the array of antiferromagnetic nanocrystals, including the bicrystal defect shown by the arrow, proving that the array acts as an extended bar magnet (believed to orient the bacterium in the earth's field). (Reprinted with permission from Ref. [6], Copyright 1998 AAAS.)
Figure 6.9 STM images of Au (111) partially covered (lower right) with C
60
molecules obtained with (a) metal tip and (b) C
60
tip over the same area. Gold atoms (
α
) and a small gold cluster (
β
) of two or three adatoms are discernible. The arrow in the bottom right shows a site where a C
60
molecule has been removed by the STM tip. Schull
et al.
[7] Figure 1.
Chapter 7: Physics-Based Experimental Approaches to Nanofabrication and Nanotechnology
Figure 7.1 Steps in formation of a suspended silicon plate above a silicon chip are shown in fields labeled (a) through (f) in this Figure (After Ref. [2].)
Figure 7.2 Single-electron transistor used to sense vibration of freely suspended crystal beam [4, 5]. (b) A “crystal beam” is suspended and free to vibrate at 116 MHz. It is metalized and acts as gate electrode in the FET transistor, denoted “island.” The “island” is connected to source and drain electrodes (right and left) by Al–AlO
x
–Al tunnel barrier contacts. At fixed gate bias and fixed source drain voltage, motion of the gate electrode induces charge on the island, and sensitively controls the source-drain current. (a) Schematic diagram of electrical operation of the device. A motion of the bar by 2 fm (about the radius of an atomic nucleus) can be detected with 1 Hz bandwidth at 30 mK. (Reprinted with permission from Nature: Ref. [4], Copyright 2003 Macmillan Publishers Ltd.)
Figure 7.3 Sketch illustrative of FinFET design of field effect transistor (see Ref. [6]). The channel, of width
W
and height
H
, runs from source to drain at the top. The gate of length L surrounds the channel on three sides, giving strong electrostatic control over the carriers in the channel. The gate insulator is indicated as having thickness
t
ox
. The whole structure, in the design shown, resides on a buried oxide (BOX) layer formed on a semiconductor substrate. Gate lengths in such structures can be less than 10 nm and are projected [6] to reach 3 nm. (Figure reproduced with permission of A. Asenov, Gold Standard Simulations, Inc.)
Figure 7.4 Frequency divider [7] operates at 750 GHz, using RSFQ superconducting junctions. This device produces half-frequency output. Twice the output frequency, 2
f
out
is shown superimposed on
f
in
, both quantities referenced to left-hand scale. Bias current (abscissa) in milliamperes does not signify heat, because the resistance is zero [7].
Figure 7.5 Estimated clock rates in RSFQ (upper) and conventional Si CMOS (lower) projected to the year 2011. (After Ref. [8].)
Figure 7.6 Schematic diagrams of (a) scanning tunneling microscope (STM) and (b) atomic force microscope (AFM). For STM, the tunnel current
I
is maintained constant by tip height, controlled by
z
piezo. For AFM, the force (spring deflection) is maintained constant. (After Ref. [11].)
Figure 7.7 STM fabrication and atomic imaging of gold Au atoms and molecules Au
n
on a surface of (110) NiAl [12]. The NiAl (110) surface (a) has dimpled channels matching the 0.3 nm spacing of Au atoms in Au molecules. A 20-atom Au
20
wire is formed (b)–(f) and its density of states matched to an 1D band model (see Chapter 4) (Reprinted with permission from Ref. [12]. Copyright 2002 AAAS.)
Figure 7.9 Determination of effective mass of electrons in the 1D nanowire. (a) Fits to particle in a box wavefunctions, see text. (After Reprinted with permission from Ref. [12]. Copyright 2002 AAAS.) The effective mass for electrons in the nanowire is determined to be about 0.5, from the
E–k
curve shown in (b).
Figure 7.8 STM measurement of density of states (DOS) versus bias
V
and position along the Au
20
wire.
V
denotes energy of empty state where tunneling occurs, and d
I
/d
V
(
nA
/
V
) measures DOS. (a) Curves of d
I
/d
V
at locations seen in (b). (c) DOS versus
x
along the wire, taken at three different energies [12].
