THE USE OF A PRESSURE-SCANNED FABRY-PEROT INTERFEROMETER FOR TWILIGHT SKY AND ZODIACAL LIGHT OBSERVATIONS

(1)

HAL Id: jpa-00213238

https://hal.archives-ouvertes.fr/jpa-00213238

Submitted on 1 Jan 1967

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.

THE USE OF A PRESSURE-SCANNED

FABRY-PEROT INTERFEROMETER FOR

TWILIGHT SKY AND ZODIACAL LIGHT

OBSERVATIONS

D. Clarke, P. Hindle, N. Reay, J. Ring

To cite this version:

(2)

JOURNAL DE PHYSIQUE Colloque C 2, supplt!ment uu no 3-4, Tome 28, mars-avrill967, page C 2

-

294

THE USE OF A PRE SSURE-SCANNED FABRY-PEROT INTERFEROMETER

FOR TWILIGHT

SKY AND ZODIACAL LIGHT OBSERVATIONS

by

D. CLARKE, P. H. HINDLE, N. K. REAY and J. RING

University of Hull, Department of Applied Physics, England

Abstract. - A pressure scanned Fabry-Perot interferometer has been designed for work on the twilight sky and zodiacal light. A digital computer has been used to obtain the optimum combina- tions of interorder spacing and resolution to enable sensitive measurements of the line contrast to be made in the shortest possible observing time. The pressure stepping sequence is controlled automatically and the spectrum is recorded on punched paper tape for direct input to the computer. Rbsumb. - Un interfkromktre Fabry-Perot a balayage par pression a kt6 construit pour SCtude du ciel cr6pusculaire et de la lumikre zodiacale. Un ordinateur est employ6 pour calculer la combi- naison optimum d'intervalle entre ordres et de r6solution pour donner la sensibilite maximum dans la mesure du contraste de la raie. La variation pas a pas de la pression est contr6lCe automa- tiquement et le spectre est enregistrk sur bande perfor6e pour entrer directement dansl'ordinateur. Introduction.

-

The Fabry-Perot spectrometer, with

its simplicity and its high resolution-luminosity product is ideally suited to that class of astronomical problems involving the study of a restricted wavelength range of the spectrum of faint extended sources. The present paper is concerned with the application of the technique to two such problems.

Observations of the spectrum of the Zodiacal light can yield important astronomical information [I]. The phenomenon manifests itself as a faint cone of light along the direction of the ecliptic, seen shortly after sunset or shortly before sunrise. The visible cone can be 200 to 300 wide at the base and reaches almost to the zenith. This light is believed to be due to sun- light being scattered by inter-planetary matter - free electrons at a high kinetic temperature give a Doppler broadening to the Fraunhofer absorption lines, whilst dust particles, which are usually assumed to be in orbit around the sun, will give Doppler shifts : Figure 1 (a and b). The original solar absorption lines will have half-widths of the order of 1

A

or less whilst in the Zodiacal light the expected Doppler shift will be a fraction of an Angstrom unit and the Doppler broadening may be several Angstrom units wide. It should be possible from accurate measurements of lines profiles to decide the relative contributions from electrons and dust grains and also to measure the relative proportions of dust grains in co-rotation or contra-rotation around the sun. If the dust grains were of primeval origin one would expect that the majority of them will be co-

Undisturbed solar BO%DEPTH Fraunhofer line

"Ffllfng-fn" by electron 800'o OF scattered component

ELECTRON-SCATTEREO COMPONENT

.

l--

--2

FIG. 1 a.

Vl,

tpTH

Blue shift of line by Doppler

e f f e c t . ln e v e n ~ n g zodiacal light implies c o - r o t a t i o n of d u s t cloud

Red Doppler s h i f t in evening zodiacal light 80%

V,:-

fmplies contra -rotation of d u s t cloud

I

Doublmg a n d f i l l i n g - i n by

co- a n d c o n t r a - r o t a t i o n

FIG. 1 b.

(3)

THE USE OF A PRESSURE-SCANNED FABRY-PEROT INTERFEROMETER C 2

-

295

rotating (that is to say, in the same direction as the majority of components of the solar system rotate), whilst if the grains are of later origin, for example cometary debris, then they will be equally likely to be co-rotating or contra-rotating.

Previous investigations of the spectrum of the day sky and of twilight have shown that there are differen- ces between profiles of absorption lines from the scattered source and the original Fraunhofer lines [2]. This effect must be caused by emission or by a rapid variation of a scattering coefficient with wavelength. It is important in these observations to have extremely accurate measurements of the depth of the Fraunhofer lines ; we can use the time variation of the height of the earth's shadow after sunset or before sunrise to identify the sources of emission or scattering in the earth's atmosphere. The spectrometer required for both problems must therefore have an instrumental profile of between 0.25 and 1.5

A

and must be luminous enough to accept a field of view of several square degrees on the sky at this resolution. Even so, previous observations suggest that fluxes of 10 to 100 pho- tons s-' are the maximum to be expected in each resolved spectral element.

