• Aucun résultat trouvé

Coherent Bremsstrahlung and the Quantum Theory of Measurement

N/A
N/A
Protected

Academic year: 2021

Partager "Coherent Bremsstrahlung and the Quantum Theory of Measurement"

Copied!
19
0
0

Texte intégral

(1)

HAL Id: jpa-00247055

https://hal.archives-ouvertes.fr/jpa-00247055

Submitted on 1 Jan 1995

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

Coherent Bremsstrahlung and the Quantum Theory of Measurement

E.H. Du Marchie van Voorthuysen

To cite this version:

E.H. Du Marchie van Voorthuysen. Coherent Bremsstrahlung and the Quantum Theory of Mea- surement. Journal de Physique I, EDP Sciences, 1995, 5 (2), pp.245-262. �10.1051/jp1:1995126�.

�jpa-00247055�

(2)

Classification Physics Abstracts

03.658 78.70F

Coherent Bremsstrahlung and the Quantum Theory of

Measurement

E-H- du Marchie van

Voorthuysen

Nuclear Solid State Physics and Materials Science Centre, University of

Gromngen,

Nijenborgh 4, NL 9747 AG Groningen, The Netherlands

(Received

29 December 1993, revised 21 September 19g4, accepted 18 October

1994)

Abstract. Coherent brernsstrahlung

(CB)

is the result of inelastic Bragg scattering of elec- trons of a few hundreds of kev, a collective elfect of trie whole crystal. In

an electron microscope it is theoretically possible to deterrnine trie row of atoms where the electron

was inelastically scattered. These staternents are contradictory. Optical path calculations are made for 160 kev electrons inelastically scattered by a silicon

[Ill]

crystal and the results are compared with rneasured CB spectra. The results can be understood by application of Van Kampen's theory of quantum mechanical measurement. Experimental CB spectra tum out to be the result of

coherent scattering of atoms within the

same row, and incoherent summation of intensitie8 from dilferent rows, or, in other words, the electrons that cause CB are locabsed within atomic rows.

l. Introduction

1-1. COHERENT BREMSSTRAHLUNG.

During

collisions of electrons of a few hundreds of

kev with atoms the

following

inelastic events can occur:

i)

inner shell ionisation

resulting

in

charactenstic

X-ray radiation, ii)

outer shell ionisation and

for

excitation

resulting

in charac- teristic

peaks

in the electron energy loss spectrum

(EELS)

in the 10 eV

region,

and

iii)

emission of a

bremsstrahlung photon

when the electron orbit is

changed by

the Coulomb field of the

atomic nudeus.

In dassical

physics

an electron is a

partide

that emits

electromagnetic

radiation when it is accelerated. The

intensity

of the

bremsstrahlung

emitted

by

an electron with

velocity /3(t)

=

v(t)/c

is [1]

d~I/duJdfl

= e~

/(47r~c) /

in

x

((n

/3) x

/3) /(1

/3

n)~j

e~~(~~~'~(~)/~)dt(~

(l)

where n is the unit vector in the direction of the emitted radiation and

r(t)

the vector

connecting

the electron with the detector.

Normally

the

bremsstrahlung

spectrum is continuous, without any structure.

We consider an electron with a

trajectory

close to a row of

equidistant

atoms with interatomic distance L. This electron

undergoes

a

velocity change Ad

each time it passes a nucleus. We

IOIIRNAL OEPHYSIQUE11 T5, N° 2, PEIIRUARY 1995 4

Q Les Editions de Physique 1995

(3)

assume

(Ad(

<

fl

+

(p(,

so the

trajectory

is a

straight

fine and the time between successive radiative interactions is constant, At

=

L/(flc).

The

velocity change

/3 is zero except for t = C + mât with m

integer,

so the

frequency

spectrum of the

bremsstrahlung

is the Fourier

transform of

equidistant sharp pulses

in the time domain:

f(uJ)

= Const à[uJ

j.27rflc/(L(1

n

/3))]

,

j

= integer.

(2)

We observe that for

photons

emitted

perpendicular

to the direction of the electrons the

ground frequency

is

proportional

to the electron

velocity

and inverse

proportional

to the interatomic

distance.

If the energy

spread

and the

angular spread

of the electrons in the

incoming

beam are

sufliciently

small the electrons may not be

regarded

as localised entities, but must be

regarded

as manifestations of a wave, the wavefunction of the

electron,

which is assumed to be a

plane

wave with

wavelength

À;. The electron leaves the atom with

wavelength

Àf,Àf > À;. When we are

dealing

with an extended structure of atoms, like a

crystal,

the scattered waves from all

atoms must be added and the

intensity (the

number of electrons detected on a screen far away

per unit of area per unit of

time)

is

equal

to the modulus of the sum wave,

squared.

