Heat transport at the microscale is important for the processing of materials with a pulsed
laser. In this study, we develop a three-level finite-difference scheme for solving micro heat
transport equations with temperature-dependent thermal properties obtained based on the
parabolic two-step model. It is shown by the discrete energy method that for constant
thermal properties the scheme is unconditionally stable. Numerical results for thermal
analysis of a gold film are obtained.
1. INTRODUCTION
Heat transfer at the microscale is important for the processing of materials with
a pulsed laser [14]. Developments of high-power short-pulse lasers have matured
into several innovative technologies, including structural monitoring of thin metal
lms, laser micromachining and patterning, structural tailoring of microlms, and
laser synthesis and processing in thin-lm deposition [5]. During ultrafast heating on
metallic lms, surface temperature change is converted from the reectivity change
measured optically on the lm surface. Due to the much smaller heat capacity of the
electron gas as compared to that of the metal lattice, the temperature rise during the
rst few picoseconds is caused mainly by the hot electron gas rather than the metal
lattice. Short-pulse laser interactions with thin metal lms has been widely investi-
gated. Major focus in modeling the ultrafast process of heat transport has been
placed on the electron gas, whose temperature becomes physically meaningful after
the rst few hundred femtoseconds [118]. The physical phenomena involved in
ultrashort laser pulse interactions with solid targets are very complex and are still the
subject of many investigations. Kaganov and colleagues [6] rst evaluated theore-
tically the heating process between electrons and lattice. Anisimov and colleagues [7]
proposed a parabolic two-step heat conduction model, which was advanced by
Fujimoto and colleagues [8]. Qiu and Tien [2] derived a more general and rigorous
509
510 W. DAI AND R. NASSAR
NOMENCLATURE
A constant coecient d2x second-order nite dierence
C heat capacity Hx ; Hx rst-order forward, backward nitedif-
Ce electron heat capacity ferences
Cl lattice heat capacity DTe electron temperature change
d radiation penetration depth Dt; Dx time increment, grid size
G electronphonon coupling factor tq constant coecient
J laser uence tT constant coecient
ke electron thermal conductivity
L lm thickness
Subscripts
k k ke
e electron
N number of grid points
i spatial node
Q heat source
l lattice
R reectivity
0 initial
t time
tp laser pulse duration
T; T1 temperature Superscript
x; y; z Cartesian coordinates n time level
2. FINITE-DIFFERENCE SCHEME
The parabolic two-step heat conduction equations with temperature-dependent
thermal properties can be written as follows [17, 18]:
qTe q qTe
Ce Te ke Te ; Tl GTe Tl Q 1
qt qx qx
qTl
Cl GTe Tl 2
qt
The electronphonon coupling factor G is a material property that represents the
rate of energy transfer between the electrons and the lattices. The heat capacity of the
electrons and lattice, Ce and Cl , and the thermal conductivity of the electron, ke , are
also material properties. While all these material properties have some temperature
dependence, the electron heat capacity is a strong function of the electron tem-
perature:
Ce Ae Te 3
Values of Ae are given by Kittel [9]. The electronphonon coupling factor and the
lattice heat capacity are assumed to be constant. The temperature dependence of the
thermal conductivity during nonequilibrium heating can be approximated as
Te
ke Te ; Tl ke T0 4
Tl
The reectivity R and the penetration depth d are material properties, while the
uence J and the pulse width tp are parameters of the incident laser pulse. This
source term neglects any interaction with the substrate. Therefore, the lm must be
thicker than the radiation penetration depth to ensure that the radiant energy does
not reach the substrate [18].
