ENTE PER LE NUOVE TECNOLOGIE, L'ENERGIA E L'AMBIENTE
Dipartimento Innovazione
ET f\T- - 9k 13
EXCIMER LAMP PUMPED BY A TRIGGERED DISCHARGE
G. BALDACCHINI, S. BOLLANTI, P. Dl LAZZARO, F. FLORA, G. GIORDANO, T. LETARDI, A. RENIERI, G. SCHINA Centro Ricerche Frascati, Roma
G. CLEMENTI, F. MUZZI, C.E. ZHENG
EL.EN., (Electronic Engineering), Firenze
MASTER
DISTRIBUTION OF THIS DOCUMENT IS UNLIMITED FOREIGN SALES PROHIBITED
RT/INN/96/13 DISCLAIMER
Portions of this document may be illegible in electronic image products. Images are produced from the best available original document ABSTRACT Radiation characteristics and discharge performances of an excimer lamp are described. The discharge of the HCl/Xe gas mixture at an atmospheric pres sure, occurring near the quartz tube wall, is initiated by a trigger wire. A maximum total UV energy of about 0.4 J in a (0.8-0.9) ps pulse, radiated from a 10 cm discharge length, is obtained with a total discharge input energy of 8 J. Excimer lamps are the preferred choice for medical and material processing ir radiations, when the monochromaticity orcoherence of U V light is not required, due to their low cost, reliability and easy mantainance.
(DISCHARGE UV LAMP).
RIASSUNTO In questo iavoro vengono illustrate le caratteristiche di eccitazione e di emissione radiativa di ima lampada ad eccimeri. La scarica di eccitazione e innescata da un filo ad alia lensione e si sviluppa vicino alia parete del tubo di quarzo in una miscela di gas HCl/Xe a pressione atmosferica. Con un’energsa d; 8 J depositata dalla scarica in una lampada lunga 10 cm, si ottiene un’cnergia massima irraggiata nell’ultravioletto di 0.4 J in un impulse di 0.8-0.9 ps di durata. L’affidabihta e il basso costo tipici delle lampade ad eccimeri le rende competitive nelle appltcaziom industriali (per es. medicale e lavorazione materiali) die non richiedono monocromaticita della luce UV. INDICE
1. Introduction...... 1
2. Experimental arrangement description ...... 2
3. Measurements for discharge performances and UV-radiation output...... 4
4. Maximum attainable UV-power density and its spatial distribution .... 11
5. Discussion...... 15
6. Summary...... 20
Appendix ...... 20
References 22 EXCIMER LAMP PUMPED BY A TRIGGERED DISCHARGE
1 - INTRODUCTION
Ultraviolet (UV) radiations have been proved very efficient for initiating some chemical or physical processes, and they have already found many applications both for medical purposes and in some industrial processes. Between the two different types of UV radiation sources mainly used now, the UV-lasers can have an output brightness, much higher than that for the other type: UV -fluorescence lamps. However in some cases, especially when the processes do not require the coherence or the monochromaticity of the light, the UV-lamps, (for exam ples, mercury lamps) are often the preferred choice for applications, owing to their low cost, simplicity, reliability, and easy maintenance. Recently, due to more and more photo-initiated or photo-assisted processes being evolved, a revived interest has been stimulated for developing new UV-discharge lamps, such as various excimer lamps!1-9 !. With these excimer lamps, the spectral widths of the fluorescence radiations are much narrower than those from the mercury lamps! 10!, and this may lead to a more efficient excitation of some interesting processes. The excimer lamps can be pumped by several different discharge excitation modes, such as the microwave discharges! 3’4!, dielectric-barrier discharges! 1,5,6,9 ! the pulsed dis charges which include axial discharges! 2,7! and transverse discharges! 8!, and so on. The microwave discharge can give a quasi-continuous UV-radiation output with high efficiency, while the pump system and its energy coupling design are rather com
1 plicated, compared to the other types of the excitation modes. Using pulsed discharges along the axial direction of the lamp tube, the lamp con struction is simple, but the discharge gas pressure is usually limited in the order of 102 torrl2,7!. Thus, the total UV-radiation intensity can not be very high. When using transverse discharges with ultraviolet preionization, a UV-fluorescence output of up to several kW/cm2, measured at the distance very near to the window, has been achieved from a gas medium with pressure higher than one atmosphere®, and the lamp in this case is like a small laser discharge chamber. For dielectric-barrier discharges (microdischarges), the lamp can work at high gas pressure with a very efficient incoherent UV-output. The discharge in the lamps takes the form of many statistically distributed, both in time and in space, filaments and each filament causes a surface discharge on the surface of the inter-electrode dielectric material. Although the spacing of the discharge is limited in the order of a few mm for a typical discharge voltage of 104 V with the dielectric thickness of several mm, the discharge-distributed area can be scaled to a large dimension with different configura tions for UV-processingt1,5 ’6,9!. Some lamps directly use the inter-electrode dielectric materials, for example, quartz as the UV-output window to simplify the structure, and in this case the surface discharge etching or electrode sputtering on the dielectric surface may cause the window-transparency problem, leading to a decrease of the UV-output. It is well known that the high voltage trigger wires have been used for a long time in some flash-tubes or lamps (see for example, [11,12]), to improve the discharge stability and the energy coupling. In this work, we present some UV-fluorescence radiation results, obtained with this kind of triggered, pulsed discharges in a quartz tubes, filled with HCl/Xe mixture up to atmospheric pressure. Some data, obtained at low pressure without using triggering, are also described for comparison.
