• lectures accompanying the book: Solid State Physics: An

Introduction, by Philip Hofmann (3nd edition 2022, ISBN:

978-3-527-41410-9, Wiley-VCH Berlin.

www.philiphofmann.net
1
 README

• This is only the outline of a lecture, not a final product.

• Many “fun parts” in the form of pictures, movies and
examples have been removed for copyright reasons.
• In some cases, www addresses are given for particularly
good resources (but not always).
• I have left some ‘presenter notes’ in the lectures. They are
probably of very limited use only.
• I do actually prefer to do give the lectures on a blackboard,
using presentations as little as possible.

2
 The electronic properties of metals

• What is a metal, anyway?

• A metal conducts electricity (but some non-metals also do).

• A metal is opaque and looks shiny (but some non-metals

also do).

• A metal conducts heat well (but some non-metals also do).

We will come up with a reasonable definition!

3
 Electrical properties of metals:
Classical approach (Drude theory)
at the end of this lecture you should
understand....

• Basic assumptions of the classical theory

• DC electrical conductivity in the Drude model

• Hall effect

• Plasma resonance / why do metals look shiny?

• thermal conduction / Wiedemann-Franz law

• Shortcomings of the Drude model: heat capacity...

4
 Chapter 4
Drude’s classical theory
Electrical properties
• Theory by Paul Drude in 1900, only three years
Electrical
after the electron was discovered.
metals propertie
• Drude treated the (free) electrons as a classical
metals
ideal gas but the electrons should collide with the
stationary ions, not with each other.
2⇤
= speed
average
k
1 3
12 mvt2= 2 3kB T
2

mvt = kB T
2 3k2 BT
vt =
8 m kB T
vt =
v
so at ⇥ 10
t room
5
ms 1
⇤ temp.
m
vt 10
= ms
5 1
5 vt ⇤
emperature this result in time and plays
an average a funda
speed of vt
ity per dergounit atime
Drude’s for the
scattering
classical event electrons
theory with the ions to ui
it follows that if we ch
he ions time
is
relaxation 1/τand. plays
time τ is acalled
fundamental
the role in
relaxati
es that the probability will per unit
on time forgo
average thethe
elect
it follows that if we choose an electro
role
attering inevent
(average the
time theory.
with
between the ionsFrom
is 1/τ the
scattering
scattering events)
will on average go the1time2 τ to
τ isdefiniti
. event called
will th
be
plays a fundamental role in the theory. 3 the ne
From t
n electron at some
scattering event will bewe random
can also
mvthe =
time,
define k
t time τ ag B the
the
T m
that if wemean choose an electron 2at some 2random
o the next
we canscattering
free path λ =
also define the τ v
event
t . If
meanandthe origin
free the o
path lab
rage go the time τ to the one next
would scattering
8
assume kevent
T
that
B
me λ = τ v . If the origin of = the scattering
eventτ willago. be the With
t
timethe τ ago. relaxation
vt With
⇤ the time
relaxa
m tha
one would assume that λ ≈ 1nm so
o define
ee paththe mean freeFor
between path
two between
scattering
the description two scatter
events
of1 alm
the origin of theisscattering are ⇤ 10
vtthe 5
ms
collisions w
attering For the aredescription
thenecessary
collisions to with
of almost all the
determine ion
ho
propert
−14
assume that λ ≈ 1nm so that τ ≈ 1 × 10 s.
is necessary
m so that τ ≈ 1 × 10 to example
determine the
−14 electrical
how many
s. vt = condu
electro
example the electrical density n, i.e. the
conductance. ⌃ The of
number re
ription of almost all properties within the Dru
6
ectrons contribute to Number
Table 6.1: a process, for electrons per atom and
of conduction
Conduction
density of electron Density
selected metals.
he relevant quantity is the electron n
per volume. The electron density is ZV n(1028 m −3)
metal
#atoms Li 1 4.7
contributes ZV conduction electrons
per Na 1 2.65
etals, ZV would be 1, for the alkaline
volume K 1 1.4
here are ρm /A atoms per cubic metre Rb 1 1.15
−3 Cs 1 0.91
gm and A is A is the atomic mass
Cu 1 8.47
ron density
calculatenasis ZV ρm /A. Values Ag
of 1 5.86
6.1. Au 1 5.9
#valence Be 2 24.7
density atomic Mg 2 8.61
electrons
mass Ba 2 3.15
per atom
Fe 2 17
Al 3 18.1
Pb 4 13.2
Sb 5 16.5
Bi 5 14.1
7
 d d 1
me vF2 = EF
⇤...this must ⇤ surely
⇤ be wrong....
⇤⇤ 2 ⇤
) = exp(i r) + exp(
(+) = exp(i i r) r)= + 2exp(
cos( i r)r)⌅(r) = 2= cos(ikr (4.2
r)
• The d
electrons should dd
strongly interact d
with
e
deach other. Why
2⇤
d
⇤
don’t they? ⇤⇤ ⇤⇤ k= ⇤
) = exp(i r) ( ) exp(
= exp(i i r) r)= 2i sin( i r)r) = 2i⇥sin( (4.2
exp( r)
• The d
electrons should d d
strongly interact d d
with the = 90
lattice ⇥
ions.d
Why
h don’t they?
h h h ⌅k (⌫r) =huk (⌫r)eikr
⇥= = ⌃⇥ = == ⌅⌃ =⌅ (4.2
• Using 2m 2mE
p classical ptr 2m
statistics for the2mE
electrons
3 tr cannot
2m kT 2m
be3
k = ± kT
right.
n⇤
This is easy to see: 2 2d
⇤ ⇤
(+) = exp(i
9 d r) + exp( i r) = 2
⇥ ⌅ 6 ⇥ 10for using
condition
9
m⇥ ⌅classical
6 ⇥ 10 statistics
m d (4.2
⇤ ⇤
( ) = exp(i r) exp( i r) = 2i
⇥⇧l ⇥ ⇧ l is some d Å d (4.3
h h h
⇥ = = ⌃
de Broglie wavelength of an electron: =⌅
p 2m 2mEtr 2m 32
h h h
= =p =q for room T ⇥ ⌅ 6 ⇥ 10 9
m
p 2mEtr 3
2m 2 kB T
⇥⇧l
8
 but:

