. (fp) stands for the actual deviation (fp) = p -
; the variance 62p is defined as 62p =<(p _
. The
'
{n
(62p)U2.
standard deviation 6 p equals
100
Since expressions with averages (EL> ,
o
I -20
r
-10
I
i-
0
+10
'
I
÷20
+
~
(2)
y(mm)
This can be seen from a series expansion of <(ER Fig. 3. Position resolution and finearity of the position-sensitive anode (aperture spacing 5 mm, diameter 0.5 mm).
EL)/(ER +EL)). Experimentally, (y> correspond to
mm) to the width of the position peaks has been quadratically subtracted (fy~×p = 6Y~'eak -- 3r2) . Two features of the experimental points in fig. 4 should be noted. First, the resolutions 6YExp are clearly dependent on the position y for the deep stripes with d = 60 and 30 mm. Second, the finer the segmentation of the PSA, the better the mean resolution down to 6yEx p = 0.33 mm for d = 10 mm (cf. table 1). However, it gets worse again for d = 5 ram. In order to explain the dependence of the position resolution on the depth d, we assume the resolution to be essentially limited by two factors. Naturally, one contribution is the noise of the electronics; the other is the energy loss straggling across the PSA.
6~y =((iv _
the maxima of the peaks in fig. 3. A further series expansion then yields
1
(~
[(1 -
-- 2(1 --
Quantities of the order higher than or equal to (fiE/
<(fER)(fEL)> -
(4)
6ER f E L
The standard deviations 6ER, fiE[. and the correlation'
1.5
1.5 (~)l (~& ,D
®o
®.
--- l ,
/ I"
1.0
d =60ram = 30ram = 10 mrn = 5ram
I
/
•
7
~
~
1.0
/" /"
\.
/
N
(~ /"
~o
X',,
0.5
3.5
®.--~
0 -30
I
1 -20
I
I -10
I
I 0
I
1 10
I
I 20
i
y (ram)
,
I -20
(3)
,
I -10
i
I 0
,
I 10
I
I 20
n
0 30
Fig. 4. Experimental (left) and theoretical (right) position resolutions 6y (fwhm) for anodes no. 1 to 4.
G. Rosner et al. / A new position measurement
564
coefficient 3' will be calculated separately with respect to energy loss straggling and electronic noise. First, the contribution of the energy loss straggling is estimated. In a simple model (for details see Appendix), we calculate the individual contributions of the energy loss straggling across one stripe. These energy loss stragglings are then mixed by diffusion of the electrons drifting from the particle track to the PSA. The main result of the model is that the diffusion process strongly correlates a left segment of the PSA to neighboring right ones, thus decreasing the energy loss straggling across this segment and giving rise to a large correlation coefficient 3' > 0. Since (y) c [ - 1 , 1], 62y decreases. Actually, 62Ystr ~ 0 , if the mixing of the energy loss stragglings across two neighboring segments is complete. This will happen for sufficiently fine segmentation of the PSA. According to the derivation given in the Appendix, 62Estr
(l+yo 2
o ) d(fo)
~2Estr'L = 62Estr
(1-yo 2
o) d(yo)
62Estr'R =
(5)
"
(6)
Yo denotes the normalized, geometrical position Yo = J'l/bl ~ [ - 1 , 1], d(yo) the actual depth [eq. (A10)]. The rms energy loss straggling ~Est r across the full PSA is estimated [7] to be 62Estr =c'K(E) •
(7)
The proportional factor 9( is estimated [4] to be CK(MeV) ~ 1.563 X 10 4
ZT
+ k2Zp
(8)
with: Zp the nuclear charge number of the projectile (e.g., 32 16S),. ZT, AT the charge and mass number of the target (e.g., '~°Ar); Sp(v) the proton stopping power (MeV cm2/mg); v the projectile velocity (MeV/u); k = kargon = 0.011 MeV 1/2. For the diffusion length in eqs. (5) and (6) we note the common ansatz (9)