2 - EXPERIMENTAL ARRANGEMENT DESCRIPTION
A schematic view of the experimental set-up used in this work is shown in Fig. 1. The electric discharge modulator consists of a three-stage pulse forming network (PFN), a thyratron (EG&G, 8614, HY-5, 40 kV/5kA), and a charge inductance L\. In order to take into account the future application needs, where the modulator and the lamp may be separated for a distance for putting the lamp as near as possible to the radiated object to obtain the largest fluorescence radiation power density, a three-meter, 50-fi
2 Discharge modulator
Discharge tube
>-(- Th Trigger V
Figure 1. Schematic of the experimental setup with a three-stage PFN. C q = 0.0135 fiF, L0 = 0.15 /tH, CT = 0.6 nF, L% = 50 \iH.
cable Ca is used to electrically connect the modulator to the discharge tube. Tr is a electrically conductive wire with a diameter of ~ 1.5 mm, which lies parallel and near to the lamp tube. Tr and the capacitance Ct = 0.6 nF are used for triggering the discharge. The circuit can be operated at the repetition rate of up to ~ 100 Hz, limited by the DC charge power supply. The discharge tube is made of quartz glass with an inner diameter of 1.8 cm. Two electrodes, whose outer diameters are both 1 cm, are made of tantalum foils with thickness of 0.1 mm. Several different distances Id between the two electrodes are tried, but most of the measurements are made using ld = 10cm. The UV-radiations are generally observed at a distance of Ss = 60 cm from the center of the discharge tube. The UV signal-receiver is set at this position, consisting of a biplanar phototube (ITT FW-114A), with a 3 mm-thick coloured glass filter (Schott UG-11) Fcoi in front, together with some neutral density filters. The transmittance Tugn of Fcoi is shown in Fig. 2. All the filters are calibrated by a spectrometer (PERKIN ELMER UV/VIS/NIR Lambda 19). A 2 mm-thick colour filter WG320 and a multi-layer dielectric coated quartz mirror Mr are also used as filters in order to study the spectral distributions of the UV- radiation. The transmittances Twg320 for WG320 and Tmr for Mr are given in Fig. 2.
3 100 Twg320
Wavelength (nm)
Figure 2. Transmittances of UG11 (—), WG320 (-•-•) and Mr (------) vs wavelength A.
The UV-radiation energy is determined by convolving the phototube response S'pt'(A) with the filter transmittance and with the time-integrated emission spectra, and also taking into account the limited surface area of the phototube through a geometrical factor Q,a (see Appendix).
3 - MEASUREMENTS FOR DISCHARGE PERFORMANCES AND UV- RADIATION OUTPUT
We brief some results in Part A, obtained without using discharge triggering wire, to see what is the main limitation for a further increase of the fluorescence energy in this case. The other two parts B and C give the results obtained using the discharge triggering.
A. UV-Fluorescence Output without Discharge Triggering
The measurements in this case are done using the same setup described before, but the trigger wire Tr (Fig. 1) is kept far from the lamp, so that the influence of the field established between the wire Tr and the lamp electrodes can be neglected compared to that between the electrodes.
4 Total gas pressure Ptot (torr)
Figure 3. Total UV-radiation energy vs the total gas pressure for the gas mixtures of HCl/Xe/Ne = 1/4/10 (•), 1/4/5 (o), and 1/4/3 (A), respectively, with Vc = 20 kV, Id = 27 cm.