In a theory which gives results like this, there must

certainly be a great deal of truth.

Hendrik Antoon Lorentz

So what are these results?

9
 Drude theory: electrical conductivity
Chapter 4
we apply an electric field. The equation of motion is
3
mvt = kB T
2 dv (4.
2 me
dt
= eE
Electrical
= 10 ms 5 properties
1 o
(4.
integration gives
eEt
metals
v = t
v(t) =
m
(4. e
⌃ is the average time between collisions then the
and if
average drift speed is
eE
a=
v =m
eE⌧ 1 3 = 10 2 (4.
ms 1
eme
for E ⇡ 10Vm mv 21
=we k
getB v
T
2 t
2
eEtremember: 10 vt = 10 ms
5 1
 Drude theory: electrical conductivity
n|v|A 1s
number of electrons passing in unit time

A n|v|A
current of negatively charged electrons

en|v|A
current density
j = nv( e)
Ohm’s law
2
eE⌧ ne ⌧
and with v=
me
we get j= E
me
11
 E ⇤ 10Vm 1
(4.6)
Drude theory: electrical
v = 10 2 vms
2 v
=1
1=eE⇧ µEconductivity
= µE (4.7
v = 10 ms me (4.7)
ne 2
⇧ e⇧
ne2 ⇧ Eµ
J =Ohm’s v=
law = mµE (4.8
J = me E2 e (4.8)
mene ⌧ e⇧
ne2j =
⇧ I µ =
E
J =
J = ne ⇧ A
2
E = neµE e e) = nµEe
=m n
e m
v( (4.9
J = me E I= neµE (4.9)
m
Je= = n v( e) = nµEe
A E|n E vA|
J =j ⌅E
= =E E= (4.10
J = ⌅E = ⇤|n ⇢ vA| (4.10)
⇤ en vA
2
and we can define ne2 ⇧ en vA
⌅ = ne ⇧ = nµe (4.11
the conductivity ⌅ =eE⇧ m = (4.11)
v= m e = µE
e nµe
1 (4
m m e = 1
and the ⇤ = e
m e
⇤ = ne22 ⇧ = nµe (4.12
(4.12)
resistivity v= ne ⇧
µE nµe (4
ne22 ⇧
dv
dv e⇧ ⌅⌅ = ne ⇧
and the ame =
µ=
ame = m = =eE
me = eE m
e mee |v| = µ|E| (4.13
(4
(4.13)
mobility dt
dt me
1
eEt1
12 11
eEt
 Ohm’s law
2
ne ⌧
j= E
me
• valid for metals