Constructive and destructive interference effects occur, like in the case of elastic

scattering,

À;

= Àf, where these effects

give

rise to the well known

Bragg scattering (electron diffraction).

As a matter of

fact, practically

ail electrons that are

elastically

scattered

by

a thin

crystal

are

leaving

the

crystal

in directions where constructive interference takes

place.

For

inelastically

scattered electrons the same is true if the excess energy is

completely

removed

by

one

bremsstrahlung photon.

The Ewald

construction,

which is very useful for

visualising Bragg scattering,

con therefore be used also to illustrate inelastic coherent electron

scattering, Figure

1. Coherent

bremsstrahlung (CB)

occurs when the vector relation kf

=

k;

+g q is satisfied and (k;( >

(kf(,

k is the electron wave vector, (k( e k =

1/À,

g is a vector

connecting

two lattice

points

in

reciprocal

space, and

hq

is the momentum of the CB

photon.

The energy of the CB

photon

is [2]

~ =

hep(g~ g2/2k;)/(1 pcos9) (3)

with gz the

projection

of g on the direction of k; and the

angle

between the direction of the

photon

and k;. Effects of

channeling

radiation are

neglected

in

(3).

Diffraction effects of the

bremsstrahlung

itself with the lattice are also

neglected,

the

wavelength

of coherent

bremsstrahlung

is

generally

too

large.

When the

incoming

electron beam is

parallel

to a zone axis, all

points

of the

reciprocal

lattice

belonging

to the same Laue zone have the same value for gz,

Figures

lc and ld.

They

contribute to the same

peak

in the

bremsstrahlung

spectrum because g <

k;,

and the equations

(2)

and

(3)

are equivalent.

CB can be observed in the continuum between the characteristic

X-rays, Figure

2. The CB peaks are marked with numbers

equal

to the order of the Laue zone. Vecchio and Williams [3]

have made an extensive expenmental

study

of the optimum conditions for

producing

CB

peaks.

These conditions are: thin

specimens

of pure elements

(no intervening X-rays

from

impurities),

dean

crystal surfaces,

zone axis orientation, low temperature

(to

reduce lattice

vibrations),

and detection with a semi-conductor detector with a solid

angle

> 0.01 steradians m a direction

perpendicular

to the beam.

Sigle

and

Carstanjen

[4] observed CB from a

quasicrystal.

Coherent

bremsstrahlung

is considered to be an annoymg side effect in

analytical

elec- tron microscopy, CB

peaks

can obscure

X-ray

peaks from impurities or can lead to wrong

intensity determinations of

impurity

X-rays. But CB can also lead to an

interesting problem

of fundamental nature.

(4)

, o .

,

k~

i~ .

.

.i 1.

o ~ .

.

a) b)

i~

,

i~

k ç

~ 3 -3

)

q . ~ 2

. -2

-

)

-1

~ o

Î

-o

. .

c)

,

d)

Fig. l. Ewald constructions in the reciprocal lattice: a) elastic scattering, (k,(

= (kf(; b) inelastic scattering, (k,( > (kf(, hq is the momentum of the emitted CB photon; c) inelastic scattering, (k,( >

(kf(, k, ii a zone axis, scattering towards

a point on the

zone axis in trie third order Laue zone;

d)

inelastic scattering, (k,( > (kf(, k, ii a zone axis, scattering towards a point in the first order Laue

zone.

1.2. DEFINITION oF THE PROBLEM. One of the most

intriguing

aspects of

physics

is the

particle-wave duality.

The

question

is the electron a localised

particle

or is it the manifestation of an extended wave? can be illustrated

nicely

with the double-slit

experiment,

see for instance

Feynman

[5]. The electron flux as a function of position of the detector, as it is measured on

a screen behind the wall with the two slits, consists of two

factors,

a diffraction factor and

an interference

factor,

like m Fraunhofer diffraction. When we observe interference it is not

(5)

Si Si sumoeak Fe Fe Cu Cu

5 7

2 4 6 8 JO

energy Ioss (kev)

Fig. 2. Experimental photon spectrum. A silicon [loo] crystal at room temperature is bombarded

by 120 kev electrons parallel to an

[Ill]

axis

in a JEOL JEM 200 CX microscope. Photons are detected in the direction H

= 90° by a

Si(Li)

detector with a resolution of130 eV. The sobd angle of acceptance is 1.8 X 10~~ srad. The spectrum is dominated by peaks of characteristic X-rays that are

indicated by their chemical symbol; the height of the Si-Ka peak at 1.8 kev is 120,ooo courts. The

peaks of coherent bremsstrahlung are marked by numbers indicating the Laue zone from which they origmate.

possible

to measure or to deduce

through

which slit the

actually

observed electron

really

went.

