Efficiency of AC Plasma Display Panels R. Ganter, Th. Callegari, L.C. Pitchford, J.P. Boeuf CPAT, Université Paul Sabatier, 118 Route de Narbonne, 31062 Toulouse cedex, France

To do: fix arrows in fig. 2 fix limit = 1.0 in figs. 4 fix y-axis in fig; 5 some estimate of E/p in anode region? references

Abstract The luminous efficacy of ac plasma display panels is relatively low, on the order of 1 lm/W. We discuss in this paper the reasons for this low efficiency. We also give an estimation of the maximum discharge efficiency in producing UV photons which can be expected in “classical” plasma display panels and discuss possible ways of improvement.

I. Introduction In AC Plasma Display Panels (AC PDP's) each pixel is a micro-discharge cell where a transient plasma can be generated. The UV photons emitted by the plasma are transformed into visible photons by phosphors which are deposited on the cell walls. Most PDP's (AC PDP's) now use dielectric barrier discharges where the electrodes are covered with a thin (30 µm) dielectric layer. A square wave voltage in the 100 kHz range is applied between the line and column electrodes defining each discharge cell. The electrodes are deposited on two parallel glass plates separated by a gas gap of about 100 µm filled with a rare gas mixture (typically Xe-Ne) at a pressure in a 300-500 torr range. Xenon is used because it is a good UV emitter, and neon because it can help reduce the operating voltage. The surfaces of the dielectric layers above the electrode are covered with a thin (0.5 µm) MgO layer. The lower operating voltage in Ne is due to the large secondary electron emission coefficient of Ne ions on MgO. Because of the dielectric layer above the electrodes, the “ON state” of a discharge cell in an AC PDP is necessarily characterized by a succession of transient discharge pulses. Applying a voltage pulse above the breakdown voltage turns on a cell. The cell is kept in the ON state by applying a sustaining voltage below breakdown. This is possible because of the “memory charges” which are deposited by each discharge on the dielectric surfaces. At each half cycle of the applied voltage, the voltage induced by these memory charges adds to the voltage applied between the electrodes and a new discharge is initiated. This discharge extinguishes due to the charging of the dielectric and is reinitiated at the next half cycle. Numerical models of PDP discharge cells have been developed over the last few yearsref and have helped understand the limitation of the efficiency in these devices. In this paper we do not intend to calculate very accurately the discharge efficiency but we want to simply estimate this parameter in order to understand more clearly the limitations and to provide some guidance toward better efficiency. We will mainly discuss the discharge efficiency for producing xenon excited states. We estimate that almost all excitation events result in a UV photon, but there are some open questions concerning collisional energy losses of the xenon excited states In section II we discuss the global efficiency of an AC PDP discharge cell based on measurements and simulations results. In section III, we use a Boltzmann solver to estimate the efficiency of electron energy deposition under a uniform electric field in a Xe-Ne mixture, as a function of Xe concentration. In section IV we discuss the possible collisional excitation losses. Comments on the efficiency of steady state Townsend or glow-discharges in Xe-Ne mixtures based on simple analytic considerations are presented in Section V. In section VI the efficiency of matrix and coplanar AC PDPs is analyzed, and we discuss possible ways to improve this efficiency.

II. Global efficiency of an AC PDP discharge The measured luminous efficiency of AC PDP discharges is rather low, on the order of 1 lm/Wref. This means that only about 0.5% of the (JP - 0.4% is the number given in fig. 1) of the total electric power is converted into useful visible photons. Figure 1 shows schematic diagram of the energy conversion in a PDP cell which is consistent with what is known from experiments and simulations. In previous papers, we used a discharge model to study the electron energy balance during a transient discharge in a PDP cell in a matrix geometry (in which the discharge occurs between opposite substrates), and we found (Refxx Meunier) that typically 10% of the total electric energy dissipated in the discharge is converted to UV photons. We will discuss below in more detail below the energy balance in the discharge. Roughly speaking, we estimate that only 40% of these photons efficiently reach the phosphors deposited on the cell (this figure depends on the cell and wall geometry). Our estimate is based on Monte Carlo calculations of the radiation transport in a matrix geometry, but it is consistent with other estimates in other

