VOI,UME 19, N U M B E R 5

P H Y S I C A L , MF:VIE,W D

1 MARCH

1379

Cosmic-ray muon spectrum up to 1 TeV at 75" zenith angle

H. Jokisch, K. Carstensen,' W. D. Dau, H. 3 . Meyer,+ and 0. C. A-llkofer
Instirut f i r Kernphysik, Uiziversify ojKiel, Grrrmany
(Receibcd 23 Octnher 1978)

The spectrum of about 6 Y 10' inuonr: in the zenith angle range C = 68"--82" and the energy range
E
1-1000 GeV has been measul.ed by a Kiel-Dtsy collaboratio~lat Hamburg. The data are represented
by a simple form fit which goes as E -'.5'sec 0 for the integral spectrunl at very high energies. indicating that
the galactic cosmic-ray nucleon spectrum is flatter than E - 2 7 5 beyond 1000 GeV.

I. INTRODUCTION

Tile investigation of cosmic-ray muons at s e a

level provides information about the primary
galactic cosmic-ray nucleons in the energy
range ab{,ve 10Ii eV (where the direct measurements failj and below 10'' eV (the onset of the
more indirect extensive-air-shower method).
F o r example, the muon charge ratio i s directly
coupled to the neutron/proton ratio in galactic
cosmic-ray nuclei.' The muon spectrum and i t s
zenith-angle dependence at s e a level reflects the
parent spectra of pions and kaons in the atmosphere' and can be related to the primary galactic
nucleon spectrum by scaling the interaction properties of hadrons from the accelerator energy
range up to higher energies. [All muons above
a certain threshold energy E , a r e produced by
galactic nucleons with a median energy of about
10 E, (Ref. 3)].
Our high-statistics experiment at large zenith
angles complements the recent measurements
(of comparable counting rates) a t intermediate
angles4 and at vertical
11. EXPERIMENTAL

The experimental setup i s shown in Fig. 1 and

the main properties a r e listed irk Table I.
An a i r gap magnet was used, so the magnetic

KIEL-DESY muon spectrometer

- spark

chambers

s c ~ n t ~ i i a t ~counters
on

B-

= fbeid direct~on

field wa.s exactly known and there were no problems with multiple scattering of muons. The
spectrometer was triggered by a coincidence
of the four scintillation counters, and the muon
track was recorded in 14 wire spark chambers
which were read out by an on-line computer.
Capacitive readout was used for the chambers
in the strong s t r a y fields near the magnet, but
for the subsequent chambers core and magnetostrictive readout could be applied.
After the careful optical surveying of all
spark chambers they were realigned again by
iterative fitting of muon tracks; the iterations
stopped when the successive corrections were
l e s s than 0.1 m m . The final realignment shifts
were l e s s than 1 mm, the spatial resolution of
90% of the spark chambers was between 0.53 and
1.0 mm in the bending plane.
In order to avoid e r r o r s introduced by possible
misalignment and inefficiency (at the edges) of
the scintillators, the geometrical acceptance of
360 cm2sr was reduced to 274.7 cm2sr in offline analysis by the requirement that the calculated track had to penetrate the scintillation
counters within the reduced a r e a s and that i t s
zenith angle was in the range 68"-82". This
method was also used for the determination of the
efficiency of the two scintillation counters in the
magnet gap. The inefficiency correction for the
four-fold coincidence was +9.1% and +4.9% for
the positive and negative magnetic fields, r e s pectively.
To get a really absolute intensity an additional
correction f o r extensive a i r showers and noise
triggers was performed with the help of a timeof-flight measurement of the muons (with a
resolution of a = l nsec). The correction due to
this amounts to - 3 %. The resulting trigger
rate i s 9.7/min (Table I).
A more detailed description of the spectrometer and i t s performance i s given in Ref. 7.