Fig. 3 shows the measured total UV radiation energy of the lamp vs the total gas pressure Ptot for several different HCl/Xe/Ne gas mixtures. Clearly, the fluorescence energy increases with the total gas pressure, and the maximum pressure Ptot, which we could have for normal discharges, is determined by the onset of discharge instability. This means that for each of the three curves recorded in Fig. 3, the experimental point with the highest Ptot corresponds to a critical value, beyond which there is a strong decrease of the UV radiation energy, and sometimes even there is no breakdown of the gas between the electrodes. We can also deduce from Fig. 3 that for a fixed gas pressure Ptot and DC charge voltage Vc, the fluorescence energy increases with the decrease of the partial pressure of neon. When substituting neon with helium as a diluent gas in the HCl/Xe mixture the behavior of the fluorescence energy vs the partial pressure of helium is observed to be similar to that for the case of neon. These results suggest that under the same DC charge voltage Vc and for a fixed total gas pressure Ptot of the mixture with the same HCl/Xe mole ratio, the fluorescence energy output may be optimized with a binary gas mixture and without using neon or helium.
5 B. Discharge Performances
Using the lamp trigger circuit, shown in Fig. 1, with a DC charge voltage of 20 kV, a fairly stable fluorescence energy output can be obtained form the discharged HCl/Xe gas mixture up to one atmosphere. As in the case of low pressure discharges (Fig. 3), when the triggered discharges occur at high pressures, it is also observed that for a fixed DC charge voltage and a given total gas mixture pressure, the total fluorescence energy decreases with the increase of the diluent gas neon pressure, as shown in Fig. 4. A result similar to Fig. 4 can be obtained for helium case. Therefore, in the following we only present the result by using the HCl/Xe gas mixture.
Fig. 5 shows typical time histories of the lamp voltage V (£), current /(£), and UV fluorescence for a 500 torr total pressure gas mixture with a mole ratio HCl/Xe = 1/9, for a discharge length Ip — 10 cm with DC charge voltage Vc = 20 kV. The peak values of the voltage and current increase with the gas pressure, but they don’t change very much when the pressure Ptot increases from 500 torr to 760 torr.
100 200 300 400 500 600 700 Neon partial pressure (torr)
Figure 4. Total UV radiation energy vs neon gas partial pressure for different gas mixtures. The symbols • and o, respectively, represent the mixtures with the mole ratio HCL/Xe =1/5 and 1/10, both balanced with neon up to Ptot = 760 torr, Id = 10 cm, Vc = 20 kV.
6 Discharge voltage V (t) (5 kV/div.) Discharge current I (t) (200A/div.)
500 ns/div.
Figure 5. Time history of the discharge voltage V(t), discharge current I(t) and XeCl UV fluorescence for a 500 torr-pressure. The gas mixture is HCl/Xe == 1/9. Id = 10 cm. Vc = 20 kV.
As shown in Fig. 5, the main pulse discharge lasts for more than 1 /is, and then both V(t) and I(t) are reversed almost simultaneously. Further measurements show that after the main discharge pulse, the reversed voltage drop on the lamp is of the order of several hundreds volts, causing a long reversed current flowing the discharge loop and a long fluorescence tail up to several fj,s.
7 For the gas pressures we are interested in, the discharge energy transfer efficiency T)tr defined as the ratio of the energy deposited in the gas mixture to the total energy stored in the modulator circuit, i.e. rjtr = J0°° I(t')V(t')dt'/ (see Fig. 1) may be measured, and some of the data are shown in Fig. 6. It can be seen that in the pressure range shown in the figure, rjtr keeps almost constant.
C. UV-Energy Output Characteristics
The UV output as a function of the gas mixture composition and pressure, of the charge voltage, of the repetition rate is measured to find the optimum working point of the excimer lamp.
Fig. 7 shows the effect of the HC1 partial pressure on the time-integrated signal intensity A, caused by the UV-radiations on the phototube. Clearly, for a sufficiently high partial pressure of HC1, the total UV-radiation energy doesn’t change very much with Phci but when Phci (20-30) torr, the energy output declines fairly quickly with
= 50 torr 40 torr
24 ton- 40 torr
Total gas pressure Ptot (torr)
Figure 6. Discharge energy transfer efficiency T)tr vs the total gas pressure Ptot for HCl/Xe mixture with the discharge length Id = 10 cm and Vc = 20 kV. The reference parameters near to the error bars represent the HCl-partial pressure contained in the mixture.
8 5
Figure 7. Time-integrated values of the phototube signal intensity vs Phci with Ptot = 1 atm for the mixture HCl/Xe and Vc — 20 kV.
the decrease of Phci - When Phci is near to zero, the received time-integrated signal intensity A approaches to a non-zero value, which is obviously generated principally by xenon discharge emission!13-17].
