You are on page 1of 19

UCLA Neptune CTR Interferometer

R. Joel England
UCLA Particle Beam Physics Lab
June 2003

‡ I. Introduction
The purpose of this document is to outline the basic operating theory behind the coherent transition radiation
(CTR) interferometry technique employed at the UCLA Neptune Laboratory for diagnosing the bunch length of
the electron beam produced by the Neptune 1.6-cell gun and PWT accelerating structure. General theoretical
results are presented and then applied specifically to the Neptune interferometer developed by Uwe Happek,
which we refer to henceforth as the Uwefarometer. Particular attention is paid to understanding the theory of the
wire-grid polarizers which are used in the Uwefarometer to split and recombine the terahertz CTR radiation.

‡ II. CTR Autocorrelation Theory

ü 1. Equations for CTR at a Conducting Boundary

We consider the coherent transition radiation emitted by an electron beam (a collection of electrons) crossing a
foil. Let the phase space distribution F of the beam be written as the product of a number density and a velocity
distribution:

FHr, vL = nHrL f HvL (1)


`
The spectrum of coherent transition energy radiated into direction k with frequency w is then defined by

IIw, kM = ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ
` dU
(2)
dw dW
The explicit formula is written

IIw, kM = 9N + ° cHk L• ° nè Iw, kM • = Ise Iw, kM ; HCTR Energy SpectrumL


` ` 2 ` 2 `
(3)

where N is the number of particles, nè is the Fourier transform of the beam density,

nè Iw, k M = ‡ nHrL ‰-Â kÿr „3 r ;


` w `
k = ÅÅÅÅÅÅÅÅ k (4)
c
c is the divergence factor,

k µ Ik̀ µ vM
`
cHkL = ‡ ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ
` ÅÅÅÅÅÅÅÅÅÅÅÅÅÅ f HvL „ v
`
° k µv •
3
(5)

and Ise Iw, k M is the single-electron spectrum, which for a conducting boundary in the ultrarelativistic limit takes
`

the form
UCLA Neptune CTR Interferometer 2

Ise Hw, qL @ ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ


e2 sin2 q
4 p c H1 - b cos qL
1
2
ÅÅÅÅÅÅÅÅÅ ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ2ÅÅÅÅ ; (6)

Once the single particle spectrum is calculated for the geometry considered, obtaining the CTR spectrum for a

the frequency dependence has vanished and so IIw, k M is a constant in frequency space. Comparing the wave-
beam of such particles requires only a calculation of the Fourier transform of the beam density. Note that in (6)
`

length to the bunch length produces two regimes of interest, in which the divergence factor takes the limiting
forms,

cHkL Ø 9
` 0 l < sz
(7)
1 l > sz
and we obtain from (3) the two corresponding forms for the radiation spectrum:

IIw, kM Ø 9
` l < sz
HN + † nè §2 L I
N Ise OTR
(8)
se l > sz OTR + CTR
In the limit of ordinary transition radiation (OTR) the radiation is simply proportional to the single particle
spectrum, which reveals no features of the beam distribution. In the regime l > sz , the radiation is a superposi-
tion of OTR and a coherent spectrum (CTR) which is modulated by the square modulus of the Fourier transform
of the beam density nè . This spectrum is coherent because the frequency content of a bunched beam distribution is
peaked about the wavelength corresponding to the length of the bunch.

ü 2. Standard Bunch Length Measurement by Autocorrelation

The measurement of bunch length using coherent transition radiation is accomplished by the use of a michelson
interferometer. This device splits the pulse of CTR radiation produced by the beam at a foil with a beam splitter.
The two split pulses are then reflected from mirrors and recombined before entering a detector (such as a Golay
cell). One of the mirrors is movable so that the relative path-lengths of the two pulses are adjustable. If the
path-lengths of the two legs of the interferometer are equal, then the two pulses arrive at the detector at the same
time and their electric fields are constructively superimposed. If the path-lengths of the two legs differ by an
amount greater than the bunch length, then the two pulses arrive at different times and their fields do not construc-
tively interfere. Instead of their fields adding, it is their intensities which add, and so the total intensity measured
is reduced by half. Consequently, by measuring intensity in the detector as a function of the path-length differ-
ence between the two legs, an inteferogram is produced which effectively measures the bunch length of the beam.
UCLA Neptune CTR Interferometer 3

Figure 1
Let R and T be the reflection and transmission coefficients of the beam-splitter and assume the mirrors to be
perfect reflectors. Then the time-integrated intensity of radiation seen at the detector is given by

IHtL = ‡ † T R 8EHtL + EHt + tL< §2 „ t



(9)

where t = d ê c is the time-delay due to a path-length difference d between the two legs of the inteferometer and

E(t) is the electric field of the original pulse at the location of the beam-splitter. After a bit of manipulation,
equation (10) may be cast into the form

IHtL = SHtL + I¶ (10)


where

SHtL ª 2 † T R §2 Re ‡ EHtL E* Ht + tL „ t ;

