Documente Academic
Documente Profesional
Documente Cultură
2005 InterPACK 05
San Francisco, California, USA, July 17-22, 2005
IPACK2005-73376
Illayathambi Kunadian
Center for Applied Energy Research
University of Kentucky
Lexington, Kentucky 40511
Email: ikuna0@engr.uky.edu
J. M. McDonough
Ravi Ranjan Kumar
Department of Mechanical Engineering
University of Kentucky
Lexington, Kentucky, 40506
INTRODUCTION
Fouriers law predicts thermal disturbances propagating at
infinite velocities, implying that a thermal disturbance applied
at a certain location in a solid medium can be sensed immediately anywhere else in the medium (violating precepts of special
relativity). In order to ensure finite propagation of thermal disturbances a hyperbolic heat conduction equation (HHCE) was
proposed [1, 2]. It has been shown that in certain situations [3]
the HHCE violates the second law of thermodynamics resulting
in physically unrealistic temperature distribution such as the temperature overshoot phenomenon observed in a slab subject to a
sudden temperature rise on its boundaries. Also, since the classical and hyperbolic models neglect the thermalization time (time
for electrons and lattice to reach thermal equilibrium) and relaxation time of the electrons, their applicability to very short pulse
laser heating [4, 5] becomes questionable.
Anisimov et al. [6] proposed a two-step model to describe
the electron temperature Te and the lattice temperature Tl during short-pulse laser heating of metals. Later, Qiu and Tien [4, 5]
rigorously derived the hyperbolic two-step model from the Boltzmann transport equation for electrons. They numerically solved
the equations of this model in the context of a 96 f s duration laser
pulse irradiating a thin film of thickness 0.1m. Predicted temperature change of the electron gas during the picosecond transient agreed well with experimental data, supporting the validity of the hyperbolic two-step model for describing heat transfer
mechanisms during short-pulse laser heating of metals.
ABSTRACT
An alternative discretization and solution procedure is developed for implicitly solving a microscale heat transport equation during femtosecond laser heating of nanoscale metal films.
The proposed numerical technique directly solves a single partial differential equation, unlike other techniques available in the
literature which split the equation into a system of two equations and then apply discretization. It is shown by von Neumann
stability analysis that the proposed numerical method is unconditionally stable. The numerical technique is then extended to
three space dimensions, and an overall procedure for computing
the transient temperature distribution during short-pulse laser
heating of thin metal films is presented. Douglas-Gunn timesplitting and delta-form Douglas-Gunn time-splitting methods
are employed to solve the discretized 3-D equations; a simple
argument for stability is given for the split equation. The performance of the proposed numerical scheme will be compared
with the numerical techniques available in the literature and it
is shown that the new formulation is comparably accurate and
significantly more efficient. Finally, it is shown that numerical
predictions agree with available experimental data during subpicosecond laser heating.
Address
c 2005 by ASME
Copyright
q(r,t) + q
(T (r,t))
q(r,t)
= k T (r,t) + T
.
t
t
(2)
= 2 T.
T
t 2
t
t
(3)
Zhang and Zhao [18, 19] have employed the iterative techniques GaussSeidel, successive overrelaxation (SOR), conjugate gradient (CG), and preconditioned conjugate gradient
(PCG) to solve a Dirichlet problem for the 3-D DPL equation.
A recent study [23] has shown that applying Neumann boundary
conditions results in non-symmetric seven banded positive semidefinite matrices that are not suitable for iterative methods like
CG and PCG. This suggests, we should seek a different numerical technique.
The purpose of this paper is to present a numerical formulation based on an unsplit DPL equation. Firstly, we present the
discretization and analysis of the 1-D DPL equation by considering heat conduction in a semi-infinite slab subject to sudden
temperature rise on one of its boundaries. Stability of the numerical scheme is shown using von Neumann stability analysis.
The numerical technique is then extended to three space dimensions, and a procedure for computing the transient temperature
distribution during short pulse laser heating of thin metal films
is briefly described. The discretized 3-D DPL equation will be
solved using a DouglasGunn time-splitting method and -form
DouglasGunn time-splitting method; a simple argument for stability will be given for the split equation. The performance of the
proposed numerical scheme will be compared with the numerical
techniques available in the literature. Finally, we present results
from specific problems providing comparison with experimental
data, classical and hyperbolic heat conduction equations.