As soon as we try to detect the passage of the electron at one of the slits we cause a disturbance

m the

phase

of the wavefunction of the electron such that as a result the interference factor

turns into a constant value. The usual

terminology

is that

through

the act of

measuring

the

electron at the slit the wavefunction of the electron has

collapsed.

The same can be said for the case of electron diffraction as it is observed in the electron

microscope.

There is no way to determine which atomic row was

passed by

the electron without

destroying

the pattern of diffraction spots.

In the case of inelastic

Bragg scattering leading

to coherent

bremsstrahlung

we can observe the

photons

as well as the electrons. The

photon

spectrum contains information about a collec- tive effect of ail

crystal

atoms, 1.e. the interference between

scattering

from ail atoms. At the

sonne

time,

detection of the electrons

provides

information on the location where the

scattering

occurred. The maximum accuracy in the measurement of the location of the

scattering

centre is

given by

the

uncertainty

relation of

Heisenberg:

if the

scattering angle

is bounded within

a too narrow cone, the

uncertainty

of the transverse momentum of the electron is too small.

For the moment we make the

assumption

that the row where the electron scattered can be determined. The electron leaves the

crystal

with a

slightly reduced,

but still

quite accurately

known energy. In a

high-resolution

electron microscope it must be

possible

to make a

good

image of the

crystal

with atomic

resolution,

to mount a two-dimensional position-sensitive elec- tron detector at the

image plane,

and to deterrnme m an electronic coincidence unit for each detected

photon

which atomic row the electron

passed

while

producing

the

bremsstrahlung.

If

we assume that the coherent

bremsstrahlung

spectrum contains interference factors between

different atomic rows, these factors wùl become constant as soon as a measurement of the row is made. So the CB spectrum must

change

as soon as a measurement is made of the atomic

row where the electron that caused the

photon passed by:

the measurement of the

trajec-

tory of the electron

through

the

crystal

reduces the

intensity

of some CB peaks and increases tue

intensity

of otuer

peaks.

Tuis is tue famous wavefunction

collapse througu

trie act of

(6)

2

fi

Mfi-

j 9

3-&- W

à4

y A

~ ~ II

X X Y Y B

8

MEMORY

llit

gate

= yes

Fig. 3. The paradox. Switches down: CB photons are detected without rneasurement of the atornic row, switches up: CB photons are detected with sirnultaneous rneasurernent of the row. Manipulation

of the switch rneans annihilation and creation of pulses in the delay units, or it means that the count rate as measured in counter A depends

on the switch setting in the future.

l)

incoming electron,

energy E. 2) crystal with atomic rows aligned parallel to the incorning electron beam. 3) electron

lens, making an image on trie detector with atomic resolution. 4) outgoing electron, energy E'

=

E e. 5) position sensitive electron detector. 6) pulses with voltage proportional to the z-coordinate.

7) pulses with voltage proportional to the y-coordinate. 8) a pulse means detection of an electron, and

the atomic row it carne from is known; this information is stored in the rnemory. 9) CB photon with

energy E, 10) pulses with voltage proportional to e,

Il)

logic pulse if e is within the window as set by

the SCA. ADC: analogue to digital converter. CNT: pulse counter. COIN: coincidence unit, logic and;

pulses arriving sirnultaneously belong to the sarne event. DELAY: pulse delay unit containing many pulses running through it with constant speed and with voltage conserved. DET: photon detector.

SCA: single channel analyser; rneasures the voltage and gives output puise if the voltage is within a

specified

interval.

measurement. Now let us assume tuat tue

collapse

takes

place

wuen tue last connection is made m tue

complicated

electronic

circuitry

that is needed for tue

photon-electron

coincidence

measurement in tue

uigu-resolution

electron microscope,

Figure

3. Tue last switcues are sit- uated far from tue detectors and near to tue counter wuere tue information will be stored.

Travelling

times of electronic

pulses

from tue detectors to tue memory are

typically

10 ps, but

by

means of

delay

units tuese times can be mcreased to

milliseconds,

seconds or even uours

without any loss of accuracy.