• valid for
homogeneous
semiconductors

current I
• not valid for
inhomogeneous 0

semiconductors

• not valid for metal 0

voltage V (arb. units)
contacts to
semiconductors
13
 Electrical conductivity of materials
10 25 Superconductor

10 8 Silver, Copper
Aluminium
10
6 Lead
4
10
2
10
1 Germanium

10 -2 Silicon
σ(Ω m )

10 -4
-1 -1

-6
10
-8
10
-10
10
Glass
-12
10 Diamond
10 -14
Sulphur
10 -16
-18 Quartz
10
10
-20 PET
10 -22
14 10 -24 Teflon
 How to measure the
conductivity / resistivity
V
I I

R1 R2

Rsample

• A two-point probe can be used but the contact or wire

resistance can be a problem, especially if the sample has a
small resistivity.
 How to measure the
conductivity / resistivity

I V I

R1 R2 R3 R4

Rsample Rsample Rsample

• The problem of contact resistance can be overcome by

using a four point probe.
er 4
Drude theory: electrical conductivity
Chapter 4 50
Ag Cu
273 K Cu
T=77 K
rical properties of
6
40
σ( x10 Ω m )

5 Au

σ (10 Ω m )
-1

Electrical propert

-1
Ag Al
Al

s
4 30
-1

-1
Be
7

7
I Mg 20 Au
J= = n ve = nµEe (4.17)
metals
Na Mg
2 Fe
A K 10
Na
Li
1 Li Fe K
Rb
0
Cs
Ba nPbBivA Sb 0
Rb
Cs Ba (4.18)
PbBi Sb
1 3
5 mv
10 nt 28=
2
vAe
15 kB20T 25 5 10 (4.1)
(4.19)
15 20 25
2n( x10 m-32) n (10
28
m )
-3
2⇤
vt ⌃==10nµe
5
ms 1 (4.20)
=
(4.2) k
ne2 ⌥
⌃ v=t = 1 (4.21)
(4.3) 3
m⇧e mvt = kB T
2
2 2
line 1⇤ 1nm1 (4.4)
⇧= = (4.22)3kB T
⌃ nµe
⇧ ⇤ 1 ⇥ 10 s 14 vt =
(4.5) m
17
⇥ = /vt
 Drude theory: electrical conductivity

• Drude’s theory gives a reasonable picture for the

phenomenon of resistance.

• Drude’s theory gives qualitatively Ohm’s law (linear relation

between electric field and current density).

• It also gives reasonable quantitative values, at least at room

temperature.

18
 The Hall Effect
E
B d ⇥ 43meV
z
z

y Ed
EF = Eg
2
x
n = Ce (Eg µ)/kB T
H

x
n=
jx N + n
d i

• Accumulation of charge p = Ce
leads
µ/kB T
to Hall field EH.
• Hall field proportional to current density and B field
np = ni pi = constant
EH = RH jx Bz EH = RH Jisx B
called
z Hall coefficient
 y The Hall coefficient

z
x
eEH = eBz vx
F= ev ⇥ B and definition E H = RH j x B z
F= eE H for the steady
state we get

|-e H|
EH EH
|-e x| RH = =
j x Bz envx Bz
|-evx Bz|
v x Bz 1
= =
envx Bz ne
2
ne ⌧
electron density form Ohm’s law? j = E
me
 data are marked by the element’s names; the calcu
The Hall coefficient
Measured Hall coefficient for selected metals, normalised by the
sult as calculated from the electron densities given in Tab. 6.1.

