1. Plasma display panels: problems and their analysis via computer simulations V.P. Nagorny, V.N. Khudik, P.J. Drallos Plasma Dynamics Corporation, USA A. Shvydky University of Toledo, USA

2. Plasma Dynamics Corporation IMID'05 Introduction  Could anybody imagined 5 years ago the question: PDP or LCD for large displays?  Competition from LCD becomes really strong Rapid falling prices ( reaching $1/ sq in at retail ) Rapid falling prices ( reaching $1/ sq in at retail ) Response time (is approaching 3ms (<10ms)) Response time (is approaching 3ms (<10ms)) Color gamut (better than NTSC) Color gamut (better than NTSC) High Contrast (> 500:1  1000:1) High Contrast (> 500:1  1000:1) Large Pixel counts (~ 10M ) Large Pixel counts (~ 10M ) The viewing angle is improving The viewing angle is improving Color filter may disappear - going to LED illumination Color filter may disappear - going to LED illumination  PDP – improved power consumption, image burning problem, …, but the main problems that we had 10 years ago are still there: Efficiency, and Brightness Efficiency, and Brightness Addressing speed Addressing speed  We better do something about these problems and quick.

3. Plasma Dynamics Corporation IMID'05 How we are doing with research  Matrix PDP – quasi 1D More Xe - higher efficiency, but also higher voltage (from 1D). More Xe - higher efficiency, but also higher voltage (from 1D). Larger Secondary Emission – Good for efficiency (from 1D). Larger Secondary Emission – Good for efficiency (from 1D). Larger Gap – Good for efficiency (from 1D). Larger Gap – Good for efficiency (from 1D).  Coplanar PDP is very complicated (many new elements – essential multi-D, barrier ribs, complex electric field configuration, cathode wave, striations, …) Qualitative results – good (2D, 3D, observations) : Qualitative results – good (2D, 3D, observations) : Discharge structure and dynamicsDischarge structure and dynamics Cathode and anode waves, striationsCathode and anode waves, striations New modes of the dischargeNew modes of the discharge New types of instabilitiesNew types of instabilities … … Quantitative results – messy (lack of “tools”, difficulty due to a size): Quantitative results – messy (lack of “tools”, difficulty due to a size): Efficiency – results can’t be trusted. Codes have built-in error ~XX%.Efficiency – results can’t be trusted. Codes have built-in error ~XX%. Cathode wave – numerical diffusion is faster than the real waveCathode wave – numerical diffusion is faster than the real wave Striations – no quantitative resultsStriations – no quantitative results Observations – sketchy. No comparison/control of ALL the factors.Observations – sketchy. No comparison/control of ALL the factors. Some data are STILL ASSUMED (  Xe,Ne,  Phos, other )Some data are STILL ASSUMED (  Xe,Ne,  Phos, other )

4. Plasma Dynamics Corporation IMID'05 Can we do better?  Yes – new powerful tools (kinetic codes), allowing quantitative investigations are developed or being developed. We can now investigate properties of all kinds of PDP discharges, including addressing speed, jitter, even efficiency... and make numerical experiments, that impossible or expensive to do in a real physical system.  One still needs measurements of certain basic parameters, that affect the regime of operation of a pdp cell, so that investigation will be more focused. One also needs more detailed analysis of physical experiments, especially those made on the panel.

5. Plasma Dynamics Corporation IMID'05 New tools in action Experimenting with Kinetic Codes  Ramp discharge – dual nature, stability  Sustain discharge Striations Striations Cathode wave – speed Cathode wave – speed  Loading effects - Small vs. Large panel

6. Plasma Dynamics Corporation IMID'05 Ramp Discharge  L. F. Weber (1995-1998) I 0 =CdV/dt

7. Plasma Dynamics Corporation IMID'05 Ramp Discharge  Ideal for the test of 3D PIC/MC - Known solution (fluid theory –analytical and numerical). Faster than 3D fluid code =  i (I 0 /2e)=(  i /2e)CdV/dt ~ (1-6)10 4  Stable if: Good priming Good priming dV/dt < R max (L,  ) dV/dt < R max (L,  )

