Professional Documents
Culture Documents
net/publication/225881622
CITATIONS READS
16 2,345
1 author:
Andrey Petrin
Joint Institute for High Temperatures
97 PUBLICATIONS 376 CITATIONS
SEE PROFILE
All content following this page was uploaded by Andrey Petrin on 01 June 2014.
ELECTRONIC PROPERTIES
OF SOLID
Abstract—We suggest a general approach to considering the thermionic, field, and thermionic field emis
sions of electrons from metals. For this purpose, based on the standard model of free electrons in a metal, we
suggest a numerical method for determining the transmission coefficient through the potential barrier at the
metal–vacuum interface suitable for an arbitrary barrier. This method is free both from the approximations
based on the saddlepoint method and characteristic of the analytical models for thermionic emission and
from the approximations for the tunneling coefficient through the potential barrier characteristic of the mod
els for field emission. Based on numerical simulations, we determine the thermal effect of the emission and
ascertain that a very sharp transition from surface cooling by electron emission to heating occurs at certain
electric field and temperature. We explain the triggering mechanism of the explosive electron emission
observed during micropoint explosions by this phenomenon. The explosive emission is shown to begin when
the level of the potential barrier at the micropoint tip drops below the Fermi level in the metal.
PACS numbers: 79.40.+z, 79.70.+q, 73.40.Gk, 73.30.+y
DOI: 10.1134/S1063776109080184
314
THERMIONIC FIELD EMISSION OF ELECTRONS FROM METALS 315
E e = ( p x + p y + p z )/2m,
2 2 2 The density of states per unit volume of a metal is
known to be Z(px, py, pz) = 2/h3, where h is the Planck
where px, py, and pz are the electron momentum com constant. Since the electrons in a metal obey the
ponents along the coordinate axes and m is the elec Fermi–Dirac distribution, the number of electrons
whose momenta fall within the intervals [px, px + dpx], The surface power density of the heat being
[py, py + dpy], and [pz, pz + dpz] is released on the metal surface during the emission of
electrons is
2 dp dp dp
N ( p x, p y, p z )dp x dp y dp z = x y z
3 ∞
h EF – Ex
Q s = 4πemkT
⎧ 2 2
px + py + pz
2
⎫
–1
h
3 ∫
D ( E x )
e
× ⎨ 1 + exp – E F⎞ ⎬ .
1 ⎛ 0
kT ⎝ 2m ⎠
⎩ ⎭
E x – E F⎞
× ln 1 + exp ⎛ –
dE x
Integrating this expression over py and pz yields the ⎝ kT ⎠
number of electrons with momenta in the interval [px, ∞ (3)
px + dpx]: 4πmkT
∞ ∞ h
3 ∫
D ( Ex ) ( EF – Ex )
=
0
2
N ( p x )dp x = 3 dp x
∫∫ dp y dp z
E x – E F⎞
× ln 1 + exp ⎛ –
h –∞ –∞ dE x .
–1
⎝ kT ⎠
⎧ 2
1 px + py + pz
2 2
⎫
× ⎨ 1 + exp ⎛
– E F⎞ ⎬ .
⎩ kT ⎝ 2m ⎠
⎭
Qs is negative when the surface is cooled by the elec
tron emission and positive when the surface is heated.
Passing to the polar coordinates py = p⊥ cosγ and If W Ⰷ kT in Eq. (2), then the unity in the denom
pz = p⊥ sinγ, we then obtain inator of the integrand may be neglected and the inte
∞ gral will be simplified greatly. Integration yields
2 dp 2πp dp
N ( 0p x )dp x =
3 x ⊥ ∫ ⊥ 2
exp ⎛ F⎞ exp ⎛ –
E px ⎞
h N ( p x )dp x = 4πmkT
dp x .
0
h
3 ⎝ kT ⎠ ⎝ 2mkT⎠
–1
⎧ 2
1 px + p⊥
2
⎫
× ⎨ 1 + exp ⎛
– E F⎞ ⎬ If we assume that D(px) = D does not depend on px
⎩ kT ⎝ 2m ⎠
⎭ (which, in general, is wrong but is quite an acceptable
approximation), then we will obtain the above Rich
∞ 2 ardson–Dushman formula for the current density of
dp x d⎛
4πmkT p⊥ ⎞
=
h
3 ∫ ⎝ 2mkT⎠
the thermionic emission instead of Eq. (2):
0
2
exp ⎛ – ⎞ = ADT exp ⎛ – ⎞ ,
4πemk 2 W 2 W
J = DT
–1 3 ⎝ kT ⎠ ⎝ kT ⎠
⎧ 2 2
⎫ h
× ⎨ 1 + exp ⎛
p⊥ ⎞
exp ⎛ x – E F⎞ ⎬
1 p
⎝ 2mkT⎠ kT ⎝ 2m ⎠
⎩ ⎭ where A = 4πemk2/h3 = 1.2 × 106 A m–2 K2.