Figure 7.10 Schematic of single-molecule STM organic chemistry on a copper surface [15, 16]. The reaction is the copper-catalyzed transformation of iodobenzene C
6
H
5
I to make C
12
H
10
. This reaction is known as the Ullman reaction, here carried out at 20 K, with the usual thermal activation energy replaced by pulses of energy from the tip. The first pulse is applied in (a)–(b) to dissociate the I, leaving a phenyl group C
6
H
5
. The second pulse is applied in (e) to cause two adjacent phenyl molecules to bind. In (c), (d), and (f), the tip is used to move a molecule.
Figure 7.11 A schematic image of the AFM-type magnetic resonance force microscope (MRFM) device used to image a single electron spin [18]. The 150-nm-wide SmCo magnet attached to the tip oscillates at fixed amplitude (16 nm) at
f
c
= 5.5 kHz in the
x
-direction about 125 nm above the sample surface. The sum of the tip's dipole magnetic field and the static magnetic field (
B
z
≈ 30 mT) add up to about 106 mT, corresponding to an applied microwave resonant magnetic field supplied at
v
= 2.96 GHz. The “resonant slice” is a bowl-shaped surface that extends about 250 nm below the tip. For a spin slightly displaced from the tip's center position in the
x
-direction, the spin is flipped twice per cycle of tip oscillation, providing a cyclic force on the tip. The
x
-component of the cyclic force on the tip can be modeled as a small change in the cantilever force constant, causing a small change in its resonant frequency, which is detected by means of a reflected light beam focused on the tip. The device is mounted in a small vacuum chamber within the bore of a superconducting magnet and is operated at 1.6 K. (Reprinted from Nature with permission from Macmillan Publishers, Ltd.)
Figure 7.12 Data that support the observation of a
single electron spin
with the magnetic resonance force microscope (MRFM), an AFM-derived scanning device [18]. These plots show the spin signal (the shift in the cantilever oscillation frequency) as the sample was scanned slowly in the
x
-direction. The 19 nm shift in apparent position of the spin between (a) and (b) in the Figure was caused by changing the applied static magnetic field from 34 to 30 mT. The spatial resolution in detecting a single electron magnetic moment is seen to be on the order of 25 nm. (Reprinted from Nature with permission from Macmillan Publishers, Ltd.)
Figure 7.13 Aspects of whispering-gallery mode detection of virus particles from solution [19] as described by Frank Vollmer and Stephen Arnold, are shown in panels (a)–(c).
Figure 7.14 Plasmonic resonance of the gold-shell nanoparticle at the infrared frequency of the whispering-gallery mode evanescent field, locally enhances electric field that traps tiny dielectric particle (top of enlarged inset), in this case a small particle of biological interest, according to the authors. The method cannot identify the particle beyond its induced electric dipole moment. Trapping of the particle, independent of its chemical nature, is detected by a small shift in the absorption wavelength of the particular whispering gallery mode, as illustrated in Figure 7.13 [20].
Figure 7.15 Resonance wavelengths of gold nanoshells on silica of gold thicknesses 10, 5, and 3.75 nm at overall diameter 50 nm [21].
Chapter 8: Quantum Technologies Based on Magnetism, Electron and Nuclear Spin, and Superconductivity
Figure 8.1 Energy levels for electron (
μ
e
) and nuclear (
μ
n
) magnetic moments in a high magnetic field
B
, after [3]. (Reprinted from Physical Review Letters with permission from APS.)
Figure 8.2 Spin-polarized densities of states for elemental metals (a) Fe, (b) hcp-Co, (c) Ni, and (d) Cu.
Figure 8.3 CIP spin valve sensor [6] (see also Figure 1.5) illustrating the shorter mean free path for carriers in the antiparallel electrode magnetization case (a), corresponding to higher spin valve resistance. (The horizontal arrows suggest the magnetization direction in the electrode, and the reversal of this direction in the upper electrode indicates that the upper (soft) magnetic layer has had its direction altered by encounter with the ambient magnetic field to be measured.) The device as shown is to be imagined as vertically positioned above a track of magnetic bits in a hard disk. Minimizing sandwich thickness
t
is needed to efficiently capture the fringe magnetic field (up or down with respect to the printed page) and also to make the mean free path for spin-flip (see text) larger than the device dimension. Panel (b) shows the sensor in its low resistance state with parallel magnetizations. (Reprinted with permission from Science, copyright AAAS.)
Figure 8.4 Illustration of the Julliere model [7] of the tunneling magnetoresistance. The currents of each spin projection are proportional to products of their initial and final densities of states, and spin direction is preserved in the tunneling process. Panel (a) shows parallel magnetizations and low resistance, panel (b) shows antiparallel, high resistance case.
