Physical Phenomena in a Coplanar Macroscopic Plasma Display Cell II. Comparisons between Experiments and Models R. Ganter, J. Ouyang, Th. Callegari, J.P. Boeuf CPAT, Université Paul Sabatier, 118 Route de Narbonne, 31062 Toulouse cedex, France

ABSTRACT Measurements of infrared and visible emission in a macroscopic discharge cell similar to a Plasma Display Panel cell are analyzed using a two-dimensional fluid model of the discharge. The comparisons between experiments and models show a good qualitative agreement but the plasma spreading velocity above the cathode surface is much faster in the experiments. We find that including photoemission in the model considerably increases the agreement between experiments and models. With a well-chosen photoemission coefficient, the model reproduces the trends observed in the experiments when the gas mixture (between 2% and 10% of xenon in neon) or the applied voltage is changed. The influence of photoemission on the current rise time and on the velocity of plasma spreading above the dielectric surfaces is more important in the macro-cell than in a similar (same dimension x pressure) PDP cell because resonant photon transport does not follow the similarity laws.

1

I. INTRODUCTION A large Research and Development effort is devoted at increasing the luminance and luminous efficiency of Plasma Display Panels (PDPs) for the market of large-area television displays. A number of recent papers on diagnostics1-4 and modeling5-12 of PDP discharge cells have presented studies of the basic physical phenomena occurring in a PDP cell. An experiment on a macroscopic plasma display discharge cell has been set up in the last few years (see Ref. 13 and the companion paper - paper I14). The geometry of this macro-cell is similar to a real PDP cell but its size is typically 100 times larger and its operating pressure 100 times less than in a real cell. The similarity laws of glow discharges allow deducing the properties of a real micro-discharge PDP cell from those of a macro-cell provided that the pd product (pressure multiplied by dimension) of both cells is the same. The macro-cell is much easier to study because of larger dimensions and time scale (according to the similarity laws, dimension and time in similar discharges scale like 1/p, where p is the pressure). A parametric study of the discharge is obviously much easier on the macro-cell and we believe that the macro-cell can be very useful for 1) parametric study of the discharge and plasma properties, 2) search for conditions of higher efficiency, 3) validation of the models. However it is well known that the similarity laws can be violated by some physical processes occurring in the plasma such as recombination or three body collisions14. Keeping these limitations in mind, we still believe that the information brought by the study of the macro-cell can still help improving the performances of PDP cells. In this paper we use the macro-cell experiments to check the validity of 2D fluid models developed previously7-9. The experimental results described in paper I are compared with the model results. In section II we compare the space and time evolution of the plasma as observed in the experiments and predicted by the models. We show in section III that the discrepancies between experiments and models are due to photoemission and that very good agreement between experiments and models can be achieved when photoelectron emission is included in the model. A general discussion is presented in section III. II. PLASMA FORMATION AND DECAY IN THE MACRO-CELL AND IN THE MODELS The experimental set-up and conditions are the same as in paper I. Comparisons between experiments and models are presented here for different gas mixtures (2%, 5%, and 10% xenon in neon) and for different values of the square wave sustaining voltage. The pressure is set to 5.6 torr in all the experiments and calculations below. The 2D fluid model has been described in previous publications7-9. The two coplanar electrodes are perpendicular to the simulation domain of the 2D model while the address electrode coincides with one side of the simulation domain. The simulation domain is therefore comparable with the side view of the cell (see Fig. 1 of paper I14) and only the emission measurements from the side view of the discharge cell will be used in the comparisons. In the first part of this paper the space and time variations of the infrared and visible emission (see paper I) are compared with the calculated space and time variations of the power dissipated by electrons into xenon and neon excitation. These comparisons are only qualitative, but we expect the space and time distributions of the measured emission and the calculated power deposition in excitation to be very similar. The measured infrared emission (823.1 nm and 828.0 nm) corresponds to the decay of upper excited states of xenon to the metastable and resonant states. Since this decay is fast (~30 ns15) compared to the time scale of the discharge, the infrared emission gives a good image of xenon excitation. The visible emission (640.2 nm) measured in the experiments corresponds to the 2p9 -1s5 transition of neon. The decay of the 2p9 state is also fast16 and this