2
= 4πmkT
1 ⎛ p x – ⎞ ⎞
ln ⎛ 1 + exp – E F dp x .
h
3 ⎝ kT ⎝ 2m ⎠ ⎠ 3. THE PASSAGE OF AN ELECTRON
THROUGH A POTENTIAL BARRIER
Substituting this expression into Eq. (1) yields the cur Consider the passage of electrons through a Schot
rent density
tky barrier (see Fig. 1). This barrier is described by the
J = 4πemkT
3
potential energy
h
(2) ⎧ W + E F – e 2 /16πε 0 x – eE ext x, x > 0,
∞
E x – E F⎞ U(x) = ⎨ (4)
× D ( E x ) ln 1 + exp ⎛ – x ≤ xl ,
∫ ⎝ kT ⎠
dE z , ⎩ 0,
0
where Eext is the strength of the external electric field
2
where Ex = p x /2m
is the energy of the directed elec and xl is the smaller root of the equation W + EF –
tron motion along the x axis. e2/16πε0x – eEextx = 0.
ψ N = d exp ⎛
ip x x N⎞
ψ N + 1 = d exp ⎛
The points xn are distributed uniformly between a and ip x x N + 1⎞
, ,
b with the step Δx = (b – a)/(N + 1). ⎝ ប ⎠ ⎝ ប ⎠
We will take a finitedifference approximation of
the derivatives at the points xn in the form with px = p0 for the barrier (5).
Fig. 3. Dependence of ln(J/E2) on E–1 for the regime of Fig. 4. Dependence of ln(J/E2) on E–1 in a wide range of
field emission calculated by our method (curve 1, the electric field strengths for a cathode with a low work func
points are marked by crosses), by the numerical method tion, W = 1.05 eV. The temperature is T = 300 K. Curves
from [14] (curve 2, open circles), and using the analytical 1, 2, and 3 were obtained, respectively, by our numerical
Murphy–Good dependence [7, 14] (curve 3, filled cir method (solid curve), the numerical method from [14]
cles). The work function is W = 5 eV and the temperature (open circles), and the approximate numerical method for
is T = 300 K. ballistic emission from [14] (squares). Also shown are the
dependences 4 obtained from the Murphy–Good analyti
cal solution [7, 14] (triangles) and various emission
Eliminating the transmission coefficient, we will regimes: ballistic (I), field (II) (0.032 V/Å < E <
0.062 V/Å), thermionic field (III), and Schottky (III) (E <
obtain a boundary condition at the right boundary: 0.018 V/Å) emissions.
ψ N + 1 = ψ N exp ( ip x Δx/ប ). (9)
As a result, we have N algebraic equations (7) with Note that the problem can also be solved numeri
the boundary conditions (8) and (9). The derived cally for the potential shown in Fig. 1. It is only needed
matrix equation to approximate well the boundary condition at the
right boundary. Thus, for example, it can be assumed
∑M ψ ij j = Bi (10) that
px = p ( xN ) = 2m [ E x – U ( x N ) ] ≈ p ( x N + 1 )
J
has a tridiagonal matrix Mij . In addition,
at the right boundary, with all of the remaining formu
2ip 0 Δx⎞
B 1 = – exp ⎛
ip 0 x 0⎞
1 – exp ⎛
, B 2 = 0, las being the same. Our numerical simulations show
⎝ ប ⎠ ⎝ ប ⎠ that the transmission coefficients change (in relative
terms) by less than 10–6).
B 3 = 0, …, B N – 1 = 0, B N = 0.
Solving Eq. (10), we will find ψN and, hence, d = 4. COMPARISON OF THE NUMERICAL
ψN exp(–ipxxN/ប). The transmission probability METHOD WITH KNOWN SOLUTIONS
through the barrier can be determined from d using the The numerical method was tested on the problem
formula of electron passage through a rectangular potential
barrier with a known analytical solution [18]. A rela
D ( E x ) = ( p x /p 0 ) d .
2
tive accuracy of the transmission coefficient ~10–5–
The width of the barrier (5) shown in Fig. 2 at the 10–4 turned out to be easily achievable.
base is equal, with a high accuracy, to bw = (W + To compare our numerical method with known
EF)/eEext. After the passage through this barrier, the approximate solutions, we investigated the regime of
energy of the transmitted electrons is exactly equal to field emission. The field emission of electrons is
the energy of the electrons incident on the barrier from known to be well described by the Fowler–Nordheim
the left and does not change during the subsequent formula [3] and its generalization, the Murphy–Good
motion to the right. Our calculations were performed formula [7]. According to these formulas, the depen
for such a region a ≤ x ≤ b that the potential barrier (5) dence of ln(J/E2) on E–1 is a straight line in the range
was entirely inside this region. For this purpose, we of electric fields characteristic of the field emission.
assumed, for example, that a = –bw/10 and b = Figure 3 shows this dependence calculated by our
11bw/10. method, by the numerical method from [14], and
using the analytical Murphy–Good dependence (the The first regime is the temperaturedependent
results of the calculations by the authors of [14]). We emission (E < 2 × 109 V m–1) at which the current
see that the results of our calculations by the numerical depends on the temperature. In this regime, the elec
method suggested above agree excellently with the trons over the barrier make a significant contribution.
