AFORS-het manul

什么是

AFORS-HET

AFORS-HET

(Automat

FOR

Simulation

of

HETerostructures)

is

a

numerical

simulation

tool,

which allows to model/simulate heterojunction semiconductor devices. An arbitrary sequence of

semiconducting layers can be modelled, specifying layer properties and interface properties, i.e.

the

defect

distribution

of

states

(DOS).

At

the

boundary,

voltage

or

current

controlled

metal-semiconductor

contacts

(MS

contacts)

or

a

metal-insulator- semiconductor

contact

(MIS

contact)

can

be

chosen.

Sub-bandgap

photon

absorption

can

be

simulated

by

specification

of

optical capture cross sections. The program solves the one dimensional semiconductor equations

using

Shockley-Read-Hall

recombination

statistics,

(1)

for

thermodynamic

equilibrium,

(2)

for

steady- state conditions under an external illumination and/or bias voltage (3) for small additional

sinusoidal

perturbations

of

the

applied

bias/illumination.

Thus

the

internal

cell

characteristics

(band diagrams, local generation, recombination, currents, carrier densities and phase shifts within

the device) can be calculated for various external boundary conditions.. Furthermore, a variety of

common

characterisation

techniques

can

be

simulated,

i.e.

current-voltage

(C-V),

quantum

efficiency (IQE, EQE), impedance (IMP), voltage and temperature dependant capacitance (C-V,

C-T),

intensity

and

field

dependant

surface

photovoltage

(SPVi,

SPVv),

photo-

and

electro-luminescence (PEL), electron beam induced current (EBIC), electrical detected magnetic

resonance

(EDMR).

Also

analytical

forms

describing

a

measurement

can

implemented

and

compared

to

the

numeric

ones

(IQEanalytic)

A

user-friendly

interface

allows

to

visualise,

store

and compare all aspects within your simulations. Furthermore, general parameter variations can be

performed.

New

numerical

modules

and

new

measurements

may

be

added

by

external

users

(open-source on demand).

如何获得帮助

We tried to equip this software with a user-friendly interface, that is mostly self explaining. We

hope, that there is no need for reading the online help, since most program options are obvious, if

you

have

already

used

similar

programs

like

PC1D

or

SCAPS.

If

you

are

however

confused

anywhere in the program press F1 to gain context sensitive help for the current program window.

What

‘

s new?

version 2.0:

?calculation mode transient was included

?several major and minor bugs fixed

version 1.2:

?It is now possible to perform a multidimensional parameter fit on measurement curves

?External circuits can now be applied

?The program now distinguishes between several calculation modes (Eq, DC, AC)

?several major and minor bugs fixed

version 1.1:

?You

can

now

customise

the

program

by

choosing

different

optical

and

numerical

calculation

models

?In front and behind the cell, optical layers can be defined, that influence the generation rate inside

the structure, as

multiple coherent/incoherent reflections can be calculated, choosing the optical

model multiple reflections and coherence. It can be selected in the settings window.

?the sequence of grid points is now editable by the user via the settings window

?Optical properties of the structure are now defined by nk- files and not anymore by alpha-files.

When loading structures created with program version 1.0 be sure to check that all layers have the

right optical files associated.

?New measurents IQE analytic, SPV intensity, PES and EDMR are introduced, the old ones have

been

improved.

New

measurement

methods

can

be

added

by

an

external

user

(open

source

on

demand).

?AC calculation now is not only available during some measurements. AC voltage and frequency

can be applied anytime as external parameters

?Thermionic- Emission

and

Drift- Diffusion,

are

now

not

interface

properties,

but

different

numerical models.

?hetero

structure

files

(*.het)

and

graphical

output

files

(*.res,

*.rac,

*.iv,

*.qe,

*.iqean,

*.adm,

*.imp, *.cv, *.ct, *.spvs, *.spvv, *.spvi, *.ebic, *.pel, *.edmr, *.var, *.spek) can be opened directly

from the Windows Explorer when the file type is associated with AFORS-HET.

?The handling of complex graphs has been improved

?several major and minor bugs fixed

?compatibility: unfortunately all graphical output files

of program

version 1.0 cannot be loaded

with the new program version, due to improvements made to the file format.

Known issues and planned improvements

We

are

well

aware

that

the

program

still

needs

a

lot

of

development

work.