'dm

111. DATA ANALYSIS

All the measured events were binned by zenith

angles ( 9 ) from 68" to 82" (bin size 2') and by 30

FIG. 1. Experimental setup

19
--

1368

(91979 The American Physical Society

COSMIC.RAY MUON SPECTRUM UP TO 1 TeV AT 75'.

TABLE I. Main properties of the experimental setup.

magnet: air gap 27 XI35 cm2; deflection power: S B d l = 1.83 Tm
mdm (meanvalue) : 745 GeV/c
collecting power: 360 cm2s r , reduced to 274.7 cm2s r
total number of events: l o 6
number or momenta >1GeV/c: reduced to 590 958
zenith angle: 75" *7"
azimuth: 288" *20" (east)
detectors: 4 scintillation counters
14 wire spark chambers
rate (corrected): 9.7 min-' *5 % ( 9 5 % confidence)
-----

momentum ( p ) intervals. For each of the seven

zenith-angle intervals model s p e c t r a (four parameters) were fitted to the frequencies of the 30
momentum bins after folding in the acceptance
and resolution functions. [The acceptance function A(#, 0) depends on the geometry of the spectrometer and is constant beyond 10 G e ~ / cand
decreases to zero below 1 GeV/c. The resolution function F($- $), gives the probability that
a deflection angle $ i s measured when an angle
$,=l/p i s expected. The resolution function used
here was the experimental deflection e r r o r distribution which was calculated from the e r r o r s
of the p a r a m e t e r s of each individual event- track
fit. As the number of sparks determining a track
varies because of inefficiencies between 4 and 14
(most probable number =12) the e r r o r distribution F ( $ - $), i s not a Gaussian with a fixed standard e r r o r a,. So the commonly used t e r m
"maximum detectable momentum" (mdm) (which
i s defined by mdm Xa,=p x$) does not give an
exact description of a spectrometer. Our mdm
values range from 250 to 1200 GeV/c, s o a
"mean mdm" is given in Table I.] The X 2 values
for the s p e c t r a fits a r e in the range 19 to 30 f o r
26 degrees of freedom.
The ratios of expected frequencies to acceptance- and resolution-corrected frequencies
give the final correction factors R ( p , 0) by
which the measured frequencies per momentum
and angle bin have to be multiplied. Figure 2
shows the function R ( p ) , which is R ( p , 8 )
averaged over 0, o r , more precisely, the factor
by which the measured numbers in Table I1 have
to be multiplied to get the c o r r e c t ones.

spectra D(P, 0) and I ( p , 0), respectively.' B ( p , 0)

i s given by

p +19.8,
p+19.8,sec0

(1)

'

[where p i s in G e ~ / cand D(p, 0) is in ~ m - ~ s r - ' s e c - '

(G~v/c)-'1. The values with indices a r e the fitted
parameters a=451 i 15, b=77.2 +4.8, c =9.2 +0.5,
d=19.8+ 1.8 (degrees of freedom DF=94, X2=1.21
XDF) . I(p, 0) is given by

where I ( p , 0) i s in cm'2sr-' sec-', r=35.0

7.2

- 1

'

'

'

lor KIEL-OESY

'

i 1.1,

D~fIerent8alSpectrum

0.9

aa a7 0.6

I
10

100

1000

p Muon "lomentumlGURI

IV. RESULTS

To present the final corrected data in a simple

way, we fitted the seven differential and seven
integral histograms by phenomenological form

FIG. 2. The correction factor R(p). R(p) has to be

applied to the measured differential spectrum. The
regions of large influence by the geometric acceptance
and by the finite resolution (mdm effect) of the spectrometer are indicated by arrows.