references. Ref. UV photons are then converted into visible photons by the phosphors. Due to the difference between the UV photon energy and the visible photon energy, 25% or less (depending on the conversion efficiency of the phosphors) of the UV photon energy is converted into visible photon energy (the UV photon wave length in xenon is in the 150-170 nm range and the visible photon wave length is about three times larger). At this stage, only 1% of the initial electric energy is carried by the visible photons emitted by the plasma. A further reduction in efficiency is due to the cell geometry; only 40% of the visible photons emitted by the discharge reach the front plate. The overall efficiency is therefore less than 0.5%. This figure is consistent with measured values of luminous efficacy of about 1 lm/W (Refs.xx). The diagram of Fig. 1 is based on estimates only (except for the discharge efficiency in UV photons production which was obtained from detailed modelling), but we think that it gives a good representation of the energy conversion scheme in an AC PDP cell. (Reference Weber, Dauyou?) It is clear from this diagram that the photon collection efficiency is a key element in the overall efficiency. Another important consideration for the efficiency is the phosphor conversion efficiency, and this could be enhanced by using longer wavelength UV radiation (from different gas mixtures) to excite the phosphors. By far the most important factor in the overall energy balance is the discharge efficiency for the conversion of electrical energy to UV radiation. In this paper we focus on the limitations and possibilities of improving the discharge efficiency for production of the xenon excited states which emit the UV photons in xenon-neon mixtures. In the rest of the paper, we will therefore characterise the discharge efficiency by the parameter η defined by: η=

Energy dissipated by electrons in xenon excitation Total energy dissipated in a discharge pulse

η will be termed as the “discharge efficiency”, or “discharge efficiency for energy (or power) deposition in xenon excitation” in the following. The parameter η can be written as: η = ηelecηexcit where ηelec =

Total energy dissipated by electrons Total energy dissipated in a discharge pulse

and ηexcit =

Energy dissipated by electrons in xenon excitation Total energy dissipated by electrons

ηexcit will be termed as “fractional electron energy (or power) dissipated in xenon excitation”, and ηelec is the “fractional discharge energy (or power) dissipated by the electrons”. The fractional electron energy dissipated in xenon excitation, ηexcit , is described in section III and the efficiency for

the conversion of this excitation energy to UV radiation is discussed in Section IV. The fraction ηelec of the total energy which is dissipated by electrons is discussed in section V.

III. Electron Energy Deposition in a xenon-neon mixture Before looking at the energy balance in a real PDP glow discharge, it is very useful to study the energy deposition by electrons in a uniform electric field in a xenon-neon mixture, as a function of the reduced electric field E/N. We have used a two-term Boltzmann solver, BOLSIGref to solve the zero-dimensional electron Boltzmann equation for different gas mixtures and reduced electric fields. The form of the Boltzmann equation used corresponds to the Pulsed Townsend (volume-integrated) experimentref and is written as: e ∂F − E. = I [ F ] − νi F write in 0D form m ∂v e ∂F − Ex = I [F ] − υi F m ∂v x

where F is the normalized electron velocity distribution function. The term on the left represents ρ ρ e ρ the acceleration (a = − E ) of the electrons in the electric field E = E x eˆ x . The first term on the m right is the collision integral which is given in detail in ref. Xx. In a uniform field, electron impact ionization leads to an electron number density (volume-integrated) which increases exponentially in time. The time derivative in the Boltzmann equation is thus replaced by the last term on the right, where υi is the average ionization frequency. The only collisional processes considered are electron collisions with ground state atoms of Ne and Xe. The set of cross-sections used Ne and Xe are the same as in Ref.xxmeunier. Figure 3 shows fractional electron energy dissipated in xenon excitation, ηexcit , as a function of reduced electric field E/N, the ratio of the electric field strength to the neutral gas density, for different concentrations of xenon in neon. This coefficient is deduced from the gain and loss terms of the electron energy balance equation by: ηexcit =