It is: x = 9.2 cm the drift length to the PSA; 6 = 17.5 V/cm, the electric field strength; D/IJ = 0.26 V the drift coefficiont D over the electron mobility ]1 from ref. 8 ; consequently, o = 5.2 mm for our chamber. The correlation coefficient 3'str reads =
duo)
-
2 r = ~cK ( 1 - y ~ 6Yst
- ~ 40 -(1
+ y g t a n 2 a o ) -1/2 ) . (11)
According to eq. (11), the limiting case of full mixing, i.e., no contribution to the position resolution from energy loss straggling, holds for d~<(l_yg)(1
40 +y2otan2ao) 1/2 "
(12)
In our experiment, this will happen at Yo = 0 for d ~< 4o = 20.8 ram, for example. Therefore, the position resolutions 6YExp in fig. 4 originate exclusively from electronic noise for PSA no. 3 and 4. Another feature of the data in fig. 4 follows naturally from eq. (11), i.e., the fact that 6YExp is worst in the middle of the PSA (y or Yo ~ 0) for the roughly segmented anodes no. 1 and 2. Concerning the electronic resolution 6Ynoise, eq. (3) becomes 2
6Ynois e =
62En°ise [1 + (y)2 3,noise( 1 _ (y)2)] 2 ~ (E) • (13)
Using. similar preamplifiers on both sides of the PSA, The rms signal-to-noise ratio ( E ) / 6 E n o i s e depends on the preamplifier used, the cable capacitances Cca b ( ~ 6 0 pF), and the detector capacitances (CpsA in table 1). We measured (E)/ 6 E ~ °lse = ~ E ~ °ise -= 6 E n o i s e.
6Enois e ~ 2 0 0 .
A TSp(t))
o =
Inserting eqs. (A10), (5), (6) and (10) into eq. (3) and identifying (y) = Yo results in
(lO)
We define a noise equivalent capacitance CNEC of our preamplifier (CNEc ~ 270 pF) and a capacitance Co = CNFC + Ccab ~ 330 pF. Then, the noise on one side of the PSA is coupled to the other side according to CpSA/(CpsA + Co). Note that this coupling is anticorrelated. Therefore, eq. (4) may be rewritten ")'noise -
CpSA Cps A + C O
(14)
Since 6Enoise/(E) as well as -Tnoise become larger with an increasing number n of stripes in eq. (13), the resolution 6Ynois e gets worse the finer the segmentation of the PSA. Therefore, it will not be reasonable to choose d very much smaller than the limiting value calculated from eq. (12). Finally, it shou!d be noted that the energy loss signals derived from one side of the PSA become smaller because of coupling to the other side. This changes the sensitivity of the energy loss and, consequently, the position measurement. ( y ) n o longer
565
G. Rosner et al. / A new position measurement
ranges from - 1 to 1, since E R and E L in the generally valid eq. (2) now represent signals derived from the preamplifiers. The coupling factors f = C p s A / C p A are given in table 1. C p A ~ 2 nF denotes the input capacitance of the preamplifiers. The position signal then reads ( y ) ~ (E R) -- (E L) (E R) + (E L)
_ ((1 - f ) ER +f/~L ) -- ((1 --f)LC'L +fgR) ((1 - - f ) E R +J~L ) + ((1 --f)/~L + f e R ) "" (/~R) -- (/~L) ~'~(1 - 2 f ) y o = (1 - zy) (--~R) + (/~.L) '
(15)
where ER, EL denote the energy losses across the PSA. Obviously, (E R) + (E L) = (~'R) + (~L) = (E). Eq. (13) then reads
62Enoise { +( Y0 ~2 ¢32ynoise = 2 ( E ) ~ 1 \ 1 - 2f]
-- 3'noise I1 -- (1Y_~°2f) 21 }.
(16)
The final theoretical position deviation 6YTh (fwhm in ram) for fig. 4 reads 6YTh = 2.35bl(32ynoise + 62ystr) x/2 .