metal the Lorentz

measured force
RH in units in the opposite direction, thus
of −1/ne
Li 0.8 Ohm’s law contains e2
Na 1.2 −eEH = −eBz v.
K 1.1 2
From this we get ne ⌧
Rb 1.0 j= E
Cs 0.9 me
EH EH
Cu 1.5 RH =But for R =the sign of e
Ag 1.3 −Jx Bz H −(−e)n(−v)
Au 1.5 is important.
Be and finally
-0.2
Mg -0.4
vB z −1
RH = = .
Al -0.3 −envBz ne
Bi 38,923measuring the Hall coefficient provides d
Therefore,
to the electron densities shown in Tab. 6.1. Tab. 6.
coefficients for some of the metals in Tab. 6.1, no
value −1/ne.
What would happen for positively charged
carriers?

|-e H| |e H|

|-e x| |e x|

|-evx Bz| |evx Bz|

Imagine that we had positive carriers with a carrier den

6.3(b)). Then it is easy to show that

−1 1
RH = RH = ,
ne pe

i.e. we get2a positive RH . The notion of positive carriers doe

ne τ
 Drude theory: why are metals shiny?

• Drude’s theory gives an explanation of why metals do not

transmit light and rather reflect it.

23
 microscopic
microscopicinterpretation.
interpretation.
Look at definitions
Some again
relations from basic optics
Look at definitions again
Look at definitions again
wave propagation in matter
~ =E
E ~ 0e
i(kx !t)
~ , ~ !t) (4)
i(kx !t)
E~E == E
~
E 0ee i(kz !t)
E(z, t) = E00 e
i(kx ,, (4)
(4)
plane wave

2⇡N 2⇡N
complex
2⇡N index
k= and kN= =n+ and
i N =n+
(5)i (5)
k of 0 and N
= refraction = n + i 1.0
(5)
0 0 0.5

N complex index i( 2⇡nof

z refraction
!t) 2⇡
z
0.0

N complex index of
NE(z, refraction
t) = Eindex
complex 0e of refraction
0 e 0 -0.5

-1.0

Maxwell relation 0.6

0 20 40 60 80 100 120

Maxwell relation
Maxwell relation
N = N (!)
0.4

0.2

0.0

p p -0.2

-0.4

p p N = p ✏ = p ✏ r + i✏ i
0 20 40 60
(6) 80 100 120

Maxwell
N =relation
✏= N✏r =+ i✏✏i = ✏r + i✏(6)
i (6)
Kramers-Kronig relations
p
Kramers-Kronig
i((2⇡N/ 0 )zrelations
!t) ⇡ Z 0
i((! 0 ✏/c)z !t)
E(z, t) = Erelations
Kramers-Kronig 0e =E 10!e ✏i(! ) 0
✏r (!) = ✏r (1) + Z 0 020 0 2 d! (7)
0 ⇡02 1 !! ✏i(!!) 0
✏r (!)⇡physics
all the interesting = 1 !is✏in
Z ✏r (1)
i (!
+the)dielectric
0 0 02 function!
2
d! (7)
✏r (!) = ✏r (1) + 0 24 2 d! ! !
(7)
 Free-electron dielectric function
one electron in d2 x(t) i!t
time-dependent field
me 2
= eE0 e
dt

i!t
Ansatz x(t) = Ae

eE0
and get A= 2
me !

