Adiabatic Joule Heating of Copper from 4 K to the Melting Temperature.

Alain Guillet1,a) and Fabrice Delamarre2

1
Groupe de Physique des Matériaux, U.M.R. CNRS 6634, Université and INSA of Rouen,
Av. de l’Université, St Etienne du Rouvray, 76801 France.
2
Département de Mécanique, INSA de Rouen, Av. de l’Université, St Etienne du Rouvray,
76801, France.1

Abstract :
Considering a copper wire heated by Joule effect and the variation of its resistivity
and specific heat with temperature, we established numerical and analytical solutions
(between 293 and 1356 K for the latter) for the evolution of its temperature over time. The
Temperature vs. Time evolution follows a Lambertian function. The calculations are based
on the assumption of adiabatic heating and uniform current distribution within the wire. We
demonstrate that at very low temperature the heating rate is strongly dependent on copper
purity.
Key Words:
Copper, Joule heating, action integral, electrical resistivity, specific heat, action integral

When an electric current flows through a metallic conductor its temperature increases
due to Joule effect, the knowledge of which is important in a number of scientific and
technical fields. Consider, for examples, the copper matrix composites used for the winding
of coils for the production of very high magnetic fields1-4, the development of
electromagnetic gunnery5, the electrical explosion of wires6,7 and the thermal ageing of
solders in electronic components due to Joule heating8. Ohmic heating could also be an
energy saving solution for metals heat treatment and a new route to investigate the influence
of very high electrical heating rate on microstructures9-16. The temperature evolution of a
conductor during electrical heating is an old problem17. It requires to solve numerically a
differential equation which takes into account the energy losses due to convection,
conduction and possibly radiancy18-20. However, if the current density is high and the
annealing time short, adiabatic conditions are generally assumed, with the exception of
Zielinskia et al.21. Therefore, the heating rate and the temperature increase are numerically
                                                                                                               
a)
Author to whom correspondence should be addressed. Electronic address:
[email protected]    

  1  
 calculated by solving equation (1) which derives from an energy balance22,23, but the
analytical solution of equation (1) is lacking:
  t Tf
f
2
Dc p (T )
  " J dt  = "   dT .                   (1)  
0 T0 # (T )
 
Where, T is the temperature, t the time, J(t) the current density, D the material density,
! Cp(T) the specific heat at constant pressure and  ρ(T) the electrical resistivity. T0 and Tf are
the initial and final temperature, respectively.   Solving equation (1) either numerically or
analytically necessitates to take into account the temperature dependence of the materials
properties D, Cp and   ρ. Surprisingly, some authors13-16, consider D, Cp and   ρ as constants,
under these assumptions according to equation (1), the heating rate is given by (2):  
 
dT " 2
  = J (t) .                       (2)  
dt DCP
 
Then, following integration of (2) the temperature increase with time is given by
! equation (3):

t
" 2
T = T0 + #J (t')dt' , (3)
DCP 0

However, since Cp and ρ   increase significantly between 4 K and the melting

! temperature of the metal, this is a gross approximation, although the temperature dependence
of the density D, due to the thermal expansion, can be neglected. Indeed, according to the
authors own calculations, not detailed here, the density variation of copper between 4 and
1356 K is only 6.5% even if the temperature dependence of the volumetric coefficient of
thermal expansion is taken into account. We show below that Cp and ρ   variations are indeed
much greater than D variation. In this work we first calculate numerically the temperature
dependence of the heating rate during Joule heating of copper by taking into account the
temperature dependence of the materials properties  ρ  and CP. Then, we establish the heating
time required to reach the melting temperature starting from three different initial
temperatures : 4 K (boiling of liquid helium), 77K (boiling of liquid nitrogen) and 293K
(ambient temperature). Finally, equation (1) is first solved numerically and then analytically
for T0>293 K. A constant current density of J=109 A/m2 was chosen, typical, for example, of

  2  
 the load of a truck battery (I≈1000 A) applied to a 1 mm2 cross-section wire. The copper
properties required for the calculations are given in Table I.

TABLE I. Physical properties of copper.

