616
22 Light-to-Electricity Conversion
Fig. 22.37 Conductivity
of a n-Si/SiO 2 -diode as a
function of temperature
(1/T ). The slope of the
dashed lines is (a)
∼ 0.56 eV (≈ E g /2) and
(b) ∼ 1.17 eV (≈ E g ).
Adapted from [1851]
10
-3
10
10
10
10
-4
-5
-6
-7
-1
1000/T (1/K)
2
.
3
4
.
3
0
.
3
8
.
2
6
.
2
4
.
2
2
.
2
0
.
2
Si
(a)
(b)
diffusion (∝ n
2
i ∝ exp(−E g /kT )). The latter process can be strongly suppressed by cooling the
device.
The gate voltage V G and the surface potential s are related to each other via
V G − V FB = V i + s =
e N A w
C i
+
eN A w
2
2 s
,
(22.55)
where w is the width of the depletion layer. w will be larger than w m in thermodynamic equilibrium.
The first term in the sum is |Q s |/C i and the second is obtained by integrating the Poisson equation for
the constant charge density −eN A across the depletion layer. The elimination of w yields
V G − V FB = s +
1
C i
2e s N A s .
(22.56)
If light is absorbed in the depletion layer (process (c)), the hole (for p-Si) drifts towards the bulk
material. The electron is stored as part of the signal charge Q sig close to the oxide semiconductor
interface (Fig. 22.36b).
V G − V FB =
Q sig
C i
+
eN A w
C i
+ s .
(22.57)
As a consequence of the increase in signal charge the potential well becomes shallower (22.57). For
each gate voltage there is a maximum charge (well capacity). The maximum signal charge is reached
for s ≈ 2 B (Fig. 22.38).
In a charge coupled device (CCD) many light-sensitive MIS diodes, as described above, are fabricated in matrix form to create an image sensor. Upon application of a gate voltage they accumulate
charge depending on the local exposure to light. The read out of this charge occurs by shifting the
charge through the array to a read-out circuit. Therefore charge is transferred from one pixel to the next.
Several schemes have been developed for this task. The three-phase clocking is shown schematically
in Fig. 22.39. Other clocking schemes involve four, two or only one electrode per pixel [1853].
Since the CCD sensor has many pixels (e.g. up to 4096) along a line, the charge transfer must be
highly efficient. The transfer of charge carriers occurs via thermal (regular) diffusion, self-induced
drift and the effect of the fringing field (inset of Fig. 22.40). The time constant with which the charge
carriers move due to diffusion (in a p-type semiconductor) is
22 Light-to-Electricity Conversion
Fig. 22.37 Conductivity
of a n-Si/SiO 2 -diode as a
function of temperature
(1/T ). The slope of the
dashed lines is (a)
∼ 0.56 eV (≈ E g /2) and
(b) ∼ 1.17 eV (≈ E g ).
Adapted from [1851]
10
-3
10
10
10
10
-4
-5
-6
-7
-1
1000/T (1/K)
2
.
3
4
.
3
0
.
3
8
.
2
6
.
2
4
.
2
2
.
2
0
.
2
Si
(a)
(b)
diffusion (∝ n
2
i ∝ exp(−E g /kT )). The latter process can be strongly suppressed by cooling the
device.
The gate voltage V G and the surface potential s are related to each other via
V G − V FB = V i + s =
e N A w
C i
+
eN A w
2
2 s
,
(22.55)
where w is the width of the depletion layer. w will be larger than w m in thermodynamic equilibrium.
The first term in the sum is |Q s |/C i and the second is obtained by integrating the Poisson equation for
the constant charge density −eN A across the depletion layer. The elimination of w yields
V G − V FB = s +
1
C i
2e s N A s .
(22.56)
If light is absorbed in the depletion layer (process (c)), the hole (for p-Si) drifts towards the bulk
material. The electron is stored as part of the signal charge Q sig close to the oxide semiconductor
interface (Fig. 22.36b).
V G − V FB =
Q sig
C i
+
eN A w
C i
+ s .
(22.57)
As a consequence of the increase in signal charge the potential well becomes shallower (22.57). For
each gate voltage there is a maximum charge (well capacity). The maximum signal charge is reached
for s ≈ 2 B (Fig. 22.38).
In a charge coupled device (CCD) many light-sensitive MIS diodes, as described above, are fabricated in matrix form to create an image sensor. Upon application of a gate voltage they accumulate
charge depending on the local exposure to light. The read out of this charge occurs by shifting the
charge through the array to a read-out circuit. Therefore charge is transferred from one pixel to the next.
Several schemes have been developed for this task. The three-phase clocking is shown schematically
in Fig. 22.39. Other clocking schemes involve four, two or only one electrode per pixel [1853].
Since the CCD sensor has many pixels (e.g. up to 4096) along a line, the charge transfer must be
highly efficient. The transfer of charge carriers occurs via thermal (regular) diffusion, self-induced
drift and the effect of the fringing field (inset of Fig. 22.40). The time constant with which the charge
carriers move due to diffusion (in a p-type semiconductor) is