∑ε ν

k k

k

evd E where εk is the energy threshold of excitation level k, νk is the electron excitation frequency of this level, and v d is the electron drift velocity. (The total power absorbed by the electrons is evdE). The sum is over all electron impact excitation process from the xenon ground state which are included in the cross-section set. These processes are excitation of the resonance state 3P1, of the metastable state 3P2, of Xe** which is the sum of the 6s', 6p, 5d and 7s states, and of Xe*** which is the sum of all higher states. As expected, it is shown in Fig. 3 that as the percentage of xenon in the mixture is increases, more energy is deposited in xenon excitation. Plasma display panels generally use low concentrations of xenon in neon (from 3% to 10%) because the breakdown voltage also increases with increasing xenon concentration. This is partly due to the much larger secondary emission coefficient of neon ions on MgO protective layer compared with that of xenon ions (the dielectric layers above the electrodes in an AC PDP are covered with a protective MgO layer).

Another important feature of Fig. 3 is the E/N dependence of the fractional electron energy dissipated in xenon excitation. This quantity reaches its maximum for low values of E/N and decreases sharply for E/N about 50 Td (1 Td = 10-17 V cm2). As the electric field is increased, relatively more energy is dissipated in xenon ionization, and for still larger fields, in neon excitation and ionization. This can be seen in Figs. 4 which show the fractional energy deposition in xenon excitation, xenon ionization, neon excitation, and neon ionization as a function of E/N, for different xenon concentrations. At very low E/N, a small fraction of the electron energy is converted to thermal energy in the gas as a result of elastic collisions. From the results above it can be concluded that efficient electron energy deposition in xenon in uniform field conditions can be achieved only when the electric field is low enough, on the order of a few tens of Td. Figure 5 shows the variations of the electron mean energy as a function of E/N for three values of the percentage of xenon in neon. From Figs. 4 and 5, one can see that to achieve an efficiency of 80% in xenon electron energy deposition in a uniform electric field, the mean electron energy must not exceed 6 eV (electron “temperature” of 4 eV) in a mixture with 4% xenon. This limiting mean energy is lower for mixtures with more xenon. Although, strictly speaking, these results are valid only for a uniform electric field, they clearly show that the cathode fall of glow discharge is not appropriate for achieving a high efficiency of xenon excitation by electrons. In the cathode region of a glow discharge, electrons can reach fairly high energies compared to the sheath voltage ref (especially for large sheath voltage and low sheath length), and one can expect that a large part of this energy will be used for xenon ionization, neon excitation and ionization rather than xenon excitation. We see below that another reason for the low overall efficiency of the cathode fall region in PDP's is related to energy dissipation by positive ions. JP - where to mention elastic collisions cooling of the electron energy distribution in the afterglow.

IV. UV photon generation Although our main focus in this article is on the efficiency for generation of xenon excitation in discharges, it is necessary to discuss the relation of this quantity to the efficiency for UV generation. Several authors have pointed out that collisional losses of the xenon excited states can decrease the efficiency of PDP's, but it is not clear from these references the extent to which collisional losses of excitation are important in PDP conditions. Based on the reasonable agreement, both qualitatively and quantitatively, between experiment and our modeling results for the efficiency of UV generation calculated assuming one UV photon is generated per excitation event, we feel that collisional deexcitation losses are small for PDP conditions. This is not, however, conclusive. In this section we discuss various collisional energy loss process and their possible effect on the efficiency. The UV photons in a Xe-Ne discharge are mainly emitted by de-excitation of resonant and excimer states of xenon. The energy transfer processes leading to UV emission are shown in the diagram of Fig. 2. Most of the energy deposited through electron impact excitation of xenon during the current pulse goes into the resonant (3P1) and metastable (3P2) states of xenon, either directly or by cascading down from the higher excited states. The 3P1 and 3P2 states are then depopulated by UV photon emission from the 3P1 state or by excimer formation, Xe 2*, and subsequent UV emission from these excited molecular states. The xenon excimer states are created mainly in the afterglow in three body collisions between the metastable or resonant states of atomic xenon with two rare gas atoms. Other losses of the excited xenon atomic states are discussed below. Although questions remain concerning the efficiency for conversion of excitation energy to UV photon energy, the Xe-Ne system appears to be quite efficient for UV production in AC PDP conditions.