the dipole moment

for one electron is P (t) = ex(t)

and for a unit volume i!t ne2 E0 e i!t

of solid it is P (t) = nex(t) = neAe =
me ! 2
25
 m e ω
hand, we know the general relation between the E and D fi
Free-electron dielectric function
other hand, we know the general relation between the E
2 i!t
D = ϵϵ0 E = ϵ0 E +neP,E0 e (5.
P (t) = 2 P,
D = ϵϵ0 E = m ϵ0eE!+
P
t we use ϵ=1+ . (5.
D = ✏✏0ϵE0 E= ✏0 EP+ P
ϵ = 1 + .
our result for the polarisation P , ϵ(5.25), we obtain an expr
0 E
electric function ϵ P (t)
is and
to getour result✏for
= 1the
+ polarisation P , (5.25), we obta
2 ✏ E e i!t 2
the dielectric ne ϵ 0 0 ωP
function
ϵ=1− = 1− (5.
ϵ0 me ω 2 2 ω2 2
so the final
ne ωP
lledresult
plasma ϵ = 1ω−
frequency given
P ϵ m ω2by = 1 −
is 0 e ω2
2
ne
so-called plasma frequency
2
ωP= . ωP given by
is called (5.
me ϵ0 the plasma frequency
2
26
2 ne
 sion for the dielectric function ϵ
4⇡i
Meaning ✏(!) of the
= (1 + plasma ). frequency
(22)
2
! ϵ = 1 − 4⇡i = 1 − P
ne ω 2
✏(!) = (1 +ϵ0 me ω 2 ). ω 2
high frequency !⌧ >>
limit:function
the dielectric 1 Drude model is!
in the
with the so-called plasma frequency ωP given by
high frequency limit: !⌧ >> 1
2
2
!P 3e 2
2 ne
✏(!) = (1 with !with 2
)= ω P =(23) .
P
!2 mr 3 me ϵ0
2 s 2
This expression for the
! P
dielectric 2
function
4⇡ne
is our final
✏(!)
!P plasma frequency = (1remember ) with ! P = .
explain that metals reflect ! pvisible light? To see mthis co
! < !(5.22).
P : ✏E(z,
is We
realt)have
and i((2⇡no ✏/
negative,
to distinguish 0 )z !t)two cases: for
propagating
between
! plasma= E 0 e
frequency √
1.0
solution
negativeP number. Therefore ϵ is purely imaginary an
0.8
p
0.6

0.4
exponentially
!~ < ! :damped
✏~εis
because of E(!, x)P= E0e = E0e real
ikx
real andpenetration
~
and ! ✏xno of
i cnegative,
negative, wavethe nowave into
propaga
propagation t
strongly absorbed. Following theisFresnel
opaque equations des
0.2

metal
0 20 40 60 80

solution 10
100 120

1.0

and reflection of light at a boundary, strong absorption p

ε real is !
~ ~ ~
0.5

and positive, propagating

ikx waves
i c at
✏x the
0.0

-0.5
implies! >
that! :
the✏ positive
becauseP of E(!, x)metal
light =E
also e
strongly
0 = E
reflected
is transparent 0 e
-1.0

0 20 40
ω > ωP , on the other hand, ϵ is real and positive and (5
60 80 100 120

reflectivity of metals
27 10
 0E
plasma frequency: simple interpretation
ne2 ⌅2
! > !P : ✏ positive= 1 =1 P
s 0 m⌅
2 ⌅2
ne2 reflectivity of metals
!P = values for the plasma energy h̄⌅P
show plasmon picture
me ✏0
measured calculated
K 3.72 eV 4.29 eV
Mg 10.6 eV 10.9 eV
Al 15.3 eV 15.8 eV
Si 16.6 eV 16.0 eV
Ge 16.2 eV 16.0 eV
surface plasmon
displaced mode
electrons(longitudinal wave)

+ --
+ iqk rk qk--|z|
+
r) =
(~F 0e e --
. (23)
+
. 12--

. ions
• longitudinal collective mode of the whole electron gas
28
.
 the Wiedemann-Franz law

κ
= constant
σ

= LT
⇥

• Wiedemann and Franz found in 1853 that the ratio of thermal and
electrical conductivity for ALL METALS is constant at a given
temperature (for room temperature and above). Later it was found by
L. Lorenz that this constant is proportional to the temperature.

• Let’s try to reproduce the linear behaviour and to calculate L here.