2

visible emission should be qualitatively comparable to the calculated power deposition by electrons in neon excitation. The measured and calculated electric properties of the discharges (current pulse, charge transferred to the dielectric surfaces) are compared in the second part of this section. A. Measured emission and calculated excitation Figure 1 shows the space and time variations of the measured infrared and visible emission from the side of the macro-cell. The measurements have been made with an ICCD camera using optical filters (see Paper I14). As described in paper I, xenon is excited above cathode (negative glow region) and above anode. The plasma above anode is closer to the surface and one can see the formation of striations in the infrared emission. The simulated power deposition in xenon and neon excitation is displayed in Fig. 2. In agreement with experimental results, the model predicts practically no neon excitation above anode. The time evolution of the measured emission of Fig. 1 and simulated excitation in Fig. 2 are in reasonably good qualitative agreement. The model is based on the local field approximation (see e.g. Ref. [7]) and therefore tends to underestimate the spatial extent of the excitation. The fluid model also does not reproduce the striations in the distribution of xenon excitation. This was expected because of the local field approximation. Note that fluid models using an energy equation for the electrons do not reproduce the striations10,11 either. An important difference between model and experiments is related to the velocity of the sheath motion and plasma spreading above cathode. This can be clearly seen by comparing the measured emission above cathode of Fig. 1 with the calculated excitation of Fig. 2. The model predicts a much slower plasma spreading above cathode than the experiments. Roughly speaking, the velocity of the maximum of calculated power deposition above cathode is about 0.2 cm/µs (the coplanar electrode width is 2 cm), while the velocity of the maximum of emission is on the order of 1 cm/µs. Figure 3 shows the sheath velocity above cathode as a function of applied voltage for different xenon-neon mixtures. This velocity is defined as the averaged velocity of the peak neon emission intensity between the time it reaches a position above the inner edge of the cathode to the time it reaches a position above the middle of the cathode. We see that this velocity sharply increases as a function of voltage and xenon content in the mixture. Measurements of the current pulse show that the sheath velocity above cathode is directly related to the current rise time. The current peak occurs once the plasma has completely covered the electrode. Dividing the electrode width by the current rise time we find an average sheath velocity that is very close to the velocity deduced from the light emission (see Fig. 3). The sheath velocity deduced from the models is typically 5 to 10 times less than in the experiments (except when photoemission is included, see section III). The measured and calculated current waveforms are discussed below. B. Measured and calculated current and charge transferred The measured and calculated current through the cathode in a Xe(10%)-Ne mixture are plotted as a function of sustaining voltage in Fig. 3 and Fig. 4 respectively. The model is two-dimensional and therefore gives a current per unit length of the discharge in the direction perpendicular to the coplanar electrode gap. In order to obtain a current from the model, we multiply the calculated current per unit length by the electrode length (2cm) in the direction perpendicular to the simulation domain (Fig. 4).