The energy conversion factor from atomic excited states to UV photons is less than one because of the energy losses during the transitions from the upper excited states of xenon to the metastable and resonant levels and because of the energy difference between the metastable or resonant states and the excimer states. Based on previous work in our group, we estimate (see RefxxMeunier) that about 70% of the energy dissipated into xenon excitation is converted into UV photon energy. The fraction of the electron energy deposited in xenon excitation in typical AC PDP conditions is approximately 15% (XX Meunier). This estimate is consistent with results from the 1D and 2D models that we used in Refsxx for a matrix AC PDP cell in a Xe(10%)-Ne mixture. Because only 70% of this energy is converted to UV photon energy, we arrive at the estimate of 10% for the efficiency of the discharge in producing UV photons which appears in Fig. 1. Note finally that, as shown in Fig. 2, xenon excited states (and therefore UV photons) can be produced not only by electron impact excitation, but also through the recombination of electrons with molecular ions. We have however shown (unpublishedxx) than the production of UV photons through this mechanism is always less than 10% of the overall UV production in AC PDP conditions. Once the energy is in the excited levels, possible collisional energy losses include superelastic collisions, ionization of the excited states, radiation to the resonance and metastable levels, and neutral quenching. Trapping of the resonance radiation increases the effective lifetime of the resonance level and therefore increases the probability of collisional quenching of Xe*. Collisional quenching to the ground state (collisions with neutrals or superelastic collisions with electrons) proceeds at rates negligible with respect to the other processes listed above. Stepwise ionization represents a loss of electron energy equal to the ionization potential of xenon because very little of the electron energy spent in ionization of the excited states is recovered in the form of UV photons. The atomic ions are lost by transport to the surfaces and by 3-body conversion to molecular ions. Atomic ion recombination with electrons (which could result in xenon excited states) is negligibly slow, but the molecular ions can recombine with the cool electrons in the afterglow. As mentioned above, we estimate that at most 10% of the molecular ions recombine into the Xe* levels before being lost by transport. Stepwise excitation of the Xe* and Xe** levels to the higher excited states in the atomic manifold is another process which could reduce the efficiency for conversion of electron energy to UV radiation. Stepwise excitation followed by deexcitation back to the original level via superelastic collisions does not reduce the efficiency in PDP's. However, deexcitation via radiation or collisions with neutrals is deleterious because the energy is not recovered by the electrons. Nevertheless, most of the excitation energy remains in the xenon system with the potential for conversion to UV photons. On the basis of values reported in the literature for the collision rates and for typical PDP conditions, it appears that Xe* is quickly converted to Xe** in stepwise excitation events (at least during the current pulse and for a 100 ns or so into the afterglow until the electron energy has cooled to less than 1 eV). The Xe** is rapidly quenched back to Xe* by superelastic deexcitation, by radiation and by collisions with the neutrals. The neutral quenching and radiation channels dominate the deexcitation, especially for high concentrations of xenon, and thus each cycle through this loop can reduce the efficiency for conversion of excitation energy to UV radiation. While stepwise ionization rates are smaller than those for stepwise excitation, each stepwise ionization event costs about 10 times the overall electron energy loss in stepwise excitation followed by neutral quenching. Stepwise ionization, however, is only possible for a very short time because the electron energy cools rapidly in the afterglow due initially to stepwise excitation and then later to elastic collisions. After about 20 ns into the afterglow, the electron density is still high but the average electron energy is less than a few eV and stepwise ionization rates are negligibly very small. There are suggestions in the literature that stepwise ionization (presumably coupled with stepwise excitation) leads to a saturation in the UV output as a function of discharge current. From the