I¶ ª 2 † T R § 2 ‡ † EHtL §2 „ t ;
¶ (11)

We note that S(t) is the autocorrelation function of E(t) with itself. In the limit t Ø ¶, the field E(t+t) goes to
zero (since it is a pulse of finite duration) and so SHt Ø ¶L = 0. Consequently, I¶ represents the baseline limit
where the two split pulses do not overlap. In the opposite limit where t Ø 0, there is no delay and hence
SHt Ø 0L = I¶ . These results imply

IHtL = I¶ µ 9
2 tØ0
(12)
1 tض
There is therefore a 2-fold increase in intensity at the detector when the two legs of the interferometer are made
equal and the peaks of the two pulses overlap. However, the relationship between the full-width half-max
(FWHM) of the autocorrelation function and the bunch length  depends upon the shape of the pulse:
UCLA Neptune CTR Interferometer 4

ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ = 9
 1 rectangular
(13)
FWHM 0.75 gaussian
So the interpretation of the experimental results is dependent upon having some expectation about the shape of the
pulse.

ü 3. The UCLA Uwefarometer

Optical Description of the Apparatus

The CTR Autocorrelation scheme utilized at Neptune (pictured below) is a polarizing Michelson interferometer
built by Uwe Happek at the University of Georgia. For this reason it has been dubbed the Uwefarometer.

è!!!
Figure 2
Define x` and y` to be horizontal and vertical unit vectors at a given polarizer. Then, let e` ± = Hx̀ ± y` L ë 2 be the
polarization vectors at 45˚ to these. The first beam splitter is also a 45˚ polarizer, so if E0 is the electric field
amplitude at {1} then the electric fields reflected and transmitted at the first splitter are

è!!!ÅÅÅÅÅ e+ ; E3 = ÅÅÅÅÅÅÅÅ
è!!!ÅÅÅÅÅ e-
E0 ` E0 `
E2 = ÅÅÅÅÅÅÅÅ (14)
2 2
where we have suppressed the ‰Â w t time dependence. Now, the field E3, f hits the 90˚ polarizer which reflects and
transmits the forward-going fields

E0 ` E0 `
E4, f = ÅÅÅÅÅÅÅÅÅÅ x ; E5, f = - ÅÅÅÅÅÅÅÅÅÅ y (15)
2 2
Upon reflection from the mirrors, the beam on path {5} gains a phase delay of t due to the path-difference pro-
duced by the translator. Hence, the reflected fields are

E0 ` E0 `
E4,r = ÅÅÅÅÅÅÅÅÅÅ x ; E5,r = - ÅÅÅÅÅÅÅÅÅÅ y ‰Â w t (16)
2 2
UCLA Neptune CTR Interferometer 5

Now, the two fields recombine at the 45˚ polarizer,

E6 = HE4,r + E5,r L = ÅÅÅÅÅÅÅÅÅÅ Hx̀ - y ‰Â w t L


E0 ` (17)
2
Only the e` + component of this field is reflected into the detector. So, we have that the field amplitude at the
detector is (replacing the suppressed harmonic time-dependence)

E7 = HE6 ÿ e+ L = ÅÅÅÅÅÅÅÅ
ÅÅÅÅÅÅÅÅÅÅ H1 - ‰Â w t L ‰Â w t
è!!!
` E0
(18)
2 2
Calculation of the Autocorrelation Signal

From Eq. (18) we see that the signal at the detector is the original electric field amplitude minus its t-delayed

tor. Consequently, if EHtL = E0 HtL ‰Â w t , then we can rewrite (18) as


counterpart. This differs from the standard method outlined previously, where the two fields added at the detec-

ÅÅÅÅÅÅÅÅÅÅ @EHtL - EHt + tLD


è!!!
1
Edet = ÅÅÅÅÅÅÅÅ (19)
2 2
The time-integrated intensity in the detector is then

IHtL = ÅÅÅÅÅ ‡ † EHtL - EHt + tL §2 „ t



1
(20)
8 -¶
Consequently, instead of Eq. (10), we have

IHtL = I¶ - SHtL (21)


where S(t) is the autocorrelation function and I¶ is the baseline:

SHtL ª ÅÅÅÅÅ Re ‡ EHtL E* Ht + tL „ t ;



1
4 -¶

I¶ ª ÅÅÅÅÅ ‡ † EHtL §2 „ t ;
¶ (22)
1
4 -¶
Hence,

IHtL = 9
0 tØ 0
(23)
I¶ t Ø ¶
Now, instead of the fields adding at the detector when the two pulses overlap, they subtract. When the two pulses
are completely overlapped, the detector signal drops to zero. Note that the intensity at the reference detector is

Iref = ‡ † E2 §2 „ t = ÅÅÅÅÅ ‡ †EHtL§2 „ t = 2 I¶


¶ ¶
1
(24)
-¶ 2 -¶
Therefore, if the autocorrelation signal is normalized to the reference, then we have