3

The capacitive current due to the voltage rise has been subtracted from the measured current in Fig. 3. The calculations assume that the voltage switches from –Vs to +Vs (Vs is the sustaining voltage) in one integration time step, i.e. almost instantaneously. The corresponding capacitive current is not plotted. In the experimental results presented here the time lag of the current is larger than the rise time of the voltage so that we expect that the fact that the actual rise time is non-zero has no significant effect on the discharge current waveform and on the space and time evolution of the discharge. However we found that some weak light emission could sometimes be measured during the voltage rise. The lag time of the experimental and calculated current increases as the applied voltages decreases. This is expected since this lag time depends on the initial charged particle densities in the gap at the beginning of the voltage pulse. These initial densities correspond to electrons and ions remaining from the previous voltage pulse and we expect that they increase when the sustaining voltage increases, and hence the lag time decreases with increasing voltage. We know that the lag time predicted by the model is not accurate since we did not attempt to include in the model an accurate description of the afterglow. We are mainly interested here in comparisons between experiment and model during the current pulse. The comparisons of Fig. 3 and Fig. 4 show that the calculated current peak is significantly smaller than the measured one (2 to 3 times) but that the duration of the calculated current pulse is longer. Moreover, the shapes of the measured and calculated currents are different. For high enough sustaining voltages, the measured current has two distinct peaks that do not appear in the calculated current. Of course we made sure that the first peak seen in the discharge current was not due to the capacitive current induced by the voltage rise (again, this current has been subtracted from the curves shown in Figs. 3 and 4). The total charged transferred can be obtained by integrating the current pulse. Figure 5 displays a comparison of the charge deduced from the measurement and calculations. We see that in spite of the discrepancy in the shapes of the measured and calculated currents, the total charge transferred during the pulse is very similar. III. INFLUENCE OF PHOTOEMISSION IN THE MODELS The discrepancies between experimental and simulation results in the coplanar macro-cell in xenon-neon mixtures are larger than expected. The agreement between the measured and calculated current pulse and plasma spreading velocity in pure neon in the matrix (and coplanar ?xx) case are much better than in the mixture. In this section we first show that including a simple model of photoemission in the model considerably improves the agreement between experiments and model. We then analyze the consequences of photoemission on the dynamics of the sheath-plasma boundary and on the shape of the current waveform. A. Including photoemission in the models As mentioned in paper I, the current rise and sheath motion velocity are controlled by 1) ionization in the discharge volume, 2) ion drift velocity in the sheath, and 3) secondary emission coefficient. The plasma spreading or sheath motion velocity can be much faster than the ion velocity if the ionization rate and the secondary emission coefficient are large. The model should be able to reproduce the current rise time and the plasma expansion velocity if we assume that our estimation of the ionization rate, the ion drift velocity and the secondary emission coefficient are accurate enough. We studied the variations in the calculated current pulse for different values of the secondary electron emission coefficient due to xenon or neon ions. It was not possible to reproduce the measured current waveform for reasonable values of these coefficients. The two characteristic

4

peaks of the measured current did not appear in the simulation results. Variations of the ion drift velocity within the estimated error bars did not lead to significant change in the calculated currents. Obviously the ion velocity imposes a limit on the current rise time and plasma spreading velocity since the ions need to move from the plasma to the surface above cathode to produce secondary electrons, which multiply in the high sheath field, and contribute to the plasma motion toward cathode. This process could be considerably accelerated if the secondary electrons emitted by the surface were generated in a shorter time than the time needed for ions to cross the sheath. Photoemission is therefore an obvious possible candidate to explain the discrepancy between measurements and calculations. In order to test the possible influence of photoemission in the calculations we included this process in the model. The goal was only to validate or invalidate the hypothesis that the observed discrepancy was due to photoemission, so we developed a very simple, semi-quantitative model of photoemission in the cell. This model is described below. At a given time of the simulation the 2D fluid model provides the spatial distribution of the total xenon excitation rate by electron impact. The distribution of this rate is similar to the distribution of the power dissipated into xenon excitation shown in Fig. 2. We can assume that on the average, each xenon excited state will lead to the generation of β resonant UV photons where β is a constant coefficient. Since the upper excited states of xenon decay to the resonant and metastable state, it is not unreasonable to assume that β will be on the order of 0.1-0.5. We then assume that the UV photons are emitted isotropically and we neglect radiation trapping. Under these assumptions, the flux of photons from the discharge volume to the dielectric surfaces can be easily deduced from the knowledge of the xenon excited state production rate and the β coefficient. Assuming a given value of the photoemission coefficient γph we can deduce the flux of photoelectrons emitted by the dielectric surfaces. This model of photoemission is very approximate since it neglects imprisonment of the resonant photons due to successive emission and re-absorption. The consequences of this simplification are discussed in section IV. The purpose here is only to check whether or not photoemission is a possible explanation of the experimental results, and not to present a detailed model of photon transport in the discharge cell. Since the coefficients β and γph are not accurately known, and since only the product (γph*= β γph) is of interest in this model, we tried to adjust γph* to obtain a good match between measured and calculated current. The results are shown in Fig. 6 for a coefficient γph* =10-3. We find that for this value of γph* the model results fit very well the experimental data (as mentioned above we do not consider here the lag time but only the shape of the current pulse). The peak current and current durations are now in much better agreement in a relatively large voltage range. Moreover, for large enough applied voltages, the simulated current presents two distinct peaks which are strikingly similar to those of the measured current, the first peak becoming more important when the voltage increases. For the same value of γph* , the model reproduces well the experimental trends when the voltage is changed and when the gas mixture is changed. This can be seen in Figs. 7 and 8 where the experimental and calculated currents in a 5% Xe in Ne mixture are plotted for different values of the applied voltage, and in Figs. 9 and 10 where similar results are presented for a 2% Xe in Ne mixture. Taking into account photoemission in the model also considerably affect the space and time variations of the power dissipated in the gas. Figure 11 displays the space and time variations of the calculated power deposition by electrons in xenon and neon excitation when photoemission is included. This figure can be compared with the calculations without photoemission (Fig. 2) and with the experimental measurements of the infrared and visible emission (Fig. 1). We find that when photoemission is included the qualitative agreement between the calculations of power

