OPTICAL REVIEW Vol. 4, No. 2 (1997) 293-296

The Use of CarOusel InterferOmeter in Fourier-TransfOrm

Ultraviolet

S pectroscO py* Jyrki K. KAUPPINEN, 1lkka K. SALOMJ~, Jari O. PARTANEN and Matti R. HOLLBERG University of Turkz4 Department of Ap~)lied Physics, FIN-20014 Turku, Finland (Received October 3, 1996; Accepted December 26, 1996)

The carousel interferometer is a new type of an interferometer, which has been invented at the University of Turku. It consists 0L a beamsplitter and five plane mirrors. Four of the mirrors are mounted on a carousel, which rotates back and forth. We have modified the interferometer for use in the Fourier-transform ultraviolet (FT-UV) spectroscopy. Test measurement with plasma radiation gives favourable results. The most important property, which makes the carousel interferometer suitable for UV measurements, is its good stability in

modulation.

Key words :

ultraviolet spectroscopy, Fourier-transform spectrometers, interferometers

resolution. Larger signals mean better signal-to-noise ratio

1. Introduction

(S/N). In the IR region, where the photon noise is insignificant compared to the noise of the detector, S/N

Fourier transform spectroscopy (FTS) in the ultraviolet

region is demanding, because the short wavelengths set

can be improved by two orders of magnitude. In the UV

special requirements for the stability of the interferometer.

region, where the photon noise is dominating, the noise is

One of the typical difliculties of the traditional Michelson

by the use of cat's eye retroreflectors,1) or by back-to-back

proportional to the square root of the signal, and the improvement of S/N is one order of magnitude. The Fellgett's or the multiplex advantage of FT-IR spectroscopy is achieved, because the whole frequency

cube corner mirrors.2) However, there still remain prob-

spectrum is measured simultaneously. With gratings,

lems in the UV region, where the mirrors of the cube corners would have to be exactly perpendicular to each

narrow frequency bands are measured one after another, if only one detector is used. If the spectral region under measurement consists of M resolution widths (elements),

interferometer is the tilting of the moving plane mirror during scanning. This problem can be solved, for example,

other. The carousel interferometer,3) which is developed at

the measurement of the whole spectrum simultaneously

the University of Turku, solves the problem mentioned above, and has several properties which make it very

with a FT•IR spectrometer increase coeflicient Jl~f. This is, however, the case only in the IR

suitable for Fourier-transform ultraviolet (FT-UV) spec-

region, where the photon noise is insignificant. In the UV region the Fellgett's advantage is lost, because also the photon noise is increased by the coefficient 1/~.

troscopy. One of the basic features of the carousel interfer-

ometer is that the optical path difference is created by rotational movement. Advantages of rotational scanning have been suggested also elsewhere.4,5) Due to practical difficulties, FT-UV spectroscopy is a

The Connes or the wavenumber advantage is a consequence of the linearity of the frequency scale of a FT spectrometer. The real frequency of the whole spectrum

method which has not been widely used. Most of the FT-UV spectroscopic equipments are big and expensive vacuum interferometers, and meant for high resolution

can be determined by fixing one single point, and the scale

remains the same in repeated measurements. There is no need of separate calibration of the spectrometer, Iike in the

measurements in laboratory conditions.6,7) There are several reasons for choosing Fourier trans-

grating spectrometer. The Connes advantage is achieved in both IR and UV regions. In addition, FTS has a few other advantages, e.g. the

form spectroscopy instead of gratings, which are the conventional technique in the ultraviolet spectroscopy.

resolving power advantage.ll'l2) These advantages are already known from the FT•infrared Fourier transform spectrometers are flexible, because (IR) spectroscopy,8) and most of them can be applied in

FT-UV spectroscopy as well.9,lo) The Jacquinot or the throughput advantage is achieved,

they offer the possibility to conveniently choose the resolu-

because in FTS the whole wavefront of the beam is used

instead of buying a new grating. Additionally. FTS makes

tion simply by changing the data acquisition length, it possible to afEect the instrumental line profile by mathe-

in the measurement of all frequencies, instead of dividing the wavefront, as is done with gratings. The throughput of

matically manipulating the interferogram truncation at the ends. This well known method is called apodization.