Cuanging

tue switch

setting

would mean that in the

delay

units

(7)

immediately

some electronic

pulses

are annihilated and other electronic

pulses

are created. If

we do not accept this consequence, it follows that the

pulses

of the

photon

spectrum with a

pulse height

distribution that is

dependent

on the

position

of the switch were

already

present

long

before the switch was

manipulated:

the count rate in counter A is

equal

to the count rate

in counter B one hour later. In that case we could make a

dear-sighted telescope

with a

unique

possibility

to become rich on the

stock-exchange.

In this paper we will discuss two propositions for

solving

this

problem:

1) the act of measurement of the atomic row is not a

manipulation

with electronic

equipment,

but takes

place

at the moment that the

outgoing

electrons are focused on a screen with atomic

resolution;

2)

CB spectra do not contain interference factors between waves

coming

from different rows.

1.3. SOME REMARKS ABOUT QUANTUM MECHANICAL MEASUREMENT. In the common

interpretation of quantum mechanics

(the

sc- called

Copenhagen interpretation)

the act of

measuring

a

physical quantity

has an enormous effect on the

physical

world: the wavefunction

changes drastically,

and some other

physical quantities

cannot be determined any more with suflicient accuracy. Before the

"holy"

act of measurement our

knowledge

of the world is

only

a

coherent

superposition

of many

possibilities,

administrated with

complex numbers,

the wave- function. But after the measurement the world is a sum of

objectively existing probabilities,

administrated with

real, positive

numbers. One may read out such a

probability

(1.e.

perform

a

dassical

measurement)

and get more

knowledge

of the

world,

or one may not, it does not make any difference for the rest of the world. So there is an essential difference between quantum

mechanical measurement and dassical measurement.

The sudden

change

in the wavefunction is called the

"collapse

of the wavefunction". In the

common

interpretation

of quantum mechanics the occurrence of the

collapse

as a result of the

imposed

measurement has to be put into the

theory

as a separate

postulate:

the

projection postulate.

According

to Bour a quantum mecuanical measurement is terminated wuen tue outcome bas been

macroscopically

recorded in a dassical apparatus. Bour made a difference between tue

microscopic world,

where the rules of quantum mechanics are to be

applied,

and the

macroscopic

world,

where dassical

physics

is valid.

A

macroscopic

system

usually

consists of a very

large

number of

partides resulting

in a very

high density

of states. In a quantum mechanical measurement a

microscopic

system in a

(nearly)

pure state is

brought

into contact with the macroscopic system. As a result a

change

is

brought

about in the

macroscopic

system and

consequently

in the

microscopic

system: the

collapse

of the wavefunction of the

microscopic

system. Because of the

large

difference in

density

of states between both systems the

change

is irreversible: the

probability

that both systems return to their

original

state can be

neglected.

According

to Van

Kampen

[6] the

huge density

of states is the main feature of a macroscopic system. The rules of quantum mechanics

apply

to the system, but the

eigenvalues

of the energy operator bave an average distance ôE tuat is mucu smaller tuan tue best accuracy tuat can be obtained in an expenment.

Manipulations

on a macroscopic system influence tue

wavefunction of tue system, but tue result is mucu to small to be observable. So tue rules of dassical mechanics are valid to a very

good approximation.

Van

Kampen

defines a quantum mechanical

measuring

apparatus as a macroscopic system in a metastable state. To my

opinion

the metastable state is not a necessary condition for

a quantum mechanical

measuring

instrument. Van Kampen [6] gives a nice exarnple of a quantum mechanical

measuring

apparatus consisting of an atom m an

excited,

metastable

(8)

state and the free space

surrounding

ii. In the

electromagnetic

field an emitted

photon

finds

nearly

cc different

modes,

so the

density

of states of the

photon

is very

large.

A fast electron is able to

trigger

the metastable atom, and

consequently

its existence in the

neighbourhood

of the atom is measured in the form of an irreversible

change

m the

electromagnetic

field.

Van

Kampen

descnbes the whole process in a quantum mechanical way, so the

measuring

apparatus has a wavefunction too. The

Schrôdinger

equation works on the whole system and

as a consequence of this it is demonstrated that

triggering

the atom means destruction of the

possibility

of interference with the electron. In Section 2 of this paper I will use such an

approach

to

analyse

the

problem

mentioned in Section 1.2.

Another

example

of'a quantum mechanical

measuring

instrument is the use of a foi] of solid matter for

measuring

the existence of a fast electron. The foi] is a macroscopic system

that

undergoes

an irreversible

change

due to the energy Ioss of the electron in the foil. This

change

can be read out

electronically

if the foil is a

semiconductor,

or it can be observed

by

measuring

a rise m temperature if the foil is part of a calorimeter. When the electron has

deposited

energy in the foil this energy may be measured or not. This last measurement is a dassical measurement because the foil is a

macroscopic

system. The wavefunction

describing

the multitude of excitations in the foil is not

changed by

this measurement in such a way that this

change

is observable. The effect of this

change

on the wavefunction of the

original

fast electron is

completely negligible.