ÅÅÅÅÅÅÅÅÅÅÅÅÅ = 9 1
IHtL 0 tØ 0
(25)
Iref ÅÅÅÅ2 t Ø ¶
This normalized signal is what is plotted by the CTR analysis VI at Neptune. The reason for this is to compensate
for changes in the signal intensity due to charge fluctuations. Consequently, the normalized signal should
(ideally) resemble the curve shown below:
UCLA Neptune CTR Interferometer 6

Idealized Uwefarometer Signal

0.5

IêIref

0
t

ü 4. Fitting Function for the Autocorrelation Curve

The shape of the autocorrelation curve is determined by the longitudinal density profile of the beam. For a
gaussian beam with an rms of st , for example, the results of the preceding section give

ÅÅÅÅÅÅÅÅÅÅÅÅÅ = ÅÅÅÅÅ J1 - ‰- ÅÅÅÅÅÅÅÅ N;


IHtL 1 t
ÅÅÅÅÅÅÅ
2

4s t
2
(26)
Iref 2

è!!!
st = ssig ë 2 . This analysis, however, fails to account for the finite bandwidth of the apparatus, which tends to
For a gaussian beam then the temporal rms of the beam st is related to the rms variance of the signal ssig by

filter out low frequency content and distort the autocorrelation curve, due to diffraction and the limited aperture of
the device. This effect is discussed in [A. Murokh, Nucl. Instr. Meth. Phys. Res. A, 410, 452 (1998)]. Assuming
a gaussian filter function with a cutoff at wc = x-1 , gives us for a gaussian beam distribution a triple-gaussian form
for the detector signal:

1 i ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ Å y
ÅÅÅÅÅÅÅÅÅÅÅÅÅ = ÅÅÅÅÅ jjjj1 - ‰ 4 s + ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ
è!!!!!!!!! !!!!!!!!
ÅÅÅÅÅÅÅÅ!ÅÅ ‰ 4 Hs +x L - ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ è!!!!!!!!!!!!!!!! ÅÅÅÅÅÅÅ ‰ 4 Hs +2 x L zzzz;
!!!!!
IHtL 2 st t 2
st t 2

2 k {
t 2
- ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ - ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ
ÅÅÅÅÅÅÅÅÅÅÅÅÅÅ Å - ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ
t
2
ÅÅÅÅÅÅÅÅ t
2 2 2
ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ t
2 2 2
(27)
Iref st 2 + x 2 st 2 + 2 x 2
In accordance with this analysis, a nonlinear fit should be done to the normalized autocorrelation signal plotted as
function of delay time t with st and x as the fitting parameters and (27) as the fitting function.

‡ III. Theory of Wire Grid Polarizers

ü 1. General Considerations and Coordinate Systems

We consider the transmission and reflection of an incident electromagnetic plane wave at a grid of parallel wires
of radius a and spacing d. Upon this polarizing grid we may impose a Cartesian coordinate system {x,y,z} such
that z points along the direction of the wires, y is normal to the plane of the grid, and x lies in the plane of the grid.
This is the coordinate system adopted by [Chambers, et al., J. Phys. D: Appl. Phys., 13, 515 (1980)]. The
coordinate system {x,y,z} has also the usual corresponding spherical {r,q,f} and cylindrical {r,f,z} representa-

ity and have magnitude k = 2 p ê l where l is the wavelength of the plane wave. This wave-vector may be decom-
tions. Let k be the wave-vector of a plane wave incident upon this grid and permit it to have arbitrary directional-

posed into the components

` `
k = k r r + kz z ; k r = k sin q ; kz = k cos q (28)
UCLA Neptune CTR Interferometer 7

The projection k r of k onto the x-y plane is shown in the figure below.

x z

" k!

Figure 3
Let the angle a denote the incidence angle of the incoming radiation wave-vector k, and b the 'apparent' angle of
the wires relative to the vertical as seen from the direction of k. The spherical polar angles of the Cartesian
coordinate system (q, f) are related to a and b by

sin q = cos a H1 - sin2 a cos2 bL


-1ê2
; (29)
cos f = sin a cos b ;
`
If we say that the direction of z when b = 0 defines the 'vertical' orientation of the grid, then the rotation angle x of
the z-axis relative to this direction in the plane of the grid is given by

sec a
tan x = ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ (30)
tan b
Consequently, if the radiation is incident upon the grid at a = p ê 4 then the angle b is equal to x only when
x=0,p/2. At intermediate values x and b will be slightly different. This is shown in the graph below:

b HsolidL, x HdashedL
1.5
1.25
1
0.75
0.5
0.25
0
0 p p p
ÅÅÅÅ ÅÅÅÅ ÅÅÅÅ
6 3 2
x
UCLA Neptune CTR Interferometer 8

Note also that the angles a and g = f - p ê 2 are equal only when k has no z-component (i.e. q = p ê 2). The angle
`
a is the incidence angle measured in the plane of k and y. The angle g is the incidence angle of the projected
`
vector k r r in the x-y plane.