(x,t) =
T (x,t) T0
,
TW T0
t
,
q
x
,
q
(4)
T
.
q
(5)
with Z =
(, 0) = 0 for [0, ),
(6)
and
BRIEF REVIEW OF ORIGINS OF DPL EQUATION
The dual phase lag concept is represented in [8] as
q(r,t + q ) = kT (r,t + T ),
(, ) = 0 ,
(7)
respectively.
Expressing (5) in terms of x and t leads to
(1)
T 2 T
+ 2 Z
t
t
t
2 T
x2
2 T
.
x2
(8)
c 2005 by ASME
Copyright
where,
t
1
1
t
C4 = Z +
,
C
=
Z
+
, (13a)
6
2 x2
2 x2
t
1
2
C5 = Z +
,
(13b)
+
1
+
2 x2
t
2
t 2
+
1
+
.
(13c)
C7 = Z +
2 x2
t
n+1 n
T
T
Tmn+1 Tmn +
t m
t m
"
n+1 2 n #
2
T
T
Z
2
x m
x2 m
"
n+1 2 n #
T
2 T
t
+
.(9)
=
2
x2 m
x2 m
1 n+1
T n+1
=
3Tm 4Tmn + Tmn1 ,
t
2t
mn
T
1 n+1
Tm Tmn1 ,
=
t m
2t
2 T
1
= 2 [Tm+1 2Tm + Tm1 ] .
x2
x
STABILITY ANALYSIS
We analyze the stability of the above scheme via a von Neumann analysis. Because the Eq. (12) is a three-level difference
scheme, the von Neumann condition supplies only a necessary
(and not sufficient) stability requirement in general. For the
present problem we define vn+1
= Tmn and replace Eq. (12) by
m
the system
(10a)
(10b)
n+1
n+1
+ Tm+1
C4 Tm1
+C5 Tmn+1
1
n
n
= C6 Tm1
+ Tm+1
+C7 Tmn vnm (14)
t
(10c)
The discretization shown in Eqs. (10a)(10c) renders the numerical scheme globally first-order accurate in time and second order
accurate in space. After plugging Eq. (10) into Eq. (9) we obtain
n+1
vm
= Tmn .
(15)
1 n+1
3Tm 4Tmn + Tmn1
Tmn+1 Tmn +
2t
1 n+1
Tm Tmn1
2t
n
Z n+1
n+1
n
2 Tm+1 2Tmn+1 + Tm1
Tm+1 2Tmn + Tm1
x
n
t n+1
n
n+1
(11)
+ Tm+1 2Tmn + Tm1
Tm+1 2Tmn+1 + Tm1
=
2
2x
n
and Tm1
= eih Tmn .
(16)
Then
C4 eih + eih Tmn+1 +C5 Tmn+1
1
= C6 eih + eih Tmn +C7 Tmn vnm (17)
t
n+1
vm
= Tmn .
(18)
+C5 Tmn+1
1
n
n
= C6 Tm1
+ Tm+1
+C7 Tmn Tmn1 , (12)
t
Tmn+1 =
3
1
C6 (2 cosh) +C7 n
T
vn (19)
C4 (2 cosh) +C5 m [C4 (2 cos h) +C5] t m
c 2005 by ASME
Copyright
n
vn+1
m = Tm ,
(20)
0.990
0.002
Tmn+1
vn+1
m
"
#
C6 (2 cosh)+C7
1
C4 (2 cosh)+C5 [C4 (2 cosh)+C5 ]t
Tmn
vnm
(21)
=1
We know that the error vector must satisfy this same equation.
n
Denoting this by znm R2 , we see that zn+1
m = Czm , where
C=
"
C6 (2 cos h)+C7
1
C4 (2 cos h)+C5 [C4 (2 cosh)+C5 ]t
0.01
0.01
(22)
0.990
is the amplification matrix for the present scheme. The von Neumann necessary condition for stability is
kCk 1,
0.002
=4
(23)
where k k denotes the spectral norm. Hence, we need to calculate the eigenvalues of C to analyze the stability of the method.