From Eq. (2), we obtain
Cl qTl
Te Tl 6
G qt
Substituting Eq. (6) into Eq. (1) gives
qTl Ce Cl q2 Tl q qTl Cl q qTl
Ce Cl ke ke Q 7
qt G qt2 qx qx G qx qx
512 W. DAI AND R. NASSAR
Te x; 0 Tl x; 0 T0 9
Hx Tin Ti1
n
Tin HxTin Tin Tni1
and so on. It can be seen that d2x Tin Hx HxTin : As such, a three-level nite-dif-
ference method for solving the above problem [Eqs. (6)(10)] can be written as
follows:
Tl n1 Tl n1 1 Ce Cl n
Ce Cl ni i i
2 Tl n1
i 2Tl ni Tl n1
i
2 Dt Dt G i
1
H k n
H
x i1=2 x Tl n1
i 2Tl ni Tl n1
i
4 Dx2
1 Cl
Hx kni1=2 Hx Tl n1
i Tl n1
i Qni 1 i N 11
2 Dt Dx2 G
Cl Tl n1 Tl ni
Te n1
i i
Tl n1
i 1iN 12
G Dt
Tl 0i Tl 1i T0 i Te 0i Te 1i T0 i 0iN1 13
Tl n0 Tl n1 Tl nN1 Tl nN 14a
TEMPERATURE-DEPENDENT THERMAL PROPERTIES 513
Figure 1. Normalized electron temperatures at the front surface of a gold lm irradiated with a 0.1-ps laser
pulse at a uence of 13.4 J=m2 and three dierent meshes.
Te n0 Te n1 Te nN1 Te nN 14b
Thus, one may solve Tl ni from Eq. (11) and then obtain Te ni from Eq. (12). It
can be seen that the truncation error of Eq. (11) at i Dx; n Dt is ODt2 Dx2 .
Hence, it is more accurate than those schemes [5, 18] obtained based on the Crank-
Nicholson method, for which the truncation error is ODt Dx2 . Here, it should be
pointed out that we use a weighted average Tl n1
i 2Tl ni Tl n1
i =4 in Eq. (11)
for stability, which is similar to the idea of the CrankNicholson method. The
advantage of the weighted average can be seen in Lemma 2, next section. Further-
more, Eq. (11) is a linearized three-level nite-dierence scheme. The resulting sys-
tem is a tridiagonal linear system and hence can be easily solved.
3. STABILITY
Since Eq. (11) is a nonlinear nite-dierence scheme, the analysis for stability is
rather complicated. Here, for simplicity we show its stability only when the coe-
cients are constants. In this case, Eq. (11) becomes
514 W. DAI AND R. NASSAR
Figure 2. Normalized electron temperatures at the front surface of a gold lm irradiated with a 0.1-ps laser
pulse at a uence of 13.4 J=m2 and three dierent time increments.
Tl n1 Tl n1 tT
C i i
2 Tl n1
i 2Tl ni Tl n1
i
2 Dt Dt
k 1 tq 2
d2 Tl n1 2Tl ni Tl n1 d Tl n1 Tl n1 Qni 15
4 Dx2 x i i
2 Dt Dx2 x i i
X
N
un ; vn Dx uni vni kun k2 un ; un
i1
TEMPERATURE-DEPENDENT THERMAL PROPERTIES 515
X
N1
kHxun k21 Hxun ; Hxun 1 Dx Hxuni 2
i1
un1
i 2uni un1
i un1
i un1
i un1
i uni 2 uni un1
i 2 16
Proof:
un1
i 2uni un1
i un1
i un1
i
un1
i 2 2un1
i uni un1
i un1
i un1
i un1
i 2uni un1
i un1
i 2
un1
i 2 2un1
i uni uni 2 funi 2 2uni un1
i un1
i 2 g
un1
i uni 2 uni un1
i 2
Figure 3. Calculated electron temperature proles for a 100-nm gold lm irradiated with a 0.1-ps laser
pulse of uence of 10.0 J=m2 .
516 W. DAI AND R. NASSAR
Figure 4. Calculated lattice temperature proles for a 100-nm gold lm irradiated with a 0.1-ps laser pulse
of uence of 10.0 J=m2 .
Hxun1
i Hxuni 2 Hxuni Hxun1
i 2 17
Proof:
Hxun1
i 2uni un1
i Hxun1
i un1
i
Hxun1
i 2 2Hxun1
i Hxuni Hxun1
i Hxun1
i
Hxun1
i Hxun1
i 2Hxuni Hxun1
i Hxun1
i 2
Hxun1
i 2 2Hxun1
i Hxuni Hxuni 2
Hxun1
i Hxuni 2 Hxuni Hxun1
i 2
TEMPERATURE-DEPENDENT THERMAL PROPERTIES 517
Figure 5. Calculated electron temperature proles for a 100-nm gold lm irradiated with a 0.1-ps laser
pulse of uence of 500 J=m2 .