Figure 8.5 Curves [9] demonstrating 60.8% tunnel magnetoresistance appearing over a magnetic field range of about 1 Oe (0.1 mT) at room temperature in a tunnel junction device consisting of soft ferromagnet CoFeB, Al
2
O
3
barrier, and FeCo hard magnet. The inset shows the resistance on an expanded magnetic field scale, clarifying the expected hysteresis. (Reprinted from IEEE Transactions on Magnetics with permission from IEEE.)
Figure 8.6 Magnetic random access memory (MRAM) cell based on a magnetic tunnel junction in series with a field effect transistor for bit read selection [10]. Perpendicular bit lines above and digit lines below the magnetic tunnel select a single tunnel junction. Magnetic fields produced by simultaneous current pulses along the control lines flip the magnetization of the free layer of the magnetic tunnel junction. This memory is “nonvolatile” such that the information is retained indefinitely in the event of power loss. Such an element appears at each node of a crossbar array, as shown in Figure 8.7. (Reprinted with permission from Science, copyright AAAS.)
Figure 8.7 Top view [10] of an MRAM array, highlighting the fully selected crossbar bit (arrows and center position) and the ½ selected bits along each current-carrying write line. In toggle-MRAM, all bits are oriented at 45° with respect to the write lines. Simultaneous current pulses in the two control lines are needed to flip the magnetization of the free layer at the intersection junction, and close control is needed to prevent errors if a flip occurs with only one pulse. It is stated that these nonvolatile random access memory arrays up to 4 Mb size have been produced. (Reprinted with permission from Science, copyright AAAS.)
Figure 8.8 Top view [12] of hybrid ferromagnet–semiconductor switching device, based on the Hall effect (see Section 5.4.1). Magnetization
M
can be written into the small sheet of soft ferromagnet by means of a current pulse in an auxiliary current line, not shown. The circled plus symbols at the left edge of the ferromagnetic layer F indicate magnetic flux (local field −
B
z
) pointing down into the semiconductor. This magnetic field, up to 0.16 T, diverts current carriers in the InAs bar as shown, to produce a Hall voltage, which represents the output of the device. This device is nonvolatile and will operate if the size scale is reduced. (Reprinted from IEEE Transactions on Magnetics with permission from IEEE.)
Figure 8.9 Side view [12] of hybrid ferromagnet–semiconductor switching device, showing detail of fringe magnetic field at the “active edge” of the ferromagnetic layer F which points down through the semiconductor layer to produce the Hall voltage. The insulating layer
I
1
comprises 45 nm SiO/3 nm InAs/3 nm AlSb/25 nm AlSb. Insulating layer
I
2
is a 2-µm-thick buffer layer of AlSb. (Reprinted from IEEE Transactions on Magnetics with permission from IEEE.)
Figure 8.10 Geometry of the spin injection device [14]. (a) Scanning electron microscope (SEM) image of device with spacing
L
of the two Co ferromagnetic crossing electrodes equal to 650 nm. Current flows from electrode Co1 through tunnel junction into the Al strip, and then flows only to the left end of the Al strip. There is no bias along the right-hand side of the Al strip between the two Co electrodes. A voltmeter V monitors the potential difference between the Al strip and electrode Co2, as indicated. A voltage is seen to appear when nonequilibrium spin population diffuses along the (unbiased) Al strip the distance
L
from electrode Co1 to electrode Co2. (b) Cross section of the device, showing flow of current from ferromagnetic Co1 into Al and out the left end of the Al strip. (c) Plot of the electrochemical potential (voltage) in the device. The solid line is voltage in absence of spin injection. Dashed lines show spin-up electrons at higher potential than spin-down electrons, which reflect the energy shift of spin-up and spin-down bands in Co, suggested by Figure 8.2b. The horizontal scale,
x
-coordinate, is measured in units of the spin diffusion length, which extends to the second tunnel junction between Al and Co2, where the nonequilibrium spin population produces an open-circuit voltage
V
. (Reprinted from Nature with permission from Macmillan Publishers, Ltd.)