Three spectrometric systems were considered before our first observations.

I. Grating spectrometer. - A Bausch and Lomb grating 208 mm by 128 mm with a blaze angle of 480 was available. Mounted as a conventional monochromator it could have given a luminosity of about 0.001 cm2 steradian at the required resolution of 1.2 A, with straight slits. (Curved slits would have given an inconvenient shape when projected onto the sky.) We experimented with longer slits and a three element image slicer to try to improve both the luminosity and the shape of field on the sky, but the reflection losses disposed of most of the theoretical flux gain.

11. Pure interferometric spectrometer. - The largest optical flats available were 90 mm in diameter and so a luminosity of 0.1 cm2 steradian was possible if one order of a Fabry-Perot spectrometer could be isolated ;

with a finesse of 25 this requires a premonochromator passing a band-width of about 30

A,

with negligible transmission outside the pass-band. Calculation sho- wed that the interference filters available at the time (1962) would have allowed parasitic light from adja- cent orders to alter the line profiles obtained. (( Pepsios 1)

was being developed but was not at that time in a sufficiently advanced state to work at a mountain laboratory. A field widened Michelson spectrometer was considered and rejected for similar reasons.

111. A grating-interferometer combination.

-

The grating described in I was used as a pre-monochromator for a Fabry-Perot. The theoretical luminosity (neglecting transmission losses) was 0.02 cm2 steradian - a considerable improvement over system I.

The shape and position of the pre-monochromator pass-band was easily changed by altering the widths of the entrance and exit slits of the spectrometer and the angle of the grating respectively.

Observations.

-

In 1962 a system of the third type was used, by a joint team from the Universities of Wisconsin and Manchester, to observe line profiles in the spectrum of zodiacal light from Chacaltaya in the Bolivian Andes [3]. This observatory was chosen

because of its proximity to the equator which ensures that the ecliptic is nearly vertical and that auroral radiations are unlikely, and because of its height (18 000 ft) which ensures that the night sky is relatively free from scattered light. A 128 mm x 208 mm Bausch and Lomb grating with a blaze angle of 480 was used in a spectrometer giving a linear dispersion of 4

A

par millimeter. The spectrometer was preceded by a 9 cm diameter Fabry-Perot interferometer which accepted radiation from the sky by means of a one-mirror coelostat. The field of view on the sky was approxi- mately 20 by lo and the overall instrumental profile was about 1.2

A

wide.

A selected photomultiplier tube cooled to a temperature of

-

40 OC was used as a detector ; the flux of photoelectrons due to the zodiacal light amounted to about five per second. Some difficulty was experienced in arranging a mechanical scan of the grating spectrometer to coincide with the pressure scan in the interferometer and so the observations were made with a static grating profile trapezoidal in shape, the interferometer being scanned across the flat top through a range of about 10

A.

Because of the low flux rate and the limited number of observations obtained it was decided to repeat the observations in 1964 using an all-interferometric spectrometer system, since, by this time, improved filters were available. An interference filter with a half-width of 5

A

preceded a Fabry-Perot interferometer, incorporating 9 cm flats with the gap having defects of less than 1/50 th of a wavelength. The Fabry-Perot interferometer was scanned once again by pressure variation, using sulphur hexafluo- ride over a range of 0 to 80 Ibs per square inch above atmospheric pressure. Thus a range of up to 20

A

(4)

C 2

-

296 D. CLARKE, P. H. HINDLE, N. K. REAY AND J. RING

light scans and its effect removed from the data subse- quently. The flux of photoelectrons on this occasion amounted to fifteen counts per second, a substantial improvement over the earlier observations. A number

of profiles of the H, line were obtained and surprisin- gly, Hg was seen in emission on many of them (Fig. 2).

These observations are now being evaluated and will be'gublished in a later paper. It is clear, however, that in order to obtain good measurements of the Doppler shift in the absorption line it will be necessary to increase the resolving power of the system with a consequent loss of flux. For this reason we decided to calculate the optimum parameters for a filter and Fabry-Perot interferometer for use in observations of the Zodiacal light and twilight sky, before making further observations.

Computer Simulation.

-

A computer programme was written to simulate the output of the working instrument, pressure scanned over a range of 6

A

centred on an absorption line taken from the (( Utrecht

Atlas of the Solar Spectrum )). The combination of filter and absorption line profile (Fig. 3) is convoluted successively with a series of Airy functions with finesses and resolutions within the anticipated working range of the real instrument. With a 6

A

scanning range the formulation of the programme is such that

the filter profile must be specified over a range of 19 and the appropriate Airy function generated over 25

A.