Fig. 8 shows a plot of the total UV radiation energy vs the total gas pressure Ptot for three different mixtures, using the DC charge voltage Vc = 20 kV and the discharge length Id = 10 cm. The maximum energy output is near 0.4 J with an intrinsic efficiency V'mtT = 9%, where 77intr is defined as the ratio of the total XeCl UV fluorescence energy output to the energy deposited in the gas. Here we take rjtr = 55% (see Fig. 6).
Fig. 9 plots the variation of the full width at the half maximum (FWHM) Tp of UV-radiation intensity pulse as a function of the total gas pressure Ptot for different HCl/Xe mixtures. The pulse width Tp increases very slowly with the total pressure -Ptot- When Ptot rises from 100 torr to an atmospheric pressure, Tp increases by a factor of ~ 1.2.
The energy conversion efficiency rjc vs Vc is reported in Fig. 10 for two different HCl/Xe mixtures, where rjc is defined as the ratio of the total XeCl UV fluorescence
9 300 -
Total gas pressure Ptot (torr)
Figure 8. Total UV radiation energy vs the total gas pressure with HC1 partial pressure of 24 torr (•), 32 torr (A), and -16 torr (o), respectively, balanced with xenon. Vc = 20 kV, Id = 10
0 100 200 300 400 500 600 700 800 Total gas pressure Ptot (torr)
Figure 9. UV output intensity pulse width tf (FWHM) vs the total gas pressure Ptot for mole ratio of HCl/Xe = 0.027 (•), 0.07 (o), 0.15 (A) and 0.35 (A), respectively. Vc — 20 kV, Id = 10 cm.
10 6
5 S' 4 I 3 § £ 2 I 1
I J------1------1____l____i____i____i i i MU 0 12 14 16 18 20 DC charge voltage Vc (kV)
Figure 10. Energy conversion efficiency rjc vs DC charge voltage Vc at a total pressure of 1 atm, where (•) represent 19 torr and (A) 30 torr HC1 partial pressure both balanced with xenon. Id = 10 cm. energy to the energy stored in the modulator, i.e. in this case (see Fig. 1). The lamp can be operated in a burst mode (for a few seconds) with a repetition rate from several ten Hz to 100 Hz. In Fig. 11, we show 100 UV-radiation intensity waveforms, recorded at a repetition rate of 100 Hz. It can be seen that the waveform stability is quite good. At the repetition rate of 5 to 10 Hz, the lamp can continuously work for a fairly long time when using a ventilator for cooling the tube. It is observed in this case that the pulse energy of the lamp output gradually decreases by about 1/3 for the first 104 shots, and then keeps almost the same value up to 3 • 104 shots without any noticeable change.
4 - MAXIMUM ATTAINABLE UV-POWER DENSITY AND ITS SPA TIAL DISTRIBUTION
Compared to lasers, the lamp fluorescence radiations have a very poor quality for prop agation, and it can not be transported over a large distance without a serious energy loss. As a consequence, the sample has to be put near to the lamp, when it requires an high intensity UV radiation. Therefore, it is necessary to know the radiation intensity
11 500 ns/div.
Figure 11. UV radiation intensity oscillogram recorded continuously for 100 shots at 100 Hz rep. rate. Vc = 20 kV. Ptot — 500 torr.
distribution around the discharge tube. Neglecting the induced radiation and absorption, the UV power received per unit area at the any point P of the space can be found by summing all the contributions of spontaneous radiations coming from each point of the discharge region inside the lamp. If the radiation region is considered to have a uniform, linear distribution, the treatment can be much simplified. Without loosing the generality, the Cartesian Coor dinate System O-XY is taken in such a way that the lamp is put along the X-axis with its center at the coordinate origin O, and the observation point P(x, y) in the O-XY plane, as shown in Fig. 12. In this way, along the direction parallel with the Y-axis, the fluorescence energy radiation flux, received per unit time in the neighborhood of the point P(x,y) can be written as
1 -5L 1 W 2 Ip + (i) 4?r 1% y/lo (l-it) +(>)
where W is the total fluorescence radiation power of the lamp, and Id is the discharge length.
12 y/lj) = 0.50
Phototube
Figure 12. Fluorescence power density distribution around the lamp tube. (a) lD(x/lD,y/lD)/lD(0,y/lD) vs x/Id (b) /d (0, y/ld)/Id(0,6) vs y/lD, where the solid points represent the measured data with Id = 10 cm and Vc = 20 kV.