Figure 8.11 Observation [14] of the Johnson–Silsbee [13] effect is an open-circuit voltage change that appears when the magnetization directions in the two Co electrodes are reversed. This means that the nonequilibrium spins that diffuse from Co1 to Co2 are oppositely directed to those in electrode Co2, which they sense through the tunnel barrier. Negative voltage sections of the measured
V
/
I
at 4.2 and 293 K appear under external magnetic fields that make the magnetization directions in Co1 and Co2 antiparallel, reversing the roles of the spin-up and spin-down bands, as illustrated in Figure 8.3. The magnetization reversals were accomplished [14] by sweeping a magnetic field
B
oriented parallel to the Co electrode directions. The scheme depends on having Co1 wider than Co2, so that its magnetic field (in the absence of a larger external field) causes the magnetizations in 1 and 2 to be antiparallel. (Reprinted from Nature with permission from Macmillan Publishers, Ltd.)
Figure 8.12 Co electrodes are shown [15] contacting two ends of a carbon multiwall nanotube, with a contact spacing of about 250 nm (a). The diameter of the MWNT was about 30 nm, ranging from 10 to 40 nm in other cases. The Co electrodes depicted in panels (b) and (c) are polycrystalline Co deposited by thermal evaporation. The substrate is a semi-insulating Si wafer covered by a 200-nm-thick layer of SiO
2
. The magnetic domain size in the Co is reported as 50 nm. (Reprinted from Nature with permission from Macmillan Publishers, Ltd.)
Figure 8.13 Measurements of the differential resistance of the Co/nanotube/Co devices (a), (b) and (c), consistent with the ideas of spin-coherent transport and with a coercive field (see Figure 5.18) in the Co on the scale of 20 mT. The high-resistance regions are interpreted [15] as arising when the magnetizations of the two Co contacts are antiparallel. It is believed that the coercive fields of 50 nm domains in polycrystalline-evaporated Co films are of this order. The magnitude of the resistance change is interpreted within the TMR formula given in connection with Figure 8.10, as consistent with spin polarizations in the Co on the order of 34%, which would give a resistance change of 21%. The observed change of resistance, 9%, is consistent with the above if the spin scattering length is shorter than the free length of the nanotube, giving the estimate for the spin scattering length of 130 nm. (Reprinted from Nature with permission from Macmillan Publishers, Ltd.)
Figure 8.14 Universal magnetic dot logic gate [16] [17]. The central nanomagnet chooses its magnetization direction (up or down, because of its elongated shape) based on the
sum
of the magnetic fields from similar input nanomagnets 1, 2, and 3. The
output
simply is antiparallel to the central magnet, and thus represents the
negative
of the majority input. This is simple magnetostatics, playing with bar magnets, whose dipole magnetic fields lead to antiparallel near-neighbors in the
x
-direction, and parallel (ferromagnetic) near-neighbors in the
y
-direction. The only quantum aspect of the situation is the formation of the ferromagnetic domains, which is a profoundly quantum effect, as discussed in relation to Figure 5.20. (Reprinted with permission from Science, copyright AAAS.)
Figure 8.15 (a) Transmission electron microscope (TEM) images [17] of planar nanomagnets of nominal dimensions with 30 nm thickness of permalloy film. Horizontal spacing between (vertically elongated) nanomagnets is 25 nm. The horizontally oriented magnets on the left-hand side are drivers, and those on the right-hand side read the output. An externally applied clocked
B
field arranges the driver magnet directions of magnetization. The idea is that the driver magnets are big enough to be dictated by the applied fields, while the smaller magnets respond only to their near-neighbors. Arrows suggest local magnetization directions and how these orientations were driven by the inputs. (b) Magnetic force microscope (MFM) image indicating the magnetization directions in the same structure as indicated in (a). The location of the nanomagnets is drawn superimposed on the MFM data. (A larger set [17] of MFM images demonstrates the full majority logic gate functionality of the device.) It is stated that the magnetization switching time in these nanometer-scale magnetic dots is about 100 ps. It is stated that this majority gate logic scheme would have an inherent operating speed of 100 MHz, with energy dissipation (heating) of about 1 eV per switching event. On this basis, an array of 1000 gates would have a total dissipation of 0.1 W, assuming each device switches once per cycle.
Figure 8.16
V–I
measurements [18] on a DNA-templated quantum interference device at extremely low temperature, 0.285 K (such that the whole structure can become superconducting at ), show hysteretic switching between fully superconducting states and two resistive states, which are generated by application of slightly different perpendicular magnetic fields (about 0.25 and 0.5 mT). In order to generate these curves, the bias current was swept from large negative values to large positive values and return. (Reprinted with permission from Science, copyright AAAS.)