29
 v = 10 ms (4.7)
the Drude jmodel, q 2=The ⇥⇥T Wiedemann
the ratio of thermal Franzand law (4.127)conduct
electrical
y calculated. ne ⇤⇧
The ⇤
J= E thermal conductivity is that (4.8)of a classical g
described m 8 k B T
by an equation similar to (5.17), only using corresp
vt = e estimated thermal (4.128)
conductivity
ties (heat 2
ne ⇧ capacity, ⌅ speed m and mean(from afree
classical
path) ideal
forgas)
the electr
J= E
1 2 = neµE (4.9)
1is given by (6.4). The result is
ectrical⇥m conductivity
= vt ⌃ cv = ⇤vt cv
e (4.129)
3 E 3
J = ⌅E = 3 κ 3 kB2 (4.10)
Cv = ⇤ NA kB = T = LT, (4.130)
2 σ 2 e2
ne2 ⇧
⌅= nv t ⇤k B (4.11)
=e
is just the ⇥Wiedemann-Franz
m law with the only problem (4.131) being
not no anything
small by about me the
2
about,
a actual but
factorquantumfortunately
of two.mechanical these
Drude, however,cancel about. Work
had made a m
result is
details
nv ⇤
⇤kof=m this is
v 2left to the reader.
k m 4k 2
T The final (4.12)
result is
actor
=
t of Btwo ne
= in
2 ⇧ this calculation
B
= B such
= LT that L came out
(4.132) almost
ore 2nehis 2theory
⌃ was 2e2 in impressive
κ ⌅e 2 2 2quantitative agreement with
π kB
dv = T = LT.
ental
amedata.= It = eEbe noticed
meshould σ e2 Drude’s(4.13)
3 that mistake was rathe
dt
d to do with the scattering probabilities of the electrons) and th
8
s givesdiscovered
adily L = 2.45 eEt 13
×by 10other
WΩ K −2
which
researchers. agrees very well with the
v(t) = (4.14)
this is 3, more or less....
tal30 data for many me metals.
 Comparison of the Lorenz number to
experimental data
at 273 K
metal 10-8 Watt Ω K-2
Ag 2.31
Au 2.35

1 2 Cd 2.42
= vFCu
⌅ cv 2.23 (7.59)
3
Mo 2.61
2 ⇥2/3
h̄ 2N
Pb 2.47
2
vF = 2 3⇥ Pt 2.51
(7.60)
me V
Sn 2.52
⇥3/2 ⇥1/3
2
⇥ 2 1 2me W h̄ N3.04
= kB T 2 3⇥ 2
(7.61)
3 2⇥ h̄ 2 Zn
2m e V2.31

2 2
⇥ kB
= T = LT L = 2.45 10-8 Watt Ω(7.62)
K-2
31 ⇤ 3 e2
 Failures of the Drude model

• Despite of this and many other correct predictions, there are some
serious problems with the Drude model.

32
 J= = ve = tem
Atn room nµE
A
Drude theory: electrical conductivity
• Onen vA assumes
1 3
50 Cu
T=77 K mvdergo
nt =
2
vAeakBscatt
T
2 2
time and pla
40 vt ⌃==10 nµe
line
5
ms 1
it follows th
σ (10 Ω m )

2
-1

Ag Al ne ⌥
30 ⌃will
v=t =on avera
m⇧e
-1

scattering ev
1⇤can1nm1also d
7

20 Au
⇧= =we
Mg Fe
Na ⇧ ⇤λ1⌃= ⇥ τ10vnµe.14
t If s t
10 Li
K one would E as
Rb J = ⌃E =
Cs Ba PbBi Sb eE ⇧
0 a=
For
h̄ 2 the
d 2 mdescrip
⇥ e
5 10 15 20 25
is necessary =to E⇥
eEt de
n (10
28
m
-3
) 2mv e=dx
2
at =
example the m elect
e
⇥ = A sin(kx x)
density n, eE⇧ i.e. the
2 v2= 2 2 2
h̄ kx by
calculated m h̄e assu
⌅ nx
33 E= =
 I
J =
Failures of = n ve = nµEe (4.
A the Drude model:
metals the mean free path
n vA (4.
1 3
mvnt =
2
vAekB T (4.
21 23
t = (4.1)
2
⌃ =
mv nµe
vt2 = 10 ms
5 2
kB T1 (4.
(4.
vt = 10
nems
5
2
⌥
1
(4.2)
⌃ v=t = (4.
m⇧e
vt = (4.3)
⇧
1⇤⇤1nm
1nm 1 (4.4) (4.
⇧= = (4.
⇧ ⇤1⌃1⇥⇥1010nµe ss (4.5) (4.
14
⇧⇤ 14