13 Fig. 12 (a) shows the fluorescence power density Id as a function of the distance x/Id for — 0.5,0.75,1.0, and 1.25, where Id is normalized to have a unit value at x = 0. Clearly, the larger the distance y, the better the uniformity along the direction parallel to the tube. For a fixed distance y from the tube, the power density Id at the observation point decreases with the distance \x\, but when |x| < ^Id and y > \Id, Id does not change very much. For example, at the points of x = ±\Id and y = \Id, Id still have about 2/3 of its central maximum value (y = 0). Fig. 12 (b) shows the fluorescence radiation power density I'd as a function of the distance y/lo for x = 0, where the curve is calculated according to the formula (1) with Id being normalized to have a unit value at y = 61 d, and the solid points indicate the measured data, which are obtained, using the lamp with Id = 10 cm and Vc = 20 kV. Clearly, the experimental data coincide very well with the calculated results if the distance y/lD is not too small. The deviations of the data from the calculated ones for small values of y/lo may be due both to the non perfectly uniform, linear emission of the lamp, and to the large absorptions suffered by some oblique incident radiations coming from the regions near the two ends of the tube. Obviously, with the decrease of y/lDt these oblique radiations have to pass thicker and thicker lamp tube wall and phototube-window for reaching the phototube cathode. Using a ready-made aluminum-alloy reflector with an aperture width of Wa = 10 cm, originally designed for another application, it is possible to bring a part of backward radiations along the positive direction of O-Y axis, as shown in Fig. 13. In this way the fluorescence intensity signal, measured at the distance of y — 60 cm and x = 0, is increased by a factor of F = 3.5. Obviously, as long as the distance S2 between the lamp and the radiated sample is large enough compared to the outer diameter
14 ventilation
Signal - receiver
Reflector
Figure 13. Schematic diagram of the lamp-reflector system.
5 - DISCUSSIONS
A. Energy Measurement Error Estimation
It is rather complicated to give a accurate value for the error of XeCl UV-energy output data given in this work, as the energy is not directly measured by a energy-meter, but determined by a convolution of the phototube response with some other wavelength- dependent parameters or quantities, such as the XeCl emission spectra, the filter trans- mittances and so on. It seems that there is a good experimental basis for the time- integrated XeCl emission spectra which may be roughly considered to have a similar shapes and similar wavelength distribution ranges, especially for XeCl (C-A) transi tions, as we did in this workl25,26,27 b With this assumption, the errors for the energy determination mainly come from the calibration of the phototube sensitivity, and from the measurements for transmittances of the various filters we used. In this way, the relative error for XeCl UV-energy output data can be estimated to be about ±20%.
15 B. Trigger Processes of the Discharges
Due to the peaking role of Ct for the voltage drop V(t) on the lamp, V(t) increases very quickly at the beginning of the discharge. For a typical gas mixture used in this work with a DC charging voltage Vc = 20 kV, the voltage peak can reach about 30 kV with a rise time in the order of 102 ns (Fig. 5). The breakdown value of the voltage for the gas in the lamp depends on many factors, such as the gas composition and its partial pressures, the rise time of the applied voltage or its time-history during the pre-breakdown period, the average and local values of the field, and'so ont18l. Some measurements show that for a typical gas mixture we used, when Vc is higher than ~ 10 kV, a stable discharge between the lamp electrodes can be obtained. For Vc < 10 kV, the discharge becomes unstable, and sometimes there is no breakdown between the two electrodes, but some spark channels can be observed both between the electrodes and the internal surface of the tube, and between the trigger wire and the external surface of the tube. In this case, these spark discharges are just dielectric barrier discharges wit h each channel corresponding to a single transit breakdown or streamer breakdown. Each channel is connected with a constricted electrode spot at the metal cathode and a surface discharge spread on the quartz surfaces, as described in [1,5,6]. The above phenomenon may imply that the ignition process of the triggered dis charge may occur as follows. After the thyratron Th being triggered, a highly non- uniform field between the two electrodes and the trigger wire Tr is established. When this field is high enough, the microdischarges which assumes spark channel or filamen tary structures are initiated in the region between the electrodes and the trigger wire. These quart?-harrier discharges, on one hand serve as an extended cathode with an abundant electron emission, on the other hand provide a lot of UV-photo emissions for preionization of the gas mixture in the lamp. With the electric charge build-up on the quartz surface, the field in each filament decreases, but before all the filamentary discharges are choked, the main discharge channel between the two electrodes is formed. The discharges occur with a little diffused form, and due to a strong radiant component of the field, caused by the disposition of the trigger wire Tr, they are near to that part of the tube wall where there is the trigger wire outside of the tube. In some works, large area surface discharges along the interface between solid and gaseous dielectrics are investigated, being used as effective plasma electrodes, or simply as UV-radiation sources in CO2, excimer, and chemical pulse HF laser system^15 ’19-23!. The possibility was also demonstrated of utilizing a surface discharge in an excimer
16 flash lamp of high repetition rate (up to ~ 10 kHz) and emitting UV-radiation of surface power density of ~ 0.3 W/cm2 l15 l. Instead of using a wire, some other shapes, such as the metal slab, a cylindrical foil, and so on were used as the “trigger electrodes” in these large area surface discharge devices. In a sense, the operation principle of the triggered discharge, studied here, is similar to that of these devices, but it is superior to them for its very simple structure.
