Sumrio

1 Teoria da Elasticidade 2
2 Concentrao de Stress 3
2.1 Abertura b em r = 0. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2 Superposio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3 Critrio de ruptura 7
3.1 Comportamentos dos Materiais . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.1.1 Puro elstico, ou quebradio . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.1.2 Viscoelasticidade . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.1.3 Plasticidade . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.1.4 Damage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.1.5 Fadiga . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
4 Elementos Finitos 11
5 Elementos Discretos 13
5.1 O mtodo de elementos-discretos aplicado de sistemas granulares . . . . . . . . . . . 14
6 Fragmentao 16
6.1 distribuio de Poisson . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
6.2 Fragmentao instantnea . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
6.3 fragmentao contnua . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
7 Simulao computacional de transies de fase estruturais em sistemas clssicos 20
7.1 Fases estruturais . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
7.2 O mtodo de Monte-Carlo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
1
1 Teoria da Elasticidade
Strain

ij
=
1
2
_
u
i
x
j
+
u
j
x
i
+
u
i
x
j
u
i
x
j
_

dx
2
i
= (
ij
+ 2
ij
)dx
i
dx
j
dx

i
=
_
1 + 2
(i)
dx
i
dx

i
(1 +
(i)
)dx
i
dx

i
dx
i
dx
i
<< 1
dV

dV
dV
= tr()
Stress

f
i
=

ij
x
j

F
i
=
_
D
f
i
dV =
_
D

ij

j
da

ij
x
j
+f
(e)
i
= 0

ij
n
j
= P
i
Relaes Termodinmicas

w =
ij

ij

dU = TdS +
ij
d
ij
dF = SdT +
ij
d
ij

ij
=
_
F

ij
_
T
Lei de Hooke

F =
1
2

2
ll
+
2
ij

ij
= 2
ij
+
ll

ij
2

ij
= 2(
ij

1
3

ll

ij
) +K
ll

ij
K = ( + 2/3)
nesses termos a energia livre de Helmoltz ca
F = (
ij

1
3

ll

ij
)
2
+
1
2

2
ll
de forma que K > 0 e > 0.
Mdulo de Young

xx
=
yy
=

2( +)

zz

zz

zz
=

zz
2(1 +)

P
E

=
E
1 +
=
E
1 +
1
1 2

ij
=
E
1 +
_

ij
+

1 2

ll

ij
_
e temos E 0 e 1 0, 5
Equao u

2
u +
1
1 2

u) =
2(1 +)
E

f
(e)
2 Concentrao de Stress
2.1 Abertura b em r = 0.
Condio imposta, vetor de Burguer
C no sentido anti-horrio, passando pela origem
_
C
du =

_
C
du
i
=
_
C

u
i
d

l =
_
S
C
(

u
i
) da = b
i
3
- 1.0 - 0.5 0.5 1.0
- 1.0
- 0.5
0.5
1.0
b

u
i
= b
i
(x)(y)

klm

2
u
i
x
l
x
m
= b
i

k
(x)(y)

2
u
i
x
i
x
j

2
u
i
x
j
x
i
= (

b)
j
(x)(y)
Equao a ser resolvida

ij
x
j
= 0

2
u
j
x
i
x
i
+
1
1 2

2
u
i
x
j
x
i
= (

b)
j
(x)(y)

2
u +
1
1 2

u) = (

b)(x)(y)
Soluo

u
x
(x, y) =
b
2
1
2(1 )
_
(1 2) ln r
x
2
r
2
+
1
2
_
u
y
(x, y) =
b
2
1
2(1 )
_
2(1 ) arctan
y
x

xy
r
2
_
4
u
y
(x, y) u
x
(x, y)

tr
yy
=
E
4(1
2
)
bx(x
2
+ 3y
2
)
(x
2
+y
2
)
2

tr
xx
=
E
4(1
2
)
bx(x
2
y
2
)
(x
2
+y
2
)
2
0 10 20 30 40
20
10
0
10
20

yy
(x, y)
0 10 20 30 40
20
10
0
10
20

xx
(x, y)
de qualquer forma

tr
yy
=
E
4(1
2
)
b
x
em y = 0
5
- 1.0 - 0.5 0.5 1.0
-1.0
-0.5
1.0
2.2 Superposio
contribuio da trinca

tr
yy
=
E
4(1
2
)
_
L
L
db()
x
db() =
db()
d
d
equao integral que determina b(x)

ij
n
j
= G(x)
tr
yy
+
e
yy
G = 0
supondo cavidade elptica
b() = b
_
1

2
L
2
db() =
b
L
2
_
1

2
L
2
d

tr
yy
=
E
4(1
2
)
_
b
L
__
L
L
(/L)
_
1 (/L)
2
d(/L)
(x/L) (/L)
<< x L << L
= b
2
/L o raio de curvatura
resolvendo a integral

tr
yy
=
E
4(1
2
)
_
b
L
_
_
1
x
_
(x L)(x +L))
_
6
1.0 1.2 1.4 1.6 1.8 2.0
0
1
2
3
4
5
xL

y
y
<< x L << L
deve ser somado a
e
yy
para obter, em (x L) = = b
2
/L

yy
/
0
= 1 +
_
L
2
3 Critrio de ruptura
Critrio de Grith, onde dE
1
energia liberada, dE
2
= 2
s
dA a energia de superfcie e dW o
trabalho exercido pelo agente externo,
dW dE
1
dE
2

G
dW dE
1
dA
2
s
para o caso de uma cavidade elipsoidal de raio maior L e raio menor b, numa situao idealizada
de um material innito sujeito a uma tenso constante , perpendicular ao raio maior. . .
E
1


2
2E
L
2
b
E
2

s
Lb
G
c
=

2
c
2E
L
s
ou

c

2E
s
L
entretanto, esse critrio no leva em conta a concentrao de tenso na extremidade da trinca,
em geral,

ij
K

ij
()

r
onde K e
ij
() dependem do campo de tenso no innito, e pode-se estabelecer o critrio
atravs do fator de intensidade K, numa situao simplicada:
G
c

1
E
_
K
2
Ic
+K
2
IIc
+ (1 +)K
2
Ic
_
7
a teoria da elasticidade no descreve bem a trinca na regio prxima ponta da trinca.
3.1 Comportamentos dos Materiais
3.1.1 Puro elstico, ou quebradio
A
O
c
3.1.2 Viscoelasticidade
A deformao de um slido causa dissipao de energia e a resposta do slido no imediata,
mas atrasada. A situao pode ser modelada introduzindo propriedades do material que sejam
dependentes do tempo, por exemplo

ij
=
_
t

C
ijkl
()
kl
(t )d
O
t
3.1.3 Plasticidade
Plasticidade um processo mais complexo que resulta em uma deformao irreversvel. No caso de
plasticidade perfeita a curva tenso deformao como a esquematizada na gura:
8
A B
C
D
O P
Para modelar o fenmeno, precisamos introduzir , =
P
+
e
, e para conhecer o estado do
sistema e a evoluo do sistema:
1. se <
P
, ou =
P
com < 0
P
= 0 e =
e
=

E
2. se =
P
com > 0 =
e
= 0 e
P
=
Em trs dimenses a tenso um tensor, e pode haver mais de um tipo de carga, deve-se nesse caso
denir um domnio no qual o material elstico, fora do qual ocorre escoamento plstico. Isso
feito atravs da introduo do potencial plstico f(), como cisalhamento leva a plasticidade muito
muito mais rapidamente, f() = f(s()), com s = tr()I. Na base dos auto-valores do tensor
s, o domnio ser um cilindro cujo eixo paralelo linha que corresponde a presso isosttica de
cisalhamento puro (s
1
+s
2
+s
3
= 1), a escolha de von Mises simplesmente a equao do cilindro
f(s())
_
tr (s
2
) R
onde R =
_
2/3
P
, por exemplo para uma tenso uniaxial, como descrito abaixo:
Digamos que
=
_
_
0 0 0
0 0 0
0 0
zz
_
_
s =
_
_

1
3

zz
0 0
0
1
3

zz
0
0 0
2
3

zz
_
_
R =
_
2
3

P
Se
f() 0
tr (s
2
)
R
2
1
ocorre escoamento plstico.
1. f() < 0
P
= 0
2. f() = 0
(a) aponta no sentido do interior do domnio (diminuindo f).
(b) aponta para fora do domnio. Nesse caso escolhemos

P
=
f

e obrigamos permanecer na superfcie do domnio D.

f

= 0
9
de onde
f

C
_

t

_
= 0
e nalmente
=
_
f

C
t
_
_
f

C
f

_
A ideia por ser generalizada para incluir o caso em que o domnio elstico muda com a aplicao
de tenso. Nesse caso estende-se
f(s())
_
tr (s X)
2
R
onde agora X() e R() so dependentes do tempo para levar em conta o processo de dissipao de
energia ao mover, expandir ou contrair o domnio de regime elstico. X seria o centro do domnio
e R o raio (note que ambos tem dimenso de tenso). Pode-se mostrar que
X =
E
h

P
R =
E
h
p
E
h
a dissipao da energia, e p a deformao plstica acumulada
p =
_
t
0
_
tr(

2
p
) d
3.1.4 Damage
Danos causados no material levam a uma reduo no seu mdulo elstico, sem necessariamente
causar deformaes plsticas.
A
B
C
D
O
d
Se para o material com dano
= C
e aps o dano
=

C
10
podemos escrever
=

CC
1
(I D)
1
.
sendo que o tensor D pode ser aproximado por uma constante multiplicada pelo tensor identidade.
Aqui tambm se precisa denir um domnio D dentro do qual a resposta do material elstica. No
caso de tenso uniaxial simples, denindo D = (), temos:
1. <
d
, ou =
d
e < 0
= (1 D)C
todo resto constante.
2. =
d
e > 0

d
=

D =

= (1 D)C

DC
3.1.5 Fadiga
Fadiga o mecanismo de ruptura que leva falha de um material quando esse submetido a uma
dada carga por um longo tempo. Normalmente nada aparentemente acontece, at que o material se
rompe catastrocamente de um momento para outro. O processo de ruptura por fadiga bastante
complicado, e sua modelagem deve incluir o acmulo de dano, e outros mecanismos como healing.
Na modelagem de DEM o acmulo de dano na escala mesoscpica diretamente incorporado,
escolhe-se uma regra de quebra que leva em considerao que as barras podem se quebrar por
extenso, exo, ou toro quando, por exemplo, a quantidade
f(t) =
_

c
_
2
+
max(|
i
|, |
j
|)

c
1.0 > 0
que corresponde ao potencial de von Mises descrito acima. Pare se levar em considerao o acumulo
de dano na escala microscpica introduzimos a quantidade
p(t) = f(t) +f
0
_
t
0
e
(tt)/
f(t