E
J = ⌃E eE = (4.
a = eE ⇧ (4.6)
a = me (4.
2 2 me eEt
⇥=
• h̄ v d= at
Is there something wrong with=eEt
2m dx
v e= at
2
the
meE⇥
= path?
seemingly sensible (4.7) (4.
(4.
assumption for the mean freeeE⇧me
=
⇥ = Av sin(k x x) (4.8) (4.
me
eE⇧
2 v2= = 2 2 2 (4.9) (4.
h̄ kx mh̄e ⌅ nx
v 34 µE
wn below, 20-nm-deep trenches 1.1

LETTER
ction. The samples in Fig. 2 were0.5
0 0.2 0.4 0.6 0.8 1
0.9
0 5 10 15 doi:10.1038/nature12952 20
. 7). The samples in Fig. 3 were L (μm) L (μm)
ar), andExceptional
then in ultrahigh vacuum
ballistic 2 transport
1.08
R2p in epitaxial graphene
dots, Fig. 3a) or at 1,150 6C (all 1.04
nanoribbons

R (h/e2)
7
tion). The natural-step 1
ribbon 2
1.00 Tejeda , Muriel Sicot , AminaTaleb-Ibrahimi , An-Ping Li ,
Jens Baringhaus *, Ming Ruan *, Frederik Edler , Antonio 1 3,4 3 4 5

Zhigang Jiang2, Edward H. Conrad2, Claire Berger2,6, Christoph Tegenkamp1 & Walt A. de Heer2

Resistance, R (h/e2)
R4p
0.96
1 2 3 4 5
1.5
Graphene nanoribbons will be essential components in future gra-
phene nanoelectronics1. However, in typical nanoribbons produced
surface of electronics-grade
L (μm)
no graphene
SiC wafers6,7,14–16 (see Figs 1a and 2), so that
patterning is required to produce nanosized ribbons. More
b I+
from lithographically patterned exfoliated graphene, the charge car- precisely, to prepare
I the ribbons shownV below, 20-nm-deep I trenches
riers travel only about ten nanometres between scattering events, !
were etched along the SiC [1100] direction. The samples in Fig. 2 were
U–
resulting in minimum sheet resistances of about one kilohm per annealed at 1,600 6C for 15 min (ref. 7). The samples in Fig. 3 were
square2–5. Here we show that 40-nanometre-wide d U+
graphene nanor- heated at 1,300 6RC1 in Ar (4 3 R102 25 mbar), and
R3 then in ultrahigh
R4 vacuum
6,7
ibbons epitaxially grown on silicon carbide are single-channel room- (UHV) for 15 min at 1,100 uC (green dots, Fig. 3a) or at 1,150 6C (all
1 μm
temperature ballistic conductors on a length scale greater than ten
micrometres, which is similar to the performance of metallic carbon
others—see Supplementary Information). The natural-step ribbon7

nanotubes. This is equivalent to sheet resistances below 1 ohm per

I– 1
square, surpassing theoretical predictions for perfect graphene8 by a Rsample Rsample b Rsample I+
2 μm In neutral graphene ribbons, we show
at least an order of magnitude. U–
that transport is dominated by two modes. 15 kV One is ballistic and tem-
d U+
perature independent; the other is thermally activated. Transport 1 μm
is protected from back-scattering, possibly reflecting ground-state I–
properties of neutral graphene. At room temperature, the resistance 2 2 μm
15 kV