The characteristic polynomial is
2
0.01
0.01
1
x
C6 (2 cos h) +C7
1
+
= 0 (24)
C4 (2 cos h) +C5
[C4 (2 cos h) +C5] t
0.990
0.002
C6 (2 cosh)+C7
C4 (2 cosh)+C5
r
C6 (2 cosh)+C7
C4 (2 cosh)+C5
2
4 [C
=7
4 (2 cosh)+C5 ]t
.
2
(25)
We must determine the larger of these, and from the requirement
0.01
max(|+ |, | |) 1,
0.01
(26)
1
x
establish permissible bounds on t and x. Since it is very tedious to solve Eq. (25) analytically, we solve it numerically by
plugging different combinations of t and x for different wave
numbers . Figure 1 shows the distribution of for different values of t and x at wave numbers = 1, 4, 7 and Z = 10. From
the figure we can see that max(|+ |) 1. It is also found that
max(| |) 1 even though we have not shown it here. Tests
were conducted for different values of Z and it was found that
the stability requirement Eq. (26) is satisfied for all the values
of Z. This suggests that the proposed numerical scheme is unconditionally stable. The stability requirement is also met for
Z = 0 (hyperbolic case) and Z = 1 (parabolic case) implying that
the numerical scheme is unconditionally stable and the stability
analysis can be applied for both parabolic and hyperbolic models.
and
at wave
q 2 T 1 T
S
1
(2 T )
2
S
+
.
+
T
+
q
T
t 2
t
t
k
t
4
(27)
c 2005 by ASME
Copyright
1
Experiment, Qui and Tien
Experiment, Brorson et al.
Num. DPL Model
Analytical, Chiu
Num. CV wave model
Num. Diffusion model
0.8
1R
S(r,t) = 0.94J
t p
L
Lx 2
(x 2 ) + (y 2y )2 z 1.992 | t 2t p |
(31)
exp
2ro2
tp
0.6
where Lx and Ly are the length and width of the metal film respectively, and ro is the thickness of the laser beam. Applying
trapezoidal integration to Eq. (27) we obtain
0.4
0.2
0.5
1
time (ps)
1.5
#
"
i 1 T n+1 T n
q h n+1
n
T
Ti, j,k +
i, j,k
t i, j,k
t i, j,k
h
i
T 2 T n+1 2 T n
n i
t h 2 n+1
=
T
+ 2 T
2 "
#
n+1
n
n
Si,n+1
t Si, j,k + Si, j,k
j,k Si, j,k
(32)
+ q
+
k
2
2
Figure 2. Front surface transient response for a 0.1m gold film. Comparison between numerical (explicit and implicit schemes), analytical
= 1.2 104m2 s1 , k =
, T = 90ps, q = 8.5ps.
315W m K
Just as was done for the 1-D case, above, we apply secondorder backward difference for the time derivative at n + 1 and
a centered difference for the time derivative at n. Second-order
derivatives in space are approximated using the usual centereddifference scheme. Thus we have
S(x,t) = 0.94J
1R
x 1.992 | t 2t p |
exp
tp
tp
= 13.7J/m2, t
(28)
= 96 f s (1 f s = 1015s)
I(t) = I0 e
a| tt |
p
with a = 1.992,
tp 0
I(t) = I0 e
= 4 ln(2)
= 2.77,
T
t
n+1
i, j,k
n
i, j,k
2 T
i
1 h n+1
3Ti, j,k 4Ti,nj,k + Ti,n1
j,k
2t
i
1 h n+1
Ti, j,k Ti,n1
=
j,k
2t
1
Ti+1, j,k 2Ti, j,k + Ti1, j,k
2
x
2 T
1
= 2 Ti, j+1,k 2Ti, j,k + Ti, j+1,k
y2
y
1
2 T
= 2 Ti, j,k+1 2Ti, j,k + Ti, j,k1
2
z
z
x2
(29)
(33a)
(33b)
(33c)
(33d)
(33e)
gives an excellent autocorrelation of laser pulse with experimental results compared to the one used by Qiu and Tien [4, 5]:
2
tt
p
T
t
n+1
n+1
n+1
n+1
C4 Ti,n+1
j,k +C5 Ti+1, j,k + Ti1, j,k +C6 Ti, j+1,k + Ti, j1,k
n+1
n
+
T
(34)
+C7 Ti,n+1
j,k+1
i, j,k1 = F ,
(30)
5
c 2005 by ASME
Copyright
where
n
n
F n = C8 Ti,nj,k +C9 Ti+1,
j,k + Ti1, j,k
+C10 Ti,nj+1,k + Ti,nj1,k +C11 Ti,nj,k+1 + Ti,nj,k1
q n1
T
+ tG
t i, j,k
1
1
t
t
C5 = T +
,C6 = T +
,
2
2 x
2 y2
t
1
C7 = T +
,
2 z2
q
1
+
2(C5 +C6 +C7 )
C4 =
t
t
t
1
1
C9 = T +
,C10 = T +
,
2
2 x
2 y2
1
t
C11 = T +
,
2 z2
1 2q
2(C9 +C10 +C11 ),
C8 =
+
t
" n+1
#
n
Si,n+1
Si, j,k + Si,n j,k
j,k Si, j,k
+ q
G =
.