X
N
Dx d2x Tl n1
i 2Tl ni Tl n1
i Tl n1
i Tl n1
i
i1
X
N1
Dx HxTl n1
i 2Tl ni Tl n1
i Hx Tl n1
i Tl n1
i 18
i1
and
X
N
Dx d2x Tl n1
i Tl n1
i Tl n1
i Tl n1
i
i1
2
HxTl n1 Tl n1 19
1
518 W. DAI AND R. NASSAR
Figure 6. Calculated lattice temperature proles for a 100-nm gold lm irradiated with a 0.1-ps laser pulse
of uence of 500 J=m2 .
X
N1 X
N
Dx ui ui1 vi1 Dx ui ui1 vi
i2 i1
X
N1 X
N1
Dx ui ui1 vi1 Dx ui ui1 vi because u0 u1 ; uN uN1
i1 i1
X
N 1
Dx Hxui Hxvi 20
i1
Figure 7. Calculated electron temperature proles for a 500-nm gold lm irradiated with a 0.1-ps laser
pulse of uence of 104 J=m2 .
Theorem 1. Suppose that Tin and Sni are solutions of the scheme, Eqs. (15),
(13), and (14a), with different initial values, and different source terms,
respectively. Let eni Tin Sni : Then for any n in 0 n Dt t0 , eni satisfies
where 2 1 2
Fn tT en1 en krx Hx en1 en 1 22
2
and ex is the dierence of corresponding source terms. Here, rx Dt=Dx2 .
Hence, this scheme is unconditionally stable with respect to the initial condi-
tions and the source term.
Proof: It can be seen from Eq. (15) that eni satises
en1 en1 tT
C i i
2 en1 2eni en1
2 Dt Dt i i
k 1 tq 2 n1
d2 en1 2 eni en1 d e en1 eni 23
4 Dx2 x i i
2Dt Dx2 x i i
520 W. DAI AND R. NASSAR
Figure 8. Calculated lattice temperature proles for a 500-nm gold lm irradiated with a 0.1-ps laser pulse
of uence of 104 J=m2 .
en1
i en1
i
0 24
By Lemmas 13, we obtain
2 2tT n1 2 2
Cen1 en1 e en en en1
Dt
Dt 2 2
k 2
Hxen1 en 1 Hxen en1 1
2 Dx
TEMPERATURE-DEPENDENT THERMAL PROPERTIES 521
tq
Hx en1 en1 2
1
Dx2
2 Dten; en1 en1 25
Substituting
the above inequality into Eq. (25) to eliminate the term
2
Cen1 en1 , further dropping out the last term on the left-hand side multiplying
both sides by Dt, and using the notation Fn, we can simplify Eq. (25) as follows:
Fn Fn 1 Dt3 kenk2
X
n
F0 Dt3 kexk2
x1
4. NUMERICAL EXAMPLE
To demonstrate the applicability of the numerical method, we investigate the
temperature rise in a thin gold lm. Thermal properties are chosen to be k0 315
W=m K, Cl 2:5 106 J=m3 K, G 2:6 1016 W=m3 K, and A 70 J=m3 K2 . We
further chose tp 100 fs, d 15:3 nm, and R 0:93 for the laser source term Q in
Eq. (5).
In calculation, we rst chose three dierent meshes of 100, 200, and 400 grid
points with the time increment Dt 0:005 ps for the nite-dierence scheme to test for
stability. Figure 1 shows the change in temperature (DTe =DTe Max on the surface
of the gold layer when L 100 nm and L 20 nm with a uence of J 13:4 J=m2 :
Compared with that displayed in [17] (see Figure 3 in [17]), they agree very well with
those obtained based on the parabolic two-step model. Also, it can be seen from Figure
1 that mesh size had no signicant eect on the solution. This implies that the scheme is
grid-independent. For these three meshes, Dx 1:0 106 mm, 0:5 106 mm, and
0:25 106 mm were chosen for L 100 nm, respectively. This implies that the mesh
ratios, Dt=Dx, are 5.0 103 , 1.0 104 , and 2.0 104 , respectively. We then chose three
dierent time increments, 0.005, 0.01, and 0.02 ps, and Dx 1:0 106 mm. Figure 2
shows the change in temperature (DTe =DTe Max) on the surface of the gold layer
when L 100 nm with a uence of J 13:4 J=m2 . It can be seen from Figure 2 that
there is no signicant dierence in the solution. Figures 3 and 4 plot the electron
temperature proles and lattice temperature proles for a 100-nm gold lm irradiated
with a 0.1-ps laser pulse at a uence of J 10:0 J=m2 ; respectively. Both of these
proles agree well with those obtained by the parabolic two-step model (see Figures 1
and 2 in [17]). Figures 5 and 6 plot the electron temperature proles and lattice
522 W. DAI AND R. NASSAR
temperature proles for a 100-nm gold lm irradiated with a 0.1-ps laser pulse at
a uence of J 500:0 J=m2 ; respectively. Again, both of these proles agree well
with those obtained by the parabolic two-step model (see Figure 5 in [3]). Figures
7 and 8 plot the electron temperature proles and lattice temperature proles for
a 500-nm gold lm irradiated with a 0.1-ps laser pulse at a uence of
J 104 J=m2 ; respectively.