5

deposition of Fig. 11 and the emission measurements of Fig. 1 is much better. The spreading of the plasma above cathode is now considerably faster than without photoemission (compare Fig. 2 and Fig. 11). Figure 12 shows the time variations of the position along the surface of the maximums of xenon emission (experiments) or xenon excitation (models) deduced from the experiments and the model with and without photoemission. The velocity of plasma spreading can be deduced from this plot. We see that there is a good match between experiments and model when photoemission is included, but the model without photoemission gives plasma spreading velocities which are 5 to 10 times lower. We also see in Fig. 1 that the plasma is not as close to the dielectric surface above the coplanar electrodes than in Fig. 2 (without photoemission), which is coherent with the experimental observations. Also, the collected current on the address electrode (not shown here) is larger when photoemission is included and is in much better agreement with the experimental measurements. B. Analysis of the consequences of photoemission We discuss here the effect of photoemission on the plasma spreading and on the shape of the current waveform. Photoemission accelerates the plasma spreading above cathode because, as mentioned above, the secondary electrons are generated much faster on the dielectric surface above cathode than when only ion induced secondary emission takes place. This leads to a faster growth of the discharge current. Since the charging of the surface is faster when photoemission is important, the current pulse duration is shorter. It is interesting to note that the total charge transferred (integral of the current) during the current pulse as predicted by the model (Fig. 3) does not change too much when photoemission is included. The model also help explaining the particular shape of the current waveform which presents two well separated peaks in the experiments, and in the model when photoemission is included. The first current peak is related to the fast plasma growth. During this first current rise the number of charged particles collected by the surface is small (ions have not yet arrived on the surface, and the photoelectron current is small). However the fast growth of the plasma induces a fast increase of the electric field above the cathode surface. This is because the cathode sheath contracts while the plasma expands. It is therefore clear that the first current peak is associated with the displacement current on the cathode surface which is due to sheath contraction and motion above the cathode surface. This peak is not present when photoemission is not included in the calculations because the plasma growth is much slower in that case and the displacement current on the cathode surface is small. The second peak of the current corresponds to the collection of ions on the dielectric surface above cathode. IV. DISCUSSION We have shown in the previous section that photoemission seems to play a non negligible role in the plasma growth in our conditions. We briefly discuss here the question of photon transport in the conditions of the macro-cell experiment and in real PDP cells. As mentioned above and in paper I similar experiments have been performed at high pressures, in real PDP cells. The plasma spreading velocity that can be deduced from these measurements is close to the velocity predicted by the model without photoemission, i.e. on the order of 0.1 cm/µs. In the macro-cell the plasma spreading velocity is much faster and the only way to reproduce this velocity (and the measured current) in the models is to include photoemission. Let us focus on the UV photons emitted by the plasma and which have a non zero probability to reach the dielectric surface above cathode before the end of the current growth. Those photons are