JOKISCH, CARSTENSEN, DAU, MEYER, A N D ALLKOFER

TABLE 11. Differential intensities
Momentutn
P (GeV/c)

Intensity
[cm-'sr-' s e c - ' ( ~ e ~ / c ) - ~ l

s=57.3*2.9, t=11.4*0.4, u=8.9*0.8, DF=101,

,y2=1.22x~F.
The differential spectrum D(p, 8 ) is not the
derivative dZ/dp of the integral spectrum I ( p , O ) ,
but the expression D i s much simpler than dZ/dp
and gives the same good fit results.
The fits were restricted to momenta beyond
30 GeV/c where geomagnetic deflection and
multiple scattering of the muons in the atmosphere a r e completely negligible. The three
t e r m s in D(p, 8 ) and I ( p , 8 ) correspond to pion
decay (proportional to sec0/p), pion production
spectrum (proportional to p-Y) and muon decay
with energy loss (3rd t e r m ) , respectively. The
pion production spectrum's exponent y was not
taken a s a fit parameter, because it has already been showng that y =2.57 0.03 gives the
best fit to our data a s well a s to both other high
statistics experiments at s m a l l e r zenith angle^.^-^
The relative deviations of the measured spect r a to the corresponding form fit a t zenith angle
0 were averaged over 0 and shown in Figs. 3
(differential) and 4 (integral) a s black dots. The

Error

Measured
number

(%)

maximum deviations which occur for the extreme zenith angle bins a r e also given.
Our form fit represents remarkably well even

r20

--

-20

-LO

f
-

75O

YALE-BNL

75'

0adwhai.Slephens

---- ---+r

p Muon Momentumat Sea ievel IGDV/cl

FIG. 3 . The relative differences of measured (or

calculated) differential spectra to the form fit a t 75"
zenith angle.

COSMIC-RAY MUON SPECTRUM U P T O 1 T e V A T 75O...

19
-

*2O

-.
------

-10

-40

1371

refer to 0=75" but include all data from 68' to

82". This was possible by modulating the 75'
form fit with the average deviation curves from
Figs. 3 and 4. The e r r o r s a r e standard e r r o r s
of the counting rates. The e r r o r of the absolute
calibration is given in Table I (rate). To visualize
the steepness of the muon spectrum, the integral
data (Table 111) a r e shown in Fig. 5 together
with the form fit and a pure power-law spectrum.

Range of Fit

pThe Yale-BNL data (Fig. 2), although of

V. DISCUSSION

100

1000

P Muon Momentum at Sw Leuel l W / c l

FIG. 4. The relative differences of measured integral

spectra to the form fit at 75" zenith angle.

the vertical data of the Durham ~ p e c t r o m e t e r , ~

which would show up in Fig. 3 on the 69" curve
(>30 GeV/c). The Yale-BNL datai0 for 8 =75"
a r e also given for comparison in Fig. 3.
Our actual measured intensities are presented
in Tables I1 (differential) and I11 (integral). They

poorer statistics, are in good agreement with our

data concerning the shape of the spectrum between 100 and 600 G ~ V / C . Their high-energy
points may not be well corrected for the resolution (mdm effect). As they use a hardware momentum selection at 100 GeV/c, their two lowest
points should not be given too much weight. Remembering the small e r r o r of our absolute rate,
there remains a systematic deficit of about 1257,
for the Yale-BNL data.
In Fig. 2 is also shown the most recent theo-

TABLE 111. Integral intensities.

Momentum
P (Ge~/c)

Intensity
( ~ msr-'
- ~ sec-')

Error
(%)

Measured
number

10

100
Muon Momentum [GeV/cl

l
l
1000

FIG. 5. The integral muon spectrum at 75". The

dashed curve represents the form fit; the straight line
shows the asymptotic behavior of the spectrum.