)dt

4 Elementos Finitos
Equao para resolver. . .

2
u +
1
1 2

u) =
2(1 +)
E

f
(e)
Escreve aproximao
u(x) = a
i

i
(x)
Escolhe uma mesh
0 L
11
Escolhe uma base para expanso

i
(x) =
_

_
x
i+1
x
x
i+1
x
i
x element
i
0 x / element
i
com essa escolha. . .
u(x) = u
i

i
(x)
Determine os coecientes u
i
que minimizem a energia
E
1
=
_
D
( u(x))( u(x))dV
E
2
=
_
D
f
e
(x) u(x)dV +
_
D
P(x) u(x)da
no caso 1D
= E
d u
dx
portanto
= Eu
i

i
(x)
= u
i

i
(x)
e temos
E =
1
2
K
ij
u
i
u
j
+F
i
u
i
com
K
ij
= 2
_

i
(x)
j
(x)dV
F
i
=
_

f
e
(x)

i
(x)dV +
_

P(x)

i
(x)da
para minimizar a energia
E
u
i
= 0,
que se resume a resolver o sistema linear de N
nodes
equaes.
K
ij
u
j
= F
i
a soluo u(x) = u
i

i
(x) ser melhor tanto melhor for a base
12
5 Elementos Discretos
No mtodo de elementos discretos (DEM), o material substitudo por um conjunto de partcu-
las que interagem entre si atravs de leis locais pr-estabelecidas. O esquema geral do mtodo de
elementos discretos muito similar ao mtodo de Dinmica Molecular (MD), em ambos os casos
as foras que atuam em cada partcula so calculadas e as equaes de Newton resolvidas numeri-
camente. No entanto, enquanto em MD pretende-se representar o sistema no nvel atmico, DEM
tem validade em escalas muito maiores, utilizando parmetros obtidos a partir da teoria da elas-
ticidade para cada elemento individualmente. Em DEM as partculas tem extenso nita, e os
desenvolvimentos necessrios para dinmica de corpos rgidos so adotados.
A forma, distribuio de tamanho, organizao inicial das partculas, bem como a escolha das
foras de contato e coeso so obviamente muito importantes em DEM.
fora de contato

F
c
i
=

f
ov
ij
+

f
at
ij
proporcional rea de sobreposio entre os polgonos, ou dada pela fora de contato
de Hertz entre duas esferas.

f
ov
ij
=
_

_
E
p
(1
2
)
A
ij
n polgonos, 2D
4
3
E
p

(1
2
)

3/2
ij
r
ij
esferas, 3D
fora de atrito proporcional velocidade relativa, e frico de Coulomb.

f
at
ij
= m

n
v
rel,n
ij
min
_
m

t
|v
rel,t
ij
|, |f
ov
ij
|
_

t
fora de coeso
Barras

F
b
i
=

f
elo
ij
+

Q
z,b
ij

f
elo
ij
= E
b
A
b

ij
r
ij

Q
z,b
ij
= E
b
I
(
z
i
+
z
j
)
L
2
e
b
y

M
z,b
i
= E
b
I
(
z
i

z
j
)
L
e
b
z
+
_

Q
z,b
i
|r
ij
| e
b
x
_

M
x,b
i
= G
b
I
tor
(
x
j

x
i
)
L
e
b
x
onde I =
R
4
4
para um cilindro de raio R, e I
tor
=
R
4
2
regra de ruptura
p(t) =
_

th
_
2
+
max (|
i
|, |
j
|)

th
1.0 0
q(t) = p(t) +
_
t
0
e
(tt

p(t

)dt

1
13
distribuio de falhas
P(
th
) =
k

o
_

th

o
_
k1
exp
_

th

o
_
k
_
,
As equaes de movimento
m
i
d
2
r
i
dt
2
=

F
i
d

L
i
dt
=

M
i
discretizao no tempo
Verlet
r(t +t) = r(t) +t v(t) +
t
2
2
a(t)
r(t +t) = 2r(t) r(t h) +t
2
a(t)
v(t) = [r(t +t) r(t t)]/2t
Leap-Frog
v(t +t/2) = v(t = t/2) +dt a(t)
r(t +t) = r(t) +t [v(t) + (t/2)a(t)]
= r(t) +dt v(t +t/2)
caso necessrio
v(t) = v(t +t/2) (t/2)a(t)
Predictor-Corrector
P(r) : r(t +t) = r(t) +dt v(t) +t
2
k1

i=1
p
i
a(t + (1 i)t)
P(v) : t v(t +t) = r(t +t) r(t) +t
2
k1

i=1
p

i
a(t + (1 i)t)
com esses valores calculados, pode se estimar a(t +t) e renar o resultado
C(r) : r(t +t) = r(t) +dt v(t) +t
2
k1

i=1
c
i
a(t + (2 i)t)
C(v) : t v(t +t) = r(t +t) r(t) +t
2
k1

i=1
c

i
a(t + (2 i)t)
os coecientes calculados para se obter da ordem O(t
k+1
)
5.1 O mtodo de elementos-discretos aplicado de sistemas granulares
uidization: quando um material granular, como areia seca, passa de um estado slido para
um estado uido dinmico, esse processo ocorre quando um uido, gs ou lquido, feito
passar pelo material granular.
clustering instabilities aglomerao devido a frico
size segregation
14
l
Tabela 1: Micro- and macroscopic material model properties.
DEM - microscpico:
Barras:
mdulo de Young E
b
/G
b
6 GPa
comprimento mdio L 0.5 mm
dimetro d 0.5 mm
strain crtico
0
0.02 -
ngulo exo crtico
0
3

parmetro Weibull k 3/10 -
Elementos:
mdulo de Young E
p
3 GPa
razo de Poisson 0,3
dimetro D
1
0.5 mm
densidade 3000 kg/m
3
Placa de impacto:
Mdulo de Young E
w
70 GPa
Interao:
coeciente de atrito 1 -
coeciente de atrito
n
0.25 s
1
coeciente de atrito
t
0.05 s
1
Resolvedor:
incremento de tempo t 1e-8 s
nmero de elementos N
p
22013 -
nmero de barras N
b
135948 -
frao de slido 0.65 -
dimetro elementos D 16 mm
DEM Macroscpico:
Mdulo de Young E 7.4 0.5 GPa
Razo de Poisson 0.2 -
Densidade 1920 kg/m
3
resistncia
c
110 MPa
Comparao:
DEM FEM
velocidade do som 2210 100 2270 20 m/s
tempo de contato 31.4 31.4 s
15
6 Fragmentao
Observam-se que fragmentos resultantes do processo de fratura apresentam leis de potncia
nas suas distribuies de tamanho.
Leis de potncia invarincia de escala
Fratura e fragmentao transies de fase.
Reviso dos fundamentos:
6.1 distribuio de Poisson
Tipos de defeitos: de volume, de superfcie e de borda.
Distribuio de defeitos:
defeitos distribudos aleatoriamente no domnio, independentemente do tipo e um do
outro.
distribuio independente do campo de tenso aplicado.
a probabilidade de n falhas no subdomnio t T do material dada pela distribuio binomial.
P(n|t) =
M!
(M n)!n!
(p)
n
(1 p)
Mn
M!
(Mn)!n!
o nmero de possibilidades de escolher n falhas em M, sendo todas equivalentes
(indistinguveis). Tomando p = t/T a probabilidade de 1 falha em t, M o nmero total de
falhas, e chamando = M/T, temos
P(n|t) =
M!
(M n)!n!
_
t
M
_
n
_
1
t
M
_
Mn
=
M!
(M n)!M
n
(t)
n
n!
_
1
t
M
_
Mn
no limite M, T com contante . . .
P(n|t) =
(t)
n
n!
e
t
Note que
M
M!
(M n)!
M
n
e
lim
n
_
1
x
M
_
M
e
x
Assim o nmero mdio de defeitos no subdomnio t
< n|t >=

n=0
n
(t)
n
n!
e
t
= t
A probabilidade de ter 1 defeito entre t e t +dt
p(t)dt = P(0, t)P(1, dt) = e
t
dt
16
6.2 Fragmentao instantnea
a probabilidade de ser formar um fragmento com volume total entre v e v+dv, rea supercial
entre s e s +ds e borda entre l e l +dl
dp(l, s, v) = P(0|l, s, v)P(1, dl, ds, dv)
= P(0, l)P(0, s)P(0, v)
_
P(1, dl) +P(1, ds) +P(1, dv)
_
= e
Q
dQ
com
Q =
l
l +
s
s +
v
v
o nmero de fragmentos com volume total entre v e v +dv, rea supercial entre s e s +ds e
borda entre l e l +dl
dn(l, s, v) =
V
0
v
e
Q
dQ
o volume acumulado dos fragmentos com volume at v, superfcie s e borda l ser dado por
V (l, s, v) =
_
l
0
_
s
0
_
v
0
vdn(l, s, v)
= V
0
_
1 e
Q
_
y(l, s, v) V (l, s, v)/V
0
=
_
1 e
Q
_
usando
l =
l
r
s =
s
r
2
v =
v
r
3
e
l

l
1/k
1
,
s

s
1/k
2
e
v

v
1/k
3
, obtemos
p(r)dr = dp(l, s, v) = e
Q
dQ
p(r) =
_
1
k
1
+
2r
k
2
+
3r
2
k
3
_
exp
_

_
r
k
1
+
r
2
k
2
+
r
3
k
3
__
.
e
y(r) =
_
1 exp
_

_
r
k
1
+
r
2
k
2
+
r
3
k
3
___
e
n(r)dr = dn(l, s, v) =
V
0
v
dp(l, s, v)
n(r) =
V
0

v
1
r
3
_
1
k
1
+
2r
k
2
+
3r
2
k
3
_
Exp
_

_
r
k
1
+
r
2
k
2
+
r
3
k
3
__
para r pequeno y(r) r

com = 1, e uma mudana de varivel de tamanho r para massa

m r
3
em n(r) leva concentrao c(m) m
(2/D)
, com D = 3.
17
6.3 fragmentao contnua
sequncia cronolgica de fragmentaes instantneas locais
c(m, t) concentrao de fragmentos com massa entre m e m+dm no instante t.
c(m, t)
t
= a(m)c(m, t) +
_

m
c(m

, t)a(m

)f(m|m

)dm

a(m)dt probabilidade de que um fragmento de massa m se quebre no intervalo de tempo

dt
a(m

)f(m|m

)dm

dt probabilidade de que um fragmento de massa m

se quebre produ-
zindo um com massa m no intervalo de tempo dt
Caso simplicado, quebra binria de um sistema homogneo
sempre um fragmento de massa total m+m

se quebra em exatamente dois,

m+m

m outro m

F(m, m

)dt a probabilidade disso acontecer

a(m) =
_
m
0
F(m

, mm

)dm

a(m

)f(m|m

) = 2F(m, m

m)

c(m, t)
t
= c(m, t)
_
m
0
F(m

, mm

)dm

+ 2
_

m
c(m

, t)F(m, m

m)dm

sistema homogneo,
F(m, m

) (m+m

)
1
depende somente da massa total do fragmento inicial, 1, fragmentos maiores tem mais
probabilidade de quebrar.
F(m

, mm

) = m
1
a(m) = m

F(m, m

m) = m
1
camos com

2
c(m, t)
mt
= m

c(m, t)
m
( + 2)m
1
c(m, t)
que pode ser resolvida analiticamente com a substituio c(m, t) = A(t)e
B(t)m