6

necessarily resonant photons because 1) in the low pressure case the production of excimer photons is small (three body collisions are not important), and 2), in the high pressure case, the formation of excimer state Xe2* is too slow as shown in the following. The three body collisions involving atomic xenon excited states with neon and xenon atoms are: Xe* + Xe + Ne
Xe2 * + Ne (rate 4.1 10-32 cm6s-1 for the resonant state Xe*(3P1), and 1.35 1032 for the metastable state Xe*(3P2))17 , and Xe* + Xe + Xe
Xe2 * + Xe (rate 1.55 10-31 cm6s-1 for the resonant state Xe*(3P1), and 8.53 10-32 cm6s-1 for the metastable state Xe*(3P2))17 . A rapid calculations shows than in our mixtures, UV photons emitted by the xenon excimer states have no chance to be emitted and reach the dielectric surface above cathode before the end of the current growth. * 3 For example using the rate of the first reaction for three body collisions involving Xe ( P1), we obtain a

time constant around 1 µs in a 10% Xe-Ne mixture at 560 torr. The current rise time is on the order or less than 50 ns under these conditions1,18xx. Therefore the probability that UV photons from the excimer states reach the surface above cathode is small in the high pressure and low pressure case. If we know look at the probability of resonant photons reaching the surface during the current rise, we have to consider radiation trapping. The natural lifetime of the xenon resonant state is 3 ns17. The Holstein theory19 shows that the apparent lifetime, which is the results of successive emissions and re-absorptions is related to the real lifetime (in the case of a simple cylindrical geometry) by:

[π k (ν 0 ) D ] τ τa = =τ g B where ta is the apparent lifetime, t the natural lifetime, g the escape factor, k(n0) the absorption frequency in the center of the line, B=1.15, and D is the diameter of the cylinder. This formula must be slightly modified in gas mixtures20,21 . For D on the order of 100 µm, (typical distance between dielectric surfaces above cathode and anode in a PDP cell) the apparent lifetime is about 100 times larger than the natural lifetime, i.e. 300 ns in our conditions5,12 . The current rise time is on the order of 50 ns in a coplanar real PDP cell so that the probability of resonant photon reaching the surface during the current rise is small. Since the apparent lifetime given in the equation above scales as the square of the cell dimension, and our macro-cell is 62.5 times larger than a real cell, we expect an apparent life time about 8 times larger in the macro-cell than in the real cell, i.e. on the order of 2.5 µs. The current pulse duration in similar discharges scales as 1/p, i.e. we expect the current rise time to be 62.5 times larger in the macro-cell. Therefore it appears that the apparent life time is not much larger than the expected duration of the current rise in the macro-cell conditions, which means that some resonant photons are much more likely to reach surface during the current rise in the macro-cell than in a real PDP cell. The discussion above shows that photoemission during current rise may be much more important in the low-pressure macro-cell than in a similar, high pressure, PDP cell because resonant photon transport does not follow the classical glow discharge similarity laws. The difference between the high pressure and low pressure cells with respect to photoemission could also be due to a different state (cleanliness, purity) of the MgO surface which would lead to different photoemission coefficients. The MgO surface used in the macro-cell was prepared in the same way as in real PDPs, but the dielectric surfaces in a real PDP undergo a number of processes which were not used for the macro-cell. Also the discharge chamber was heated to only 200 ° C before the experiments while temperature as high as 400 ° C are used during the fabrication process of PDPs. We also noted that measurements and simulations of the plasma formation and decay, and of the current pulse were in much better agreement in pure neon. The plasma expansion velocity measured 1/ 2

7

in a macro-cell with a matrix electrode configuration in pure neon was in good agreement with the models without including photoemission. Estimation of the apparent life time of resonant UV photons from the 1s states in neon give similar values than in neon, and it is therefore not clear why photoemission does not need to be included in the model to obtain a good match with the experimental results. V. CONCLUSION