1372

J O K I S C R , CABSTENSEN, DAU, MEYER, AND A L L K O F E R

19

comes flatter at primary energies beyond 1 TeV.

retical calculation of muon spectra which was

done especially for 75" zenith angle.'' The calculation started with the primary galactic nucleon
spectrum and solved the pion and kaon transport
equation rigorously. The theoretical curve represents remarkably well our data from 10 to
100 GeV/c, but seems to fade out at higher
energies. The same effect is seen by comparing
the vertical spectrum, calculated in the same
way,I2 to measured vertical spectra, a s shown
by the Kiel group.' As Badhwar and Stephens
use the spectral index - 2.75 for their primary
nucleon spectrum (an extrapolation from direct
measurements at lower energies) one can conclude that the galactic nucleon spectrum be-

This work was part of the experiment PI08

performed at DESY (Deutsches Elektronen Synchrotron), Hamburg. We like to thank our colleagues A. Bzcker, C. Grupen, and W. Stamm
for their kind collaboration during our common
experiment. It is also a pleasure to acknowledge
the technical assistance of the DESY groups F51
and S2. We thank the DESY administration as
well a s the Deutsche Forschungsgemeinschaft
for financial support.

*Now a t Abteilung Medizinische Statistik und Dokumentation, University of Kiel, Germany.

?Now at Fachbereich Physik, Gesamthochschule Siegen,
Germany.
' 0 . C. Allkofer, K. Carstensen, W. D. Dau, and
H. Jokisch, Phys. Rev. Lett. 41, 832 (1978).
'w. D. Dau, K. Carstensen, and H. Jokisch, in P r o ceedings of the Fourteenth International Conference
on Cosmic R a y s , Munich, 1975, edited by Klaus
Pinkau (Max-Planck-Institute, Munchen, 1975). Vol. 6,
p. 1931.
3 ~ K.
. Gaisser, J. Geophys. Res.79, 2281 (1974).
4 ~ H.
. Burnett, G. E. Masek, T. Maung, E. S. Miller,
H. Ruderman, and W. Vernon, in Proceedings of the
Thirteenth International Conference on Cosmic R a y s ,
Denver, 1973 (Colorado Associated Univ. P r e s s ,
Boulder, 1973), Vol. 3, p. 1764.
5C. A. Ayre, J. M. Baxendale, C. J. Hume, B. C.
Nandi, M. G. Thompson, and M. R. Whalley, J. Phys.
Gl-, 584 (1975).
6 ~ G.
. Thompson, R. Thornley, M. R. Whalley, and
A. W. Wolfendale, in Proceedings of the 15th International Conference on Cosmic R a y s , Plovdiv, 1977,
edited by Christo Christov (Institute for Nuclear Re-

s e a r c h , Sofia, 1977), Vol. 6, p. 21.

'K. Carstensen, Ph.D. thesis, University of Kiel,
1978 (unpublished).
*s.Miyake, in Proceedings of the Thirteenth International Conference on Cosmic R a y s , Denver, 1973
(Colorado Associated Univ. P r e s s , Boulder, 1973),
Vol. 5, p. 3658.
C. Allkofer, K. Carstensen, W. D. Dau, H. Jokisch,
and H. J. Meyer, in Proceedings of the 15th International Conference on Cosmic R a y s , Plovdiv, 1977
edited by Christo Christov (Institute for Nuclear Res e a r c h , Sofia, 1977), Vol. 6 , p. 38.
'OL.
Leipuner, R. Larsen, L. Smith, R. Adair, B. Higgs,
H. Kasha, and R. Kellog, in Proceedings of the Thivteenth International Conference ca Cosmic R a y s , D a v e r , I973 (Colorado Associated Univ. P r e s s , Boulder,
1973). Vot. 3 , p. 1771.
llG. D. Badhwar and S. A . Stephens, in Proceedings
Cosmic R a y s ,
of the 15th International Confe-ceon
PTovdiv, 1977, edited by Christo Christov (Institute for
Nuclear Research, Sofia, 1977), Vol. 6, p. 200, and
private communication, 1978.
1 2 ~ D.
.
Badhwar, S. A. Stephens, and R. L. Golden,
Phys. Rev. D 15,820 (1977).

ACKNOWLEDGMENT