The equation used to generate the Airy function is

where I = transmitted intensity, N = finesse,

J = resolution (Width of function at

3

intensity),

I = wavelength.

Equation (I) is obtained by substituting,

in the Airy function equation and normalising to the position of maximum transmission. A further refine- ment of the programme permits the wavelength of the peak transmission of the filter to be shifted with respect to the absorption line before convolution with the Airy function. This enables the influence of temperature and tilt of the interference filter to be assessed quantitatively.

(5)

THE USE OF A PRESSURE-SCANNED FABRY-PEROT INTERFEROMETER

Results. - Figure 4 shows the expected profile of H p obtained when scanned at resolutions of 0.1, 0.25 and 1.0

A

for a finesse of 25. The repeating

-3 0 '3.

A,&--

C O N V O L U T E D PROFILES FOR Hg

pattern of the 0.1

iP

curve is produced by more than one order of the Airy function passing under the cen- tral minimum of the filter x line curve. From the sets of convolutions a contrast surface (contrast being defined as the ratio of maximum to minimum intensity in the convoluted profiles) has been plotted (Fig. 5)

giving an overall picture of the way in which contrast varies with N and J. If reproductible results of contrast measurement are required, points on the surface corresponding to high rate of change of contrast should be avoided.

Figure 6 shows the variation of measured line contrast with the temperature of the order sorting filter. The calculation has been performed for a finesse of

25 and a resolution of 0.25

A.

This combination will be shown to be near the optimum working values for

l t t . * a t a - . ~ # ~ o

-6 - 5 -4 -3 - 2 - 1 0 1 1 3 4 5 6

DEG. C. VARIATION OF C O N T R A S T WITH TEMPERATURE OF THE FILTER

the system. The filter drift coefficient has been taken as 0.18

A

degrees C-I ; the value quoted for the

(( Thin Film Products )) 5

A

pass-band filter. It can

be seen that if the contrast is to be measured to an accuracy of 1

%,

the filter temperature must be controlled to better than 0.1 OC.

Optimisation of observing time. - An important consideration in the design of an optical system to be used for the observation of faint sources is the mini- misation of observing time required to obtain a given accuracy of measurement, viewed in the light of the physical capabilities of the apparatus.

The photometric m accuracy required to detect a small change in the position or shape of an absorption line (e. g. Doppler shifts in Zodiacal light or the

(( filling-in )) of twilight sky lines) is a rapidly varying

function of the measured line contrast ; the latter being a function of the finesse and resolution of the Fabry-Perot. The total transmitted flux through the system depends upon the finesse and resolution of the Fabry-Perot and the shape of the pass-band of the order sorting filter. Bearing in mind these considera- tions we must choose an optimum working value of finesse and resolution.

(6)

C 2

-

298 D. CLARKE, P. H. HINDLE, N. K. REAY AND J. RING

when compared with its solar counterpart under the same experimental conditions may be defined as

OL = RB

-

Rs

1 - RB (3)

20 where R, and R, are the ratios of the (( bottom of the 400

-

line 1) intensity to the continuum intensity, for the

-.

+!

observed source and the sun respectively. Equation (3) ₃ is valid only if a is sensibly wavelength independent 4

2

25

over the wavelength region of the few Angstrom

_g

₃₀₀

_-

units studied. Partial differentiation of eqn (3) gives F us the fractional error in a in terms of the errors in R,

_$

₃₀

and Rs. Assuming that Asis much better known than RB, 2

since many observations of this quantity can be made,

3

the equation for the error in a becomes, 2 _D

gmo-

25

8 (1

+

4

- -

_--.

a a(l - R,) (4)

where ARB is the fractional error in R,. It can be seen _{100 -}

that the photometric accuracy required to obtain a given accuracy in the measurement of a depends

strongly on both a and the line contrast R,. The accuracy required of R,, just to detect a given a is

obtained by putting 6ala = 1 in equation (4). 0 ₀ _.1 _.2 _*3 _.4 _.5 _.6 _.? ₄ _.9 _1.0 A computer programme has been written which _Reso/ut/on_A.U

calculates the line contrast that would be measured by _FIG._7. the Fabry-Perot system for a given interference

filter profile, a given absorption line and a series of _130- values of finesse and resolution of the interferometer.

At the same time it calculates the flux through the 120

optical system at the wavelength points to be used in

the measurement of contrast for each set of values of

-.

110 finesse and resolution. The computer evaluates, using ..?.