published results, it is not established for what conditions either of these stepwise processes affects the efficiency in the xenon system in PDP conditions. An accurate evaluate of the efficiency of conversion of excitation energy to UV photon generation would require a solution of the space and time dependent rate equations for the xenon levels coupled to the continuity equation for the electron density. The utility of such a detailed approach is limited by uncertainties in the energy transfer reaction rates and relevant cross sections.

V. Efficiency of Townsend and glow discharges In section II above, we only discussed the energy dissipated by electrons in the gas volume. In a discharge both electrons and positive ions gain energy from the field and release this energy through collisions with neutral atoms and with the walls. In this section we estimate the contribution of positive ions to the total energy balance in a Townsend discharge, and in the cathode and positive column regions of a glow discharge. The positive column and cathode regions will be considered separately. Only steady state discharges will be discussed but the general conclusions can be applied to the transient glow discharges used in AC PDP's. The electrical power absorbed by the electrons and ions in a discharge is given respectively by r r r r Pe = ∫ J e .E d 3r and Pi = ∫ Ji .E d3 r where Je and Ji are the electron and ion current densities. In PDP discharge conditions the energy absorbed by the electrons is subsequently dissipated through inelastic collisions with neutral atoms while the ion energy is dissipated through elastic and charge transfer collisions with neutral atoms and by collisions with the walls. The energy dissipated by ions therefore represents a loss of efficiency in PDP's since it leads to heating of the gas and the surfaces. A. Positive column In a (axially?) uniform positive column the electron and ion current densities can be written simply as

ρ

ρ

ρ

ρ

J e = enµe E et J i = enµi E where µe and µi are the electron and ion mobilities. The contributions of electrons and ions to the total power dissipation PT are therefore respectively: µi µe Pi = PT and Pe = PT µe + µi µe + µi Since the ion mobility is typically 100 times lower than the electron mobility, the contribution of ions to the overall energy balance in a positive column is therefore negligible (ηelec : 1 )xx and one can write in this case: ηPositiveColumn ≈ ηexcit A positive column of a discharge in a xenon-neon mixture can therefore be very efficient in producing xenon excited states if the electric field is chosen (by adjusting the charged particle losses, i.e. tube radius etc…) low enough, as discussed in section III. This is well known for fluorescent lamps where very high efficiency can be achieved if the electric field is adjusted to optimize electron excitation of the resonance lines of mercury. ref B. Townsend discharge We consider now a simple steady state 1D Townsend (uniform electric field) discharge. The electron and ion current densities in the gas gap can be written as: and J i ( x ) = J T − J e ( x) where the total current density JT, J e ( x ) = J e ,0 exp(αx ) J T = J e ,0 exp(αd ) , is constant and the electron and ion current densities at the cathode are related by

J e ,0 = γ J i (0) . d is the gap length, J e ,0 the electron current density on the cathode surface, α is the ionization coefficient (assumed here to be constant and dependent on the reduced electric field), and γ is the secondary electron emission coefficient due to ion impact (we neglect other emission processes and assume that γ is a constant). 1 In a Townsend discharge the self-sustaining condition M = exp(αd ) = 1 + must be satisfied. γ In these conditions, the fractional power dissipated by electrons is: d