It is

only

the deed of

positioning

the foil in the

trajectory

of the electron that influences the wavefunction of the

electron, nothing

else.

A related

theory

is the modal

interpretation

of quantum mechanics of Dieks [7].

Another

approach

is that of Zurek [8]. In

practice,

macroscopic systems can never be isolated

completely

from the outside world.

Leakage

of energy leads to loss of coherence between different states, or, m other

words,

the

off-diagonal

eleménts of the

density

matrix

disappear

and we are left with the trace elements: real

positive

numbers

havmg

the

meaning

of dassical

probability.

In ail mentioned

approaches

there is no need for a

boundary

between

microscopic

and macro- scopic. In

principle,

the rules for the

microscopic

world

(quantum mechanics)

are valid

always,

but very soon the rules of dassical

physics

may be used without any loss of accuracy, The step quantum mechanics - dassical

physics, induding

the

"collapse

of the wavefunction" is

just

a consequence of the

growing complexity

of trie system as time goes on and the

projection

postulate

can be left out. The

change

in the wavefunction has

nothing

to do with a deed of

measuring by

some human

being.

In a real quantum mechanical calculation the

probability

=

(probability amplitude(~

must be calculated as soon as the quantum mechanical system has started to

develop

to a macro-

scopic stage. The outcome would not be different if one would follow the

development

of the wavefunction any

further, although

the calculations would soon become

impossible

due to the tremendous

complexity

of a

macroscopic

system. So the outcome is

independent

on the

arbitrary

choice when to calculate the

probability, provided

it is not done to

early.

In the

"paradox"

of

Figure

3 the wavefunction has

developed

into a macroscopic stage when the

pho-

ton has

dissipated

its energy in the

photon

detector and the electron has

deposited

energy in

trie

position

sensitive electron detector.

Any

further

manipulations

do not influence the wave-

function in such a way that the consequences on the microscopic system are measurable. So

creationlannihilation

of electronic

pulses

does not occur, and a

dear-sighted telescope

cannot be built.

(9)

2. The Model

2.1. GENERAL. In order to test the

validity

of the

propositions

in Section 1.2 a

simple

model is made that can be

applied

in computer simulations.

Following

Van

Kampen

[GI the

complete

system is described

by

a wavefunction. We will make a full quantum mechanical

description

of the system in

Figure

3

induding

the

detectors,

but

exduding

the electronics.

Because of the

complexity

we have to restrict ourselves to a rather

symbolic presentation

of the wavefunction components. A ngorous treatment of

subsystems

is given

by

Dieks [7] where the biorthonormal

decomposition

of

subsystems plays

an

important

rote.

The total wavefunction is the

product

of wavefunctions of the

following

components: an

incoming

electron, a

crystal,

space

containing

an electron lens wherein the

outgoing

electron is

moving,

an electron screen that can be put in the focal

plane

of the lens or in the

image

plane,

the

electromagnetic field,

a

photon detector,

and a wall

surrounding

the whole set-up

completely.

The focal point of the lens is m the middle of the

crystal.

We

apply

the

thin-crystal limit,

so Bloch waves are

neglected.

The

complete

wavefunction is written as

x=alil>14l>lS>lE>lD>lW>. (4)

with

ahj " 1: at the intersection of row h with

plane j

there is an atom,

ahj = 0: at the intersection of row h with

plane j

there is no atom.

il >

incoming electron,

which

develops

in time from:

(il

>=

(il+

>: a wave

packet

with mean wave vector k; and mean energy

E,

to

(il

>=

(il~

>: no

incommg

electron.

(4l >: the

outgoing electron,

which

develops

in time from

(4l >= (4l~ >: no

outgoing electron,

to

4l >= (4l(~~ >:

outgoing

electron emitted

by

atom

h,j

with wave vector k

=

kf,

and

finally

back to

Ill

>= (4l~ >.

(S

> the screen, which

develops

m time from

(S

>=

(S~

>: the screen, not hit

by

an

electron,

to

(S

>=

(S(~~

>: the screen, hit at position x

by

an electron with wave vector k

coming

from

atom

h, j.

(E

>: the

electromagnetic field,

which

develops

in time from

(E

>=

(E~

>: empty

electromagnetic field,

to

(E

>=

(El

>: a

bremsstrahlung photon

with energy E has left the

crystal.