Figure 8.17 (a)DNA-templated two-nanowire SQUID-like device [18]. Two strands of DNA are stretched across a trench etched into SiN/SiO
2
on a Si chip. As suggested by the pictorial representation, long DNA strands are simply deposited (in a drop of water) on the SiN layer, above the trench; in the process of drying, they self-assemble to pull the crossing sections tight across the trench, normal to the trench orientation. Subsequently, the molecules and the banks are coated with superconducting Mo
21
Ge
79
. Resulting are two weak links (MoGe-coated DNA strands) spaced by dimension 2
a
about 595 nm across a gap of width
b
about 120 nm, creating a (SQUID interferometer) area 2
ab
, which is normal to an applied
B
field, in the
z
-direction. The included magnetic flux is nominally 2
abB
, but is found in the experiment to be much larger, because [18] the
B
field penetrates significantly into the “few nanometers thickness” MoGe contacting electrode “leads,” whose width is 2
l
. The field penetrates the leads to a distance on the order of 9–15 µm. Current flows to the device through these MoGe contacts of nanometers thickness and width 2
l
. (b) Scanning electron microscope image of the device shows that the crossing wires (superconducting weak links), as located by the stretched DNA strands, lie accurately straight and parallel across the trench. (Reprinted with permission from Science, copyright AAAS.)
Figure 8.18 Resistance versus magnetic field (perpendicular to device) at temperatures 1.2–1.9 K. Each oscillation represents a change in flux through the device of one flux quantum . The solid lines are generated [18] by a model and assume a period
B
for the oscillations of 456 μT. This period is far short of the conventional value , which is explained [18] in terms of penetration of magnetic flux into a much larger area, due to the extremely thin nature of the contacting MoGe leads. The lack of zero resistance in this nanoscale superconducting device is attributed to “phase slips” in the narrow wires, an effect well known in superconducting nanowires. (Reprinted with permission from Science, copyright AAAS.)
Figure 8.19 The simplest stage [19] of the Josephson junction logics. (a) Equivalent circuit and scheme of its operation. The input current is the signal plus the bias, and when the sum exceeds the critical current for the junction, it switches (passes one flux quantum) and the current (abruptly) is sent to the load resistor. How this happens is different when using (b) an underdamped (unshunted) and (c), (d) an overdamped (shunted) junction. The situation in (b) was found eventually to be detrimental, in that it was “latched,” and hard to get out of. The use of the “shunted junction” (c) avoids the latching, but now the only message of the junction switching is the voltage pulse, sent down the transmission line, as shown in (d). This is the present state of the art, keeping track of the SFQ pulses. The curves shown of voltage and phase versus time are the result of sophisticated numerical calculations [19], which are believed to be realistic. (Reprinted from IEEE Transactions on Applied Superconductivity with permission from IEEE.)
Figure 8.20 A sensitive
V
/
F
ADC [20] based on the
V
-flux transfer characteristics of the SQUID and followed by an RSFQ binary counter. The train of SFQ pulses coming from the DC SQUID device is counted by an RSFQ counter of the type described in Section 7.5. (Reprinted from Proceedings of IEEE with permission from IEEE.)
Chapter 9: Silicon Nanoelectronics and Beyond
Figure 9.1 (a) Schematic of device and instrumentation, using a charged AFM tip (see text) to image the quantum point contact (QPC) conductance d
I
/d
V
, at small fixed source–drain voltage. (b) Quantum steps are observed in the device conductance d
I
/d
V
at 1.7 K (with no tip present) as a function of the voltage
V
g
applied to the point contact gates, a geometrical constriction. At −1.3 V, the flow is shut off. The first mode of electron flow, corresponding to d
I
/d
V
= 2
e
2
/
h
, has onset at
V
g
= −1.2 V. The inset shows the geometry of the point contact gates. As they are made less negative, the effective width of the channel between them is increased, allowing additional modes of electron flow through the constriction [2]. (Reprinted with permission from Science, copyright AAAS.)
Figure 9.2 Images of d
I
/d
V
in units of
G
0
of the
n
= 1,2 (upper panels) and
n
= 3 (lower panels) modes of electron propagation through the point contact constriction [2]. (a), (b) and (c) Images of electron flow on the left-hand and right-hand sides of the constriction. (d),(e), (f) Calculated wavefunctions for modes 1 to 3, respectively. (g), (h), and (i) Plots of the current intensity measured as a function of angle. (d),(e),(f), (j), (k) and (l) Corresponding models. (Reprinted with permission from Science, copyright AAAS.)