(17)
The agreement with the data is satisfactory. The systematic trends with d and y are reproduced correctly. The resolutions with anodes no. 3 and 4 are limited by the electronic noise. According to eq. (12) the contribution from energy loss straggling is negligible. It becomes important with anodes no. 1 and 2. As expected, anode no. 3 shows the best position resolution. 3.2. Linearity o f the position measurement
n
( r i - li ) g i
n
i=1
_ cos o~ ~ (ri _ l i ) E i . Ed i=1
Y -
(19)
S (r i + l i) Ei i=1 Here, E i denotes the energy loss across the ith stripe, E the total energy loss across the PSA. For 142 MeV 32S, ao = 1.4 °, and anode no. 3, eq. (18) yields a value of k <~ 10 -4 at any point Yo. This is negligible. The experimental nonlinearity ,,y ~Exp in fig. 5, as usual, is determined to be the deviation of the data with respect to a regression line. Though Xy Exp, indeed, is very small for anode no. 3, it is significantly larger than expected from eq. (18). A large systematic dependence of kyZXpony appears for anode no. 1. The reason becomes obvious from the simple model of fig. 6. The shadow with depth d*, which stems from the entrance window and the voltage divider, moves the starting point of the particle track further down the position-sensitive anode. This changes the position information [cf. eq. (1)] in a nonlinear way. Since the nonlinearities caused by the geometry are small, (L~R-- L~L)/(~R + ~L) ~Yo. ~ , L stands for the energy losses across the total right and left parts of the PSA. With the actual energy losses ER, L = L~R,L --E~, L, eq. (1) becomes
(~
- E~)
-
(~
- ED
Y - (~R -- E~) + (L~ - E l ) "
(20)
Assuming the differential energy loss dE/dz to be constant yields ~R + ~L = nd(dE/dz)/cos ot and E~ + E~ = d*(dE/dz)/cos a. A series expansion of eq. (20), and ignoring terms in [(E~ + E~)/(E R + EL) ] 3 results in X = Y - Y o = 1 + ~-~]~--~(Yo -- 1 ) + 2 E L c ° s ~ nd
Though, in principle, systematic nonlinearities of a measurement can be corrected for off-line, it is preferable to have none. Beside tedious computer work, they cause changes in the sensitivity of the experiment. Optimizing the geometry of the positionsensitive anode, we estimated the linearity to be best for the sawtooth configuration of fig. 1. Following the arguments in sections 2.1 and 3.1, and assuming that (dE/dz)i = Ei/d = constant, the geometrical nonlinearity ~. reads (for symbols see fig. 1)
X=y-yo,
with
(18)
(2~) Four cases have to be distinguished, resulting in different values of E~ cos a. Eq. (21) yields X ~< 0 for d* <~d/2 and y>~y* : E~ coso~ = 0
y
•
L7L. C O S O ~ ~
d,_(l+yO)(d
.[.
l+yo]-' Zl ]
.
Further X > 0 for d/2< d*'( d and y >~y* : E [ cos a =½d(1 - Y o )
,
y
(41+ o 1 Z1 !
•
566
G. Rosner et al. / A n e w position measurement
2.0
ii
-2.0
t
f
i i
t
,, L
i J
h i h i
J
t
--
i L
1.0
/
t
~ i
i
/ ,'
i
L i i
,
,
• "*,,,,,,,,,,. ,
®,,..-- t ..x ~
,.<
/
,,A
@
, /;
-
0 UJ:>~
/'
i
i
i
1.o / /
/
i
v
/
/ #i s
/
,
"
.
.4-+
,
~..../..
"i "\',-.-."/i
®
,
/
i
E
v
° R
,
,' ,I
/ ,,
-1.0 _
,,
/
',
Ii
/
,
#
~
i 'w"
-2.0
/
L
f~
(~1
d =60ram
(~ •
d=3Omm
(~)e
d=lOmrn
(~)I
d= 5ram
I -20
I -10
I 0
I 10
I 20
I -20
-1.o
-
-2D
t i i
/
#
-
-30
-
/
I -10
I 0
I 10
I 20
30
y(mm) Fig. 5. Experimental (left) and theoretical (right) nonlinearities Xy for anodes no. l to 4.