, resultando
em
dB
dt
= 1 B(t) = (t +s)
(t +s)
dA
dt
=
2

A(t) = (t +s)
2/
18
e ento
c(m, t) = (t +s)
2/
exp
_
(t +s)m

_
m
1
1
m
1
8
5 10 50 100 500
0.5
1.0
5.0
10.0
50.0
t
c

m
,
t

Populao de pequenos m aumenta devido quebra de maiores, eventualmente comea dimi-

nuir quedo esses comeam a quebrar em menores ainda.
t=1
t=8
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
5
10
15
m
c

m
,
t

medida que o tempo aumenta, predominncia de pequenos fragmentos

19
7 Simulao computacional de transies de fase estruturais em
sistemas clssicos
VOLUME 73, NUMBER 5 PHYSICAL REVIEW LETTERS 1 AUGUsT 1994
Plasma Crystal: Coulomb Crystallization in a Dusty Plasma
H. Thomas,
*
G. E. Morfill, and V. Demmel
Max Pla-nck Inst-itut fu
rEx'traterrestrische Physik, 85740 Garching, Germany
J. Goree'
Department of Physics and Astronomy, The University of Iowa, Iowa City, Iowa 52242
B. Feuerbacher and D. Mohlmann
DLR, Institut fur Raumsimulation, 51140Kiln, Germany
(Received 25 January 1994)
A macroscopic Coulomb crystal of solid particles in a plasma has been observed. Images of a cloud
of 7-p, m
"dust"
particles, which are charged and levitated in a weakly ionized argon plasma, reveal
a hexagonal crystal structure. The crystal is visible to the unaided eye. The particles are cooled by
neutral gas to 310 K, and their charge is &9800e, corresponding to a Coulomb coupling parameter
I & 20700. For such a large I value, strongly coupled plasma theory predicts that the particles should
organize in a Coulomb solid, in agreement with our observations.
PACS numbers: 52.25.-b, 64.70.-p, 94.10.Nh
The search for model systems of crystalline structures
to study phase transitions was initiated by Wigner in the
1930s with the theory of the Wigner crystal [1]. Since
that time experimental verification has been achieved for
several specific systems. On the atomic scale these are
ion crystals [2

4] and electron crystals [5] and on macro-

scopic scales colloidal crystals in aqueous solutions [6,7].
Each of these systems has advantages and disadvantages
for the detailed study of the phase transition of inter-
est, such as formation, growth, and melting of crystalline
structures.
Here we report an experimental observation of a macro-
scopic Coulomb crystal formed from a dusty plasma. We
term this structure a "plasma crystal.
"
Because of its
macroscopic size, and the ease of photographing it and
controlling it over a wide range of parameters (including
plasma density, temperature, neutral gas, and particle size),
the detailed study of plasma crystals could well lead to a
better understanding of phase transitions. Based on theo-
retical calculations this was first suggested by
Ikezi [8].
Dusty plasmas have been the subject of intensive study
for astrophysics and plasma-aided manufacturing. In
space, dusty plasmas are ubiquitous [9

ll], including
interstellar clouds, circumstellar and protoplanetary
ac-
cretion disks, nova ejecta, and planetary magnetospheres.
The thermodynamics, chemistry, and electromagnetic
evolution of these systems are affected and sometimes
dominated by
dust. In microelectronics fabrication,
particles can grow in the plasmas in surface processing
reactors, and remain electrically suspended there until they
fall to a surface and contaminate it [12,13]. Thus it is not
surprising that astronomers and industrial researchers have
investigated many physical processes [12

18] including
especially the problem of dust charging [19

22].
A dust particle, like any surface exposed to a plasma,
is charged by collecting ions and electrons, or by
photo-
emission and secondary emission. In the absence of elec-
tron emission, the charging process is dominated by
the
incoming flux of electrons and ions from the surround-
ing plasma. The charge on an isolated particle in the
plasma is
Q~
=
C@where C
= 4m.
d'or
is the capaci-
tance, r the particle's radius, and the steady-state surface
potential 4, is calculated by equating the ion and elec-
tron fluxes [21,22]. For nondrifting Maxwellian distribu-
tion functions characterized by
ion and electron tempera-
tures T; and T4,is the solution of [1

(e4, /kttT;) j
=
(m;ksT, /m, ksT;)' 'exp(e4, /kttT, ),
while for drifting dis-
tributions a more general relation is satisfied [19]. These
expressions are suitable for an isolated particle, but 4,
and
Q~
are diminished below this level for particles
with sufficient number density n
to deplete the elec-
tron density naswill happen unless P && 1, where
P
=
695T, rn~/nwith T, in eV and r in p, m [20,22, 23].
Each particle is surrounded by
a Debye cloud of the op-
posite charge, characterized by a shielding length A. For
an infinite nondrifting plasma, A is the "linearized" De-
bye length (1/2A;
+ 1/A2)
'12
[24], where A; (( A, for a
typical discharge with cold ions (T;
T, ). According to
the Bohm sheath criterion for a dust-free discharge, at the
edge the ions drift at approximately c,
=
(k&T,
/m;)'l~,
so they enter the sheath with the same energy as elec-
trons. Accordingly A might be modeled more accurately
ash=A, .
The transition from random aggregates of dust in
plasmas to a more ordered structure requires special
conditions. Early investigations on plasma crystalliza-
tion were done in the context of one-component plas-
mas (OCP). In an OCP, such as a pure ion plasma,
overall confinement is effected by
a fixed "neutralizing
background,
"
e.g.,
electromagnetic fields. The thermo-
dynamics is described by
the Coulomb coupling pa-
rameter, I
=
Q/4neokBT~b,
which is
.
the ratio of the
652 0031-9007/94/73 (5)/652 (4)$06.00
1994 The American Physical Society
VOLUME 73, NUMBER 5 PHYSICAL REVIEW LETTERS 1 AUGUST 1994
0 0 ~ ~ 0 ~ ~
~ ~
~ ~ 0
~ ~ ~ 0 ~ ~ ~ ~ 0 ~ ~ ~ ~
0 ~ 0 ~
~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~
0
~ ~ ~ ~ ~ ~
~ ~ ~ ~ 0 0
~ ~ ~ ~ ~ 0 ~ ~ 0 ~
~ 0
~ ~ ~ ~ 0 0 ~
~ ~ ~ ~ ~ ~ ~ 0
~ ~ ~ ~ ~ 0 ~
~ ~ ~ 0 ~ ~
0
~ 0 ~ ~
~ 0 ~ ~ ~ ~ ~ ~
0 ~ ~ ~ ~
~ 0
0 0 ~ 0 ~ 0 ~ ~
~ ~
~ ~ ~
~ ~ ~
~ ~
0 ~ ~ ~ ~
~ ~ ~ ~ 0
~ ~ ~ ~ ~
~ 0 ~ ~ ~ ~ ~ 0 ~
~ ~ ~ ~
~ ~ ~ ~ 0
~ 0 ~ ~
~ ~
~ ~ ~ ~ ~
0 0 ~
~ ~ ~ ~
~ ~
~ ~ ~ ~ ~ 0
~ ~ ~ ~
~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ 0
~ ~
~ ~ ~ ~ ~ ~
~ ~ 0 ~
~ ~
~ ~ 0
~ ~ ~
'~ ~ ~
~ ~
~ ~
~ ~ ~ ~ ~
~ ~ ~
'~ 0
~ ~ ~
~
I
~ ~ ~ ~ ~ ~ ~
~ ~ ~ ~ ~ ~ ~
~ ~
~ ~ ~ ~ ~
0 ~ ~ ~ ~ ~ ~
~ ~
~ ~ ~ ~
~ ~
~ ~ ~ ~ ~ ~ ~ ~ ~
~ l 0
~ ~ ~ ~ ~ ~ ~
~ 0 ~ ~
~ ~ ~ 0
~ ~
~ ~ ~
'~ ~ ~ ~ ~
~ ~ ~ ~ ~ ~ ~ ~
~ ~
~ ~
~ ~ ~ ~ ~ ~ ~
~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ 0
~ ~ ~ ~ 0
~ ~
~ ~ ~ ~
~ ~
I I ~ ~ ~ ~
50
.

(a)
a 40
C3)
6$
c 30
8
K
20
=
10
=
0
0
.I]il I
cell
~ experiment
Q random
Ulnnn. n
0.1 0.2
area (mm2)
60 = (b)
Cb
m
40
Q)
K 20
60 t t t i i & s
~
& s s t s
FIG. 3. Voronoi analysis of particle locations from the image
shown in Fig. 2.
5 6 7 8
number of sides
9 l0
cells (see Fig. 3). The cells are mainly six sided, with
nearly equal areas and uniform spacing, as shown in
Fig. 4. The mean interparticle distance b, (lattice con-
stant) is 250 p, m. The measurements were compared with
a synthetic random particle distribution using the same
properties

number and density

as in the experiment.
This simulates a dusty plasma in its "gas phase.
"
The
comparison is shown in Fig. 4. It is obvious that the ob-
served particle structure is more crystalline than gaslike.
The video showed the individual particles oscillating gen-
tly about fixed equilibrium centers in what appeared to be
Brownian motion. Occasionally a grain diffused through
the structure, visibly perturbing the positions of the near-
est particles.
In order to estimate the coupling parameter I we need
to know the particles' kinetic energy (or temperature)
and charge. The particles are cooled by the neutral
gas. Based on frame-by-frame tneasurements of the
mean particle velocity, we estimate the particle kinetic
temperature to T=310 K, which is close to room
temperature. A comparable temperature was measured
by Boufendi et al. [13]using laser-Doppler velocimetery;
their particle size and I were much smaller than ours.
We estimate the plasma parameters and the grain
charge in the following way. We assume room tem-
perature ions and other parameters that are typical
[27] of low-power rf discharges: k~T,
=
3
~
I eV,
and plasma density n;
=
10 cm, accurate to a fac-
tor of 4. (In this experiment, we did not measure
any of these parameters. )
We calculated the potential
4, for two extreme cases of the ion drift speed, zero
and casdiscussed above. Correspondingly, we find
(

7.2
~
2.0 V)
~
4
~
(

11.7
~
3.9 V) for an isolated
grain. For our discharge the parameter P lies in the range
0.08
~
P
~
1.2, where the uncertainty arises from the
uncertainty of n;. Correspondingly, the charge is reduced
f I I I
l
I I I I
l
I
40
.