Ma θDb,(24) Tmc ρ293Kc,(25) αd,(25) Cp(293K)e Df
(kg/mole) (K)   (K)   (10-8Ω.m)   (K-1)   (J/kg/K) (kg/m3)
0.0635 315 1356 1.724 0.00393 393 8920
a  is  the  atomic  mass,  b  is  the  Debye  temperature,  c  is  the  melting  temperature,    

d  is  the  temperature  coefficient  of  resistivity,  e  is  the  specific  heat  et  T=293  K,  

f  is  the  density  

 
For metals, both free electrons and phonons contribute to the specific heat, although
the latter is largely preponderant at high temperature27. The phonons contribution to the
specific heat at constant volume (in J/mole/K) is given by equation (4)27:
  3
# T & #"D /T x 4 e x &
CV (T ) = 9R% ( % * x 2
dx ( . (4)
$ " D ' $ 0 (e )1) '
 
There is no exact solution for this integral, which must be solved numerically.
! However, at high temperature the molar specific heat is close to CV≈3R (Dulong and Petit
Law). At low temperature (T/θD<<1) a T3 law is obeyed27,28, then equation (5) follows :  
 
# T &3
Cv (J / mole / K ) = 1450% ( ) 4.65.10*5 T 3 .
ph.
(5)
$" D '
 
Furthermore, there is also an electronic contribution to the specific heat26,27 which is
! only significant at low temperature (equation 6):

Cvel. (J / mole / K ) = 5.06 "10#4 T . (6)

However, solving (1) requires the knowledge of CP, the molar specific heat at constant
!
pressure. At low temperature the difference between Cp and CV is small28 but it cannot be
neglected a priori for these calculations. The relationship between CV and CP is given by
(7)26-28:
 
9# 2VT
CP " CV = . (7)
$

!   3  
 with :
α: linear thermal expansion coefficient
V: molar volume
T: Temperature
β: Bulk modulus
 
According to Newham26 the quantity A for copper (A=1.55x10-5 mole/J) being constant over
the temperature range 100-1200K, then equation (7) can be written as (8),:
 
2
CP − CV = ATCP . (8)

Solving (8) for Cp, there is only one positive solution (9):
€
 
1" 1" 4ATCv
Cp = . (9)
2AT
 
Finally, the total specific heat at constant pressure in J/kg/K is simply obtained by
! dividing (9) by the atomic mass M (in kg/mole) of copper (Table I). The evolution of Cp and
Cv from 4K to the melting temperature of copper appears in Fig. 1. Clearly, CP becomes
significantly different from CV only above 300K.

FIG. 1. Specific heat of copper at constant pressure (Cp) and at constant

volume (Cv).  

  4  
  
Furthermore, Fig. 1 reveals that between 293 K and the melting point, the specific
heat varies linearly with temperature and can be approximated by equation (10):
3R
CP =
M
[
[1+ β (T − 293)] = 393.1 1+1.9 ×10−4 (T − 293) . ] (10)

For copper, between 77 and 1356K, the resistivity varies linearly with temperature
€
and is classically25 written as (11), where ρ293K and  α  are given in Table 1:  
 
ρ (T ) = ρ 293K [1+ α (T − 293)] . (11)

Matula30 used the following variant (Eq. 12) of the Bloch-Grünensein equation to
€
calculate the phonons contribution to the resistivity between 4K and the melting temperature :

% BT # $ CT p ( # D $ CT
" ph. (T ) = A'1+ + D( D ) *+( ). (12)
& # D $ CT T ) T

With :
!  
4 x
z 5ez
"(x) = $ dz . (13)
x5 0 (e z #1)2
 
And, for copper30:
! A=1.809.10-8  Ω.m, B=-6.0.10-3, C=0.0456, D=-6.476.10-4, p=1.84.  
 
According to the Matthiassen rule, impurities and defects (atoms in solid solution,
dislocations, vacancies) contribute also to the resistivity through   ρi. Then, the total resistivity
can be written as (14):  
 
" (T ) = " ph. (T ) + "i. (14)
 
The defects contribution is independent of temperature and led to a plateau at low
!
temperature called Residual Resistivity. The resistivity-temperature curves between 4 and
1356 K copper with different purities are shown in Fig. 2 on a log-log scale which highlights
the low temperature region. These data were extracted from29,30. The phonon contribution

  5  
 was calculated with equations (12) and (13) to which a residual resistivity of  ρi=2.10-11Ω.m
was added29.  
 