C. Xenon Discharge Emission
The time-integrated signal intensity recorded for Phci = 0 (see Fig. 9) is believed due to the so called third continuum emissions from discharged xenon gas. For xenon pressure Px e over 102 torr, the third continuum is usually made up of two continuum bands centered close to A™ax = 270 nm and A™** = 390 nm respectively i17l. The intensities of the continua in general increase fairly quickly with the pressure, and seem near to be saturated when Px e = (1 — 1.5) atmt 17l. The third continuum of xenon gas can be obtained by various excitation methods, such as e-beam or a a-particle excitation^ 14’16 ’17], discharge pumping^13’24] and so on. As the spectrum of the emitted light for xenon depends in a complicated way on electron and ion densities, and also on their energies, which may be directly influenced by the pumping methods and their intensities, there are some differences about the data of the wavelength AJ1^ and A™ax , observed. To determine the radiated energy of xenon discharge, we need to know its emission spectrum. Some evidences indirectly show that the spectrum of xenon discharges in our experiment may be near that obtained in [14], where a 700 torr xenon gas is excited with incident a-particles. Using this spectrum, we can achieve fairly good coincidences between the calculated and measured data for the ratios Amr/A and Awg32o/A, as shown in Table 1, where Amr and Awg32o are the time-integrated signal intensities, respectively, when using Mr and WG320 as the filters.
Table 1. Comparisons of the measured values of Amr/A and Awg32o/A with the calculated results when using the spectrum of xenon reported in [14].
Measured values Calculated values Amr/A 0.20 0.18 Awg32o/A 0.58 0.58
Using the spectrum reported in [14], the total radiated energy E\e in the wavelength
17 range of (215-380) nm can be calculated from the measured time-integrated signal in tensities A, and the result is Exe == 0.25 J with t)intr. = 6%.
D. Estimation of the Energy Percentagefor XeCl (B-X) and (C-A) Transitions
The spectral measurements for HCl/Xe mixture excited by a-particles, show that when the concentration of HC1 exceeds 1%, the emissions of the third continuum of xenon (290 nm - 390 nm) disappear, and in this case the spectrum is mainly dominated by XeCl fluorescence emissionsf16 ^. Therefore, it is reasonable to assume that for an atmospheric pressure HCl/Xe mix ture with Phci ^ 10 torr, the discharge-excited emission spectra are mainly constituted by XeCl (B-X) and (C-A) transitions. Since XeCl (B-X) and (C-A) emissions extend from 300 nm to 315 nm with the peak at 308 nm, and 320 nm to 380 nm with the peak at 345 nm [25-27], respectively, it is possible to use the WG320 filter for estimating the energy fractions generated by (B-C) and (C-A) transitions respectively.
0.8
< 0.6 COc3 g0.4 < n 0.2
0 120 160 PHC1 (torr)
Figure 14. Ratio 7 of the time-integrated signal intensity by using WG320 to that without using WG320 as a function of Phci ■
In Fig. 14 we show the measured ratio 7 = Awgzio/A as a function of Phci , where Awgzzo and A are the time-integrated signal intensities with and without using WG320 as a filter respectively. By using 7 = A/A = 0.54, and convoluting the filter response with the signal-receiver response and with the relative spectral intensities of the band
18 (B-X) and (C-A), the ratio E&x/Eca of the energies contained in (B-X) to that of (C- A) transitions can be calculated. Noticing that XeCl (B-A) and (C-A) spectra almost coincide to each other, if we use the brancing ratio of 0.067 for (B-A)/(B-X) P&l, the energy percentages for (B-X), (C-A) and (B-A) transitions can be deduced. The same measurements as above described are also made with the mirror Mr, instead of WG320, as the filter. The final results obtained by these two different filters are given in Table 2.
Table 2. Energy fractions for transitions XeCl (B-X), (C-A) and (B-A).