(c)
e 30-
C5
U)
20
O
&10
-'
nn [lll
0 0.1 0.2
distance to next
I I I
l
I I I I
l
'1
la. .
0.3 0.4
neighbor (mm)
FIG. 4. Distribution of (a) Voronoi-cell areas, (b) number
of Voronoi-cell sides, and (c) near-neighbor distances. The
experimental data (black columns) are from the image in Fig. 2.
Data for a random-point distribution (white columns) is shown
for comparison.
by 6% to 449o from the isolated particle case [20,22].
Using these values, we estimate the grain charge to lie
between the extreme values

9800
~
Q~/e
~
27300.
The shielding length is estimated to lie between the
"linearized" Debye length and A, 52
~
A
~
405 p, m,
resulting in 4.8
~
~
~
0.6. The Coulomb coupling
parameter I is then estimated to be & 20700. Our
observation of a crystalline structure is consistent with
this large value of I .
The crystal forms easily at these parameters, because
the large charge assures strong interparticle Coulomb
forces, while the neutral gas cools the particles to a low
temperature. %'hen the rf power was raised, however,
we found that particles moved more violently, and many
appeared to have no equilibrium positions, so the cloud
qualitatively appeared to be liquidlike. However, we did
not observe a distinct phase transition of the cloud as a
whole. One problem in observing a phase transition is
that we view only a single two-dimensional layer, and a
change of the plasma conditions (e.g., heating the cloud
by increasing the rf power) leads to a displacement of
654
letters to nature
NATURE | VOL 412 | 16 AUGUST 2001 | www.nature.com 717
coupling constant. More general conditional evolutions of the environment E can be of
course considered.
Following entanglement, the state of the system alone is described by the reduced
density matrix obtained from jF
SE
i by a trace over the environment:
r
S
Tr
E
jF
SE
ihF
SE
j jaj
2
j ih j zab*j ih 2j z*ab*j 2ih j jbj
2
j 2ih 2j
Disappearance of the off-diagonal terms signies perfect decoherence. In the {j i; j 2i}
basis this is guaranteed when the overlap z he

je
2
i Trje
2
ihe

j disappears. It is
therefore natural to measure effectiveness of decoherence by the magnitude of the overlap
of the two conditional states of the environment, which in turn determines the degree of
suppression of the off-diagonal terms in r
S
.
This two-paragraph `crash course' is no substitute for a more complete discussion of
decoherence
68
. We have swept a number of issues under the rug. Foremost among them is
einselectionthe emergence, in the course of decoherence, of the preferred set of pointer
states that habitually appear on the diagonal of r
S
essentially independently of the initial
states of either E or S. Stability of these pointer states (rather than the diagonality of the
density matrix in some basis) is the key to the role played by decoherence in the transition
from quantum to classical
28,18
.
Another important subject avoided here is the likely situation when the state of the
environment is represented by a mixture r
E
. Detailed discussion of this case (treated
extensively before
68,1719
, although not from the point of view of the sub-Planck
structures) is beyond the scope of this paper, but the basic conclusion is easy to state: the
estimated magnitude of the smallest displacement leading to orthogonality is still given by
equation (17). That is, it is still related to the smallest scales compatible with the classical
action A associated with r
E
. Off-diagonal terms of r
S
are suppressed by z TrU
2
r
E
U

,
which is the relevant generalization of equation (16). This expression can be expanded for
simple displacements U
6
D
6
in the limit of small shifts to recover equation (17), with
the only difference arising fromthe fact that nowthe mixture r
E
must be used to obtain the
averages (hx
2
i Trx
2
r
E
, etc.).
Note that the size of the structures in phase space depends on the representation of the
quantum state. For instance, one does not expect scale a of equation (3) in the Husimi
function, that is, in the Wigner function smoothed on a Planck scale. This is no surprise
and does not affect our conclusions concerning decoherence, as only the Wigner function
yields scalar product through the simple equation (14).
Note that the discussion throughout the paper is set in one spatial dimension. Generali-
zation to d dimensions is as conceptually straightforward as it is notationally cumbersome.
The structure saturates in volumes of a
d
, etc. This has little effect on decoherence, as it
depends on displacements that yield orthogonality, and these are still d < a.
Received 25 September 2000; accepted 25 June 2001.
1. Heisenberg, W. U

ber den anschaulichen Inhalt der quantentheoretischen Kinematik and Mechanik.

Z. Phys. 43, 172198 (1927); The physical content of quantum kinematics and mechanics (Engl.
Trans.) in Quantum Theory and Measurement (eds Wheeler, J. A. & Zurek, W. H.) (Princeton Univ.
Press, Princeton, 1983).
2. Zurek, W. H. Decoherence, chaos, quantum-classical correspondence, and the algorithmic arrow of
time. Phys. Script. T76, 186198 (1998).
3. Zurek, W. H. Pointer basis of a quantum apparatus: Into what mixture does the wavepacket collapse?
Phys. Rev. D 24, 15161524 (1981).
4. Zurek, W. H. Environment-induced superselection rules. Phys. Rev. D 26, 18621880 (1982).
5. Joos, E. & Zeh, H. D. The emergence of classical properties through the interaction with the
environment. Z. Phys. B 59, 229 (1985).
6. Zurek, W. H. Decoherence and the transition from quantum to classical. Phys. Today 44, 3646
(1991).
7. Giulini, D., Joos, E., Kiefer, C., Kupsch, J., Stamatescu, L.-O. & Zeh, H. D. Decoherence and the
Appearance of a Classical World in Quantum Theory (Springer, Berlin, 1996).
8. Zurek, W. H. Decoherence, einselection, and the quantum origin of the classical. Rev. Mod. Phys.
(in the press); also as preprint (quant-ph 010527) at hhttp://xxx.lanl.govi (2001).
9. Haake, F. Quantum Signatures of Chaos (Springer, Berlin, 1991).
10. Casati, G. & Chrikov, B. Quantum Chaos (Cambridge Univ. Press, Cambridge, 1995).
11. Hillery, M., O'Connell, R. F., Scully, M. O. & Wigner, E. P. Distribution functions in physics:
Fundamentals. Phys. Rep. 106, 121167 (1984).
12. Berry, M. V. & Balazs, N. L. Evolution of semiclassical quantum states in phase space. J. Phys. A 12,
625642 (1979).
13. Korsch, H. J. & Berry, M. V. Evolution of Wigner's phase-space density under a nonintegrable
quantum map. Physica D 3, 627636 (1981).
14. Zurek, W. H. & Paz, J. P. Decoherence, chaos, and the Second Law. Phys. Rev. Lett. 72, 25082511
(1994).
15. Berman, G. P. & Zaslavsky, G. M. Condition of stochasticity in quantum non-linear systems. Physica
(Amsterdam) 91A, 450 (1978).
16. Habib, S., Shizume, K. & Zurek, W. H. Decoherence, chaos, and the correspondence principle. Phys.
Rev. Lett. 80, 4361 (1998).
17. Caldeira, A. O. & Leggett, A. J. Path-integral approach to quantum Brownian motion. Physica 121A,
587616 (1983).
18. Paz, J. P. & Zurek, W. H. in Les Houches Lectures Session LXXII (eds Kaiser, R., Westbrook, C. and
David, F.) 533614 (Springer, Berlin, 2001).
19. Braun, D., Haake, F. &Strunz, W. A. Universality of decoherence. Phys. Rev. Lett. 86, 29132917 (2001).
20. Hannay, J. H. &Berry, M. V. Quantization of linear maps on a torusFresnel diffraction by a periodic
grating. Physica 1D, 267290 (1980).
21. Caves, C. in Physical Origins of Time Asymmetry (eds Halliwell, J. J., Perez-Mercader, J. &Zurek, W. H.)
4777 (Cambridge Univ. Press, Cambridge, 1993).
22. Miller, P. A. &Sarkar, S. Signatures of chaos in the entanglement of two coupled quantumkicked tops.
Phys. Rev. E 60, 1542 (1999).
23. Braginsky, V. B. & Khalili, F. Y. Quantum nondemolition measurements: the route from toys to tools.
Rev. Mod. Phys. 95, 703711 (1996).
24. Karkuszewski, Z., Zakrzewski, J. & Zurek, W. H. Breakdown of correspondence in chaotic systems:
Ehrenfest versus localization times. Preprint quant-ph/0010011 at hhttp://xxx.lanl.govi (2000).
Acknowledgements
This research was supported in part by the National Security Agency. I thank A. Albrecht,
N. Balazs, C. Jarzynski, Z. Karkuszewski and J. P. Paz for useful chaotic conversations.
Correspondence and requests for materials should be addressed to W.Z.
(e-mail: whz@LANL.gov).
.................................................................
Crystalline ion beams
T. Scha tz, U. Schramm & D. Habs
Sektion Physik, LMU Munchen, D-85748 Garching, Germany
..............................................................................................................................................
By freezing out the motion between particles in a high-energy
storage ring, it should be possible
14
to create threads of ions,
offering research opportunities beyond the realm of standard
accelerator physics. The usual heating due to intra-beam colli-
sions should completely vanish, giving rise to a state of unprece-
dented brilliance. Despite a continuous improvement of beam
cooling techniques, such as electron cooling and laser cooling, the
ultimate goal
5
of beam crystallization has not yet been reached in
high-energy storage rings. Electron-cooled dilute beams of highly
charged ions show liquid-like order
6,7
with unique applications
8
.
An experiment
5
using laser cooling
9,10
suggested a reduction of
intra-beam heating, although the results were ambiguous. Here we
demonstrate the crystallization of laser-cooled Mg
+
beams circulat-
ing in the radiofrequency quadrupole storage ring PALLAS
11,12
at a
velocity of 2,800 ms
-1
, which corresponds to a beam energy of
1 eV. A sudden collapse of the transverse beam size and the low
longitudinal velocity spread clearly indicate the phase transition.
The continuous ring-shaped crystalline beam shows exceptional
stability, surviving for more than 3,000 revolutions without
cooling.
In ion storage rings
13
, gaseous ion beams are heated through
scattering within the beam. In combination with the varying
focusing and bending elements, this mechanism couples part of
the beam energy into the random ion motion. Cooling increases the
phase-space density of the beam, and further amplies the scatter-
ing rate
14
. In high-energy rings, this vicious circle can only be
overcome by increasing the number of focusing sections
15
, or, in
practice, by choosing a sufciently low density of stored ions
5
.
Recently, space-charge-limited densities have been reached
10
, but
no beam crystallization has been observed so far, in striking
contrast to the routine generation of elongated ion crystals at
rest in linear
16,17
and ring traps
12,18
. As an illustration, in Fig. 1 we
present images of ion crystals at rest
12
, gained with our storage
ring PALLAS (Paul laser cooling acceleration system), described
below. Ions remain ordered because their mutual Coulomb
repulsion overcomes their mean kinetic energy. The overall Cou-
lomb repulsion is compensated by an external parabolic trapping
potential w. The formation of the crystalline structure is well
understood
4,1720
. It develops from a linear chain of ions over a zig-
zag band to three-dimensional helices when the linear ion
density
19
l is increased, either by adding more ions or by reducing
the conning potential.
The difference between the behaviour in storage rings and traps
might be caused by the predicted excitation of ion crystals passing
through the periodic bending and focusing sections of a ring
2022
.
To bridge this gap and experimentally elucidate the conditions
necessary for obtaining crystalline beams in high-energy rings, we
2001 Macmillan Magazines Ltd
letters to nature
NATURE | VOL 412 | 16 AUGUST 2001 | www.nature.com 717
coupling constant. More general conditional evolutions of the environment E can be of
course considered.
Following entanglement, the state of the system alone is described by the reduced
density matrix obtained from jF
SE
i by a trace over the environment:
r
S
Tr
E
jF
SE
ihF
SE
j jaj
2
j ih j zab*j ih 2j z*ab*j 2ih j jbj
2
j 2ih 2j
Disappearance of the off-diagonal terms signies perfect decoherence. In the {j i; j 2i}
basis this is guaranteed when the overlap z he

je
2
i Trje
2
ihe

j disappears. It is
therefore natural to measure effectiveness of decoherence by the magnitude of the overlap
of the two conditional states of the environment, which in turn determines the degree of
suppression of the off-diagonal terms in rS.
This two-paragraph `crash course' is no substitute for a more complete discussion of
decoherence
68
. We have swept a number of issues under the rug. Foremost among them is
einselectionthe emergence, in the course of decoherence, of the preferred set of pointer
states that habitually appear on the diagonal of r
S
essentially independently of the initial
states of either E or S. Stability of these pointer states (rather than the diagonality of the
density matrix in some basis) is the key to the role played by decoherence in the transition
from quantum to classical
28,18
.
Another important subject avoided here is the likely situation when the state of the
environment is represented by a mixture r
E
. Detailed discussion of this case (treated
extensively before
68,1719
, although not from the point of view of the sub-Planck
structures) is beyond the scope of this paper, but the basic conclusion is easy to state: the
estimated magnitude of the smallest displacement leading to orthogonality is still given by
equation (17). That is, it is still related to the smallest scales compatible with the classical
action A associated with r
E
. Off-diagonal terms of r
S
are suppressed by z TrU
2
r
E
U