of both modes is found to increase abruptly at a particular length—

the ballistic mode at 16 micrometres and the other at 160 nano-
G (e2/h)

metres. Our epitaxial graphene nanoribbons will be important not

0.5
only in fundamental science, but also—because they can be readily
produced in thousands—in advanced nanoelectronics, which can
1
make use of their room-temperature ballistic transport properties. c Vb d
The energy spectrum of a graphene ribbon with length L and width I– I+ 2 μm
W is approximately given by
r"
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
np#2 "mp#2
0
0.1 1 10
d
!
En,m ~+h!c z ð1Þ
W L L (μm)
 Failures of the Drude model: electrical
conductivity of an alloy
Cu Au
12

10
resistivity (10-8 Ωm)

4
Johansson C. H. and Linde J. O.,
2 Ann. Phys. 25, 1 (1936)
0 20 40 60 80 100
content of Au atoms (%)

• The resistivity of an alloy should be between those of its

components, or at least similar to them.

• It can be much higher than that of either component.

36
 E2 e)
⌅
nµ(
1 1 J = ⌅E =
Failures of the Drude model: heat capacity
⇤= = 1 ⇤ E 1 (4.22)
⌅ nµe 3 ⇥ NAJkB=T ⌅E + 6=⇥ NA kB T =
2 1 ⇤2
consider the classical energy forEone mole of solid kB T in a heat
J = ⌅E = 1 21 1 (4.23)
bath: each degree of freedom contributes
⇤ 6⇥ N A kB T k+
with BT
6 ⇥ NA kB T =
1 2 21 2 9
1 3 ⇥ NA kB T + 6 ⇥ NA kB T = NA kB
kB T 3 ⇥ 2 1 1 21 1 (4.24)
29
energy2 9N⇥A kBNTA+ +N
kB6T⇥heat 6⇥ NA=kB T
Acapacity
kB T NA=k
12 2 12 2 2
1 1 6 ⇥ N1A9kB T + 6 ⇥ N A kB T = 6NA kB
monovalent 3 ⇥ NA kB T + 6 ⇥ NA kB6 T⇥ 2 = N NkA kTB+ T 6 ⇥2 1 N9 k T(4.25)= 6NA k
2 2 2A B ANBA kB
1 2 1 2 2 15
1 1 9 ⇥ N1A kB T + 6 ⇥ N 1 A kB T = 15 NA k
divalent 6 ⇥ NA kB T + 6 ⇥ NA k9B⇥ 2T =N6N k ATkB+T 6 ⇥ 2 N 6N k T = 2 N
(4.26)
k
2 2 2
A B
2
A B A B
2
A
9
1 1 15 N9 A kB 15
9 ⇥ NA kB T + 6 ⇥ NA kB T = NA kB T2 NA kB NA(4.27) kB
trivalent 2 2 2 2 2
6NA kB
9 6NAh̄k2B d2 ⇥(4.28)
el. transl. 2ions
N k
A B vib. 15
15
NA2m kB dx2 = E⇥
6NA kB 2 NA keB (4.29)
2
• Experimentally, one finds 15 a value of about 3N A
3N
⇥kB= at
k
A sin(k
room x x)
2 2 (4.30)
N k A B
temperature, independent 2 of
A the
B number of valence h̄ k h̄ 2 2 2
⇥ n
2 E 2= x
= x
electrons (rule of Dulong and Petit), as if the
h̄ ⇥2 2m
2delectrons do
2m 2
2 h̄ d ⇥= E⇥
e e L
not contribute at all. h̄ d 2
⇥37 2m dx 2 = E⇥⇥
= E⇥ 22m dx 2
e 2 (4.31)
2
 Many open questions:

• Why does the Drude model work so relatively well when

many of its assumptions seem so wrong? In particular, the
electrons don’t seem to be scattered by each other or by
the lattice ions. Why?

• How do the electrons sneak by the atoms of the lattice?

• Why do the electrons not seem to contribute to the heat

capacity?

• Why is the resistance of an disordered alloy so high?

38