Indices h,

j

are

dropped

because of the

large wavelength

of the

photon:

h and

j

cannot be measured from

El

>, so it does not

depend

on

h, j. Finally (E

>

develops

back to

(E

>=

(E~

>.

D >: trie

photon detector,

which

develops

from

ID

>=

(D~

>: trie

photon

detector is not hit

by

a

photon,

to

ID

>=

IDI

>: the photon detector is hit

by

a photon with energy E.

W > the watt around tue

crystal.

Starting

witu an electron in tue incoming bearn:

x =

£ahJ(il+

> (4l~ >

(S~

>

(E~

>

(D~

>

(W~

>

(5)

hJ

(10)

tue wavefunction

develops

under influence of tue Dirac

equation.

Tue electron is

inelastically

scattered

by

atom

h, j

and a

bremsstraulung photon

witu energy E is emitted. Tue interaction between tue fast electron and tue atom tuat causes

bremsstraulung

in tue energy region of

interest

(a

few

kev)

is assumed to be located in a small

region

around tue atomic nudeus.

Tue Born

approximation

is

used,

we

drop

tue

unperturbed incoming

wave. After some time tue electron wave

packet

is between tue

crystal

and tue screen, and tue

photon

wave

packet

is

between the

crystal

and the detector:

X "

£ ahj(il~

>

(4l(~~ >

(S~

>

(El

>

(D~

>

(W~

>

(6)

hjke

The electron hits the screen at position x and the

photon

hits the detector or the wall

x =

£ ahj(4f~

> (4l~ >

(S(~~

>

(E~

>

IDI

>

(W~

>

hjkxe

+

£ ahj(il~

> (4l~ >

(S(~~

>

(E~

>

(D~

> (W~+ >

(7)

hjkxe

The second term is not important for the measured

photon yield.

The wavefunction has

developed

now into a macroscopic stage: excitations on the electron

screen and in trie

photon

detector. Therefore we can calculate now the

probability

that a

photon

of energy E is detected in the

photon

detector and that the electron has hit the screen:

Î(E)

"

x(E)x*(E)

"

£

tlhjtlh,j, < iÎ~ iÎ~ >< 4Î~ 4Î~ >

hh' jj' kk' xx'

<

S(~~~ S(~,~,~,

>< E~ E~ ><

Dl Dl

>

(8)

After a certain time we know that

(il

>, (4l >, and

(E

> are

really

empty, that is the electron

is for sure not in front of the

crystal

and not between the

crystal

and the screen, and the photon is not between the

crystal

and the detector. Therefore the

probabilities

< il~ il~ >,

< 4l~ 4l~ >, and < E~ E~ > are

equal

to one.

Each part of the screen forms a separate macroscopic system, so there is no coherence between the wavefunctions in different parts,

only

terms x = x' survive in the summation [7].

<

Dl Dl

> is trie detection

efliciency

of trie

photon

detector which we make

equal

to 1.

The

shape

of trie CB spectrum becomes:

Î(E)

"

~

~hjah'j' <

~~kx ~~j'k'x

~ ~~~

hh'jj'kk'x

with (k( a constant

value, depending

on the energy of the

incoming

electrons and the energy loss e.

If the screen is

positioned

in the focal

plane

of the lens there is a one-tc-one

correspondence

between trie wave vector k and the

position

vector on the screen x. Therefore terms with

k

#

k' do not survive [7]. So trie CB spectrum becomes

Î(E)

"

~

tlhjtlh'j' <

Sljk ~~j'k

~

"

~

ÎXk(E)Î~

(1°)

hh,jj,k k

xk(E)

#

£

tlhj

S~k

>,

(11)

hj

(11)

with the electron screen in the focal

plane;

if the electron lens is removed and the electrons hit the wall of the vacuum chamber far away we expect the same result.

Equation (10)

means

that for each value of k the

probability amplitudes

for electrons

coming

from ail atoms must be added and alter that the

intensity

must be calculated. The CB

intensity

is the sum of the

intensities for ail k vectors.