Fluid models have been used to compare the space and time evolution of the infrared and visible emission from xenon and neon measured in a macroscopic PDP, cell with the calculated distribution of power deposition in xenon and neon excitation. The measured and calculated current waveforms have also been compared, for different values of the applied voltage and for different xenon-neon mixtures. Both experiments and model show that the electric energy above anode is large enough to significantly excite xenon but too small for neon excitation. We also find that the plasma expansion and current rise are faster in the experiment than in the simulation. Also the measured plasma expansion is much faster in the macro-cell than in a real PDP cell. The model results for the plasma spreading velocity are in relatively good agreement with the measurements in real PDPs, but not in the macro-cell. A much better agreement between experiments and model in the macro-cell can be obtained when photoemission is included (in a simplified way) in the model. Both plasma spreading velocity and current rise time are in much better agreement, and the trends with changes in applied voltages and gas mixtures in the experiments and in the simulations are identical. The existence of two well separated peaks in the measured current under some conditions is also reproduced by the models when photoemission is included. The differences in plasma spreading velocity in the macro-cell and in real PDP cells could be due to the fact that radiation trapping does not follow the scaling low. The time taken by the resonant photons to reach the dielectric surfaces scales as the square of the discharge dimension, i.e. as 1/p1/2 for “similar” discharges. Therefore it appears that UV photons do not reach the dielectric surface fast enough (i.e. before the end of the current rise) in a real PDP cell. Another possible reason for the differences between the macro-cell and a real PDP cell with respect to photoemission effects could be that the state of MgO surface in the maco-cell is not identical as in a PDP cell. ACKNOWLEDGEMENTS

This work has been supported partly by Thomson Plasma and by the French Ministry of Research and Technology in the frame of the RMNT program.

8

REFERENCES

[1] K. Tachibana, S. Feng, T. Sakai, J. Appl. Phys. 88, 4967 (2000) [2] T. Yoshioka, L. Tessier, A. Okigawa, K. Toki, Journal of the SID, 8, 204 (2000) [3] C. K. Yoon, J. H. Seo, K.-W. Whang, IEEE Trans. Plasma Science 28, 1029 (2000) [4] M. Sawa, H. Uchiike, K. Yoshida, Journal of the SID, 8, 163 (2000) [5] J. Meunier, Ph. Belenguer, and J.P. Boeuf, J Appl Phys 78, 731 (1995). [6] Vera [7] J.P. Boeuf, C. Punset, A. Hirech, and H. Doyeux, J. Phys. IV France, 7, C4-3 - C4-14 (1997) [8] C. Punset, J.-P. Boeuf, and L.C. Pitchford, J. Appl. Phys. 83 1884 (1998) [9] C. Punset, S. Cany, and J.P. Boeuf, J. Appl. Phys. 86, 124 (1999) [10] S. Rauf and M. J. Kushner, J. Appl. Phys. 85, 3460 (1999) [11] G.J.M. Hagelaar, M.H. Klein, R.J.M.M. Skijkers, G.M.W. Kroesen, J. Appl. Phys. 89, 2033 (2001) [12] G.J.M. Hagelaar, M.H. Klein, R.J.M.M. Skijkers, G.M.W. Kroesen, J. Appl. Phys. 88, 5538 (2000) [13] Th. Callegari, R. Ganter, J.P. Boeuf, “Diagnostics and Modeling of a Macroscopic Plasma Display Panel Cell”, J. Appl. Phys. 88, 3905 (2000) [14] R. Ganter, J. Ouyang, Th. Callegari, J.P. Boeuf, companion paper, J. Appl. Phys [15] W.J. Alford, J. Chem. Phys. 96, 4330 (1992) [16] neon [17] J. Galy, K. Aouame, A. Birot, H. Brunet, and P. Millet, J. Phys B 26, 447 (1993) [18] experimental current in PDP [19] T. Holstein, Phys. Rev. 72, 1212 (1947) ; T. Holstein, Phys. Rev. 83, 1159 (1951) [20] Mikoshiba [21] N. Sewraj, J.P. Gardou, Y. Salamero, P. Millet, Phys. Rev. A 62, 052721 (2000)