The

following

improvements and bugfixes are being developed at the moment:

New developments:

-

a periodic modulation of the light intensity will be available

-

correlated dangling bond defect distributions will be implemented

-

a numerical model for Schottky-Bardeen contacts will be implemented

-

Fermi-Dirac statistics will be implemented

-

the defect-pool model to treat a-Si:H layers will be implemented

0

bugfixes:

-

phase

shifts

in

the

AC-voltage

cell

results

should

be

continuous

and

not

distracted

by

+-180?phase jumps (minor bug)

Numerical Models

AFORS-HET builds a discrete set of gridpoints at which the semiconductor equations are being

solved. There are 4 different types of gridpoints (bulk point, interface point, first interface point

and last interface point). Even if a gridpoint is called an interface point, it is never located exactly

at

the

interface,

but

always

a

very

small

distance

away

from

it.

This

distance

can

be

specified

within the

numerical_settings.

So

for

an

heterojunction

interface,

there

are

actually

two

interface

points

belonging to that interface, one at each side of the heterojunction. For each of these different types

of

gridpoints

different

differential

equations

/

boundary

conditions

and

eventually

modified

routines

for

solving

the

resulting

discrete

equations

have

been

implemented.

A

set

of

those

routines is called numerical model. By applying a different numerical model to a point you change

the

way

the

program

calculates

at

this

point.

You

can

choose

the

models,

in

the

windows

for

editing layers and interfaces (first, last or any other). New numerical models may be added by an

external user (open source on demand). At the moment the following models are implemented:

bulk numeric models:

There are two numerical models for a bulk gridpoint:

Standard:

Within

this

numerical

model,

Poissons

equation

and

the

electron/hole

equation

of

transport

is

stated

in

a

discritized

form,

together

with

(1)

the

partial

derivatives

of

these

equations,

(2)

a

routine to solve locally for the resulting discrete equations.

Crystalline-Silicon:

Within

this

numerical

model,

the

standard

numerical

model

is

modified

within

the

routine

for

solving locally for the resulting discrete equations, in order to account for impurity scattering and

carrier-carrier

scattering

within

crystalline

silicon.

That

is,

the

electron/hole

mobilities

will

no

longer

treated

to

be

constant

within

a

layer,

but

they

will

(iteratively)

depend

on

the

local

electron/hole particle densities within the cell.

interface numeric models:

There are two numerical models for an interface gridpoint:

Drift- Diffusion:

The

transport

across

the

heterojunction

interface

is

modelled

by

drift-diffusion,

in

complete

analogy to the bulk. In order to do so, an interface layer is assumed, with the thickness given by

the

distance

of

the

two

interface

gridpoints

specified

in

numerical

settings.

Additional

interface

states

can

be

specified,

which

will

result

in

additional

interface

recombination.

Within

the

interface layer, all the layer properties are linearely transformed from one semiconductor to the

second semiconductor.

Thermionic emmision:

Alternatively, the transport across the heterojunction can be modelled by thermionic emission over

the energetic barrier of the heterointerface. Additional interface states can be specified, which will

result in additional interface recombination. These interface states can interact with both adjacent

semiconductors, thus charge carriers can transverse the heterointerface via defect states from one

semiconductor to the other.

first interface numeric models:

So far, there are two numerical models for the first interface gridpoint:

Metal/semiconductor Schottky contact:

The

front

contact

is

treated

as

a

metal/semiconductor

Schottky

contact.

That

is,

the

difference

between the metal work function and the electron affinity of the adjacent semiconductor defines

an energetic barrier for the current flow from the semiconductor into the metal. Interface states are

not considered, if they are specified, they will be ignored.

Metal/insulator/semiconductor MIS contact:

The front contact is treated as a metal/insulator/semiconductor contact. Thus there is no current

flow into the front contact. The corresponding insulator capacity has to be specified. Additional

interface defects can also be specified, resulting in an enhanced interface recombination and also

in a modified band banding in thermodynamic equilibrium.

last interface numeric models:

At the moment, there is only one numerical model for the last interface gridpoint:

Metal/semiconductor Schottky contact:

This is exactly the same as stated in the first interface gridpoint

transient mode:

Since version 2.0. AFORS-Het offers a transient calculation mode. Each numerical model has to

provide

additional

functions

that

solve

the

problem

under

transient

conditions.

For

more

information on the numerical implementation of the transient mode click here.

Defining a structure

The first step when starting AFORS-HET is usually to define the structure you want to simulate. A

structure

always

consists

of

a

front

contact,

a

back

contact,

and

a

variable

number

of

layers

in

between (at least 1). Between all these items are interfaces, which are by default disabled (drift

diffusion transport across the interface). Since version 1.1 a structure furthermore contains optical

layers, which define light absorption, reflection and transmission at the cell contacts. Since version

1.2

external

circuits

can

also

be

defined,

i.e.

a

serial

resistance

Rs,

a

serial

capacitance

Cs,

a

parallel resistance Rp and a parallel capacitance Cp.