Figure 9.3 Stubbed quantum waveguide device [3]. Model of four-gate electron stubbed tuner is shown. Upper and lower portions show outlines of the four electrodes labeled S
1
, T, S
2
, and B. The curves labeled
y
t
and
y
b
are theoretical estimates of the upper and lower boundaries of the electron gas under typical bias conditions. The source and drain, as in the device of Figure 9.1, again located to the left and to the right, are biased with a small voltage, and the conductance d
I
/d
V
is measured and is also calculated. In this device, the electron gas is at a depth of 80 nm, Fermi energy of 8 meV, and very high electron mobility and long mean free path. (Reprinted from Physical Review B with permission from APS.)
Figure 9.4 Calculated conductance d
I
/d
V
curves as function of gate voltage [3]. Curves (1) and (2) differ in the details of the biasing of the four electrodes of the gate. These curves clearly show the two plateaus for
n
= 1 and
n
= 2 as were observed in Figure 9.1. Here, the finely defined minima interrupting the basic plateaus in d
I
/d
V
are believed to arise from interference of the main propagating electron modes from left to right, with more closely spaced electron states resulting in confinement within the four-gate structure. Such fine structures are observed (not shown) at very low temperatures (0.09 K), but disappear from experimental measurements at higher (but still cryogenic) temperatures (2.5 K). (Reprinted from Physical Review B with permission from APS.)
Figure 9.5 Inset shows the geometry of single-wall nanotube (SWNT), with attached source and drain ohmic contacts, mounted above an insulating layer on a conducting degenerately doped Si wafer which acts as gate electrode. The measured (
V
= 0) conductance d
I
/d
V
is shown (solid curve) versus applied gate voltage
V
g
plotted in the range −8 V to about 1 V. The dotted curve shows a sinusoidal function with the same average period. The length of the nanotube is about 200 nm [4]. (Reprinted from Nature with permission from Macmillan Publishers, Ltd.)
Figure 9.6 Conductance (d
I
/d
V
) interference images, with gray-scale calibration shown on the right [4]: (a) from a 530 nm device and (b) from a 220 nm device. Inset in (b) shows that the spacing of the oscillations with gate voltage is inversely proportional to device length,
L
. The oscillating features are more accurately described as dips in conductance, with spacings
V
c
3.5 meV for the 530 nm device, and 6.5 meV for the 220 nm device. These
V
c
values are deduced from the locations (see arrows) along the
V
-axis where adjacent positively and negatively sloping lines intersect. The inset also shows a solid line with a slope equal to
hv
F
, where
v
F
= 8.1 × 10
5
m s
−1
is the Fermi velocity in the nanotube. (Reprinted from Nature with permission from Macmillan Publishers, Ltd.)
Figure 9.7 Optical micrograph [5] of an SWNT “chemi-capacitor.” The region between the widely spaced electrodes is covered with an optically transparent but electrically continuous network of SWNTs (shown in the inset atomic force microscope image). The capacitance was measured between this top surface (the conducting nanotube network) and the underlying conducting Si substrate, using an AC method. The capacitance is sensitive to the effective nanotube diameter, which is slightly increased by (
E
-field-induced) adsorbates (The widely spaced electrodes were interdigitated to allow simultaneous measurement of the network resistance, but this was not employed in the present work, and the two large-scale electrodes were electrically shorted.). (Reprinted with permission from Science, copyright AAAS.)
Figure 9.8 Capacitance changes on a 0.1% scale are clearly registered with the adsorption of a typical polar molecule, DMF (dimethylformamide) administered in 20 s doses at the indicated concentrations [5]. The DMF molecule has a relatively large fixed dipole moment, 3.62 D. (Reprinted with permission from Science, copyright AAAS.)
Figure 9.9 (a) Schematic of carbon nanotube cross-bar array as RAM [6]. Orthogonal control lines terminate (b) at the edges of the array. The
m
ground-level nanotubes are arranged accurately forming a grid lying directly on the (oxidized) ground plane surface. The
n
elevated nanotubes are supported, midway between nanotube intersections, by an accurately located square array of
mn
support pillars. The tube diameters may be around 1 nm, and the pillar spacing may be as small as 20 nm. Once an upper tube approaches closely enough to the crossing nanotube immediately beneath, it can lock into a near-neighbor (small tunneling resistance) configuration by short-range van der Waals attraction. A subsequent electrical voltage pulse (simultaneous along the
m
and
n
control lines) can restore the upper tube to its relaxed, linear, and high-resistance configuration. (Reprinted with permission from Science, copyright AAAS.)