8. Plasma Dynamics Corporation IMID'05 Ramp Discharge  3D cell, 3D PIC/MC simulation vs. 3D Fluid

9. Plasma Dynamics Corporation IMID'05 Ramp Discharge  1D cell, Ramp Rate = 1V/us, ~ 3300 (3D PIC/MC)

10. Plasma Dynamics Corporation IMID'05 Ramp Discharge  1D cell, ~15000 (3D PIC/MC)

11. Plasma Dynamics Corporation IMID'05 Ramp Discharge  1D cell, ~15000 (3D PIC/MC)

12. Plasma Dynamics Corporation IMID'05 Ramp Discharge  Fluctuations in equilibrium  N ~ N 1/2 N  For ~ 10000,   N i ~100 (1%) - fluid approximation seems good (99.7% less than 3  )

13. Plasma Dynamics Corporation IMID'05 Ramp Discharge  In the Townsend discharge N i is the result of a balance, rather than equilibrium (  N ~ N 1/2 ): N i  N e =  N i  N i = (  N i ) exp(  L) = N i N i  N e =  N i  N i = (  N i ) exp(  L) = N i N e =  N i + (  N i ) 1/2  N i + ( N i /  ) 1/2 (sec. emis.) N e =  N i + (  N i ) 1/2  N i + ( N i /  ) 1/2 (sec. emis.) N e =  N i  N i +  N i (avalanche),  N i ~ ( N i /  ) 1/2 N e =  N i  N i +  N i (avalanche),  N i ~ ( N i /  ) 1/2  N i ~ ( N i /  ) 1/2 >> ( N i ) 1/2  N i ~ ( N i /  ) 1/2 >> ( N i ) 1/2  For the Ramp discharge PDP cell is statistically small.

14. Plasma Dynamics Corporation IMID'05 Ramp Discharge  Ramp discharge may be statistically unstable. V cell ~10 -5 cm 3 n > n min ~10 5 cm -3 E < E max ----------------------------------- Fluctuations lead to diffusion between “fluid trajectories”. Oscillations increase or decrease until large oscillation occurs. E max

15. Plasma Dynamics Corporation IMID'05 Ramp Discharge  1D cell, 3D PIC/MC :  Xe - dependence  Statistical Instability is powerful – it works even when and  are large. One needs external source to restart discharge.

16. Plasma Dynamics Corporation IMID'05 Ramp Discharge  Exoemission (1D test cell – initially empty, 3D PIC/MC, =15000,  Xe = 0.001, 7% mix ) =15000,  Xe = 0.001, 7% mix )

17. Plasma Dynamics Corporation IMID'05 Ramp Discharge  Exoemission - 3D cell initially empty, 3D PIC/MC, ~ 60000

18. Plasma Dynamics Corporation IMID'05 Ramp Discharge Summary  Ramp discharge (RD) exhibits dual nature – fluid and statistical.  For the Ramp discharge PDP cell is statistically small. RD is statistically unstable unless external source of electrons is present. Average of simulations of 100 cells with different random seed numbers (no exoemission) showed much more fluid-like behavior, with much smaller fluctuations  ramp measurements based on LINE current or/and light integration, can miss the statistical part of the ramp discharge behavior. Experiments on macrocell also miss statistical effects (very large N).  In low-Xe mixtures some stabilization may be from metastables, but in hi-Xe mixtures only exoemission can stabilize the ramp.  With larger  Xe, the required level of exoemission may be smaller.  Still no reliable data on  Xe and exoemission rate, as well as  Xe,Ne in the presence of exoemission – data critical for stability, efficiency, addressing, lifetime !!!!

19. Plasma Dynamics Corporation IMID'05 Sustain Discharge – Experiments  Current physical observations do not distinguish between different mechanisms of the processes. Special experiments would be very expensive to make, but 2D/3D PIC/MC simulations can do the job. Mesh: 135x60x22

20. Plasma Dynamics Corporation IMID'05 Sustain Discharge – Striations  Guess: Anode striations caused by the ion-sound waves, or/and modulation of the surface charge on the cathode. Freeze ions above the part of the anode - Striation are even more pronounced, and stay.  They are not a result of ion motion, have nothing to do with ion sound. Freeze ions above the part of the anode - Striation are even more pronounced, and stay.  They are not a result of ion motion, have nothing to do with ion sound. Used uniform charge distribution, or no charge on the cathode Used uniform charge distribution, or no charge on the cathode

21. Plasma Dynamics Corporation IMID'05 Sustain Discharge – Striations  Guess : Anode striations – is a plasma ionization wave; the cathode wave – is the surface charging wave (there is no way to distinguish between these mechanisms just from observation). Reality – Almost opposite: 1. experiment with d /  =10 -4 (metal). 1. experiment with d /  =10 -4 (metal).