J e Edx d Pe ∫0 1 = = αd ∫ eα y dy PT J T Ed de 0 Therefore, P 1 − e −α d Pi and ηelec = e = = 1 − ηelec PT αd PT The fractional power dissipated by electrons in a Townsend discharge is plotted as a function of αd in Fig. 6. Since the self-sustaining condition provides a relation between αd and γ , the corresponding value of γ is also indicated on the upper horizontal axis in Fig. 6. This result shows that the contribution of ion dissipation increases when the secondary emission coefficient decreases. For a value of the secondary emission coefficient of 0.5 (which is an upper value in PDP's, and in most cases) the maximum power dissipated by electrons is 60% of the total electric power (i.e. 40% is dissipated by ions). Therefore even if the electric field can be chosen low enough to have a large xenon excitation efficiency, let say ηexcit = 80% , the global discharge efficiency η will be less than 50% in a steady state Townsend discharge. C. Cathode region of a glow discharge Now, let us consider a steady state, 1D glow discharge, without positive column (i.e. including cathode sheath and negative glow). In this case, almost all the discharge power is deposited in the high field sheath region; very little is deposited in the low field negative glow. Thus, the ratio of the ion current to the total current in the cathode sheath determines the fractional power absorbed by the ions, and this ratio is determined mainly by γ, the secondary electron emission coefficient at the cathode. In order to simplify the analytical treatment, we assume that all the ionization occurs outside the sheath in the negative glow plasma (this becomes realistic for abnormal glow discharges, i.e. large sheath voltage, small sheath length). In that case, the ion current density is constant in the sheath and the fractional power dissipated by ions can be written as: Pi ∫ J i Edx J i (0) = : PT ∫ J T Edx JT Since J i (0) =

1 J (0) γ e

with

J T = J i (0 ) + J e (0)

, we have:

J i (0) =

JT . Therefore: 1+γ

Pi 1 P γ = , and ηelec = e = PT 1 + γ PT 1 + γ We see that, as in the case of the Townsend discharge, the fractional power dissipated by electrons decreases when the secondary emission coefficient decreases. For an ideal value of

γ = 0.5 , the fractional power dissipated by electrons according to this simple model is 33%, i.e. 67% of the total power is wasted by ions in the sheath. This simple analysis tends to overestimate the contribution of ions to the total power dissipation since we assumed a constant ion current in the sheath. However this number is not very far from the estimate of the ion power deposition obtained with a much more sophisticated approach based on a self-consistent model of an AC PDP discharge refmeunier. This detailed discharge model gave a power dissipation by ions equal to 60% of the total electric power dissipation in the discharge. Figure 7 shows the fractional power dissipated by electrons as a function of γ deduced from the equations above in the cathode region of a glow discharge and in a Townsend discharge (same as Fig. 6). These results illustrate clearly that the relatively large fractional power deposited in the ions in Townsend and in short (no positive column) glow discharges is the reason for their low efficiency for the production of UV radiation. We discuss in the next section some particular features of the discharges used for AC PDP's with emphasis on xenon excitation efficiency.

VI. Matrix and coplanar plasma display panels Two main type of electrode configurations are used in conventional AC PDP's: matrix (or double substrate) and coplanar (or single substrate) configuration. In the matrix structure the discharge occurs at the intersection of two perpendicular electrodes covered with a dielectric layer (~30 µm) and separated by a gas gap of about 100 µm. In the coplanar structure, the main discharge takes place between parallel electrodes deposited on the same plane, covered with a dielectric layer, the distance between them being on the order of 100 µm. The electrode width is typically between 100 µm and 300 µm. In the coplanar case, a third electrode, orthogonal to the two parallel electrodes is used to address the cells. The discharges in matrix and coplanar electrode configurations have been simulated in previous publications ref. In this paper we simply discuss some aspects of the discharges in these configurations, with emphasis on the xenon excitation efficiency. In the conditions of typical PDP's the relatively low pd product (on the order of 5 torr.cm, corresponding to p~500 torr and d~100 µm) does not allow the formation of a long positive column and the structure of the discharge is typical of that in a glow discharge including only a cathode sheath and a negative glow (the minimum sheath length in a typical matrix PDP discharge is between 10 and 20 µm, i.e. a large part of the total gap length). Therefore the mean energy of the electrons responsible for energy deposition in these discharges will be controlled by the large sheath electric field, and, as we have seen above, this is not good for efficient electron energy deposition in xenon. The simulations ref give a discharge efficiency in xenon excitation, ηexcit of typically 15% in a matrix configuration. However, in spite of the fact the there is no room for an efficient “classical” positive column in an AC PDP discharge, it is interesting to note that the xenon excitation and UV emission is not produced exclusively in the sheath region of these discharges. This has been shown in the simulations ref and also has been demonstrated in experiments ref. The presence of the dielectric layers in an AC PDP discharge leads to a spreading of the plasma along the dielectric surfaces above the electrodes. This spreading is due to the progressive charging of the dielectric which induces electric potential gradients along the dielectric surfaces. The electrons on the anode side, and the ions on the cathode side, are pulled by the electric field parallel to the surface induced by the charging. The simulations show that this electric field is sufficient to provide relatively important xenon excitation while the plasma spreads along the surface above anode. This is illustrated in Fig. 8 which shows the instantaneous power deposition in xenon excitation in a matrix and in a coplanar configuration, at times close to the time of maximum discharge current. The power deposition in xenon excitation in the