an interferometer can be as much as two orders of magnitude larger than with a grating with the same

On the whole, the possibility of using various mathematical methods in handling the spectral data is an important feature of FTS. Our goal has been to construct an FT-UV spectroscopic

*Presented at 1996 International Workshop on Interferometry (IWI '96), August 27-29, Saitama, Japan. 29 3

294 OPTICAL REVIEW Vol. 4, No. 2 (1997)

J.K. KAUPPINEN et al.

equipment, which is mobile, easy to use, and cheap to build. The equipment can be used in low-resolution applications. We are convinced that there is a growing need for such an equipment. One typical field of use would be the control of heavy metals in the environment, which is very laborious by other methods.

o( Ideg

2. The Carousel Interferometer The optical layout of the carousel interferometer is shown in Fig. 1. The interferometer consists of the beamsplitter B and the five plane mirrors Ml' M2' M3' M4 and M5' The beamsplitter and the end mirror M5 are fixed. The

four mirrors Ml ' M2' M3' and M4 are mounted on a carousel, which rotates back and forth around the axis A of the carousel. The axis is perpendicular to the plane of the figure. The angle y determines the directions of the side mirrors Ml and M2, and the angle fi tells the directions of the mirrors M3 and M4' The incoming beam hits the beamsplitter at the angle 6. The transmitted beam travels in the left section of the interferometer first to the mirror Ml' then to M3' and to the end mirror M5' The beam hits M5 perpendicularly, if the condition 2y+2p=e is fulfilled, and returns the same path back to the beamsplitter. The reflected beam from the

Fig. 2. Optical path difference x of a carousel interferometer of typical dimensions (y=8.5', ~=14', the distance between the side mirrors Ml and M2 is 9 cm, and the distance between the end mirror M5 and mirrors M3 and M4 is 7 cm) as a function of the rotation angle c! '

tion of the rotation angle a! can be calculated.3) It is an almost linear function of the rotation angle. The optical path difference of a carousel interferometer with typical

dimensions is shown in Fig. 2. In practice, the maximum rotation angle is restricted by the falling off of the beam

interferometer. The optical path difference of the two

from mirror M3 Or M4 Or end mirror M5' The value of the maximum angle is of the order of 5'-10', depending on the chosen dimensions and angles of the mirrors. An optical

beams is created by the rotation of the carousel with the

path difference of, for example, I cm is easily achieved, and

mirrors M1' M2' M3' and M4' If the carousel is rotated

this is already enough to give a resolution, which is typically used in low-resolution FT-IR or UV spectros-

beamsplitter follows a similar path in the right side of the

counterclockwise, the optical path of the beam in the left side of the interferometer is shortened and the path in the right side is lengthened. The functioning of the carousel interferometer is explained in detail in Ref. 3). The optical path difference of the two beams as a func-

copy.

The carousel interferometer contains only plane mirrors, which can easily be prepared with sufficient accuracy

for the short UV wavelengths. Mechanical distortions due

to bending or twisting of the mount are efficiently eliminated: the beamsplitter and the end mirror are fixed on the same metal body in the irnmediate vicinity of one

/.~ Jl.__

_,_l~ P\..

;} ~~f~

M4

M8

~jll :~_= 7

7+b;:= j

7;il ~

1~~~~ l{

-1= ~~7

!1-

another and the rotation axis A, thus minimizing their movements with respect to each other. The four rotating mirrors are fixed to the same body, which is connected to

the surroundings only through the rotation axis; the danger of the bending of this carousel is minimal. Because the optical path difference is made by rotation, and not by

linear movement, even distortion by vibration is effectively reduced. By modifying the optical layout of the carousel interferometer it is possible to build several variations of rotating interferometers. The principle is always the same, only the

A M5

path of the light and the amount of the mirrors are

Me

Mi

changed .

3. Ultraviolet Spectroscopy 3.1 Stability e

B Fig. 1. Optical layout of the carousel interferometer. B is a beamsplitter and rvl:1-M5 are plane mirrors. Mirrors Ml~M4 rotate back and forth around the axis A.