22. Plasma Dynamics Corporation IMID'05 Sustain Discharge – Striations  2. “No Cathode” experiment, “frozen” ions. Results: a) S ~ Q e d/(  E), v f ~ I e (t) b) No significant difference between narrow (0-1eV), and wide (0-8eV) EDF of the source S. Ion density (top) Charge Deposition Rate Ion density (side) Potential (side)

23. Plasma Dynamics Corporation IMID'05 Sustain Discharge – Striations  Guess: There is about 12V ( I Xe ) between striations  voltage across PC is about 12V x N str. Reality – the voltage between striations decreases with time. Typically it starts with 12-13V, but by the time the last striation appears, the voltage between first ones may be just 2-3 V. Voltage across PC continue to decrease after the anode is completely covered by electrons. Reality – the voltage between striations decreases with time. Typically it starts with 12-13V, but by the time the last striation appears, the voltage between first ones may be just 2-3 V. Voltage across PC continue to decrease after the anode is completely covered by electrons.

24. Plasma Dynamics Corporation IMID'05 Sustain Discharge – Striations  Guess: Striations of PDP sustain discharge similar to those that occur in DC positive column.  Reality: PDP striations appear in the process of charging dielectric, ion motion is not important for their formations, and they disappear due to an ion motion. In the DC positive column stationary striations require ion motion.

25. Plasma Dynamics Corporation IMID'05 Sustain Discharge – Striations  The formation of the striations follow the charging of the anode dielectric and formation of the conducting channel near the surface. The process of charging the surface limits the speed of the appearing of the striations.  Ions inside the channel do not actively participate in the creation of dense areas – they may be considered immobilized. It is ionization by electrons, and the consequent decreasing of the electric field in the area makes a striation.  Speed increases with thicker dielectric d/  (same current), larger current, lower normal electric field above the anode.

26. Plasma Dynamics Corporation IMID'05 Sustain Discharge – Striations Striations near the address electrode

27. Plasma Dynamics Corporation IMID'05 Sustain Discharge – Striations

28. Plasma Dynamics Corporation IMID'05 Sustain Discharge – Striations

29. Plasma Dynamics Corporation IMID'05 Sustain Discharge – Striations

30. Plasma Dynamics Corporation IMID'05 Sustain Discharge – Cathode Wave  Cathode wave speed investigation There is not enough experimental data that would allow distinguish one mechanism of the wave propagation from another. Speed changes during propagation, hard to say why. There is not enough experimental data that would allow distinguish one mechanism of the wave propagation from another. Speed changes during propagation, hard to say why. Fluid codes as well as kinetic Boltzmann codes suffer from numerical diffusion, rising from convective term or. Numerical diffusion of ions, is especially important, since it is responsible for artificial spread of ions ahead of the real cathode wave. Fluid codes as well as kinetic Boltzmann codes suffer from numerical diffusion, rising from convective term or. Numerical diffusion of ions, is especially important, since it is responsible for artificial spread of ions ahead of the real cathode wave. Our previous 3D simulations have shown that the wave speed is of the order of the ion velocity in the cathode area, but we didn’t obtain self-similar behavior, since discharge dynamics affected the speed. Our previous 3D simulations have shown that the wave speed is of the order of the ion velocity in the cathode area, but we didn’t obtain self-similar behavior, since discharge dynamics affected the speed.