,
which is the relevant generalization of equation (16). This expression can be expanded for
simple displacements U
6
D
6
in the limit of small shifts to recover equation (17), with
the only difference arising fromthe fact that nowthe mixture r
E
must be used to obtain the
averages (hx
2
i Trx
2
r
E
, etc.).
Note that the size of the structures in phase space depends on the representation of the
quantum state. For instance, one does not expect scale a of equation (3) in the Husimi
function, that is, in the Wigner function smoothed on a Planck scale. This is no surprise
and does not affect our conclusions concerning decoherence, as only the Wigner function
yields scalar product through the simple equation (14).
Note that the discussion throughout the paper is set in one spatial dimension. Generali-
zation to d dimensions is as conceptually straightforward as it is notationally cumbersome.
The structure saturates in volumes of a
d
, etc. This has little effect on decoherence, as it
depends on displacements that yield orthogonality, and these are still d < a.
Received 25 September 2000; accepted 25 June 2001.
1. Heisenberg, W. U

ber den anschaulichen Inhalt der quantentheoretischen Kinematik and Mechanik.

Z. Phys. 43, 172198 (1927); The physical content of quantum kinematics and mechanics (Engl.
Trans.) in Quantum Theory and Measurement (eds Wheeler, J. A. & Zurek, W. H.) (Princeton Univ.
Press, Princeton, 1983).
2. Zurek, W. H. Decoherence, chaos, quantum-classical correspondence, and the algorithmic arrow of
time. Phys. Script. T76, 186198 (1998).
3. Zurek, W. H. Pointer basis of a quantum apparatus: Into what mixture does the wavepacket collapse?
Phys. Rev. D 24, 15161524 (1981).
4. Zurek, W. H. Environment-induced superselection rules. Phys. Rev. D 26, 18621880 (1982).
5. Joos, E. & Zeh, H. D. The emergence of classical properties through the interaction with the
environment. Z. Phys. B 59, 229 (1985).
6. Zurek, W. H. Decoherence and the transition from quantum to classical. Phys. Today 44, 3646
(1991).
7. Giulini, D., Joos, E., Kiefer, C., Kupsch, J., Stamatescu, L.-O. & Zeh, H. D. Decoherence and the
Appearance of a Classical World in Quantum Theory (Springer, Berlin, 1996).
8. Zurek, W. H. Decoherence, einselection, and the quantum origin of the classical. Rev. Mod. Phys.
(in the press); also as preprint (quant-ph 010527) at hhttp://xxx.lanl.govi (2001).
9. Haake, F. Quantum Signatures of Chaos (Springer, Berlin, 1991).
10. Casati, G. & Chrikov, B. Quantum Chaos (Cambridge Univ. Press, Cambridge, 1995).
11. Hillery, M., O'Connell, R. F., Scully, M. O. & Wigner, E. P. Distribution functions in physics:
Fundamentals. Phys. Rep. 106, 121167 (1984).
12. Berry, M. V. & Balazs, N. L. Evolution of semiclassical quantum states in phase space. J. Phys. A 12,
625642 (1979).
13. Korsch, H. J. & Berry, M. V. Evolution of Wigner's phase-space density under a nonintegrable
quantum map. Physica D 3, 627636 (1981).
14. Zurek, W. H. & Paz, J. P. Decoherence, chaos, and the Second Law. Phys. Rev. Lett. 72, 25082511
(1994).
15. Berman, G. P. & Zaslavsky, G. M. Condition of stochasticity in quantum non-linear systems. Physica
(Amsterdam) 91A, 450 (1978).
16. Habib, S., Shizume, K. & Zurek, W. H. Decoherence, chaos, and the correspondence principle. Phys.
Rev. Lett. 80, 4361 (1998).
17. Caldeira, A. O. & Leggett, A. J. Path-integral approach to quantum Brownian motion. Physica 121A,
587616 (1983).
18. Paz, J. P. & Zurek, W. H. in Les Houches Lectures Session LXXII (eds Kaiser, R., Westbrook, C. and
David, F.) 533614 (Springer, Berlin, 2001).
19. Braun, D., Haake, F. &Strunz, W. A. Universality of decoherence. Phys. Rev. Lett. 86, 29132917 (2001).
20. Hannay, J. H. &Berry, M. V. Quantization of linear maps on a torusFresnel diffraction by a periodic
grating. Physica 1D, 267290 (1980).
21. Caves, C. in Physical Origins of Time Asymmetry (eds Halliwell, J. J., Perez-Mercader, J. &Zurek, W. H.)
4777 (Cambridge Univ. Press, Cambridge, 1993).
22. Miller, P. A. &Sarkar, S. Signatures of chaos in the entanglement of two coupled quantumkicked tops.
Phys. Rev. E 60, 1542 (1999).
23. Braginsky, V. B. & Khalili, F. Y. Quantum nondemolition measurements: the route from toys to tools.
Rev. Mod. Phys. 95, 703711 (1996).
24. Karkuszewski, Z., Zakrzewski, J. & Zurek, W. H. Breakdown of correspondence in chaotic systems:
Ehrenfest versus localization times. Preprint quant-ph/0010011 at hhttp://xxx.lanl.govi (2000).
Acknowledgements
This research was supported in part by the National Security Agency. I thank A. Albrecht,
N. Balazs, C. Jarzynski, Z. Karkuszewski and J. P. Paz for useful chaotic conversations.
Correspondence and requests for materials should be addressed to W.Z.
(e-mail: whz@LANL.gov).
.................................................................
Crystalline ion beams
T. Scha tz, U. Schramm & D. Habs
Sektion Physik, LMU Munchen, D-85748 Garching, Germany
..............................................................................................................................................
By freezing out the motion between particles in a high-energy
storage ring, it should be possible
14
to create threads of ions,
offering research opportunities beyond the realm of standard
accelerator physics. The usual heating due to intra-beam colli-
sions should completely vanish, giving rise to a state of unprece-
dented brilliance. Despite a continuous improvement of beam
cooling techniques, such as electron cooling and laser cooling, the
ultimate goal
5
of beam crystallization has not yet been reached in
high-energy storage rings. Electron-cooled dilute beams of highly
charged ions show liquid-like order
6,7
with unique applications
8
.
An experiment
5
using laser cooling
9,10
suggested a reduction of
intra-beam heating, although the results were ambiguous. Here we
demonstrate the crystallization of laser-cooled Mg
+
beams circulat-
ing in the radiofrequency quadrupole storage ring PALLAS
11,12
at a
velocity of 2,800 ms
-1
, which corresponds to a beam energy of
1 eV. A sudden collapse of the transverse beam size and the low
longitudinal velocity spread clearly indicate the phase transition.
The continuous ring-shaped crystalline beam shows exceptional
stability, surviving for more than 3,000 revolutions without
cooling.
In ion storage rings
13
, gaseous ion beams are heated through
scattering within the beam. In combination with the varying
focusing and bending elements, this mechanism couples part of
the beam energy into the random ion motion. Cooling increases the
phase-space density of the beam, and further amplies the scatter-
ing rate
14
. In high-energy rings, this vicious circle can only be
overcome by increasing the number of focusing sections
15
, or, in
practice, by choosing a sufciently low density of stored ions
5
.
Recently, space-charge-limited densities have been reached
10
, but
no beam crystallization has been observed so far, in striking
contrast to the routine generation of elongated ion crystals at
rest in linear
16,17
and ring traps
12,18
. As an illustration, in Fig. 1 we
present images of ion crystals at rest
12
, gained with our storage
ring PALLAS (Paul laser cooling acceleration system), described
below. Ions remain ordered because their mutual Coulomb
repulsion overcomes their mean kinetic energy. The overall Cou-
lomb repulsion is compensated by an external parabolic trapping
potential w. The formation of the crystalline structure is well
understood
4,1720
. It develops from a linear chain of ions over a zig-
zag band to three-dimensional helices when the linear ion
density
19
l is increased, either by adding more ions or by reducing
the conning potential.
The difference between the behaviour in storage rings and traps
might be caused by the predicted excitation of ion crystals passing
through the periodic bending and focusing sections of a ring
2022
.
To bridge this gap and experimentally elucidate the conditions
necessary for obtaining crystalline beams in high-energy rings, we
2001 Macmillan Magazines Ltd
letters to nature
718 NATURE | VOL 412 | 16 AUGUST 2001 | www.nature.com
constructed the low-energy radio frequency (r.f.) quadrupole stor-
age ring PALLAS
11
, which resembles a linear Paul trap, bent in a
circle
18
(see Fig. 2). Sixteen individual drift tubes enclose the
quadrupole rods and can be used to transport and position ions
along the orbit
12
. Care has been taken to keep potential distortions
on axis below100 meVand to provide perfect grounding of the drift
tubes to facilitate ion acceleration and to reduce heating due to
longitudinal excitations. Radial connement is achieved by apply-
ing an r.f. voltage U
0
cost of a typical amplitude U
0
200 Vand
frequency 2p 36:3 MHz between the quadrupole ring elec-
trodes. Stored ions experience a harmonic pseudo-potential
wr w
0
r
2
=r
2
0
with a depth of w
0
eU
2
0
=4m
2
r
2
0
4 V, where e
and m stand for the ion charge and mass of the
24
Mg
+
ions and
r
0
2:5 mm for the aperture radius. The oscillation of ions in this
conning pseudo-potential corresponds to the betatron oscil-
lation
13
of ions in a high-energy ring, and the period of the time-
varying potential corresponds to its discrete focusing sections. For
the given parameters, the number of betatron oscillations per round
trip Q, which characterizes the focusing conditions, amounts to
Q 50. The ion velocity spread is reduced by means of standard
laser cooling, exploiting the Doppler-shifted 3s
2
S
1/2
3p
2
P
3/2
transi-
tion of the moving
24
Mg
+
ions.
Ions are accelerated by the light pressure of the continuously
tuned co-propagating laser (force +F
1
). The nal beam velocity of
v 2;800 ms
21
is dened by the counter-propagating and thus
decelerating laser (-F
2
), kept at xed frequency. The longitudinal
velocity spread of the beam is reduced by the dispersive character of
the combined laser force (F
1
2 F
2
). The typical behaviour of a
circulating cold ion beam is shown by the dotted curve in Fig. 3. As
the detuning of the co-propagating laser is decreased, the uores-
cence rate increases. Then, as it approaches the resonance, the forces
start to compensate and the rate drops off again. Increasing the
conning potential markedly changes the behaviour (solid curve).
Here the uorescence at rst follows the previous curve, then
decreases abruptly, and subsequently rises to a sharp peak whose
width is dominated by the saturation-broadened linewidth of the
transition. This signature of an abrupt decrease has been previously
observed in ion traps
23
and indicates the phase transition to the
crystalline state. After a slight reduction of the potential, the phase
transition can no longer be resolved (dashed curve). This behaviour
suggests that the ions are conned strongly enough to form the
string but do not experience unnecessary r.f. heating, which is
responsible for the visibility of the phase transition.
Additionally, the phase transition is pinpointed by a sudden
decrease of the transverse beam size (Fig. 4a, b). A comparison of
the spatial prole with that in Fig. 1a species the formation of a
circulating ion string. The number of particles in the string (18,000)
was deduced fromthe overall uorescence yield and denotes a mean
interparticle distance of d <20 mm (l 0:4). The beam radius j
r
corresponds to an initial transverse temperature of T
'
30 K for
the gaseous and T
'
,0:4 K (resolution-limited) for the crystalline
beam. The longitudinal velocity spread of the beam was probed
opposite to the cooling section (see Fig. 2). The line proles in
Fig. 4c and d reveal a width of the velocity distribution of 4 ms
-1
(T
k
50 mK) for the cold gaseous beam and less than 1 ms
-1
laser (
1
)
probe
cooling
cooling
fluorescence
detection
photomultiplier 1
intensified
CCD camera or
oven for
Mg-atoms
2r
0
10 mm
electron gun
ring electrode
drift tube
r.f.+
r.f.+
r.f.
r.f.