$100- equation (4) with 8ala = 1, the integration time

??

required in the measurement of R,. Values of a were

₆

90- chosen to be 0.5, 1.0 and 2.0

%

of the continuum level -a

2

for a line scanned at a resolution of 0.25

A

and a 9.80-

finesse of 25. The LX values for any other combination

s

of finesse and resolution are evaluated assuming that ,270 2 the component added to the source spectrum is equi-

₃

_so

valent to an additional flux of energy which is inde- Q

pendent of wavelength. 50

Results. - The programme was run for the Hg

absorption line and a 5

A

pass-band filter for finesse It0 values of 20, 25 and 30 and resolutions running from ₃₀ 0.1 to 1.0

A.

Figure 7 is a graph of observing time

plotted against resolution for the three values of finesse 20

mentioned above and for a values of 0.5 and 1

%.

The observing time is in arbitrary units proportional to the real observing time. This is necessary because

l o - 0

the actual flux received at the detector will depend

- 20 - - -

-

- -

-

10% 2 5 \ \

L\/

20 20% I 3 0 1 I I .I .2 . 3 .G

(7)

THE USE OF A PRESSURE-SCANNED FABRY-PEROT INTERFEROMETER C 2

-

299

ged portion of the bottom region of figure 7 with the a = 2

%

value included.

The existence of a well defined minimum observing time illustrated in the above two figures demonstrates the value of performing such calculations before real observations are commenced. It can be seen that the observing time required just to detect a given a depends strongly on the resolution employed ; the branching of the curves with finesse shows, as one would expect, that the highest attainable values of this parameter should be used.

The new spectrometer. - Based on these calculations and on experience gained as a result of the Chacaltaya experiments a new Fabry-Perot interferometer using 9 cm flats has been constructed. (Shown schematically in figure 9). As a result of the previously mentioned calculations a temperature controller has been designed to maintain the ambient temperature of the interferometer to

4

1/50 degrees C.

A beam splitter provides a compensation channel to record time variations in atmospheric transparency.

group, out of 120 sample points, across the 25 maximum scanning range of the instrument, to be selected. The instrument steps automatically, integrates on each step for some pre-set time and punches out the signal and compensation channel intensities onto paper tape. The scanning sequence can be re-cycled a pre-set number of times.

This summer, the interferometer will be used at the high altitude observing station at Testa Grigia (Cervi- nia), under the computed optimum conditions for the measurement of the Hg absorption line contrast in

the twilight sky, and for the observations of the Hg emission in the night sky. Doppler shift measurements in the Zodiacal light will be attempted later.

[I 7 BEGGS (D. W.), BLACKWELL (D. E.), DEWHIRST (D. W.),

WOLSTENCROFT (R. D.), M. N. R. A. S., 1964,

127, pp. 319-328. M.N. R. A. S., 1964, 127,

pp. 329-341. M. N . R. A. S., 1964, 128, pp. 419- 430.

compensa ion

n-ch.nnel

~o+o..ltipl.r _{l$_h_t_frpm}_m:rc)lr~ viewing porthole

I

r n -

I \ f i l t e r ( Hat I T 1 t

concave _&pressure _vessel _signal

1 I \ I , , : -IF--? ,, . .r

.

1\ field lens ,~ha""$l ,

_I 1 .

rnlrrOr _{i n n e r box [ternperoture controlled to 1/50}_oC]

The signal detector is thermoelectrically cooled to

-

30 degrees C to improve the signal to noise ratio of the system.

The etalon has been milled from a solid cylinder of steel, thus eliminating all screwed sections that might give rise to creep and subsequent misalignment of the flats. In addition, to ensure that the etalon will remain in adjustment for long periods of time it is intended to use optically contacted spacers.

To facilitate observations, a fully-automatic, digital- output, electronic system has been incorporated [4].

A modification of this system enables any one or a

[2] GRAINGER (J. F.), RCNG (J.), Nature, 1962,193, pp. 762- 763.

[3] RING (J.), CLARKE (D.), JAMES (J. F.), DAEHLER (M.), MACK (J. E.), Nature, 1964, 202, pp. 167-168. [4] HINDLE (P. H.), IBBETT (R. N.), J. Scient. Znstr., 1966,

43, pp. 209-214.

INTERVENTIONS

(8)

C 2

-

300 D . CLARKE, P. H. HINDLE, N. K. REAY AND J. RING

J. RING. - The Michelson used in the visible will have the same SIN as a scanning Fabry-Perot inter- ferometer of the same size if it includes the same spectral elements. But if one is only interested in a few points in the spectrum the Michelson suffers the Fellgett disadvantage since it generally gives a lower SIN for a given observing time than the Fabry-Perot interferometer.

J. F. JAMES.

-

The Fellgett advantage does not disappear completely if the photon count rate is very low. The dark counts from the photomultiplier are detector noise and it is possible, with a Michelson, to get a Fellgett advantage of 2 or 3.