If the screen is

positioned

in the

image plane

and if the lens makes an

image

of the

crystal

with atomic resolution, electrons

coming

from different atomic rows hit the screen at different

positions

x. Therefore there is a one-tc-one

correspondence

between the row index h and the position index x and terms with h

#

h' do not survive in the summation. So the CB spectrum

during good imagmg

of the electrons is

Î(E)

"

£

tlhjtlhj' <

~~k ~~'k'

~

~ ~~~J ~~k ~Î~

~~~~

hjj'kk' ~ J~

Equation (12)

means that for eacu row tue

probability amplitudes

for electrons coming from different atoms wituin the row

going

in different directions k must be added and after that the

intensity

must be calculated. The CB intensity is the sum of the intensities for ail rows.

An alternative way to calculate the CB spectrum under the

focusing

condition is

using Huygens' principle

for

calculating

the intensity distribution in the

image plane

out of the

amplitude

distribution in the focal

plane, equation (11),

and sum the intensities for the whole

image plane.

If there is no difference between

(10)

and

(12)

,

proposition

2 in Section 1.2 is true. If there is a

difference, proposition

1 may be true, and we are able to influence the CB spectrum with

manipulations

on the

outgoing

electrons at a

large

distance from the

crystal,

an

example

of a

delayed-choice experiment

[9]

2.2. CALCULATIONS. From the

equations (10), (11),

and

(12)

we learn that the coherent

bremsstrahlung yield

of

photons

with energy E

depends

on

£~~ ahj(S(~

>. This is the

prob- ability amplitude

that the electron screen is hit

by

an electron with energy Ebeam E that

was scattered from an atom at the intersection of row h and

plane j

in the direction

k/

k (, summed over all atoms h, j. An

elegant

way to perform such a calculation is using the

analogy

between

optics

and wave mechanics [Si. This

analogy

is an

example

of

Feynman's path integral

method

[loi.

At a number of

photon energies

E the

following

calculations were done. For each atom and each

scattering

direction a

phase

p is

calculated,

see

Figure

4,

P=alk>1+blkfl> (13)

in which it is assumed that the

phase

shift in the melastic scattering process is

independent

of the direction of the

outgoing

wave; k; and kf = k are the wave vectors of the incoming and

the

outgoing electron, respectively.

For each spot x on the focal

plane, corresponding

to wave vector k, the total

probability amplitude

is calculated:

Xx(E)

"

£

e~~~~,

(14)

hj

which is

equivalent

to

equation (11).

The

intensity

of CB

photons

with energy E and with the electron screen at the focal

plane

is

proportional

to

I(E)

=

L lxx(E)l~ (15)

(12)

ç lrow

h

F

a l' atom

plane j

b

1 F' focal plane

1,

Fig. 4. Definition of optical paths, not to scale.

with the summation extended over all x values with

scattering angle

8 smaller than a critical

angle 8~r.

Huygens' principle (Refs. [Il], Eq. (3.31)

and

[12], Eq. (1.12))

is used to calculate the

intensity

distribution in the

image plane

from the

probability amplitude

distribution in the focal

plane, equation (14).

The calculations of

equations (14)

and

(15)

must be made for ail directions of kf where the

single-atom scattering amplitude

is

large.

The scattenng

amplitude

is taken as a step function:

1 for

scattering angle

8 < 8~r and 0 for 8 > 8~r.

Sommerfeld [13] gave the

angular

distribution for electrons that lost some of their kinetic energy in the

bremsstrahlung

process for the non-relativistic case. The intensity is maximal at 8

= 0 and is reduced to half maximum at Hi/2 "

((k;( (kf()/(k;(.

For 160 kev electrons

losing

3 kev

Ri

/2

" 0.5°. This is the sonne order of

magnitude

as the deflection

angle

of the

first order

Bragg

diffracted

peak,

0.85° for the situation descnbed below. Sommerfeld made

use of the pure Coulomb field of the atomic nudei, without

taking

mto account the

screening

effect of the

surrounding

electrons.

Koch and Motz [14] made a

compilation

of

bremsstrahlung formulae, taking relativity

into ac-

count. From their formula

IBS, using

the Born

approximation

and the Thomas-Fermi

screening function,

the

angular

distribution of

inelastically

scattered electrons can be

calculated, Fig-

ure 5. The much

larger

value of

Hi

/2, 8°, is caused

by

the

screening effect,

which restricts the domain of

impact

parameters to smaller values.

Calculations were made for 160 kev electrons

bombarding

two silicon

crystals

with the

il11]

axis

along

the electron beam, a thin

crystal,

2 x 2 x 4 nm,

containing

840 atoms on 105 different atomic rows, or a thick

crystal,

2 x 2 x 40 nm,

containing

8400 atoms on 105 atomic rows.

(13)

3

g

g

~ 2 ai

_oc

à Î

w

Ô

0 10 20 30

inefastfc scattering angle (degrees)

Fig. 5. Dilferential cross section