ü 2. The Cutoff Frequencies for the Uwefarometer

The Bragg condition for the wire grid for arbitrary incidence reads

Jkx + ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ N = k r 2
2pn 2
(31)
d
Noting that k x = k sin q cos f and k r = k sin q we have that the lowest (n = 1) critical wavenumber kc is given by

2pêd
sin q H1 + cos fL
kc = ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ
ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ (32)

rotation angle of Polarizer 1 and let x - p ê 4 be the rotation angle of Polarizer 2. The incidence angle is assumed
Let us consider the two polarizers of the Uwefarometer, which have a relative rotation of p/4. Then, let x be the

fixed at a = p ê 6. Below, we plot the resulting cutoffs for the two polarizers as a function of x.

wc1 Hps-1 L @solidD , wc2 Hps-1 L @dashedD


25

20

15

10

0 p p
0 ÅÅÅÅ ÅÅÅÅ 3p p
4 2 ÅÅÅÅÅÅÅÅ
4
xHradL

ü 3. Reflection and Transmission Coefficients of a Wire-Grid Polarizer

The Uwefarmoter uses a pair of wire-grid polarizers to split the CTR pulse and recombine it. We wish to consider
the reflection and transmission properties of such a polarizer. We can write the complex reflection and transmis-
sion amplitudes as

r = r ‰Â yr ; t = t ‰Â yt (33)
For the explicit formulas for these quantities, we will rely upon [Edward, J. Appl. Phys. 23, 605 (1952)] wherein

k < p ê d for the case where the polarization is parallel to the wires. The results are functions of k and a when the
the magnitudes and phases of the reflectivity and transmission amplitudes are found in the limits a << d and

plane of incidence is orthogonal to the plane of the grid. To generalize to arbitrary k, we make the substitutions
k Ø k r and a Ø g, noting that k r and g are the corresponding magnitude and incidence angle of the wave-vector
in its projection upon the orthogonal x-y plane.
UCLA Neptune CTR Interferometer 9

ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ ; cos yr = - r H1 + Ri 'L


"################################
H1 + Ri 'L + HX p '######## + X#i###### 'L2#
1
rHkL = ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ
ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ
2

"####################################
è!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!! r 1 - Hr2 H1 + Ri 'LL
(34)
2

1 - r2 H1 + Ri 'L
tHkL = 1 - r2 - 2 r2 Ri ' ; tan yt = ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ
ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ
ÅÅÅÅÅÅÅÅ
Å ÅÅÅÅÅÅÅ
ÅÅÅÅÅ

Here Ri , Xi , and X p are real and imaginary parts of the complex impedances

Zi = Ri + Â Xi ; Z p = 1 + Â X p (35)
where Zi is the resistive impedance of a single wire and Zp is the grid impedance in the limit of infinite conductiv-
ity. Primes denote multiplication by 60 p d cos g. These quantities may be obtained from the formulas

Ber HqL Bei' HqL - BeiHqL Ber ' HqL + ÂBerHqL Ber ' HqL + Â BeiHqL Bei ' HqL
è!!!ÅÅÅÅÅÅÅ 9 ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ =
HBer ' qL2 + HBei ' qL2
Rs
Zi = ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ
ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ
ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ
ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ
pa 2

X p ' = -k r d sin f 9‚ cosHn d k r cos fL Y0 Hn k r dL + ÅÅÅÅÅÅ ln H0.890 k r aL=


¶ (36)
1
p

where q = aHw m sL1ê2 and Rs = Hw m ê 2 sL1ê2 , Y0 is the zeroth order Bessel function of the second kind, and
n=1

`
the wires. If we neglect the resistivity of the wires HZi Ø 0L, the reflection and transmission coefficients become
k r = k ÿ r = k sin q is the magnitude of the component of the incident wave-vector normal to the direction (z) of

"################
1 + HX p 'L####2#
1
rHkL = ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ ; cos yr = - r

è!!!!!!!!!!!!!
(37)
è!!!!!!!! !!!!!ÅÅÅÅ
r
tHkL = 1 - r2 ; tan yt = ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ
ÅÅÅÅÅÅÅÅ
1 - r2

(expressed via f and q) are required. Let us take the example where the grid is vertical Hb Ø 0L. In this limit, (29)
Note that these quantites are dependent upon the vector quantity k, meaning both its magnitude k and direction

yields

sin q = 1 ï q = p ê 2
(38)
cos f = sin a
Plotting r as a function of frequency w for fixed a = 10 mm, d = 100 mm, with q = p ê 2 and several values of a,
we obtain
UCLA Neptune CTR Interferometer 10

p p
a= 0, ÅÅÅÅÅÅ , ÅÅÅÅÅÅ
8 4
1

0.8

0.6
r
0.4

0.2

w Hps-1 L
0 2.5 5 7.5 10 12.5 15 17.5

Under these conditions, the critical wavelength wc = c kc from equation (32) becomes

d H1 + sin aL
2pc
wc = ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ (39)

For the present value of d and the values of a in the graph above, we have

a wc Hps-1 L
------ -------

pê8
0 18.83

pê4
13.61
11.03
These match the cutoffs observed in the graph.