Transitions B-X C-A B-A Energy percentages (%) 49 ± 17 48 ± 17 3 ± 1
E. Life time problem
It seems that for the geometryof the lamp used in this work, there are two different fac tors which influence the excimer lamp life time. One is the sputtering of the electrodes, which continuously deposits metal atoms during the discharges on the internal surface on the gas vessel, resulting in a reduced UV transparency of the quartz wall and the other is the effect of the gas degradation which arises from impurity generation caused by prolonged contact of HC1 with the inside of the lamp and by the combined action of UV radiation and the electrical discharge. Really, because both the quartz and the electrode materials employed in our measurements are not well compatible to HC1, the excimer lamp is observed to have quite a large decrease of the UV fluorescence output, even after being filled with a fresh gas mixture for only a few days operation. It appears to be difficult to find a electrode material which is both compatible with HC1 and little sputtering influence of discharge, but it is possible to design a new structure of the electrodes using a metal HCl-compatible and having a geometry which could control the discharge sputtering effect on the quartz wall. 6 - SUMMARY
The following points summarize the main results which are obtained by using a triggered discharge for the UV-lamps filled with atmospheric HCl/Xe gas mixtures.
1. The measurements show that for the same total gas pressures with the same DC charge voltages, the UV-radiation energy of the discharge for the mixtures of HCl/Xe/He or HCl/Xe/Ne is lower than that for the case when keeping the same pressure ratio of HCl/Xe, but without using diluent gas He or Ne.
2. A total UV-radiation energy of ~ 0.4 J can be achieved in a discharge length of 10 cm with the intrinsic and total efficiencies of ~ 9% and 5% respectively.
3. The UV-radiation pulse duration tf is not very sensible to the gas mixture pressure. Generally, rp is in the range of (0.7-0.9) ns, when the pressure Ptot is lower than 1 atm.
4. At a distance of (6-8) cm from the lamp tube, the maximum attainable pulse power and averaged power per unit area on the radiated sample can be estimated to have the values of about 0.7 kW/cm2 and 0.05 W/cm2 (in burst mode) respectively.
5. Some preliminar irradiation experiments were performed to dessicate industrial- used paints for protectivity the furnitures. The results show that the drying effect over a 50 cm2 surface is completed after 1000 shots at 30 pps (that is about half a minute irradiation time). This means that a number of lamps in parallel can dry a medium-size painted furniture in time-window of interest for industrial purposes.
APPENDIX
The signal intensity Vs(t) in units of volt, caused by the incident UV-radiations on the signal receiver, can be expressed by
%(() = /dA f(-V) ' Spr(A)' ^gii(A) - Zpr , (A-2) where ■S'pr(A) is the wavelength-dependant sensitivity of the phototube (Amp./W), Zpt is the phototube output impedance (f2), and /(A,t) is the spectral intensity of UV-radiations (W/nm).
20 After integrating the two sides of the above formula over time t, we obtain the time-integrated signal intensity A as
A = J dX-U(X) • Spr(X) • Tugii(A) • f2^ • Zpt , (A-3) where [7(A) = J dt ■ I(X,t) is the time-integrated spectral intensity (J/nm). Obviously, the total radiated energy can be calculated as
E — J dX- [7(A) , ' (A-4) if [7(A) is known. The XeCl fluorescence spectra is mainly composed of two parts, generated by XeCl
(B-X) and (C-A) transitions respectively. Suppose U b-x (A) and U c -a (A) are the time- integrated spectral intensities, respectively, for XeCl (B-X) and (C-A) transitions, their time-integrated signal intensities can be correspondingly written as
Ab x = j dX • U b-x (A) ■ Spr{A) • Tugn(A) • Qa • ■Z'pt , (A-5) and f dX ■ U c -a (A) • Spt(X) • Tugii(A) • Qa • Zpt , Ac-A = (A-6) with
Ab-x + Ac -a = A (A-7) and UB-x(A) + Uc-A(A) = [7(A) . (A-8)
If the shapes of the spectral distributions of the XeCl (B-X) and (C-A) transitions are known, their absolute values can be readily deduced from the measured values for Ab-x and Ac-a according to the formulae (A-5) and (A-6). Therefore the radiated energies can be easily obtained by integrating U b-x (A) and U c -a (A) over the wavelength.