1
1
5

m
m
laser (
2
)
laser (
1
)
photomultiplier 2
Figure 2 Axial and radial cut through the r.f. quadrupole storage ring PALLAS. The
length of the orbit enclosed by the ring-shaped quadrupole electrodes is 0.36 m.
Sixteen drift tubes are distributed around the ring, which can be powered individually.
The drift tubes at the two opposite locations for laser cooling and laser probing are
highlighted. To load the ring,
24
Mg atoms are ionized inside the trapping volume by
electron bombardment.
cold beam (
0
= 6.3 V )
crystalline beam (
0
= 15 V )
phase transition
crystallizing beam (
0
= 17 V )
100
80
60
40
20
0
150 100 50 0
Relative detuning ( /2)
F
l
u
o
r
e
s
c
e
n
c
e

r
a
t
e

(
k
H
z
)
Figure 3 Fluorescence rate of the ion beam as a function of the frequency detuning of the
co-propagating laser. The detuning is given in terms of half the natural transition linewidth
( 2p 342:7 MHz). The laser is slowly (4 s) tuned towards the resonance with the
counter-propagating laser at xed frequency, which denes the nominal ion velocity of
2,800 ms
-1
. The beam contained 18,000
24
Mg
+
ions, as deduced from the uorescence
rate which was calibrated against the ion crystal at rest. The marked discontinuity in the
solid curve (arrow) is characteristic of the phase transition to the crystalline beam.
z [m]
4
2
= N/z a
a
r
= 5.2 m

0
0 100 200
300 400
0
= 2.8 V
0
= 6.3 V

b
Figure 1 Images of ion crystals at rest in PALLAS. False colours reect the uorescence
intensity of individual ions. The ions are longitudinally conned in a weak static potential
12
which is generated by the two drift tubes, highlighted in Fig. 2. The crystal becomes more
complex when the linear ion density l N=z 3a, N denoting the number of particles
and a the WignerSeitz radius
19
, increases with lowered conning potential w
0
from a to
b or when it increases stepwise along the axis z because of the weak longitudinal
connement, as illustrated in b. The radius j
r
5:2 mm of the linear ion string mainly
reects the overall spatial resolution (integration time 0.4 s).
2001 Macmillan Magazines Ltd
Sistemas modelo para estudo de transies de fase em estruturas cristalinos.
Sistemas clssicos (a
1/3
)
=
h

2mk
B
T
<< a
20
VOLUME 83, NUMBER 23 P HYS I CAL RE VI E W L E T T E RS 6 DECEMBER 1999
Structure of a Three-Dimensional Coulomb Crystal in a Fine-Particle Plasma
Yasuaki Hayashi*
Department of Electronics and Information Science, Kyoto Institute of Technology, Matsugasaki, Sakyo-ku, Kyoto 606-8585, Japan
(Received 25 January 1999)
The structure of a three-dimensional Coulomb crystal formed with 1.4-mm particles grown in a
plasma was analyzed by observation from the top and the side using two CCD video cameras at the
same time and at the same position. The crystal structure was generally face-centered orthorhombic
with (001) planes parallel to an electrode. Under coexistence with the close-packed structure in a one-
dimensional external force, the transition between the two structures occurred through the slip of planes
perpendicular to the force direction with constant interplane distance.
PACS numbers: 52.25.Zb
Small particles become negatively charged in a plasma
due to the difference in mobility between ions and elec-
trons, and they collectively interact, i.e., become strongly
coupled when their charge is high enough. It has been
predicted that they can form a crystal structure, that is,
Coulomb crystal, under proper conditions even if the
Coulomb interaction energy is reduced by the Debye
shielding in a plasma [1]. Observations of the crystal or-
dering were reported in 1994 by three groups indepen-
dently [24] and many experimental results on Coulomb
crystals or ordered structures have been published since
then [59]. Coulomb crystals in ne-particle plasmas are
good models for real atomic crystals, and many physi-
cal phenomena, e.g., the liquid-solid phase transition, are
expected to be claried through the investigation of the
crystals because their scale is large enough to be easily
observed with the use of a video camera.
Most Coulomb crystal structures that have been reported
to be observed in ne-particle plasmas are two dimen-
sional, i.e., simple hexagonal, and some are three dimen-
sional such as body-centered cubic (bcc) or face-centered
cubic (fcc) [3,6,10]. The observed bcc structure was that
with (110) or (001) planes parallel to electrodes, perpen-
dicular to the direction of the electric eld of the electrode
sheath. In order to obtain accurate relative lattice con-
stant values, three-dimensional observation of the crystal
at one time is indispensable, because ne particles in those
structures shift their positions in several seconds to change
the structures. In this work, the three-dimensional struc-
ture of three-dimensional Coulomb crystals is investigated
through the observation from the top and the side with the
use of two CCD video cameras at the same time and at the
same space position.
An experimental system was constructed as shown in
Fig. 1. The system consists of a vacuum reaction cham-
ber, an argon-ion laser (wavelength: 488 nm), two CCD
video cameras, and a rotating analyzer for the measure-
ment of the Mie-scattering ellipsometry [11,12]. The vac-
uum chamber is a cube with a side length of 10 cm.
Viewing windows are provided in the ve faces of the
chamber. Two are for the incidence and outgoing of
argon-ion-laser light. Two others are for the observa-
tion of Coulomb crystals from the side and the top. The
other one is for the monitoring of the particle diameter by
Mie-scattering ellipsometry. 13.56 MHz rf was applied
to an electrode of 4 cm diameter with the chamber wall
grounded. A ring of 3 cm inner diameter was put on the
electrode in order to effectively trap ne particles above it
by forming a potential bucket. Gas inlet and exhausting
ports were provided close to each other so that particles
were not transported by gas ow.
Fine particles were prepared on the seeds of ultrane
carbon particles, which were injected at the rst stage of
the growth, in a 20% methane/argon plasma under the
conditions of 5 W rf power and 40 Pa (0.3 Torr) pressure.
Spherical carbon particles can be grown through coating
of hydrogenated amorphous carbon on the seeds by the
dissociation of methane gas in the plasma [2,12]. Size
distribution of such particles was found from the analysis
of Mie-scattering ellipsometry to be shifted log normal
on a linear scale with geometric standard deviation of
1.5 and geometric mean size of 50 nm for seed particles
[2]. Using Mie-scattering ellipsometry, the diameter of
FIG. 1. Schematic of the experimental system.
4764 0031-90079983(23)4764(4)$15.00 1999 The American Physical Society
VOLUME 83, NUMBER 23 P HYS I CAL RE VI E W L E T T E RS 6 DECEMBER 1999
growing particles was monitored, and the growth was
stopped at the time of the diameter of 1.4 mm, at which
three-dimensional Coulomb crystals of ne ordering were
formed in former experiments [13,14], by the decrease
of rf power to 1 W. For the above size distribution,
98% of the particles are in the size of 1.4 6 0.05 mm.
Particle arrangement above the center of the rf electrode
was observed from two directions through convex lenses
by CCD video cameras. A laser light beam of 1.4 mm
diameter was directed and the brightness of the CCD
cameras was adjusted in order that the bright images were
taken for ne particles in a few of the lowest layers and in
several vertical layers. Since the time was also recorded
in both top and side images, the images of the same time
were possible to be certied later.
Fine particles spread over the entire region in the poten-
tial bucket are suspended about 5 mm above the electrode
in the luminous region of the negative glow. The result
of measurement of potential distribution using a Langmuir
probe in pure argon plasma proved that ne particles were
in the quasineutral transition region as in a former experi-
ment [15] according to the Bohm sheath criterion. The
top and side images, which were taken at the same posi-
tion and at the same time, of a three-dimensional Coulomb
crystal formed by 1.4 mm carbon particles are shown in
Fig. 2. The relative horizontal positions of ne particles
in the top view were adjusted so as to agree with those in
the side view at a crystal-grain boundary. The accuracy
of the relative positions is estimated to be a few vertical
planes. Bright spots in the top view image indicate par-
ticles in the lowest layer and comparatively dimmed or
small spots indicate those in the second lowest layer, be-
cause particles were illuminated in the off axis of the laser
beam of the Gaussian distribution under them. From the
correspondence of particle arrangement in the top view
with that in the side view, particles are found to be aligned
in the perpendicular direction of the side view plane. This
means that each bright spot in the side view shows par-
ticles piled up in the direction (see the right region of the
gure). From these facts, particle positions in each layer
were identied in the top view by manual pointing on the
image taken into a computer. In Fig. 3(a), they are indi-
cated by closed circles for those in the lowest layer and
open circles in the second lowest layer corresponding to
the top view of Fig. 2. It is found that the particles in
the second lowest layer are above the center of two par-
ticles in the lowest layer. In the top view of Fig. 2, some
particles in the lowest layer are seen to be piled up by
the particles that are in the third lowest layer. And, by
the correspondence to the side view, it is also found that
particles in the third or fth lowest layer are arranged at
the same horizontal positions as in the lowest layer and
that they are arranged periodically in the vertical direc-
tion as well as in the horizontal direction. As a result, it
can be realized that the crystal structure is similar to the
body-centered cubic (bcc) structure with (110) planes par-
FIG. 2. Top and side video images of a three-dimensional
Coulomb crystal taken at the same time and the same position.
allel to the electrode. The three-dimensional structure of
a unit cell of the crystal is depicted in Fig. 4. The lattice
constants a , b, and c are determined from the average
in the image to be 106, 157, and 165 mm, respectively.
The relation among these sizes is
p
2 a , b , c; there-
fore the crystal structure is not strictly bcc but generally
face-centered orthorhombic (fco). The top view and the
side view in Fig. 2 show (001) planes and planes per-
pendicular to [110] axes of the structure, respectively. A
(001) plane of a unit cell of the structure is indicated by
a dotted square in Fig. 3(a). In the video images, at least
eight horizontal layers in the side view and tens of unit
cells in one crystal grain in the top view were observed.
The Wigner-Seitz radius ra is determined from the sizes
of the lattice constants to be 54.7 mm. When the Debye
length lD is 20 mm, which was calculated from measured
plasma density of 4 3 10
9
cm
23
by a Langmuir probe
for a pure argon plasma and estimated ion temperature
of 350 K, k ralD 2.7. When uncertainty is taken
into account for plasma density of 2 8 3 10
9
cm
23
and
ion temperature of 300500 K, the k value ranges from
1.6 to 4.1.
The averages and standard deviations of the lattice
constants for 11 data during 10 s around the time for
Fig. 2 were 106 and 2.6 mm for a, 159 and 4.1 mm for
4765
=
_