ü 4. Reflection and Transmission Coefficients for the Uwefarometer

The various angles for each of the two polarizers of the Uwefarometer are given below. Note that in spite of the
appearances in the cartoon drawing above, the angle of incidence is not p/4 but rather p/6.

Angle Polarizer 1 Polarizer 2

pê6 pê6
----- ------- -------

pê4
a
x 0

pê2
b 0.714 0

pê3
q 1.209
f 1.183
wc 14.6 ps-1 12.5 ps-1
Plotting the magnitudes and phases of the reflection and transmission amplitudes for both polarizers we obtain the
graphs below.
UCLA Neptune CTR Interferometer 11

Polarizer 1: rHsolidL,tHdashedL Polarizer 2: rHsolidL,tHdashedL


1 1
0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.2

w Hps-1 L w Hps-1 L
2.5 5 7.5 1012.51517.520 2.5 5 7.5 1012.51517.520

Polarizer 1: yr HsolidL,yt HdashedL Polarizer 2: yr HsolidL,yt HdashedL


p p

p p
ÅÅÅÅ ÅÅÅÅ
2 2

0 0
w Hps-1 L w Hps-1 L
0 5 10 15 20 0 5 10 15 20

ü 5. The Matrix Notation

The polarization state may be propagated through an optical system using a matrix formalism.

Let us define a coordinate system (x,y) transverse to the plane of the optical component being considered. Let 1
and 2 denote the regions before and after the device and let subscript r and t distinguish reflected and transmitted
waves. The incoming, outgoing, and reflected electric fields may then be written.

Ei = E0 Hx̀ + yL ;
`
Et = E0 Ht1 x + t2 yL ;
` `
Er = E0 Hr1 x + r2 yL ;
(40)
` `

where t1,2 and r1,2 are the transmission and reflection coefficients for (x,y) polarization. The S-matrices for this
device may therefore be written

St = J N ; Sr = J N
t1 0 r1 0
(41)
0 t2 0 r2
We can then write that for transmission and reflection respectively,

Et = St Ei ; Er = Sr Ei (42)
Suppose for example that t2 = r1 = 1 and t1 = r2 = 0. Then,

St = J N ; Sr = J N ï Et = E0 y ; Er = E0 x
0 0 1 0 ` `
(43)
0 1 0 0
That is, the optic is a perfect polarizer, allowing only the y-polarized light to transmit. If this polarizer is rotated
by an angle g relative to the coordinate axes (x,y), then the new S-matrix for either transmission or reflection may
be obtained from the old one by permitting a unitary rotation matrix UHgL to act upon it:
UCLA Neptune CTR Interferometer 12

S Ø UHgLT S UHgL ; UHgL = J N


cos g -sin g
(44)
sin g cos g
For example, if g = -p ê 4 then

ij ÅÅÅÅ12 - ÅÅÅÅ12 yz i ÅÅÅÅ1 ÅÅÅÅ12 yz


St Ø jjj 1 zz ; Sr Ø jjj 2 zz
z j 1 z
k - ÅÅÅÅ2 { k ÅÅÅÅ2 ÅÅÅÅ12 {
(45)
ÅÅÅÅ12
Hence, the reflected and transmitted fields for a rotated polarizer are

Er = ÅÅÅÅÅÅÅÅÅÅ Hx̀ + yL = ÅÅÅÅÅÅÅÅ


è!!! ÅÅÅÅÅ e+ ; Et = ÅÅÅÅÅÅÅÅÅÅ Hx̀ - yL = ÅÅÅÅÅÅÅÅ
è!!!
E0 ` E0 ` E0 ` E0 `
ÅÅÅÅÅ e- (46)
` è!!!
2 2
where e± = Hx̀ ± yL ë 2 are orthogonal unit vectors at right angles to the original (x,y) axes. Note that (43) are
2 2
`
the same results obtained in equation (14) for transmission through the rotated polarizer.

‡ IV. Applications to the Neptune CTR Interferometer (Uwefarometer)

ü 1. Transmission Efficiency of the Uwefarometer: Ideal Polarizers

Let us do the rest of the analysis for the Uwefarometer using the matrix notation. We can split the path from the
source to detector into the following steps:

E0 Ø St Ø E3 Ø N Ø E6 Ø Sr Ø E7 (47)
t
Here, S corresponds to transmission through the 45˚ polarizer, N is the combined action of the 90˚ polarizer and
two mirrors, and Sr corresponds to reflection from the 45˚ polarizer (this time coming from the opposite direc-
tion). Let us operate in the basis of e` ± . Then,

St = J N; Sr = J N ; e+ = J N ; e- = J N
0 0 1 0 ` 1 ` 0
(48)
0 1 0 0 0 1
If N was not rotated by p/4 relative to the chosen basis, then its transfer matrix would be M = St St ‰Â d + Sr Sr .
Including the rotation U by -p/4, then it reads

N = U T M U ; M = HSt St ‰Â d + Sr Sr L (49)
Now, we have that the total transport matrix is

1 i 1 - ‰Â d 0 zy
Stot = Sr N St = ÅÅÅÅÅ jj z
2 k 0 0{
(50)
è!!!
Hence, with E0 = e- E0 ë 2 , we obtain
`

ÅÅÅÅÅÅÅÅÅÅ H1 - ‰Â d L e+
è!!!
E0 `
E7 = Stot E0 = ÅÅÅÅÅÅÅÅ (51)
2 2
This is the same result we found in equation (18).