anode region (due to dielectric charging and plasma spreading) is much more efficient than in the cathode region (classical sheath region) because of the lower electric field. This has been recently confirmed by several experimental measurements showing that the ratio of xenon (infrared) to neon (visible) emission is much larger in the anode region than in the cathode regionref. (some idea of the E/P in this region near the anode?)

VII. Possible improvements of PDP discharge efficiency The limitation of efficiency due to large power absorption by ions in the sheath of glow discharges is an important drawback of the AC matrix and coplanar PDP's which are presently developed in the industry (and of DC PDP's previously developed). Figure 7 shows that it would be rather difficult to decrease the dissipation by ions in the sheath below 50% in any Townsend discharge or glow discharge (without positive column). The fractional power dissipated by electrons in the discharge, ηelec is therefore less than 50% and the maximum theoretical power deposition in xenon excitation, η, is also below 50%. How close to the 50% limit can be achieved for the power deposition in xenon excitation η strongly depends on the electron kinetics and on the space and time variations of the electric field in the cell. It is clear from the results of section II that one should avoid large sheath electric field (small sheath lengths) because this leads to large electron mean energy and inefficient xenon excitation. On the other hand we have seen in section V that xenon excitation in the anode region is more efficient because of the low electric field parallel to the surface induced by the charging of the dielectric surfaces. It seems therefore reasonable to seek for conditions where the plasma spreading along the dielectric surface above anode is enhanced. We have also seen that, as is well known in the lamp industry, a positive column plasma can be much more efficient for excitation of specific states if the electric field is well adjusted. Research groups in Japan have studied the possibility of using positive column plasmas in PDP's. The good efficiency of lamps is also due to the fact that the length of the tube can be much longer than the length of the cathode sheath so that the contribution of the cathode region (not efficient) to the overall efficiency can be kept very low. (NOTE! hot cathodes don't have the same limitations and lamps are hot cathodes.) This is more difficult in PDP's where, because of resolution or photon collection issues the dimensions of the cell cannot be much longer than 1 mm (i.e. the pd product is less than 50 torr.cm). On the other hand fast addressing becomes more difficult for large pd products. Finally other kinds of discharge excitation must also be investigated. A very attractive discharge regime is the RF regime where the frequency is large enough to confine the plasma. In this regime the plasma can be sustained even if the secondary electron emission is zero and with low applied voltage amplitude. This is an ideal situation for efficient xenon excitation because the average electric field and electron energy can be kept low. On the other hand selective addressing is possible in these conditions because the minimum RF sustaining voltage is much smaller than the RF breakdown voltage. The feasibility and higher efficiency of RF PDP's has been recently demonstrated at LG Electronicsxx. Simulations ref have also confirmed the significant increase in efficiency in the RF regime and the possibility of addressing. The calculated discharge efficiency in xenon excitation η was shown to reach values as large as 80% in RF PDP's, vs less than 25% in AC PDP's. Other issues must however be solved (uniformity of the voltage along the electrodes at high frequency, cross-talk …) before the RF PDP's can be considered as a good candidate to replace AC PDP's. LCP: Anything that increases the ratio je/jT in the sheath increases the efficiency. secondary discharges..? Or anything that increase E in regions where there is an electron current increases the efficiency. The gamma that controls the efficiency is an effective gamma resulting from Xe+ and Ne+. One could increase the effective gamma by increasing the ratio Ne+/ Xe+ (for a constant mixture).