There are several factors, which affect the stability of a

spectrometer. The noise in electronics and in detectors (scintillation noise, photon noise) cause instabilities. We shall concentrate on the mechanical stability of an interfer'

ometer. Stability in modulation is a very important property of

J.K. KAUPPINEN et al. 295

OPTICAL REVIEW Vol. 4, No. 2 (1997) an interferometer. It becomes crucial, if measurements are

used as the sample clock. If samples are taken at every

made in the UV region, where the wavelengths are short.

zero-crossing of the laser signal, wavenumbers up to 15802

A very significant source of instability is the variation of

cm~1 can be reached. However, in the UV region these

temperature. This may change the dimensions of the

frequencies are too low to produce an unaliased spectrum. We have to divide the wavelength of the laser in several parts. Eight sample points per every wavelength of He-Ne

interferometer, and thus decrease modulation. Most FT-IR spectrometers are thermally stabilized. The thermal stabili-

zation is not, however, sufficient for applications, where the intensity must remain stable in strongly varying conditions. This is the situation, if an interferometer is used as an analyzer. The stability should be improved by solutions which make the construction itself tolerate the changes of temperature better. We studied the stability of the carousel interferometer

in changing temperatures. The interLerometer was placed

in a thermally controlled box, where the temperature varied between 20'C and 40'C. The modulation of the interferometer and the tilting of the optical components were measured continuously. The latter was done interferometrically by the use of interference fringes of a He-Ne laser.

First we investigated the behavior of the end mirror and

the beamsplitter in changing temperatures. The way in

laser will lead us to a spectrum with the wavenumber range 0-63209 cm~1, which in wavelengths means the region above 158 nm.

We tested two methods for dividing the laser wavelength. First one was based on the use of a phase-10ckedloop. It is an integrated circuit, which measures the input frequency (here the frequency of the laser interference signal) and gives the wanted multiple of the frequency as

an output. In this method problems may occur, if the velocity of the moving mirrors changes. This changes the frequency of the laser interference signal, respectively. During every cycle of the input signal the phase-10cked-

100p gives a frequency based on the preceding period, which may result in clocking errors, if the mirror velocity does not remain exactly constant. As a second solution we built sample clock electronics, which was based on special

which they are fixed in the interferometer has a large effect

integrated circuits. It measures the incoming signal and

on their response to temperature variations. The clearly best results were obtained by using mechanical fixing. We

simultaneously gives a multiple of the frequency. The

found that the beamsplitter and the end mirror, when

and the changes of the mirror velocity cannot cause

properly mounted, are practically immune for changes of

clocking errors. Mirrors with a flatness of 1/20 Iaser wavelength had to

temperature. No changes in the interference pattern of the

laser could be seen. This means that thermal expansion cannot tilt them and thus decrease the modulation.

advantage of this solution is, that it functions in real time,

be used in order to enhance the modulation in UV

fixing requires special attention. For this reason we devel-

wavelengths. The mirrors were made of A1-coated silica. The material of the plate beamsplitter was UV enhanced quartz (Suprasil 1). The wavelength region of the beamsplitter was 190-670 nm, which, in practice, determined the wavelength region of our measurements.

oped a method to mount and adjust the mirrors in pairs (MI and M3 together, and M2 and M4 together). An imppr-

4. Experiments

For the stability of the carousel the most critical point

turned out to be the fixing of the mirrors Ml~M4 to the carousel. Since the mirrors have to be adjustable, their

tant Lactor is, how the pairs are attached to the carousel.

Our method seems to work very well, and we succeeded to

We have tested the carousel interferometer in several measurements in the infrared, visible and ultraviolet

very adequate for infrared region. At present we are

regions, and the test results have been promising. An important application of the carousel interferometer in the UV is the analysis of heavy metals by inspecting the radiation emitted by plasma. Figure 3 shows an example of experiments, where we measured the emission spectra of a

continuing the work with the UV region stability of the equipment. Testing of different construction materials is

mercury argon Hg(Ar) Iamp. The spectrum is an average of four single measurements. The resolution of the mea-

an essential part of this work. Already now it is obvious, that the carousel interferometer has several good Leatures for use in FT spectroscopy.