ü 2. Transmission Efficiency of the Uwefarometer: Non-Ideal Polarizers

We consider now the case where the polarizers of the inteferometer are not ideal but have transmission and
reflection amplitudes that are functions of frequency. Now we have that the transport looks like
UCLA Neptune CTR Interferometer 13

E0 Ø P Ø E3 Ø N Ø E6 Ø A Ø E7 (52)
where P and A are the transport matrices of the first polarizer for transmission and reflection respectively,

P=J N ; A=J N
p1 0 a1 0
(53)
0 p2 0 a2
and N is given as before by (46) but in accordance with (38) we use the more general forms

St = J N ; Sr = J N
t1 0 r1 0
(54)
0 t2 0 r2
We then have that the unrotated matrix for the second polarizer and mirror combination is

i r21 + ‰Â d t12 yz
M = jj z
k {
0
(55)
0 r2 + ‰Â d t22
2

The total transfer matrix is then

Stot = Sr HU T M UL St = ÅÅÅÅÅ J N
1 a1 p1 Q+ a1 p2 Q-
(56)
2 a2 p1 Q- a2 p2 Q+
Where we have defined

Q± ª Hr21 + ‰Â d t12 L ± Hr22 + ‰Â d t22 L (57)


The so-called response function is then given by

R = ÅÅÅÅÅ ‚ †HStot Lij §2 = ÅÅÅÅÅ 8HA1 P1 + A2 P2 L †Q+ §2 + HA2 P1 + A1 P2 L †Q- §2 <


1 1
2 i, j 8 (58)

where Ai ª †ai §2 , Pi ª †pi §2 . Note that

†Q± §2 = Q0,± + Q1,± ‰Â d + Q1,± ‰- d


êêê
(59)
where

Q0,± ª H†r1 §4 + †r2 §4 + †t1 §4 + †t2 §4 L ± I r22 rê 21 + r21 rê 22 + t22 t 1 + t12 t 2 M ;


ê2 ê2

Q1,± ª Hr1 2 ± r2 2 L Ht1 2 ± t2 2 L;


* (60)

with bars denoting complex conjugate. Consequently, R becomes

êê
R = R0 + R1 ‰Â d + R1 ‰- d (61)
where

R0 ª ÅÅÅÅÅ 8HA1 P1 + A2 P2 L Q0,+ + HA2 P1 + A1 P2 L Q0,- <;


1
8
R1 ª ÅÅÅÅÅ 8HA1 P1 + A2 P2 L Q1,+ + HA2 P1 + A1 P2 L Q1,- < ;
(62)
1
8
Results (58) and (59) are now equivalent to equation (6) of [Chambers, Mok, and Parker]. In the case of ideal
polarizers, we have that P2 = A1 = t2 = r1 = 1 and P1 = A2 = t1 = r2 = 0. Then,

1 1
Q0,± = 2 ; Q1,± = ±1; R0 = ÅÅÅÅÅ ; R1 = - ÅÅÅÅÅ ; (63)
4 8
UCLA Neptune CTR Interferometer 14

and so

Rideal = ÅÅÅÅÅ H2 - ‰Â d - ‰- d L = ÅÅÅÅÅ † 1 - ‰Â d §


1 1 2
(64)
8 8
This matches the result we would obtain by inserting (47) into (55).

ü 3. Transmission Efficiency of the Uwefarometer: With Simplifying Assumptions

Now, let us assume that the two polarizers are identical but not ideal and that transmission and reflection are the
same for each from either the front or the back. Then, p j Ø t j and a j Ø r j . Further assume that t2 = r1 and that
r2 = t1 . In that case,

Polarizer 1 : t2 = r1 ª r ; r2 = t1 ª t ;
(65)
Polarizer 2 : t2 = r1 ª r' ; r2 = t1 ª t ' ;
So, the matrices read

P=J N ; A=J N ; St = J N ; Sr = J N
t 0 r 0 t' 0 r' 0
(66)
0 r 0 t 0 r' 0 t'
The resulting transfer matrix is then

1 i H1 + ‰Â d L r t Hr£ 2 + t £ 2 L H1 - ‰Â d L r2 Hr£ 2 - t£ 2 L yz
Stot = ÅÅÅÅÅ jjjj zz
z
2 k H1 - ‰Â d L t 2 Hr£ 2 - t£ 2 L H1 + ‰Â d L r t Hr£ 2 + t £ 2 L {
(67)