0
h
mv
f
B
(v)dv
f
B
(v)dv = 4
_
m
2k
A
BT
_
3/2
v
2
exp
_

mv
2
2k
B
T
_
cristalizao de Wigner
=
Q
4
0
ak
B
T
<
c
Suspenses coloidais consistem de partculas com dimetro da ordem de 10
8
a 10
4
, dispersas
em uma soluo uida. Suspenses aquosas de polystyrene e polymethylmethacrylate so
bastante utilizadas experimentalmente. As macro-partculas num uido atmico tem massa
muito maior que as dos componentes do uido, e portanto seu movimento pode ser considerado
desacoplado (10
14
a 10
12
s para os tomos e 10
9
a 10
5
s para os coloides.
21
A primeira contribuio para fora entre os colides vem de uma atraode van der Walss
dipolo-dipolo induzido, que sozinha levaria o sistema a uma oculao irreversvel.
estabilizao estrica
Hard Sphere
Softsphere
V (r
ij
) =
_
_
_
4
_
_

r
ij
_
12

_

r
ij
_
6
_
+ r
ij
< 2
1/6

0 r
ij
2
1/6

Estabilizados por carga (solvente), contra-ons

V (r
ij
) =
Z
2
e
2
exp (r
ij
)
4r
ij
Congurao do sistema para uma dada temperatura pode ser obtida por
Dinmica Molecular (Elementos Discretos)
Monte Carlo
Minimizando a energia calculada analiticamente para uma dada congurao (em
T = 0)
22
7.1 Fases estruturais
Generic properties of a quasi-one-dimensional classical Wigner crystal
G. Piacente,* I. V. Schweigert,

J. J. Betouras,

and F. M. Peeters

Department of Physics, University of Antwerp (Campus Drie Eiken), B-2610 Antwerpen, Belgium
Received 24 June 2003; published 30 January 2004
We studied the structural, dynamical properties and melting of a quasi-one-dimensional system of charged
particles, interacting through a screened Coulomb potential. The ground-state energy was calculated and,
depending on the density and the screening length, the system crystallizes in a number of chains. As a function
of the density or the conning potential, the ground state congurations and the structural transitions between
them were analyzed both by analytical and Monte Carlo calculations. The system exhibits a rich phase diagram
at zero temperature with continuous and discontinuous structural transitions. We calculated the normal modes
of the Wigner crystal and the magnetophonons when an external constant magnetic eld B is applied. At nite
temperature the melting of the system was studied via Monte Carlo simulations using the modied Lindemann
criterion MLC. The melting temperature as a function of the density was obtained for different screening
parameters. Reentrant melting as a function of the density was found as well as evidence of directional
dependent melting. The single-chain regime exhibits anomalous melting temperatures according to the MLC
and as a check we study the pair-correlation function at different densities and different temperatures, which
allowed us to formulate a different melting criterion. Possible connection with recent theoretical and experi-
mental results are discussed and experiments are proposed.
DOI: 10.1103/PhysRevB.69.045324 PACS numbers: 64.60.Cn, 64.70.Dv, 61.46.w
I. INTRODUCTION
Recently there has been a great deal of interest in meso-
scopic systems consisting of interacting particles in low di-
mensions or conned geometries. A class of quantum aniso-
tropic systems exhibiting stripe behavior appears in the
quantum Hall effect,
1
in oxide manganites, and in high-T
c
superconductors,
2
where electronic strong correlations are re-
sponsible for the formation of these inhomogeneous phases.
Another class of conned quasi-one-dimensional Q1D ge-
ometries appears in many diverse elds of research and some
typical and important examples from the experimental point
of view are: electrons on liquid Helium,
3,4
microuidic
devices,
5
colloidal suspensions,
6
and conned dusty plasma.
7
A major phenomenon which is expected to occur in
charged particles interacting via a Coulomb or screened Cou-
lomb potential is Wigner crystallization WC Ref. 8 at low
enough temperatures and densities when the potential energy
overwhelms the kinetic energy. Indeed, evidence of such a
type of transition was found very recently
3
in experiments on
electrons on the surface of liquid Helium, where the elec-
trons were conned by metallic gates and exhibited dynami-
cal ordering in the form of laments. This particular experi-
ment posed many interesting questions regarding the nature
of the transition to WC, its density dependence, and the melt-
ing. Furthermore, the considered system has been proposed
as a possible step towards the realization of a quantum com-
puter with electrons oating on liquid Helium.
9
In this paper, as a rst step towards the understanding of
the behavior of these systems, we start with a two-
dimensional system consisting of an innite number of
charged particles and we impose a parabolic conning poten-
tial in one direction. The particles interact with a Yukawa-
type potential where the screening length is an external pa-
rameter. Physically, it can be adjusted, e.g., by the gate
voltage that connes the electrons. The combination of the
interaction among particles and the external potential leads
to a rich structural phase diagram as a function of the screen-
ing length and the density n of the system. The structural
units at temperature T0 K) are parallel chains of particles
the number of which depends on the values of and n. The
transition from one conguration to the other can be obtained
via a rst- or a second-order transition.
Before proceeding further, we should comment on the
possibility of two-dimensional crystalline order. According
to the Mermin-Wagner theorem
10
there is no true long-range
crystalline order in two dimensions. However, this theorem
is only strictly valid when the potential falls off faster than
1/r and in the thermodynamic limit. When the same argu-
ments of the theorem are applied to a large but nite system,
no inconsistencies arise from the assumption of crystalline
order. Thus any system that can be studied in laboratory or in
computer simulations can exhibit crystalline order.
11
On the
other hand, short-range order is expected to form even in the
thermodynamic limit.
In a related work
12
which discussed the temperature
equilibration of a one-dimensional Coulomb chain, two dif-
ferent equilibration temperatures were assigned (T

and T

),
reecting the different behavior of the modes due to the
strong connement.
The WC in strictly one-dimensional and in the quantum
regime was studied by Schultz.
13
He found that for arbitrarily
weak Coulomb interaction the density correlations at wave
vector 4k
F
decay extremely slowly the most slowly decay
term is exp(cln x)].
Other remarkable work on the quantum transport and pin-
ning in the presence of weak disorder, where it was shown
that quantum uctuations soften the pinning barrier and
charge transfer occurs due to thermally assisted tunneling, is
described in Ref. 14.
In addition to the structural properties, it is instructive to
study the normal modes of these kind of anisotropic systems.
PHYSICAL REVIEW B 69, 045324 2004
0163-1829/2004/694/04532417/$22.50 2004 The American Physical Society 69 045324-1
23
As guras abaixo mostram a energia do sistema e a derivada para = 1
If the particles crystallize in a single chain, the minimum
energy is obtained when the particles are placed on the y
axis, where the conning potential is zero. In this case the
linear density is n

e
r
0
/a and the x coordinate of the par-
ticles are x
i
ia, with i0,1,2, . . . , . The energy
per particle is
E
1
n

e
j 1

1
j
exp j /n

e
. 3
The case of Coulomb interaction is treated using the Ewald
summation method so that the summation over long distance
can be done effectively. Following the standard
procedure
2124
we obtain for E
1
,
E
1
0
n

e
2

lim
x0

j
2e
2j x

1
j /2n

j 0

2
n

e
2
xj
2

1
n

2
n

e
2
x
2

e
2
1
x

,
4
where
1
(x)
x

dt exp(t
2
)1/t,
2
(x)
/xerfc(x), and erfc(y)12/
0
y
e
t
2
dt.
The rst summation contains a divergent term at j 0
coming from the lower limit of the integration in the function

1
(x0). This divergence is remedied if we subtract the
interaction energy E
b
of the negatively charged particles
with the positive background which also diverges logarith-
mically in one dimension. In that case we can proceed using
the limit lim
x0
x
1
erf(x)2/:
E
1
E
1
0E
b

e
2

j 0
2
1
j /2n

j 0

2
n

e
2
j
2

e
. 5
In the two-chain conguration the particles crystallize in two
parallel lines separated by a distance d and displaced by a
distance a/2 along the y axis. The energy per particle in this
case is
E
2

c
2
n

e
2

e
2

j 1

1
j
exp 2j /n

e
2

j 1

exp 2 j 1/2
2
c
2
/n

j 1/2
2
c
2
, 6
FIG. 1. The energy per particle as a function
of density for 1.
FIG. 2. The zero-temperature structural phase diagram.
GENERIC PROPERTIES OF A QUASI-ONE- . . . PHYSICAL REVIEW B 69, 045324 2004
045324-3
energy per particle for each chain. This is computed by con-
sidering a particle at a particular chain and taking into ac-
count all the interactions with the rest of the particles. The
cases of interest are the congurations for which it is pos-
sible to distinguish internal from external chains and may be
related to the difference in the melting behavior which is
discussed in Sec. V. The interesting observation is that in
every case the energy per particle is larger in the external
chain than the internal ones.
This asymmetry reects the fact that for each particle re-
siding in an external chain the gain in energy due to the
conning potential is higher than the difference in the Cou-
lomb energy due to the lack of symmetric neighboring
chains, as compared to a particle residing in an internal
chain. E.g., for a three-chain system where the middle chain
is the zeroth and the external ones are denoted by 1 and
1, we have for the energy of two particles:
E
1
E
0
E
conf , 1
E
Coulomb, 1,1
E
Coulomb, 1,0
0,
7
where E
Coulomb, ,
denotes the Coulomb energy of a particle
residing in chain interacting with the particles in chain
and E
conf ,
denotes its conning energy.
In the case of the rst density regimes where the four-
chain structure is optimal this difference is not large due to
the fact that the internal distance is lesser than the external
one. On the contrary, the difference is much larger in the
second regime of the four-chain structure. Another interest-
ing observation is that as we approach the limit of Coulomb
interactions (1) the energy difference tends to vanish and
the system behaves isotropically.
B. Structural transitions
We have seen that by increasing the density, the system
changes its conguration, in other words it undergoes a
structural transition. It is a natural question to study the
order of these transitions. For this purpose the derivative of
the energy with respect to the density was calculated which
is shown in Fig. 5 for the case of 1. For every value of
the inverse screening length only the transition between the
one- and the two-chain congurations is continuous and all
the others are discontinuous. This conclusion agrees with the
results of Fig. 3, where discontinuous changes of the lateral
position of the particles correspond to rst-order transitions.
The transition 12 is a zig-zag transition
25
Fig. 6. The
transition 24 occurs through a zig-zag transition of each
of the two chains accompanied by a shift of a/4 along the
chain, which makes it a discontinuous transition Fig. 6. In
principle, these kind of almost zig-zag transitions are pos-
sible for three-, four-, ve-, and six-chains to result into six-,
eight-, ten-, and twelve-chain structures, respectively. Actu-
ally, these were observed during the numerical simulations,
especially for very small value of , but they represent meta-
stable states and are not the most energetically favored con-
gurations.
C. Limit of short-range interaction and large density
In order to make the connection with the regime where
the hard-core potential can be used as a working hypothesis,
we investigate the limit a. It can be shown that the varia-
tion of the distances between chains can be neglected and in
the limit where m
0
2
W
2
q
2
/(a) (W is the width of the
strip, following the spirit of the hydrodynamic consideration
FIG. 5. The derivative of the energy with respect to the density
for 1. Only the transition from one to two wires is continuous
second order the rest are rst order.
FIG. 6. The mechanism of the structural tran-
sitions 12 chains and 24 chains.
GENERIC PROPERTIES OF A QUASI-ONE- . . . PHYSICAL REVIEW B 69, 045324 2004
045324-5
A gura abaixo mostra todo diagrama de fase
If the particles crystallize in a single chain, the minimum
energy is obtained when the particles are placed on the y
axis, where the conning potential is zero. In this case the
linear density is n