 
 
 
 
 
 
 
 
 
 
 
FIG. 2. Resistivity of copper with different purities.
 
There is no significant difference between the experimental work of Jensen et al.30,
Matula calculations29, and, for T>293K, the temperature coefficient of resistivity given by
Delomel25. At low temperature (T<60 K) linearity is no more observed, rather it decreases
with a TN variation with 3<N<528. Then, resistivity varies according to equation (15):
 
" (T ) = KT N + " i . (15)
 
Then, between 4 and 60 K, best fit for the data of Fig. 2 is given by equation (16) :
!
 
1.724.10$8
" (T ) # 7.88.10$17 T 4 + . (16)
RRR

Where RRR is the Residual Resistivity Ratio, i.e. RRR=ρ293K/ρi.   According to the
! differential form of equation (1), if the current density J is kept constant, the temperature
dependence of the heating rate (HR) is given by (17):
 
dT J 2 " (T )
= . (17)
dt DC p (T )
 

!   6  
 The heating rate versus temperature curves are shown Fig. 3, revealing three
interesting features. First, it demonstrates that even if J is constant, the heating rate is not
constant over the duration of the current pulse when the temperature dependence of the
materials properties Cp and ρ  are taken into account. Second, the heating rate between 4 and
60 K, the amplitude of the HR variations is several orders of magnitudes and is strongly
dependant on the copper purity. Third, in this temperature range, the heating rate decreases to
a minimum and then increases up to the melting temperature.  
 
 
 
 
 
 
 
 
 
 
 

FIG. 3. Heating rate vs. temperature curves of copper during joule Heating.
(J=109A/m2).
 
Equation (17) can be written as (18) which highlights the respective contribution of
phonons and impurities to the heating rate:
 
dT J 2 # " ph. (T ) "i &
= % + (. (18)
dt D $ C P (T ) C P (T ) '
 
With  (19):  
!  
" dT % J 2 ( ph. (T )
$ ' = . (19)  
# dt & ph. D C P (T )

 
! and  (20):  

  7  
  
" dT % J 2 (i . (20)  
$ ' =
# dt &imp. D C P (T )

Fig. 4 reveals the influence of each contribution to the heating rate. Above 77K, the

! influence of impurities is clearly negligible.

 
 
 
 
 
 
 
 
 
 
 

FIG. 4. Heating rate vs. temperature curve of copper during adiabatic Joule
heating. This figure hightlights the respective contribution of phonons and
impurities (ρi=0.002  µΩ.cm) to the heating rate.  
 
The temperature increase during annealing is calculated by solving equation (21). If J
is independent of time it follows :

Tf t
DC p (T )
# " (T )
dT = # J 2 dt = J 2t . (21)
T0 0
 
The numerical solution of (21) is given in Fig. 5 with three initial temperatures: 4 K,
! 77 K and 293 K and J=109 A/m2.
 
 
 
 
 

  8  
  
 
 
 
 
 
 
 
 

FIG. 5. Temperature vs. time curves of copper during Joule heating.

considering three different initial temperatures (ρi=0.002µΩ.cm).  

On a semi-log scale, it is clear that if 77 K<T0<293 K, the temperature vs. time curves
are linear. Then, for practical purpose, the following fit can be used to calculate the
temperature evolution during Joule heating (22):
T ≈ T0 exp(kt) . (22)
The best fit for T0=77K and 293K is k=16.28 s-1 and 16.47 s-1, respectively.
 
€
We can search for an analytical solution of (21) for T>293K by considering the
temperature variation of Cp and   ρ. Introducing Cp(T) (eq. 10) and   ρ(T) (eq. 11) in the left
hand side of equation (21), one obtains (23) :  
 
T T
DC p (T ) 3RD[1+ $ (T % 293)] 3RD T [1+ $ (T % 293)]
# dT = # M" 1+ & (T % 293) dT = # 1+ & (T % 293) dT. (23)
293 " (T ) 293 293 [ ] M " 293 293[ ]
 
Setting x=T-293, we integrate (24):
! ΔT
1+ βx  βΔT α − β 
∫ 1+ αx dx = α
+  2  ln(1+ αΔT )
 α  . (24)
0
Inserting (24) in (23) and considering the right hand side of equation (21) we obtain (25):