31. Plasma Dynamics Corporation IMID'05 Sustain Discharge –Cathode Wave  “Experimental” setup for the CW speed investigation, using 3D/2D PIC/MC codes. Cathode, V = 0 Dielectric,  plasma Anode, V=V a dielectric: d, 

32. Plasma Dynamics Corporation IMID'05 Sustain Discharge – Cathode Wave self-similar behavior. Dielectric thickness d/  =10 , V a = 500V,  = 0.5 (pure Ne); Resolution 600x320 (field).  S,  / C  nini nene CW speed investigation (Fully-Conservative 2D PIC/MC) CW speed investigation (Fully-Conservative 2D PIC/MC) t = 90.5ns V=2.2km/s

33. Plasma Dynamics Corporation IMID'05 Sustain Discharge – Cathode Wave self-similar behavior. Dielectric thickness d/  =10 , V a =500V,  =0.5 (pure Ne); Resolution 600x320 (field).  S,  / C  nini nene CW speed investigation (Fully-Conservative 2D PIC/MC) CW speed investigation (Fully-Conservative 2D PIC/MC) t = 183 ns v=2.2km/s

34. Plasma Dynamics Corporation IMID'05 Sustain Discharge –Cathode Wave  Effect of diffusion on the cathode wave structure D e =D i =0 D e, D i ≠ 0

35. Plasma Dynamics Corporation IMID'05 Sustain Discharge – Cathode Wave  Speed of the CW, v CF (when it propagates) is about the ion speed near the surface in front of the CW, v CF ~ v i. Diffusion and the angle at which electric field enters the dielectric is very important for the propagation of the Cathode Wave. It is ironic that in the conditions of a pdp ( d /  ~ a few microns) the cathode wave charges dielectric capacitor almost completely, so it may seem that it is a charging wave.  v CF increases with Thinner dielectric: d /  Thinner dielectric: d /  Larger electric field near the tip. Larger electric field near the tip.

36. Plasma Dynamics Corporation IMID'05 Single Cell vs. Row of Cells  Display line has ~ 3000-6000 cells; Number of lines ~500-1000.  Wires have resistance, inductance, capacitance ( R, L, C ) – which may be important, when many cells work together. These parameters ( R, L, C ) change when one moves from 42 to 60in panel or from 840 to 1920 pixels/line even if pixels or wires stay the same.  Interaction between cells may strongly affect the work of individual cell. Voltage applied to a cell may differ from the desirable. This may change the “rules of the game” - a cell design or driving waveform may work good in one panel and not so good in the other one, or in the same panel when many pixels are turned ON.  Current measurements do not show the voltage and the current through an individual cell. What can we do about it?

37. Plasma Dynamics Corporation IMID'05 Single Cell vs. Row of Cells  Simplification: All lines work together, all lines are identical. Single charge between metal plates - difficult Many charges - simple Q 

38. Plasma Dynamics Corporation IMID'05 Single Cell vs. Row of Cells Single row – difficult: One has to consider interaction with many rows Many rows – simpler: one may consider one row with correct boundary conditions

39. Plasma Dynamics Corporation IMID'05 Single Cell vs. Row of Cells From original 18 matrix elements in a cell C 11 - C 33, L 11 - L 33, only 6 or 7 are independent (due to a symmetry of a cell), which for sustain discharge can be organized in just five combinations: C 1 - C 3,L A, L B. From original 18 matrix elements in a cell C 11 - C 33, L 11 - L 33, only 6 or 7 are independent (due to a symmetry of a cell), which for sustain discharge can be organized in just five combinations: C 1 - C 3, L A, L B. Transformers A and B :  V A =L A d/dt(I 1 - I 2 ),  V B =L B d/dt(I 1 +I 2 - I 3 ) ( I Positive direction )

40. Plasma Dynamics Corporation IMID'05 Single Cell vs. Row of Cells  Single cell

41. Plasma Dynamics Corporation IMID'05 Single Cell vs. Row of Cells  Many cells, medium load (probably most important)

42. Plasma Dynamics Corporation IMID'05 Single Cell vs. Row of Cells  Many cells, large load

43. Plasma Dynamics Corporation IMID'05 Summary  Once again, what it’s gonna be: LCD or PDP? – My answer: It depends. It depends on the future progress of PDPs.  The next generation of PDPs will be simpler, much more efficient, but we need to accelerate our progress NOW. We need a breakthrough, especially on efficiency and addressing speed. Using kinetic codes for making numerical experiments where it’s difficult to make or analyze a real ones, is essential. Today’s computer codes and computer power are good enough for this task.