2
2
Observe that this is now in the general form required for construction of a multilevel Douglas and Gunn time splitting [21].
We now split Eq. (38) into three equations corresponding to x, y
and z directions:
0
n
n+1
n+1
1 2C50 Ti,n+1
+C
T
+
T
5
j,k
i+1, j,k
i1, j,k = S
n+1
n+1
0
2C60 Ti,n+1
j,k +C6 Ti, j+1,k + Ti, j1,k = 0
0
n+1
n+1
2C70 Ti,n+1
+C
T
+
T
7
j,k
i, j,k+1
i, j,k1 = 0.
(35)
(36a)
(36b)
(I + Ax )T (1) = Sn Ay T n Az T n
(36c)
(I + Ay )T (2) = T (1) Ay T n
(I + Az)T
(36d)
(36e)
(3)
=T
(2)
Az T ,
(36g)
(37)
C5
q ,
+ t
C7
C70 = 1
q .
+ t
(I + Ay )T
(2)
(I + Az )T
(3)
(41c)
n+1
=T
(1)
=T
(2)
=T
(3)
(43a)
(43b)
(43c)
+T ,
(43d)
where,
(41b)
0
n+1
n+1
1 2C50 2C60 2C70 Ti,n+1
j,k +C5 Ti+1, j,k + Ti1, j,k
n+1
0
n+1
n+1
n
+C60 Ti,n+1
+
T
+C
T
+
T
7
j+1,k
i, j1,k
i, j,k+1
i, j,k1 = S , (38)
C50 =
(41a)
n+1
n+1
0
(42a)
T
+
T
+C
I + Ax = 1 2C50 Ti,n+1
5
i+1, j,k
i1, j,k
j,k
0
n+1
n+1
I + Ay = 1 2C60 Ti,n+1
(42b)
j,k +C6 Ti, j+1,k + Ti, j1,k
0
n+1
n+1
(42c)
I + Az = 1 2C70 Ti,n+1
j,k +C7 Ti, j,k+1 + Ti, j,k1
(36f)
(I + Ax )T (1) = Sn (I + A)T n
Fn
q ,
1
+ t
C6
C60 = 1
q ,
+ t
(40c)
where
to obtain
Sn =
(40b)
(40a)
(39a)
(39b)
(39c)
6
c 2005 by ASME
Copyright
311
311
300
300
DPL Model
DPL Model
311
311
300
Hyperbolic Model
300
Hyperbolic Model
311
311
300
Parabolic Model
300
Parabolic Model
results of Brorson et al. [22] and Qiu and Tien [4, 5] corresponding to the front surface transient response for a 0.1m thick gold
film. The laser heat source term given by Eq. (28) is used for
this purpose. The thermal properties ( = 1.2 104m2 s1 , k =
315W m1 K 1 , T = 90ps, q = 8.5ps) are assumed to be constant. The temperature change is normalized by the maximum
value that occurs during the short-time transient. The results
from the present numerical scheme compare well with experimental and analytical results. The CV wave and the parabolic
models neglect the microstructural interaction effect in the shorttime transient, rendering an overestimated temperature in the
transient response as seen in the figure.