T DC p (T )
€ 3RD
# dT = 2 [$%&T + ($ ' % )ln($&T +1)] = J 2t . (25)
293 " (T ) $ M" 293

Then setting (26):

!
  9  
 α 2 Mρ 293K 2
X= J t +1 , (26)
3Rβ
 
and  (27):  
€  
Y = αΔT +1 ,                       (27)  
 
and  (28):  
€
 
α−β
A=( ) ,                       (28)  
β
 
we have (29):  
€  
 X = [Y +
  A ln(Y   )] = ln(Y   A eY )                 (29)  
 
The  inverse  function  of  (29)  can  be  found  using  Mathematica  (30):  
!
" X%
$eA '
Y = AW $ ' , (30)
  $ A'
# &
 
Where   W(z)   is   the   Lambert   function   which   satisfies,   W(z)exp(W(z))=z,   this  
!
function  have  several  applications  in  physics31.  The  temperature-­‐time  equation  can  be  
written  as  (31):  
 
 
& " $ % ) 35& % ) -& " 2 M, 293 2 )& % )075
"#T +1 = ( +W 4( + exp/( J t +1+( +28 , (31)
' % * 56' " $ % * .' 3RD% *' " $ % *159

Finally, we have (32) :

!  
% " # $ ( 24% $ ( ,% " 2 M+ 293 2 (% $ (/64 1
T (K ) = 293+ ' W
* '3 exp
* .' J t +1*' *17 # . (32)
  & "$ ) 4
5& " # $ ) -& 3RD $ )& " # $ )048 "  
However, if the temperature variation of Cp is neglected (β=0 in eq. (10)), the
integration of (21) leads to a much simpler equation (33)  :  
!

  10  
  
1 - &#M$ 293K t ) 0
T (K ) " 293+ / exp(
# . ' 3RD
% J(t)2 dt + ,12 .             (33)  
0 * 1
 
If J is constant during the electrical pulse, (33) can be expressed as (34):
!
1 , %#M$ 293K 2 ( /
T (K ) " 293+ . exp J t * +11 . (34)
# - '& 3RD ) 0

The heating rate is therefore (35):

!
* $t '-
, d & # J(t)2 dt ) /
dT M" 293 , % 0 ( / *0M" 293 t -
dt
=
3RD , dt
exp
/ ,+ 3RD # J(t)2 dt / . (35)
0 .
, /
  + .

Then, knowing the temperature dependence of the materials properties and the current
! density, the temperature evolution during Joule heating can be calculated. These calculations
can be extended to others materials than copper with the exception of ferromagnetic metals
(Fe, Ni, Co) which display a strong discontinuity of the specific heat at the Curie
temperature28.

  11  
References  
 
1   P.H.  Frings  and  L.  Van  Bocksal,  Physica  B  211,  73  (2003).  
 
2   J.D.  Embury  and  K.  Hen,  Current  Opinion  in  Solid  State  and  Materials  Science  3,  304  
(1998).  
 
3   E.   Snoeck,   F.   Lecouturier,   L.   Thilly,   M.J.   Casanove,   H.   Rakoto,   G.   Coffe,   S.   Askénazy,   J.P.  
Peyrade,   C.   Roucau,   V.   Pantsyrny,   A.   Shikov   and   A.   Nikulin,   Scripta   Mat.   38,   1643  
(1998).  
 
4   A.   Lagutin,   K.   Rosseel,   F.   Herlach,   J.   Vanacken   and   Y.   Bruynseraede,   Meas.   Sci.   and  
Technol.  14,  2144  (2003).  
 
5   A.  E.  Zielinski,  S.  Niles  and  J.  D.  Powell,  J.  Appl.  Phys.  86,  3943  (1999).  
 
6   V.I.   Oreshkin,   S.A.   Barengolts   and   S.A.   Chaikovsky   in   Proceedings   of   the   28th  
International   Conference   on   Phenomena   in   Ionized   Gases,   Prague,   Czech   Republic,   July  
15-­‐20,   2007,   edited   by   J.   Schmidt,   M.   Simek,   S.   Pekarek   and   V.   Prukner   (Institute of
Plasma Physics,  Prague,  2008).  
 