So the response function is

R = ÅÅÅÅÅ 9G- †1 - ‰Â d § + G+ †1 + ‰Â d § =
1 2 2
(68)
8
where

G- = °r£ 2 - t £ 2 • I† r2 § + † t2 § M ;
2 2 2

G+ = 2 °r£ + t£ • † r t§2 ;
(69)
2 2 2

ê
Using the property r t + t rê = 0, it can be shown that

°r£ ° t £ • = Ir£ ± t£ M = 9
1
H2 r
2 2 2 2 2 2
£2 2 (70)
- 1L
Hence,

G- = 2 r4 - 2 r2 + 1
G+ = 2 I2 r£ - 1M Hr2 - r4 L
2 2 (71)

For the same parameters as Figure 4, we plot G± :


UCLA Neptune CTR Interferometer 15

G- HsolidL,G+ HdashedL
1

0.8

0.6

0.4

0.2

w Hps-1 L
2 4 6 8 10 12 14

Note that G- represents the filter function which acts upon the part of the signal response which interests us,
namely the part corresponding to destructive interference. The smaller but nonzero G+ contribution represents a
secondary contribution to the signal due to constructive interference of undesired polarizations which were
transmitted through the system because of the non-ideal transmission and reflection characteristics of the
polarizers.

Now we derive R directly from the results of the previous section. We find that

Q0,± = 2 9 H1 - 2 r L ; Q1,± = 9 H1 - 2 r L ;
£2 2 £2 2
(72)
1 -1
And

P1 = A2 = t2 = 1 - r2 ; P2 = A1 = r2 (73)
Hence,

R0 = ÅÅÅÅÅ 9r4 + Hr2 - 1L - 2 r2 Hr2 - 1L I1 - 2 r£ M =;


1 2 2 2
4
R1 = ÅÅÅÅÅ 9- r4 - Hr2 - 1L - 2 r2 Hr2 - 1L I1 - 2 r£ M = ;
(74)
1 2 2 2
8
Note that these quantities are related to G± plotted above by

R0 = ÅÅÅÅÅ HG+ + G- L ; R1 = ÅÅÅÅÅ HG+ - G- L


1 1
(75)
4 8
We plot R0 and R1 in the graph below.
UCLA Neptune CTR Interferometer 16

8†R1 §HsolidL, 4†R0 §HdashedL


1

0.8

0.6

0.4

0.2

w Hps-1 L
2 4 6 8 10 12 14

note that the above plot is only correct on the far left side where w < p c ê d = 9.4 ps-1 and consequently, the R1
Note that the relevant cutoff frequency is the smaller of the two, which in this case is that of Polarizer 2. Also

function does not vanish at the cutoff. In order to correctly plot the response function out to the cutoff, a more
general theory must be used to calculate the resistivity r.

Let us now consider G± as functions of the overall rotation angle of the two-polarizer combination. In order to
function as an interferometer, we require only that the two polarizers be rotated relative to each other by p/4. As
demonstrated in Section 2, however, the cutoffs and therefore the transmission curves of the system depend upon

in the plane of the grid for Polarizer 1 and let x - p ê 4 be the corresponding angle for Polarizer 2.
the overall orientation of the two-polarizer system. Let x be the angle of the wires from the vertical as measured

G- HsolidL,G+ HdashedL, x=0.00p 8†R1 §HsolidL, 4†R0 §HdashedL, x=0.00p


cpêdw1w2 cpêdw1w2
1 1
0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.2

w Hps-1 L w Hps-1 L
5 10 15 20 5 10 15 20

From the above animation, we see that in the region where the theory is strictly valid Hw < cp ê dL the response
function does not change much as the two polarizers are simultaneously rotated.

ü 4. Generalized Interferogram of the Uwefarometer

The intensity of the CTR radiation emitted at the foil is given by

IHwL = Ise N 2 † f HwL§2 = †EHwL§2 (76)


where f HwL is the Fourier transform of the longitudinal beam distribution function, Ise is the single-particle
spectrum, and E is the Fourier transform of the electric field. We assume therefore that the Electric field ampli-
tude at the entrance to the interferometer has the same frequency structure as the beam:

è!!!!!!
EHwL = E0 f HwL; E0 = N Ise (77)
UCLA Neptune CTR Interferometer 17

If the response function is given by (72)

R = ÅÅÅÅÅ 9G- †1 - ‰Â d § + G+ †1 + ‰Â d § =
1 2 2
(78)
8
we interpret this to mean that the signal at the detector is given by

†E0 §2
IHtL = ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ ‡ 9G- † f HwL H1 - ‰Â w t L§ + G+ † f HwL H1 + ‰Â w t L§ = „ w
2 2
(79)
8
The S-matrix for the reference signal is simply