Particle ~
~z
/
I
\
Yi" /'
u \
/ /
\ \ \ \
/ \
/ // /
i ii
\
/ // i iI
\ \ \ \
// Target
,Z 1
Eq. (21) has been evaluated with a constant shadow depth of 48 mm for anodes no. 1 to 4. In order to be able to compare the calculated nonlinearity with the experimental one (3`~xp in fig. 5), a regression line has also been fitted to 3`. This has been done within the sensitive width of the PSA. The sensitive region ( - 0 . 6 ~
Fig. 6. Influence of the shadow of the entrance window (hatched area) on the linearity of the position measurement (see text).
567
G. Rosner et al. / A new position measurement
Table 1 Quantities characterizing the position-sensitive anodes (PSA), like the depth d of one stripe, the capacitance CpsA of the PSA, the experimental (Exp) and theoretical (Th) coupling factors f, mean position resolutions 6~ (fwhm) and nonlinearities ~y.
Anode
d (mm)
CpSA (pF)
fExp (%)
fTh (%)
-6YExp (mm)
-6YTh (mm)
~Exp (~am)
~yTh (mm)
1 2 3 4
60 30 10 5
31 34 82 154
1.6 2.7 4.1 7.7
1.6 2.7 4.1 7.7
1.05 0.58 0.33 0.44
0.95 0.62 0.35 0.42
1.23 0.24 0.07 0.03
0.93 0.24 0.17 0.09
of the electrons on the anode, because of the influence of the voltage divider across the entrance window on the path of the electrons. Therefore, the nonlinearities described above, to some extent, will be inevitable. However, the maximum nonlinearities rapidly get smaller with decreasing d, since nd = constant= 120 mm in eq. (21) and d * < d . Consequently, d might be chosen to be smaller in the vicinity of the entrance window. From the partly conflicting tendencies of position resolution and linearity we would estimate anode no. 3 to be the best choice (for our chamber). For this anode, position resolution (fwhm) plus nonlinearity amounts to "q3.4 mm within the full angular acceptance of the chamber.
4. Conclusion It has been shown that the position-sensitive AE anode described in this paper yields precise position measurements with high linearity. The experimental setup is very simple to realize. The influence of energy loss straggling and electronic noise on the position resolution has been described. Therefore, an optimizing prescription could be given. In further experiments, the PSA turned out to work quite well with large area (A~2 = 125 msr) ionization chambers, too [9]. We consider the PSA to be equally adequate for detecting lighter ions in ionization chambers used in magnetic spectrographs.
3.3. Influences on energy, energy loss, and time resolution
Appendix
No negative effects on energy and energy loss resolution were observed by using the position-sensitive anode as first AE anode of the IC. The energy loss resolution is governed by the energy loss straggling across the PSA, which does not depend on the segmentation. Therefore, the additional electronic noise of the PSA, caused by one more PA, plays no role. We measured ~iAE/AE= 3% for elastically scattered 142 MeV 325 ions. The main contributions to the energy resolution are given by the inhomogeneity and energy loss straggling in the entrance foil. The resolution obtained was 5E/E = 0.3%. Since the second coordinate was measured by the IC, the PPAC could be optimized with respect to time resolution. With the setup described in section 2.2 we measured a signal-to-noise ratio of better than 200. The rise time of the signals was less than 2 ns, the time resolution 150 ps.
Here we estimate the resolutions ~Estr,R, ~Estr,L, and the correlation factor 3'str caused by energy loss straggling across the position-sensitive anode (PSA). The main feature of the simple model is the mixing of original energy losses E* or energy loss stragglings ~E*. This is assumed to happen by diffusion as the electrons drift down to the PSA. Thus, the measured energy losses E and their resolution 6E will differ from the original ones. Fig. 7 illustrates this process, o stands for the diffusion width; l and r denote the actual depths of a left and right spike of the PSA. The individual spikes are divided into three parts, depending on whether or not they are affected by the diffusion process. This division is projected onto the particle track. The calculations will be carried out for the right part of the PSA. They are analogous for the left part. Consider, for example, the mixing z o n e ( E L ) i 3 - -
568
G. Rosner et al. / A new position measurement (E~)il (E~)i 2 (E~L)I3 (E~)il .~.,__,~,._A__
(E~)i2 ^
(E~)i3 ,,.._,,_~
Z ~'~ ionization track
\
Frisch grid
segmented anode
~,,...w__> (EL)il
(EL)i2 (EL)J3 (ER)il
(ER)i2
(ER)i3
Fig. 7. Section through the position-sensitive anode along the particle track. Hatched areas denote the diffusion regions (see text).