7   V.S.  Sedoi  and  Y.F.  Ivanov,  Nanotechnology  19,  145710  (2008).  
 
8   B.Y.  Wu,  M.O.  Alam,  Y.C.  Chan,  J.  of  Elec.  Mat.  37,  469  (2008).  
 
9   J.  Campbell  and  H.  Conrad,  Scripta  Met.  Mat.  31,  69  (1994).  
   
10   H.  Conrad,  Mat.  Science  and  Eng.  A  287,  227  (2000).  

11   W.  Zhang,  M.L.  Sui,  Y.Z.  Zhou  and  D.X.  Li,  Micron  34,  189  (2003).  
 
12   Y.  Zhou,  S.  Xiao  and  J.  Guo,  Materials  Letters  58,  1948  (2004).  
 

  12  
13   Z.  Xu,  G.  Tang,  S.  Tian  and  J.  He.,  Mat.  Science  and  Eng.  A  424,  300  (2006).  
 
14   H.  Lina  ,Y.  Zhao,  B.  Zhao,  L.  Han,  J.  Ma,  Q.  Jiang,  ISIJ  International  48,  1647  (2008).  
 
15   Z.J.  Wang  and  H.  Song,  Trans.  Nonferrous  Met.  Soc.  China  19,  s409  (2009).  
 
16   E.I.  Samuel,  A.  Bhowmik  and  R.S.  Qin,  J.  Mater.  Res.  25,  1020  (2010).  
 
17   E.  Verdet,  Théorie  mécanique  de  la  chaleur,  Vol.  2,  (Masson,  Paris,  1872),  p.  197.    
 
18   H.S.  Carlslaw  and  J.C.  Jaeger,  Conduction  of  heat  in  solids,  (Oxford  Science  
Publications,  Clarendon  Press,  Oxford,  1990),  p.  149.    
 
19   M.  Nivoit,  J.-­‐L.  Profizi  and  D.  Paulmier,  Int.  J.  Heat  Mass  Transfer.  24,  707  (1981).  
 
20   V.T.  Morgan  and  N.  G.  Baron,  J.  Phys.  D  :  Appl.  Phys  19,  975  (1986).  
 
21   A.E.  Zielinskia,  S.  Niles  and  J.D.  Powell,  Journal  of  Applied  Physics  86,  3943  (1999).  
 
22   R.   Kratz   and   P.   Wyder,   Principles   of   pulsed   magnet   design,   (Springer,   Berlin,   2002),  
p.27.  
 
23   S.  Rembeczki,  M.  Sc.  thesis,  University  of  Debrecen,  Debrecen,  Hungary  (2003).  
 
24   N.  W.  Ashcroft  and  N.  D.  Mermin,  Physique  du  Solide,  (Editions  de  Physique,  Les  Ulis,  
2002),  p.537.  
 
25   J.-­‐C.  Delomel,  Techniques  de  l’Ingénieur,  vol.  K7102v2  (2009).  
 
26   R.E.  Newham,  Properties  of  Materials  Anisotropy,  Symmetry,  Structure,  (Oxford  
University  Press,  Oxford  2005)  p.  43.  
 
27   M.   Gerl   and   J.-­‐P.   Issi,   Physique   des   Matériaux   vol.   8,   (Presses   Polytechniques   et  

  13  
 Universitaires  romandes,  1987).  
 
28   R.E.  Smallman  and  R.J.  Bishop,  Modern  Physical  Metallurgy  and  Materials  
Engineering  »,  (Butterworth-­‐Heinemann,  1999).  
 
29   R.A.  Matula,  J.  Phys.  Chem.  Ref.  Data  8,  1147  (1979).  
 
30   J.E.  Jensen  .  W.A.  Tuttle,  R.B.  Stewart,  H.  Brechna  and  A.G.  Prodell,  Selected  cryogenic  
data  notebook,  Vol.  II,  Sections  X-­‐XVIII  (Brookhaven  National  Laboratory,  Upton,  NY  
1980),  p.  X-­‐E-­‐5.  
 
31   R.M.  Corless,  G.H.  Gonnet,  D.E.G.  Hare,  D.J.  Jeffrey  and  D.E.  Knuth,    Advances  in  
Computational  Mathematics  5,  329  (1996).  
 

  14