Click on the Button

慏

efine Structure?in order to create, load or modify a structure. All the items

of the structure (contacts, layers, interfaces, external circuits) are displayed here. By clicking on an

item you can change it

抯

properties (e.g. click on a layer, if you want to change its properties).

The simplest structure possible (1 layer and no interfaces) is offered if you start the program or if

you select the button

慛

ew Cell?

Press

the

buttons

under

the

label

慉

dd

Layer?to

add

new

layers.

The

button

labeled

慹

lectric?will add a new electric layer. This new layer will be placed between the last layer and the

back contact. Now, you either have to specify the material properties manually, or you load a layer

that already exists. The same procedure works with the optical layers. Press

憃

ptic front?or

憃

ptic back?to add an optic layer in front of or behind the structure. Click the optic layers to edit

their properties. Since version 1.1 you can also change the sequence of layers by clicking on the

arrows in front of the structure. If you have specified the structure click on

慜

K?and the program

will start calculating the Eq equilibrium state for your structure.

Furthermore you have the possibility of saving the structure to a file (*.het) and loading previously

saved structures (save/load buttons). There are already some structures included with the program

so you might want to load and modify them for your purpose.

External Parameters

On

the

left

side

of

the

main

program

window

you

can

see

the

external

parameters.

They

are

divided into 3 subgroups:

temperature, illumination and boundary conditions.

Each time you start

the program, the external parameters are reset to defaults.

Illumination

First

you

must

decide

if

illumination

should

be

turned

on

or

off

(darkness).

If

illumination

is

turned

on

you

can

define

the

incoming

light,

which

consists

of

two

components

，

that

can

be

individually

turned

on

and

off:

a

spectral

component

and

a

monochromatic

component.

If

both

components are enabled the incoming light will be the sum of both components. Furthermore you

can decide between front side and back side illumination. To view the complete incident spectrum

use the

憇

pectra?button in the main window .

The spectral component can be directly defined by an incident file (*.in), specifying the number of

photons

at

each

wavelength

of

the

incident

illumination.

These

numbers

will

be

additionally

multiplied with the factor called

times you have to enter. The default file

慉

?with the default factor times=1 describes the

average incident irradiation of the sun in middle Europe in summer at noon.

The

generation

of

electron-hole

pairs

within

the

semiconductor-layers

will

then

be

calculated

using the spectral absorption coefficient alpha [cm^-1] of the layers. Only super-bandgap photons

with E>Eg Opt will be absorbed due to the layer parameter alpha and generate electron- hole pairs,

photons with E

have to switch on the optical capture cross sections cno and cpo in the Editing Defect Menu.

The spectral component can also be indirectly defined by just loading a generation file (*.gen).

These

files

contain

information

about

how

many

electron-hole

pairs

are

generated

at

a

certain

position

within

the

heterostructure.

So

you

may

like

to

use

other

programs

to

compute

a

more

accurate generation profile, accounting for surface texture and multiple internal reflections. Note

that, if you load generation files, settings like front contact absorption or reflection are ignored.

You can create your own incident (*.in) and/or generation (*.gen) files, but be sure, that they have

the right file format (see example files), otherwise the program might run into trouble. There are

no

restrictions

concerning

the

spacing

of

the

data

points

in

these

files,

as

linear

interpolation

routines

will

be

used.

Per

definition,

the

spectral

illumination

is

zero

at

a

wavelength

larger

or

lower than the maximum or minimum wavelength specified in these files.

The other light component is a monochromatic light source (laser light), which is defined by it

抯

wavelength, it?

s intensity and it

抯

spectral width. If you have a generation file loaded, the generation caused by

the laser is nevertheless calculated individually and added to the generation implied by the file.

Temperature

The temperature is uniform throughout the device. By default it is set to 300 K. No temperature

dependence

of

any

material

parameter

is

involved

in

the

calculations.

It

could

however

be

implemented

by

programming

an

numerical

module

for

the

material

under

consideration,

for

example an extension of the numerical module crystalline silicon (open- source on demand).

Boundary Conditions

The back contact of the heterostructure is always assumed to be a

metal- semiconductor contact

(MS

contact).

The

front

contact

can

be

either

a

metal-semiconductor

contact

or

a

metal- insulator-semiconductor

contact

(MIS

contact).