Figures 36 show the comparison of transient temperature
distribution caused by a pulsating laser beam of 200nm diameter heating the top surface of the gold film (500nm long and
wide and 100nm thick) at various locations of the film every
0.3ps, predicted by DPL, hyperbolic and parabolic heat conduction models. The energy absorption rate given by Eq. (31)
is used to model three-dimensional laser heating. As explained
c 2005 by ASME
Copyright
311
309
311
300
300
300
DPL Model
DPL Model
309
311
311
300
300
300
Hyperbolic Model
Hyperbolic Model
311
309
311
300
300
300
Parabolic Model
Parabolic Model
formed to test the convergence of the numerical solution. The numerical technique was then extended to three dimensional geometry, and a numerical procedure for computing the transient temperature distribution during short pulse laser heating of thin films
has been presented. The discretized 3-D microscale DPL equation has been solved using DouglasGunn time-splitting method
and -form DouglasGunn time-splitting method. The present
numerical scheme, employing the -form Douglas-Gunn timesplitting outperforms all the numerical techniques known to us
in terms of computational time taken to complete the simulation.
REFERENCES
[1] Vernotte, P., 1958, Les panadoxes de la theorie continue de
lequation de la chaleur, C.r.acad.Sci.Paris, 246, pp.31543155.
[2] Cattaneo, M.C., 1958, Sur une forme de lequation de
la chaleur eliminant le paradox dune propagation instantanee, Comptes Rendus Hebd. Seances Acad. Sci., 247,
pp.431-433.
c 2005 by ASME
Copyright
Numerical techniques
N=21
N=41
N=51
N=101
Explicit Scheme
4.88
147.62
450.26
7920.00
GaussSeidel
14.14
253.42
627.03
11343.06
Conjugate Gradient
12.33
124.83
270.3
3614.69
D-G time-splitting
9.24
82.44
165.76
1506.38
-form D-G
8.54
70.5
140.92
1344.4
Table 1.
[3] Y. Taitel, On the Parabolic, hyperbolic and discrete formulation of the heat conduction equation, International Journal of Heat and Mass Transfer, 15, 369 (1972).
[4] T. Q. Qiu, C. L. Tien, Short-pulse laser heating on metals,
International Journal of Heat and Mass Transfer, 35, 719
(1992).
[5] T. Q. Qiu, C. L. Tien, Heat transfer mechanisms during
short-pulse laser heating of metals, ASME Journal of Heat
Transfer, 115, 835 (1993).
[6] S. I. Anisimov, B. L. Kapeliovich, and T. L. Perelman,
Electron emission from metal surfaces exposed to ultrashort
laser pulses, Sov. Phys. JETP, 39, 375 (1974).
[7] D. Y. Tzou, Macro-to-Microscale Heat Transfer: The
Lagging Behavior, Taylor and Francis, Washington, DC
(1996).
[8] D. Y. Tzou, A unified approach for heat conduction from
macro-to micro scales, ASME Journal of Heat Transfer,
117, 8 (1995).
[9] D. Y. Tzou, The generalized lagging response in smallscale and high-rate heating, International Journal of Heat
and Mass Transfer, 38, 3231 ( 1995).
[10] D. Y. Tzou, Experimental support for lagging behavior in
heat propagation, J. Thermophysics and Heat transfer, 9,
686 (1995).
[11] Guyer, R. A., and Krumhansl, J. a., 1966, Solution of the
linearized Boltzmann equation, Physical Review, 148, pp.
766-778.
[12] W. Dai, and R. Nassar, A finite difference method for solving the heat transport equation at the microscale, Numer.
Methods Partial Differential Equations, 15, 698 (1999).
[13] W. Dai, and R. Nassar, A compact finite difference scheme
for solving a one-dimensional heat transport equation at the
micro-scale, Journal of Computational and Applied Mathematics, 132, 431 (2001).
[14] W. Dai, and R. Nassar, A finite difference scheme for solving a three dimensional heat transport equation in a thin
film with micro-scale thickness, International Journal for
9
c 2005 by ASME
Copyright