9

FIGURE CAPTIONS

Figure 1: Side view of the xenon infrared emission (820-830 nm) and neon visible emission (635-45 nm) at different times of a discharge pulse in the macro-cell obtained with an intensified CCD camera. The macro-cell geometry is shown in Fig. 1 of paper I1. The gap between coplanar electrodes is 0.5 cm, their width is 2 cm, and the gas gap spacing is 1 cm. The gas mixture is Xe(10%)-Ne at 5.6 torr. The applied voltage is a square wave of 190 V amplitude, rise time less than 1 µs, and 100 Hz frequency. Figure 2: Calculated power deposition in xenon (top), and neon (bottom) excitation in the conditions of Fig. 1, when photoemission is not included. Figure 3: Measured plasma spreading velocity above cathode as a function of applied voltage and for different gap mixtures. The velocity is deduced 1) (solid symbols) from the motion of the maximum light emission above cathode, and 2) (open symbols), from the current rise time (see text). Figure 4: Measured current pulses as a function of applied voltages in a Xe(10%)-Ne mixture at 5.6 torr. Figure 5: Calculated current pulses in the conditions of Fig. 5. Figure 6: Comparison of the measured current and the current calculated with and without including photoemission for an applied voltage of 240 V in a Xe(10%)-Ne mixture at 5.6 torr. Figure 7: Measured current pulses as a function of applied voltages in a Xe(5%)-Ne mixture at 5.6 torr. Figure 8: Simulated current pulses as a function of applied voltages in a Xe(5%)-Ne mixture at 5.6 torr with photoemission included in the model (γph’=10-3 ) Figure 9: Measured current pulses as a function of applied voltages in a Xe(2%)-Ne mixture at 5.6 torr. Figure 10: Simulated current pulses as a function of applied voltages in a Xe(2%)-Ne mixture at 5.6 torr with photoemission included in the model (γph’=10-3 ) Figure 11: Calculated power deposition in xenon (top), and neon (bottom) excitation in the conditions of Fig. 1 and Fig. 2, when photoemission is included (to be compared with Fig. 2). Figure 12: Time evolution of the positions of maximum infrared xenon emission above cathode and above anode from the measurements and the simulations (with and without photoemission included in the model)

10

cathode anode

823 nm 828 nm

2.1 µs

2.3 µs

2.7 µs

anode

cathode

1.9 µs

640 nm

0.5

1.0

2.0

4.0

Figure 1

11

8.0

12

cathode anode

Xe 2.7 µs

4.2 µs

7.7 µs

anode

cathode

1.2 µs

Ne

0.1

0.3

1.0

Figure 2

12

3.0

10.

6

Averagre velocity (10 cm/s)

5 2 1 Ne-Xe 2% Ne-Xe 5% Ne-Xe 10% solid : from images open : from rise time

0.5

160

180

200 220 240 Applied voltage (V)

Figure 3

13

260

Current (mA)

4

Exp 10% Xe

260 V

3 2

240 V 220 V 200 V

1 0 0

2

190 V

4

Time (µs)

Figure 4

14

6

8

260 V

Current (mA)

1.5

Model 10% Xe ' γph =0

240 V 1.0

220 V

200 V 190 V

0.5 0.0 0

2

4

6

8

10

Time (µs)

Figure 5

15

12

14

Current (mA)

5 4

Mod. '

γph =10

3

-3

Exp.

2

Mod.

1 0 -1

Xe(10%)-Ne 240 V

(no photoemission)

0

1

Time (µs)

Figure 6

16

2

3

Current (mA)

4 260 V

Exp 5% Xe

3 2

240 V 220 V

1 0 0

2

200 V

190 V

4

6

Time (µs)

Figure 7

17

260 V

Current (mA)

1.5

Model 10% Xe ' γph =0

240 V 1.0

220 V

200 V 190 V

0.5 0.0 0

2

4

6

8

10

Time (µs)

Figure 8

18

12

14

Current (mA)

1.5

220 V

Exp 2% Xe 210 V

1.0 190 V 180 V

0.5 0.0 0

10

Time (µs)

Figure 9

19

1.5

Model 2 % Xe ' -3 γph =10

Current (mA)

220 V 1.0

210 V 190 V

0.5

0.0 0

180 V

2

4

6

Time (µs)

Figure 10

20

8

10

cathode anode

Xe 1.31 µs

1.56 µs

2.0 µs

anode

cathode

1.10 µs

Ne

0.1

0.3

1.

Figure 11

21

3.0

10.

3

Mod. γph =10

'

cathode

Exp. ' Mod. γph =0

-3

2 anode

Position (cm)

4

1 0 -5

0

Time (µs)

Figure 12

22

5