e
r
0
/a and the x coordinate of the par-
ticles are x
i
ia, with i0,1,2, . . . , . The energy
per particle is
E
1
n

e
j 1

1
j
exp j /n

e
. 3
The case of Coulomb interaction is treated using the Ewald
summation method so that the summation over long distance
can be done effectively. Following the standard
procedure
2124
we obtain for E
1
,
E
1
0
n

e
2

lim
x0

j
2e
2j x

1
j /2n

j 0

2
n

e
2
xj
2

1
n

2
n

e
2
x
2

e
2
1
x

,
4
where
1
(x)
x

dt exp(t
2
)1/t,
2
(x)
/xerfc(x), and erfc(y)12/
0
y
e
t
2
dt.
The rst summation contains a divergent term at j 0
coming from the lower limit of the integration in the function

1
(x0). This divergence is remedied if we subtract the
interaction energy E
b
of the negatively charged particles
with the positive background which also diverges logarith-
mically in one dimension. In that case we can proceed using
the limit lim
x0
x
1
erf(x)2/:
E
1
E
1
0E
b

e
2

j 0
2
1
j /2n

j 0

2
n

e
2
j
2

e
. 5
In the two-chain conguration the particles crystallize in two
parallel lines separated by a distance d and displaced by a
distance a/2 along the y axis. The energy per particle in this
case is
E
2

c
2
n

e
2

e
2

j 1

1
j
exp 2j /n

e
2

j 1

exp 2
j 1/2
2
c
2
/n

j 1/2
2
c
2
, 6
FIG. 1. The energy per particle as a function
of density for 1.
FIG. 2. The zero-temperature structural phase diagram.
GENERIC PROPERTIES OF A QUASI-ONE- . . . PHYSICAL REVIEW B 69, 045324 2004
045324-3
24
Melting
anticrossing between the two branches. Although these nd-
ings conrm the previous assertion that the presence of B
does not alter the structural properties of the system it also
reveals the differences softening of the acoustic mode at the
same density, inuence on the gap between optical and
acoustical branches and on eigenfrequencies within each
branch, which are induced by the magnetic eld.
V. MELTING
A. General discussion and results
In this section we study the melting of the WC by Monte
Carlo MC simulations. After the ground-state conguration
was achieved as explained in Sec. II, the system was heated
up by steps of size T, typically T510
4
, and equili-
brated to this new temperature during 10
5
10
6
MC steps. In
Fig. 12 we show typical trajectories of particles as they arise
from our MC simulation. It is evident that there is a different
behavior of the system in the x and the y directions as may be
expected by the anisotropy in the two directions. In order to
quantify the observations, we studied rst the potential en-
ergy as a function of temperature Fig. 13. In the crystalline
state the potential energy of the system increases practically
linearly with temperature and then exhibits a very fast in-
crease in a small critical temperature range after which it
starts to increase linearly again but now with a slightly larger
slope. In the latter region the system is in the disordered i.e.,
liquid phase. The fast increase of the potential energy is
indicative of the melting of the WC. To nd the critical tem-
peratures we studied, following the spirit of Ref. 32, the
modied Lindemann parameter L
p
u
2
/d
r
2
, where u
2
is
dened by the difference in the mean-square displacements
of neighboring particles from their equilibrium sites r

0
and
d
r
is the relevant interparticle distance as we discuss below.
The quantity u
2
can be written as
u
2

1
N

i1
N
1
N
nb

j 1
N
nb
r

i
r

0i
r

j
r

0j

, 25
where means the average over the MC steps, N is the total
number of particles in our simulation unit cell and the index
j denotes the N
nb
nearest neighbors of particle i. In order to
describe more accurately the difference between the two di-
rections, we studied separately u
x
2
and u
y
2
as functions of
temperature. For the melting along the x direction, the dis-
tance d
r
is the interparticle distance a introduced in Sec. I,
while for melting along the y direction d
r
is the interchain
distance which is a function of the density n

e
.
33,34
At low temperatures, the mean-square relative displace-
ments slowly increases linearly with temperature as a conse-
quence of harmonic oscillations of the particles about their
equilibrium positions see Fig. 14. From Fig. 14 we note
clearly that this linear increase is larger in the unconned
direction than in the conned direction. In some critical tem-
perature region, u
x
2
and u
y
2
start to increase very rapidly
which is the consequence of the fact that the particles have
attained sufcient thermal energy that they can jump be-
tween different crystallographic positions. According to the
modied Lindemann criterion MLC, when L
p
reaches the
FIG. 11. The magnetic-eld dependence of the softening of the
phonon mode at the structural transition from one to two chains.
FIG. 12. Particle trajectories for 310
7
MC steps for three different temperatures and
three different values of the density at 0. 01,
1, and 3.
GENERIC PROPERTIES OF A QUASI-ONE- . . . PHYSICAL REVIEW B 69, 045324 2004
045324-9
A abaixo para analisar a temperatura de melting em direes diferentes
semiempirical critical value 0.1 the system melts. This cri-
terion was used to dene the melting temperature T
m
.
From the corresponding analysis two different melting
temperatures T
x
and T
y
can be assigned. The results are sum-
marized in the phase diagram of Figs. 15a15c for
0.01, 1, and 3, respectively. There are several interesting
features in these phase diagrams.
a The nearly Coulomb system (0.01) has a melting
temperature which is on average 1520% higher than for the
screened Coulomb interparticle interaction with 1, which
has on its turn an average melting temperature about 15%
higher than the screened Coulomb system with 3. There-
fore, we conclude that the effect of screening is to reduce the
melting temperatures;
b A reentrant behavior is observed as a function of den-
sity, the minima of the melting temperatures occur at the
values of the density where the structural phase transitions
were predicted see Fig. 2;
c There is a regime close to each structural transition
point where the system is frustrated, in the sense that it uc-
tuates between the two structures. In this regime, which we
term as frustration regime, the system makes continuous
transitions from one metastable state to the other which
strongly reduces the melting temperature;
d For 1 and 3, there is a region in density for
which the system melts rst in the unconned direction,
while it is not melted in the conned one. This regime re-
sembles the ndings of Ref. 15 in the regime termed as
locked oating solid. For the Coulomb limit there is no evi-
FIG. 13. The energy per particle as a function of temperature for
the four-chain structure with 1 and n

e
3.9. There is a fast
increase of the energy at the melting temperature.
FIG. 14. The mean-square relative displacements for 1, n

e
1.6 in the case of the two-chain conguration. The dashed hori-
zontal line corresponds to the modied Lindemann criterion MLC
in the unconned direction, while the dotted line corresponds to the
MLC in the conned direction.
FIG. 15. Melting temperature as a function of density for: a
0.01, b 1, and c 3. The insets in a and b show an
enlargement of the four-chain region, which is located between the
two- and three-chain phases.
PIACENTE, SCHWEIGERT, BETOURAS, AND PEETERS PHYSICAL REVIEW B 69, 045324 2004
045324-10
semiempirical critical value 0.1 the system melts. This cri-
terion was used to dene the melting temperature T
m
.
From the corresponding analysis two different melting
temperatures T
x
and T
y
can be assigned. The results are sum-
marized in the phase diagram of Figs. 15a15c for
0.01, 1, and 3, respectively. There are several interesting
features in these phase diagrams.
a The nearly Coulomb system (0.01) has a melting
temperature which is on average 1520% higher than for the
screened Coulomb interparticle interaction with 1, which
has on its turn an average melting temperature about 15%
higher than the screened Coulomb system with 3. There-
fore, we conclude that the effect of screening is to reduce the
melting temperatures;
b A reentrant behavior is observed as a function of den-
sity, the minima of the melting temperatures occur at the
values of the density where the structural phase transitions
were predicted see Fig. 2;
c There is a regime close to each structural transition
point where the system is frustrated, in the sense that it uc-
tuates between the two structures. In this regime, which we
term as frustration regime, the system makes continuous
transitions from one metastable state to the other which
strongly reduces the melting temperature;
d For 1 and 3, there is a region in density for
which the system melts rst in the unconned direction,
while it is not melted in the conned one. This regime re-
sembles the ndings of Ref. 15 in the regime termed as
locked oating solid. For the Coulomb limit there is no evi-
FIG. 13. The energy per particle as a function of temperature for
the four-chain structure with 1 and n

e
3.9. There is a fast
increase of the energy at the melting temperature.
FIG. 14. The mean-square relative displacements for 1, n

e
1.6 in the case of the two-chain conguration. The dashed hori-
zontal line corresponds to the modied Lindemann criterion MLC
in the unconned direction, while the dotted line corresponds to the
MLC in the conned direction.
FIG. 15. Melting temperature as a function of density for: a
0.01, b 1, and c 3. The insets in a and b show an
enlargement of the four-chain region, which is located between the
two- and three-chain phases.
PIACENTE, SCHWEIGERT, BETOURAS, AND PEETERS PHYSICAL REVIEW B 69, 045324 2004
045324-10
25
7.2 O mtodo de Monte-Carlo
faz uso do conceito de Cadeia de Markov (anlogo estocstico da mecnica clssica), caracte-
rizado pela ausncia de memria (ramdom walk, por exemplo)
one-step probabilidade de transio
P(x
0
, ..., x
n
) = P(x
n
, y
n1
. . . P(x
1
, x
0
)a
0
Algortimo
1. especica uma congurao inicial x
0
.
2. gera um novo estado x
1
3. computa a probabilidade de transio P(x
0
, x
1
)
computa variao da energia H
se H 0 aceita
se H > 0
computa exp H/k
B
T
gera R [0, 1]
aceita se R exp H/k
B
T
a equao de Langevin
dv
dt
= v +

R(t)
26