VIII. Conclusion The limitations and possible improvements of PDP discharge efficiency have been analyzed. The discharge efficiency has been characterized by the efficiency of power deposition in xenon excitation. It was shown than the main reason for the limitation of the efficiency is the large power deposition by ions in Townsend or glow discharges. In conventional AC PDP's, the power dissipation by ions is on the order of or more than 50% of the total power deposition and the calculated xenon excitation efficiency is typically between 15% and 25%. …. .. ACKNOWLEDGEMENTS This work has been supported by Thomson Plasma.

References [1] J.P. Boeuf, Th. Callegari , C. Punset, and R. Ganter, IDRC 18, pp. 209-220 (1998) [2] C. Punset, J.P. Boeuf, and L.C. Pitchford, J. Appl. Phys. 83, 1884 (1998) [3] H. Uchiike, Workshop Digest of Technical Papers, ASIA Display’98, p. 195 (1998); M. Sawa, H. igest p. 361 (1998) [4] T. Yoshioka, L. Tessier, A. Okigawa, and K. Toki, IDY99-20, pp. 113-118 (1999) Townsend & PC: S. Mikoshiba, S. Shinada, and S. Shirai, J. Appl. Phys. 58, 3720 (1985) PC + auxiliary Y. Okamoto and M. Mizushima, IEEE Trans. Electron Devices, vol. ED-27, 1778 (1980) T. Kagemya, H. Matzusaki, and M. Yokozawa, IEEE Trans. Electron Devices, vol. ED-25, 1094 (1978) S. Cany, J. Kang, C. Punset, J.P. Boeuf, Proc. International Display Workshop 1999 (IDW’99), PDPp4-1, p. 751 J. Kang, W.G. Jeon, O.D. Kim, J.W. Song, J.P. Boeuf, M.H. Park, Proc. International Display Workshop 1999 (IDW’99), PDPp1-19, p. 691

Figure Captions Figure 1: Estimation of the energy balance in an AC PDP discharge Figure 2: Schematic diagram of the excited species kinetics in a Xe-Ne mixture in PDP conditions Figure 3: Fractional electron power deposition in xenon excitation under a uniform, steady state electric field as a function of E/N Figure 4a: Cumulative diagram of the electron power deposition under a uniform, steady state electric field in a Xe(4%)-Ne mixture as a function of E/N, showing the power deposited in xenon excitation, neon excitation, xenon ionization and neon ionization. Figure 4b: Same as Fig. 4a for a Xe(10%)-Ne mixture Figure 4c: Same as Fig. 4a for a Xe(30%)-Ne mixture Figure 5: Mean electron energy as a function of E/N under a uniform electric field, for three different concentrations of xenon in neon. Figure 6: Fractional power dissipated by electrons in a Townsend discharge (the rest being dissipated by ions) as a function of αd (ionization coefficient times gap length). Since the breakdown condition must be satisfied, there is a simple relation between αd and the secondary electron emission coefficient γ (upper horizontal axis). Figure 7: Fractional power dissipated by electrons in the cathode region of a glow discharge and in Townsend discharge as a function of the secondary electron emission coefficient γ Figure 8: Calculated distribution of electron power density dissipated in xenon excitation in (a) a matrix, and (b) a coplanar electrode configuration, at times close to the time of peak current, after Ref. Xx . The gas mixture is Xe(10%)-Ne for the matrix case, and Xe(5%)-Ne for the coplanar case (see these references for the exact conditions of simulations).