d~a/dedQphatonsin8d8

as a function of the scattering angle

8 of trie electrons for inelastically scattered electrons of 160 kev on a silicon atorn producing a

brernsstrahlung photon of 3 kev that is ernitted in a direction perpendicular to the electron. bearn.

3. Calculation of CB

Spectra

with

Boundary Angle

8~r = 2.5°

In this section it is assurned that inelastic

scattering

can occur at

angles

greater than the small- est

Bragg

reflection

angles.

So for each Laue zone several

reciprocal

lattice

points

contribute to the coherent

bremsstrahlung, Figure

ld.

3,1. THE CB SPECTRUM WITH THE ELECTRON SCREEN AT THE FOCAL PLANE.

Figure

6

gives

the calculated CB spectrum

according

to

equation (15)

for a thin

crystal.

The calculated spectrum has the same features as measured spectra,

Figure

2 and [2]:

peaks

2 and 6 are

absent, peaks

4 and 8 are more intense than their

neighbours.

A CB spectrum calculated from

just

one atomic row,

containing

8 atoms, has the same

shape

as

Figure

5 and its

mtensity

is a factor of105 smaller. So the calculated spectrum, as

well as the

expenmental

spectra, must be

interpreted

as

intensity

summations of spectra from separate atomic rows; mterference terms from different rows are absent in the

wavefunction,

so for

large boundary angle

8~r

proposition

2 of Section 1.2 is true.

A peak in the CB spectrum is the result of contributions from a number of

reciprocal

lattice

points

from the same Laue zone. The energy loss is smaller if the

angle

between k; and

kf

increases, second term in

(3),

see also

Figure

1d. In order to calculate the finestructure in a Laue

peak

the thickness of the

crystal

must be increased.

Figure

7 shows the result for the

peak

at 3.3 kev. It is this g vector

broadening

that determines the width of the

complete peak Ils]

3.2. THE CB SPECTRUM wiTH THE ELECTRON SCREEN AT THE IMAGE PLANE. The

intensity

distribution in the

image plane

is calculated from the

complex amplitude

distribution in the focal

plane by application

of

Huygens' principle.

The calculated

intensity

distribution

gives

a

good image

of the atomic rows in the

crystal.

The

intensity

of CB

photons

is

equal

to the sum of the

intensity

in the

image plane.

For

all cases that were calculated the result was a

complete

absence of any difference between the

yield

of CB

photons

with the electron screen at the focal

plane (which

is

equivalent

to the absence of the

lens)

and the screen at the

image plane.

(14)

g

~ g

é

é 1

0 2 3 4

kev)

Fig. 6. -

a energy loss = CB energy. Silicon [Ill] crystal 4 nm thick.

The

peaks

are arked by nurnbers the Laue 20ne frorn which hey

originate. Large

angular

distribution

of elastically scattered electrons: Hcr = 2.5°.

4

3.0

energy foss (kev)

Fig. 7. -

crystal 40 nm.

3.3. DiscussioN.

- The result

the coherent spectrum from the way the ectrons

are

detected.

Whether

the outgomg

electrons

are focused or not is not of any relevance

roposition 1 of

Section 1.2 is not

true. Nothing pecial happens when the outgoing

electrons

are

The reason that here is no

effect is quite

obvious. The wavefunction does

contain interference terms between

probability amplitudes from

different

rows.

tomic rows

can be imagèd very

well with

inelastically scattered electrons, but a

row, that

would

destroy the

mterference,

makes

no

difference

for

the result.

The absence of

any

effect

of

the

screen position on the photon

Références

Documents relatifs

Characterization of insect pheromones might have dominated the early days of the Journal of Chemical Ecology, but exciting contributions from the multifaceted discipline of

While heavy subjects can be ended by a boundary tone or not (the realization of such a tone depending in this case on the speech rate and on the NP’s

We revisit the evidence of the existence of a long -run link between financial intermediation and economic growth, by testing of cointegration between the growth rate of real

Drawing inspiration from the Belgian Constitution of 1831, Art. 37 was last modified by the 1956 constitutional amendment. During the revision, Art. 37 was at the centre of a debate

It is shown that many characteristic features of the relaxation phenomena can be well explained qualitatively by this simple model if one assumes an appropriate

We will however show that both notions are orthogonal issues, as interference occurs as soon as actions of u and observations of v “coincide” (which is not necessarily a

Fractal dimension is used to measure the spatial arrangement (morphology) of built-up areas within a Nuts1 region (Wallonia, Belgium) and, more particularly, to test to what

We revisit the evidence of the existence of a long-run link between financial intermediation and economic growth, by testing of cointegration between the growth rate of real GDP,