21 REFERENCES
[1] B. Eliasson and U. Kogelschatz: Appl. Phys. B46 (1988), 299-303. [2] H. Kumagai and M. Obara: IEEE Transactions on Plasma Science 16 (4) (1988), 453-458. [3] H. Kumagai and M. Obara: Appl. Phys. Lett. 54 (26) (1989), 2619-2621. [4] H. Kumagai and M. Obara: Appl. Phys. Lett. 55 (15) (1989), 1583-1584. [5] B. Eliasson and U. Kogelschatz: IEEE Transactions on Plasma Science 19 (1991), 309-323. [6] B. Gellert and U. Kogelschatz: Appl. Phys. B52 (1991), 14-21. [7] T. Letardi, P. Di Lazzaro, G. Giordano, G. Schina, A. A. Carabelas: ‘Generazione di radiazione ultravioletta mediante lampade ad eccimeri’ (Generation of UV radiation by excimer lamps), presented in 75th National Conference of Italian Society of Physics, Cagliari, Sept. 28 - Oct. 3, 1989. [8] A.M. Boichenko, V.S. Skakum, V.F. Tarasenko, E.A. Fomin, and S.I. Yakovlenko: Laser Physics, 3 (4) (1993), 838-843. [9] A.M. Boichenko, V.S. Skakum, V.F. Tarasenko, E.A. Fomin, and S.I. Yakovlenko: Laser Physics 4 (3) (1994), 635-637. [10] American Electricians Handbook, ed. by C.C. Carr, McGraw Hill Book Comp., Inc., New York. Sec. 10-73 — Sec. 10-87. [11] T.H. Maiman, R.H. Hoskins, I.J. D’Haenens, C.K. Asawa, and V. Evtuhov: Phys. Rev. 123 (4) (1961), 1151-1157. [12] J.F.Holzrichter and J.L. Emmett: Appl. Opt. 8 (7) (1969), 1459-1464. [13] J.L. Emmett and A.L. Schawlow: Appl. Phys. Lett. 2 (11) (1963), 204-206. [14] P. Millet, A. Birot, H. Brunet, J. Galy, B. Pones-Germain, and J.L. Teyssier: J. Chem. Phys. 69 (1) (1978), 92-97. [15] V.M. Borisov, A.M. Davidovskii, and O.B. Khristoforov: Sov. J. Quantum Elec tron. 12 (11) (1982), 1403-1408. [16] H. Asselman, P. Hives, J. Galy, H. Brunet, and J.L. Teyssier: J. Phys. B: At. Mol. Opt. Phys. 26 (1993), 2311-2322. [17] A.Kh. Amirov, O.V. Korshunov, and V.F. Chinnov: J. Phys. B: At. Mol. Opt. Phys. 27 (1994), 1753-1771. [18] S. Bollanti, T. Letardi, and C.E. Zheng: IEEE Trans, on Plasma Science 19 (2) (1991), 361-368. [19] N.G. Basov, A.S. Bashkin, P.G. Grigor’ev, A.N. Oraevskii, and O.E. Porodinkov:
22 Sov. J. Quantum Electron. 6 (9) (1976), 1128-1130. [20] S.I. Andreev, I.M. Belousova, P.N. Dashuk, D. Yu. Zaroslov, E.A. Zobov, N.V. Karlov, G.P. Kuz’min, S.M. Nikiforov, A.M. Prokhorov, A.N. Sidorov, L.L. Chel- nokov, and M.D. Yarysheva: JEPT Lett. 21 (7) (1975), 194-195. [21] S.I. Andreev, I.M. Belousova, P.N. Dashuk, D. Yu. Zaroslov, E.A. Zobov, N.V. Karlov, G.P. Kuz’min, S.M. Nikiforov, and A.M. Prokhorov: Sov. J. Quantum Electron 6 (8) (1976), 931-934. [22] V.Yu. Baranov, V.M. Borisov, A.M. Davidovskii, and O.B. Khristoforov: Sov. J. Quantum Electron. 11 (1) (1981), 42-45. [23] V.M. Borisov, A.M. Davidovskii, S.G. Mamonov, and O.B. Khristoforov: Sov. J. Quantum Electron. 13 (5) (1981), 681-682. [24] J.L. Emmett and A.L. Schawlow: J. Appl. Phys. 35 (9) (1964), 2601-2604. [25] J. Tellinghuisen and M.R. McKeever: Chem. Phys. Lett. 72 (1) (1980), 94-99. [26] E. Quinones, Y.C. Yu, D.W. Setser, and G. Lo: J. Chem. Phys. 93 (1) (1990), 333-344. [27] D. Lo and C.E. Zheng: J. Phys. D:Appl. Phys. 20 (1987), 714-717. [28] P.J. Hay and T.H. Dunning Jr.: J. Chem. Phys. 69 (1978), 2209.
23 Edito dall' Funzione Centrale Relazioni Lungotevere Grande Ammiraglio Thaon di Revel, 76 - 00196 Roma Stampa: RES-Centro Stampa Tecnografico - C. R. Frascati
Finite di stampare nel mese di novembre 1996