surements, determined by the optical path difference, was

3. 2 Other Considerations The carousel interferometer meets the special requirements of the optical design of an UV interferometer. But

spectral lines of mercury are clearly seen at wavenumbers 17271 cm~1 and 17331 cm~1 (the yellow doublet), 18315 cm~1 (the green line), 22935 cm~1, 24752 cm~1, 27397 cm~1 (the triplet), and in the UV region at 39370 cm~1. Due to the 10wer response of silicon in the UV region, the UV Iine appears to be smaller than it should be. Also argon, which is used as a starter gas in the lamp, can be seen in the

enhance the temperature stability of the carousel. The best way to fix the mirrors is to use mechanical fixing, just as in the case of the end mirror and the beamsplitter. The stability of the carousel interferometer is already

an additional, essential problem in FT-UV spectroscopy is, how to gain a sufficient accuracy in the determination of sample points of the interferogram. In the infrared region one sample of the interferograrn is taken for every wavelength of a laser, which is used as a sample clock. This

9.3 cm~1. The detector which was used in the measurement was a silicon detector (Hamamatsu S1336-BQ). The

means, that we get an unaliased spectrum for the

spectrum (the lines between the wavenumbers 10000 cm~1 and 15000 cm~1). In the arrangement of the measurement

wavenumber region below 7901 cm~1, if a He-Ne laser is

of Fig. 3, the plasma light was collimated and lead to the

J.K. KAUPPINEN et al.

296 OPTICAL REVIEW Vol. 4, No. 2 (1997)

/ mercury '

>1

4J

.H

> (D v)

.H Ei

ril

5000 Fig, 3.

35000 40000 45000 lOOOO 15000Wavenumber 20000 25000 30000 [1/cm] Emission spectrum of a mercury spectral lamp, recorded by the carousel interferometer.

interferometer with paraboloidal mirrors. The same results have also been gained by using a fiber as a light guide.

We have also carried out real plasma measurements.

5. Conclusions The carousel interferometer appears to be most suitable for the low resolution UV spectroscopy. It is relatively

Emission spectra of radio frequency plasmas of argon and

simple and easy to prepare, and its stability is essentially

nitrogen were measured, and we found out, that the system, in principle, works. However, special attention has

better than that of other mobile interferometers. Test measurements in UV region were successful. Naturally

to be payed to the stability of the plasma flame. If the

the good properties of the carousel interferometer make it

intensity of the plasma light vibrates with a frequency that

desirable also in visible and infrared regions, and it will be

is near the frequency of the modulation of the interferometer, a large noise may be caused in the spectrum.

yzers.

installed in the GASMETTM or Quantum 7000R gas anal-

In the infrared region the main application of the carou-

sel interferometer is at the moment low resolution gas analysis. The study of stability, that was mentioned in the section 3.1., is a part of a project, in which the carousel

interferometer will be developed suitable for the GASMETTM or Quantum 7000R gas analyzer. Also for that purpose it is important that the interferometer stays stable

Ref erences

J. Connes and P. Connes: J. Opt. Soc. Am. 56 (1966) 896. E. Kyrd, J. Kauppinen, M.-L. Junttila, M. Linnavuo, P. Wallin and T. Valkeaph~: Rev. Sci. Instrum. 58 (1987) 1180. 3) J.K. Kauppinen, I.K. Salomaa and J.O. Partanen: Appl. Opt. 34 1)

2)

(1995) 6081. 4)

J.R. Sternberg and J.F. James: J. Sci. Instrum. 41 (1964) 225.

in many different kinds of conditions. The UV region sets even tighter requirements for the stability of the equip-

5)

ment, and we will continue the work to fulfil these

7)

M. Ford and A.R. Gee: Proc. SPIE 553 (1985) 365. P. Luc and S. Gerstenkorn: Appl. Opt. 17 (1978) 1327. A.P. Thorne, C.J. Harris, I. Wynne-Jones, R.C.M. Learner and

demands even better.

6)

G. Cox: J. Phys. E: Sci. Instrum. 20 (1987) 54. 8) P.R. Gri~~ths, H. Sloane and R.W. Hannah: Appl. Spectrosc. 31 (1977) 485. 10)

A.P. Thorne: J. Anal. At. Spectrom. 2 (1987) 227. A.P. Thorne: Anal. Chem. 63 (1991) 57A.

11)

J.K. Kauppinen: Appl. Spectrosc. 38 (1984) 778.

12)

J.K. Kauppinen: International Conference on Fourier Transform Spectroscopy (Durham University, England, 1983) p. 5.22.

9)