5. CONCLUSION
In this study, we develop a three-level nite-dierence scheme for solving micro
heat transport equations with temperature-dependent thermal properties, which are
obtained based on the parabolic two-step model. It is shown by the discrete energy
method that for constant thermal properties this scheme is unconditionally stable
with respect to the initial conditions and the source term. However, the stability of
the scheme with temperature-dependent thermal properties needs further study.
Numerical results for thermal analysis of a gold lm show that the scheme is ecient.
REFERENCES
1. T. Q. Qiu and C. L. Tien, Short-Pulse Laser Heating on Metals, Int. J. Heat Mass
Transfer, vol. 35, pp. 719726, 1992.
2. T. Q. Qiu and C. L. Tien, Heat Transfer Mechanisms during Short-pulse Laser Heating
of Metals, ASME J. Heat Transfer, vol. 115, pp. 835841, 1993.
3. T. Q. Qiu and C. L. Tien, Femtosecond Laser Heating of Multilayered MetalsI.
Analysis, Int. J. Heat Mass Transfer, vol. 37, pp. 27892797, 1994.
4. T. Q. Qiu, T. Juhasz, C. Suarez, W. E. Bron, and C. L. Tien, Femtosecond Laser Heating
of Multilayered MetalsII. Experiments, Int. J. Heat Mass Transfer, vol. 37, pp. 2799
2808, 1994.
5. D. Y. Tzou and K. S. Chiu, Temperature-Dependent Thermal Lagging in Ultrafast Laser
Heating, Int. J. Heat Mass Transfer, vol. 44, pp. 17251734, 2001.
6. M. I. Kaganov, I. M. Lifshitz, and L. V. Tanatarov, Relaxation between Electrons and
Crystalline Lattices, Sov. Phys. JETP, vol. 4, pp. 173178, 1957.
7. S. I. Anisimov, B. L. Kapeliovich, and T. L. Perelman, Electron Emission from Metal
Surfaces Exposed to Ultra-short Laser Pulses, Sov. Phys. JETP, vol. 39, pp. 375377,
1974.
8. J. G. Fujimoto, J. M. Liu, and E. P. Ippen, Demtosecond Laser Interaction with Metallic
Tungsten and Non-equilibrium Electron and Lattices Temperature in Thin Gold Film,
Phys. Rev. Lett., vol. 53, pp. 18371840, 1984.
9. D. Y. Tzou, A Unied Filed Approach for Heat Condition from Macro- to Micro-Scales,
ASME J. Heat Transfer, vol. 117, pp. 816, 1995.
10. D. Y. Tzou, Macro to Micro Heat Transfer, pp. 2564 and 111146, Taylor & Francis,
Washington, DC, 1997.
11. R. F. W. Herrmann, J. Gerlach, and E. E. B. Campbell, Ultrashort Pulse Laser Ablation
of Silicon: An MD Simulation Study, Appl. Phys. A, vol. 66, pp. 3542, 1998.
12. R. Rethfeld, A. Kaiser, M. Vicanek, and G. Simon, Femtosecond Laser-Induced Heating
of Electron Gas in Aluminum, Appl. Phys. A, vol. 69, pp. 109112, 1999.
13. J. K. Chen, J. E. Beraun, and T. C. Carney, A Corrective Smoothed Particle Method for
Boundary Value Problems in Heat Conduction, Int. J. Numer. Meth. Eng., vol. 46, pp.
231252, 1999.
TEMPERATURE-DEPENDENT THERMAL PROPERTIES 523