(ER)il in fig. 7. The measured deviation (~ER)il reads
since 62E[ + 62E~ = 62E * = 62E.
1 * * (6ER)il = (6EL)i3 = ~-[(6EL)i3 + (6ER)il ] .
In order to calculate the correlation coefficient %tr [see eq. (4)], one has to estimate the quantity
(A1)
Assuming the original energy loss stragglings (6E[)q, (~E~)i/, f = 1 , 2 , 3, to be uncorrelated results in 1 ~2(ER)il = ~-[~2(E~)i3 + ~2(E~)il ] .
n
(A2)
3
=
]=1
Since only (6ER)q of neighboring segments are correlated by diffusion [see eq. (A1)], only these terms contribute to the sums in eq. (A8). Applying analogous arguments as in the derivation of ~2Estr,R, we arrive at
It follows from the geometry and 3
= Z) 8 (ED, j=l
[analogously for ~2(/z'~)i] that
o 62(ER)i1 = 4117 62(E[)i + or 62(E~)i] "
(A3)
Analogously 62(ER)i2 -
3
62E
r+l
(A9)
"Ystr - 6E R 6E L
With the actual depth (see fig. 1)
r - 2o r
~2(ER)i3 = ~
(A4)
t~2 (E~)i
~2(E~)i+ 1 + ?
°
~2(E~)
;1
.
(AS)
(A10)
as a function of the geometrical positionyo = y l / b l [-1,1 ], then r + l =d(Yo), r/dO'o)
Finally, n
d(yo) = d(1 +y2o tan2o~o)w2
= (ER)/(E)
= (1 +yo)/2, I/dO'o) = (1 - y o ) / 2 , and eqs. (A6), (A7), (A9) become
3
62Estr,R = ~ ~ 62(ER)i] i:lj=l
~2Fstr,R=(1 +y 0
r - o .o2-* ~ 2 - * , - r - ° r + l (0 /SL + 0 /SR) --~+~/- 62E
O )
2
d(Yo) 82E
(All)
2
d0~o)
(AI2)
(A6) 82E
and l--0 82Estr,L = ~ -
82E,
(A7)
~'str d(Yo) °
1
--~y2o
(A13)
G. Rosner et al. / A new position measurement
References
[1] D. Shapira, R.M. Devries, H.W. Fulbright, J. Toke and M.R. Clover, Nucl. Instr. and Meth. 129 (1975) 123. [2] J.R. Erskine, T.H. Braid, and J.C. Stolzfus, Nucl. Instr. and Meth. 135 (1976)67. [3] H. Sann, H. Damjantschitsch, D. Hebbard, J. Junge, D. Pelte, B. Povh, D. Schwalm and D.B. Tran Thoai, Nucl. Instr. and Meth. 124 (1975) 509. [4] K. Kusterer, J. Betz, H.L. Harney, B. Heck, Liu Ken Pao and F. Porto, Nucl. Instr. and Meth. 177 (1980) 485.
569
[5] R. Allemand and G. Thomas, Nucl. Instr. and Meth. 137 (1976) 141. [6] B. Kolb, H. Ingwersen, G. Ihmels, E. Jaeschke, R. Repnow and Th. Walcher, Rev. Phys. Appl. 12 (1977) 1571; B. Kolb, G. Hlawatsch, G. Rosner, Th. Walcher, H. Ingwersen, E. Jaeschke and R. Repnow, Nucl. Instr. and Meth. 188 (1981) 555. [7] H. 'Schmidt-B6cking and K. Hornung, Z. Phys. A286 (1978) 253. [8] E. Mathieson and N. el Hakem, Nucl. Instr. and Meth. 159 (1979) 489. [9] G. Rosner, J. Pochodzalla, B. Heck, G. Hlawatsch, B. Kolb and A. Miczaika, to be published.