Sref = J N
r 0
(80)

so, its response function is Rref = 1 ê 2. The reference intensity is therefore still given by
0 t

†E0 §2
Iref = ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ ; (81)
2
The normalized signal is thus

ÅÅÅÅÅÅÅÅÅÅÅÅÅ = ÅÅÅÅÅ ‡ † f HwL§2 8G- H1 - cos@w tDL + G+ H1 + cos@w tDL< „ w


IHtL 1
(82)
Iref 2
If we wish to include the low frequency cutoff imposed by the collecting optics and diffraction effects, we can
make the replacement

f HwL Ø f HwL gHwL (83)


Here gHwL is the low frequency filter, which we can represent by

gHwL = 1 - ‰-x w2
2
(84)
where x-1 is the low-frequency cutoff.

ü 5. Reproduction of Results from Chambers, et al.

As a check on our analysis, we will attempt to reproduce the graph of Figure 4(a) of [Chambers, et al., J. Phys. D:
Appl. Phys., 13, 515 (1980)]. First note that the angles used are as follows:

Angle Polarizer 1 Polarizer 2

pê4 pê4
----- ------- -------

pê4 pê2
a

pê2
x

pê4
b 0.714

pê2
q 1.047
f 0.955
wc 13.8 ps-1 26.6 ps-1
The resulting transmission and reflection amplitudes for 2 a ê d = 0.2 are as follows:
UCLA Neptune CTR Interferometer 18

Polarizer 1: rHsolidL,tHdashedL Polarizer 2: rHsolidL,tHdashedL


1 1
0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.2

w Hps-1 L w Hps-1 L
5 10 15 20 25 5 10 15 20 25

Polarizer 1: yr HsolidL,yt HdashedL Polarizer 2: yr HsolidL,yt HdashedL


p p

p p
ÅÅÅÅ ÅÅÅÅ
2 2

0 0
w Hps-1 L w Hps-1 L
0 5 10 15 20 25 0 5 10 15 20 25

The reconstructed plot of Figure 4(a) is shown below. The dotted curve corresponds to 2 a ê d = 0.2, the full curve
2 a ê d = 0.3, and the dashed curve 2 a ê d = 0.4.

0.8

8†R1 §
0.6

0.4

0.2

w Hps-1 L
2 4 6 8 10 12

The dotted and solid curves appear to match fairly well with plot 4(a) of Chambers. The dashed curve does not

cies. This discrepancy is not unexpected, since our theory for the reflectivity r breaks down as a ê d approaches
match so well, although it does show the qualitative feature of the "dip" followed by a "hump" at higher frequen-

unity. Also note that all three plots have been truncated a bit before the cutoff. Technically, our results are valid
UCLA Neptune CTR Interferometer 19

The dotted and solid curves appear to match fairly well with plot 4(a) of Chambers. The dashed curve does not

cies. This discrepancy is not unexpected, since our theory for the reflectivity r breaks down as a ê d approaches
match so well, although it does show the qualitative feature of the "dip" followed by a "hump" at higher frequen-

only in the regime w < p c ê d and the plots tend to turn upward near the cutoff.
unity. Also note that all three plots have been truncated a bit before the cutoff. Technically, our results are valid

ü 6. Beam Spectrum With High-Frequency Filter

We now consider the effect of the high-frequency filter upon interferogram results using the Uwefarometer.
Below is the longitudinal profile of a beam simulated in ELEGANT and propagated through the Neptune S-Bahn
with sextupoles turned off in the simulation.

Longitudinal Density Profile

0.12
0.1
0.08
lz 0.06
0.04
0.02
0
-4 -2 0 2 4
zHmmL
This profile was chosen because it has a very high frequency content, due to the sharp spike in the center. Below
we plot the corresponding frequency spectrum F(w) with and without multiplication by the low-pass filter function
G- (w) which was obtained in Section 8. The filtered function is shown by a dashed line. Autocorrelating the
simulated frequency spectrum with itself, we obtain the simulated interferograms shown in the right-hand figure,
which are almost identical.

FHsolidL, G- F HdashedL: Sextupoles Off Interferograms: without G- HsolidL, withHdashedL


0.2
0.175 1
0.15
Ha.u.L 0.1
0.125 0.8
SHtL 0.6
0.075 0.4
0.05
0.025 0.2

w Hps-1 L
2 4 6 8 10 12 14 5 10 15 20 25 30
tHpsL

Performing a tri-gaussian fit to the data in accordance with the method of [Murokh, et al] we obtain from the
interferograms an rms sigma of st = 0.59 ps for the case where the high-frequency filter is not included, versus
0.64ps for the case where the filter is included. The removal of higher frequencies narrows the frequency spec-
trum and therefore produces an interferogram which is characteristic of a slightly longer beam. The frequency
filter therefore imposes a lower limit on the longitudinal beam size which can reliably be measured by this tech-
nique. However, in the present case the difference in sigma values is only 8%, indicating that we are not unduly
close to the saturation point.

You might also like