Murphy–Good dependence. In the limit when E 0, the curves approach the
To test our numerical method, we also investigated corresponding values defined by the Richardson–
the regime of electron emission in a wide range of Dushman formula.
electric fields. Figure 4 shows the dependence of The second regime is the field emission. Here, the
ln(J/E2) on E–1 for ballistic, field, and thermionic electrons tunneling through the barrier make a major
emissions. We considered the case of a low work func contribution.
tion, W = 1.05 eV (at temperature T = 300 K), where,
The third regime is the ballistic emission of elec
according to [14], there exists a wide range of fields at
trons [14], which arises when the barrier height
which the analytical formulas do not give a satisfactory
becomes lower than the Fermi energy. The field corre
approximation. We see from Fig. 4 that the calcula
sponding to the transition to this regime can be calcu
tions by our method describe well the dependence
lated from the equation W = –ΔU. Hence, E =
under consideration in such a wide range of field
strengths and yield more accurate results than the 4πε0W 2/e3; for copper, this field is E ≈ 1.41 × 1010 V m–1.
numerical method from [14]. Finally, the fourth regime is the saturation of the
Thus, we conclude that the emission characteris electron emission, which arises when the barrier
tics are described accurately using our method and height becomes lower than the bottom of the conduc
that the results of our calculations agree well with the tion band. The field corresponding to the transition to
analytical dependences in the ranges of their applica this regime can be calculated from the equation W +
bility. EF = –ΔE. Hence, E = 4πε0(W + EF)2/e3; for copper, this
field is E ≈ 9.2 × 1010 V m–1. The saturation current den
sity for copper is Jsat ≈ 4 × 1015 A m–2 = 4 × 1011 A cm–2.
5. RESULTS OF OUR CALCULATIONS Figure 6 shows the dependence of the modulus of
AND THEIR DISCUSSION the surface thermal power density |Qs(E)| released on a
Figure 5 shows the calculated dependence of the copper surface (more precisely, inside the metal near
emission current density on the electric field strength its surface) during the emission of electrons calculated
on a copper surface (W = 4.5 eV, EF = 7.0 eV) for three from Eq. (3) on the electric field strength. As an exam
surface temperatures. Four emission regimes can be ple, the |Qs(E)| curves corresponding to surface tem
arbitrarily identified for the range of electric field peratures of 1000 and 1300 K are shown. Here, four
strengths under consideration. emission regimes are also distinguishable. The satura
8. W. B. Nottingham, Phys. Rev. 59, 907 (1941). 16. N. D. Lang and A. R. Williams, Phys. Rev. B: Condens.
9. F. M. Charbonnier, R. W. Strayer, L. W. Swanson, and Matter 18, 616 (1978).
E. E. Martin, Phys. Rev. Lett. 13, 397 (1964). 17. A. B. Petrin, IEEE Trans. Plasma Sci. 28, 1763 (2000).
10. W. Schottky, [!]Phys. [!]Z. 15, 872 (1914).
11. L. W. Nordheim, Proc. R. Soc. London, Ser. A 121, 626 18. Z. Flügge, Practical Quantum Mechanics (Springer,
(1928). Heidelberg, 1971; LKI, Moscow, 2008), Vol. 1.
12. K. L. Jensen and M. Cahay, Appl. Phys. Lett. 88, 154 19. G. A. Mesyats and D. I. [!]Proskurovskii, Pis’ma Zh.
105 (2006). Éksp. Teor. Fiz. 13 [!](1), 7 (1971) [JETP Lett. 13 (1),
13. K. L. Jensen, J. Appl. Phys. 102, 024911 (2007). 4 (1971)].
14. V. Semet, Ch. Adessi, T. Capron, R. Mouton, and Vu 20. G. A. Mesyats, Pis’ma Zh. Éksp. Teor. Fiz. 57 [!](2), 88
Thien Binh, Phys. Rev. B: Condens. Matter 75, 045430 (1993) [JETP Lett. 57 (2), 95 (1993)].
(2007).
15. B. Lippmann and J. Schwinger, Phys. Rev. 79, 469
(1950). Translated by V. Astakhov