This

can

be

selected

by

choosing

the

appropriate NUMERICAL-MODEL. If you choose a metal/semiconductor contact, you can decide

whether the contact is voltage or current controlled. In case of an MIS contact the current has to be

zero, so there is no current control. The capacitance density of the insulator of the MIS contact is

specified in the Front Contact Editing Menu

?external voltage

(voltage controlled MS or MIS contact):

you specify the applied voltage external voltage [V] across the heterostructure.

?external current

(current controlled MS contact):

you specify the total current density external current [mA/cm^2] through the heterostructure. In

this mode the iteration is more likely to fail if the starting solution provided is far away from the

real

solution.

It

抯

a

good

idea

first

to

iterate

the

desired

current

by

using

the

external

voltage

boundary conditions and to adjust the desired current manually. If the current is in the appropriate

range, change the boundary conditions to Mscurrent.

Furthermore you can decide whether to fix the potential f front side or back side to zero. This is

only changing the energetic scale and will not affect the other results. However, the resulting band

diagrams will be correspondingly fixed front side or back side.

Aditionally to the DC boundary conditions mentioned above, periodic sinusoidal perturbations can

be applied by switching to AC calculation mode and enabling AC boundary conditions. These are

characterised by an AC frequency [Hz] and an AC voltage [V] or AC current [A/cm^2] and/or an

AC illumination [1/(cm^2 s]. Also note that in transient calculation mode the way to enter external

parameters changes slightly.

At the end of a calculation (for example if you press the calculate button), the missing quantities

external current and external voltage, which you have not specified, will be computed.

Settings

Accuracy: While the program calculates, it tries to solve the semiconductor equations. It does so

by assuming the actual results to be a first approximate solution and calculating iteratively a next

(hopefully

better)

solution.

In

an

ideal,

but

in

no

way

realistic

case,

the

equations

are

solved

exactly, that means DGLSYS:=0 at every grid point. The Accuracy is the maximum deviation of

the numerical semiconductor equations from zero for the scaled semiconductor equations [Sel, p.

208]:

Accuracy:=max(Abs(DGLSYS), {all gridpoints})

The actual Accuracy is plotted in the status bar of the main window during a calculation. If this

accuracy

is

lower

than

the

value

of

Accuracy

specified

here,

the

program

knows

that

the

calculation can be finished and further iteration is not necessary.

Min. number of iterations: Even if the actual accuracy is already lower than the specified value

after

the

first

iteration,

the

program

iterates

at

least

so

many

times

in

order

to

ensure

better

convergence.

Max.

number

of

iterations:

Even

if

the

accuracy

is

not

lower

than

the

specified

value,

the

calculation is finished when this number of iterations is reached. The actual number of iterations is

always plotted in the status bar of the main window during a calculation.

Max. number of iterations (equilibrium): Even if the accuracy is not lower than the specified value,

the

equilibrium

calculation

is

finished

when

this

number

of

iterations

is

reached.

The

reason

to

distinguish

between

the

two

maximum

number

of

iterations

is

that

an

equilibrium

calculation

often needs more iterations to converge: If you press the calculate equilibrium button, or if you

manually

change

your

heterostructure,

the

starting

solution

will

be

generated

by

the

computer

program itself and is therefore far away from the actual solution.

Delta E: Delta E defines the discretisation of the energy within the bandgap for defects.

Spacing between layers: All gridpoints are located in the bulk, never at an interface. The distance

between the two gridpoints, left and right to a heterointerface can be specified. This distance will

be used as a thickness for the interface layer, if the numerical module drift-diffusion is used. Also

the spacing between the front/back-side interface and the first/last gridpoint can be specified.

Numeric

0

incident/absorbed

photons:

The

number

of

photons

incident

on

the

heterostructure

and the number of photons absorbed within the heterostructure are calculated numerically. Define

here below which value these numbers should be treated as zero.

Edit grid points: Press this button, to open a window that lets you edit the number and position of

the structure

抯

grid

points.

Normally

the

program

calculates

the

best

amount

and

position

of

grid

points.

However, you might want to discretize some regions more precisely or save calculation time by

thinning the grid points out. You can set single points at a specific position or a lager amount of

points in a specified range. If you want to delete points, select the range and set a small number of

points there. Note, that by calculating the equilibrium the program will set a new grid, that does

not contain the changes made before by the user. You can also save and load often used grid point

discretizations to files(*.grd).

----Page Optical----

This is where you can select the optical model you want to use, in order to calculate the generation

rate. For each optical model you can adjust settings here if there are any. Refer to the online help

page of optical models for an explanation of these parameters.

----Page Measurements----

Here you can disable measurements you don

抰

want to use in order to have a cleaner program

surface. You can enable the measurements again anytime you like. Changes need a program restart

to take effect.