Journal of the Korean Physical Society, Vol. 58, No. 4, April 2011, pp. 756∼760

Optimization of the Buffer in a Flat for Brightness Improvement

Taewon Kim School of Electrical and Electronic Engineering, Chung-Ang University, Seoul 156-756, Korea

Kangwhan Lee∗ School of Information Technology, Korea University of Technology and Education, Cheonan 330-708, Korea

(Received 17 June 2010, in final form 17 February 2011)

The influence of the buffer gas pressure and composition on the performance of a flat fluorescent lamp (FFL) i.e., brightness, is investigated for a 4000 FFL. To find the optimal buffer gas condition for the 4000 FFL’s maximum performance, specifically brightness, the total gas pressure was varied from 30 Torr to 5 Torr with Ne:Ar composition ratios of 80:20, 90:10, and 0:100. As a result, when the total gas pressure and the composition were 25 Torr and Ne:Ar with a ratio of 80:20, respectively, the maximum brightness could be obtained. This agreed well with the result of a simulation that was performed with the volume-averaged (global) model of the plasma chemistry of Ne and Ar and their mixtures (including constant Hg gas).

PACS numbers: 52.80.Yr Keywords: Flat fluorescent lamp, Buffer gas, Discharge DOI: 10.3938/jkps.58.756

I. INTRODUCTION that has a cylindrical bar shape with a very small ra- dius. The CCFL has a high brightness and a long life of use. However, when a LCD uses a conventional CCFL, The flat fluorescent lamp (FFL) is a flat light source, the brightness uniformity becomes a problem. There- emitting light from a large active area. It does not need fore, to increase the brightness uniformity, the backlight a light distribution element, thus reducing the cost for unit, which uses the CCFL as a light source, needs op- the whole lighting panel. Therefore, FFLs are suitable tical members, such as a light guide panel (LGP), a dif- for large-area flat light sources, such as photolithogra- fusion member and a prism sheet. Consequently, the phy light sources and broadcast light sources. The liquid LCD device using the CCFL becomes large in size and crystal display (LCD) is also a well-known application of heavy in weight due to the optical members. To solve the FFL. The LCD, which features a thin lightweight this problem, a flat fluorescent lamp as backlight unit configuration and low electricity operability, has shown light source has been suggested. Figure 1(a) shows a rapid expansion in its range of applications with improve- 3200 flat fluorescent lamp. This FFL backlight unit offers ments in liquid crystal materials and development of fine longer lifetimes and better light uniformity compared to pixel processing techniques, and is widely used in house- that of the typically cylindrical CCFL. Another bene- hold television sets, desktop/notebook computer moni- fit is that the number of FFLs required to illuminate tors, large-sized flat panel television sets and so forth. an LCD is much less than the number of CCFLs. For However, unlike the cathode ray tube or the plasma dis- instance, a 4000 LCD TV will typically require at least play panel, the liquid crystal display panel does not emit 10 CCFLs while a FFL-based backlight system require any light by itself, but merely changes the orientation or just one FFL. Even at larger sizes, only one FFL will be the arrangement of the liquid crystal. This characteris- enough. tic of the LCD makes it need, at the rear of the liquid As the range of the application of liquid crystal dis- crystal display panel, a backlight unit to supply the light plays becomes broad, great effort has been made to im- to the information display surface uniformly. In a con- prove their electro-optical characteristics, such as view- ventional LCD device, the light supplying backlight unit ing angle, contrast ratio, gray-scale capability, and so generally uses a cold cathode fluorescent lamp (CCFL) forth. Thus, as a basic requirement, brightness improve- ment is still the critical issue. In this paper, to find ∗E-mail: [email protected]; Author to whom correspondence should the maximum brightness of a flat fluorescent lamp, for be addressed which and have been widely used as the -756- Optimization of the Buffer Gas in a Flat Fluorescent Lamp ··· – Taewon Kim and Kangwhan Lee -757-

flat fluorescent lamp. The channels are spaced apart at a regular interval. A cross section of each of the channels has a semi-elliptical shape and an aspect ratio of about 0.24. The aspect ratio is defined by the channel height divided by the channel width of the channel cross sec- tion. The shape of channel cross section and the aspect ratio are also good parameters that could be optimized in the future. The channels are connected to each other by a gas ‘pass-hole’. Through these holes, the buffer gas and the vapor can be diffused uniformly. The upper and the lower substrates are bonded with each other by a sealing paste, such as a frit including metal and glass. The frit has a lower melting point than glass. The frit is disposed along the edge portions of the upper and the lower substrate, and both substrates are com- pressed when the frit is heated so that both substrates are combined with each other. Then, air in the channels is expelled, and the discharge gas is injected into the channels. The getter, including mercury, is attached to the flat fluorescent lamp and then heated. In that case, the mercury in the getter is vaporized and then injected Fig. 1. (Color online) (a) Real flat fluorescent lamp for a and diffused into the lamp, and the getter is removed. 3200 liquid crystal display backlight unit and (b) sketch of a Usually, in the case of a 4000 flat fluorescent lamp, 80 flat fluorescent lamp including the inside layer structure. mg mercury is used in the getter. The inner surface of upper substrate except for the portion corresponding to the channels makes contact with the lower substrate due buffer gas, including mercury as a 253.7 nm wavelength to the pressure difference between the light-emitting dis- ultraviolet light generation source, we focused on the op- charge spaces and the atmosphere. The discharge gas timization of the buffer gas pressure and the composition includes argon (Ar), neon (Ne) and mercury (Hg) vapor. ratio. The paper is organized in four sections as follows: Band-shaped electrodes are made in both edges on the Firstly, in Section II, I shall briefly explain the flat fluo- outer surfaces of the upper and the lower substrates cor- rescent lamp and the processes to produce the lamp used responding to both ends of the channels. The first and in this experiment. Section III is devoted to the descrip- the second electrodes extend in a direction that is sub- tion of the measurement processes, apparatus and the stantially perpendicular to the direction in which the simulation with a volume-averaged (global) model. The channels extend. Both electrodes include a metal like measurement and simulation results are discussed in Sec- copper (Cu) having a high electrical conductivity. The tion IV. The conclusion and future work are presented first and the second electrodes are also formed on the in Section V. outer surface of the lower substrate at the same posi- tions as the upper substrate electrodes. When a dis- charge voltage is applied to the discharge gas through the first and the second electrodes, the discharge gas, II. FFL PRODUCTION PROCESSES mercury, generates ultraviolet light. The glass surfaces of the lamp which contact the discharge gases are coated The flat fluorescent lamp consists of a upper glass sub- with Y2O3 to prevent sodium ions from flowing out into strate and a flat lower glass substrate. Soda lime glass the discharge spaces. is used as both substrates. When the lower and the up- The flat fluorescent lamp further includes a phosphor per glass substrates are combined with each other, to layer and a light-reflecting layer in the case of the lower make a number of channel-shaped light emitting dis- substrate. The light-reflecting layer is formed on the in- charge spaces, the upper glass substrate is manufactured ner surface of the lower substrate. The fluorescent layer by using a glass-forming process. In detail, a flat glass is disposed on the light-reflecting layer. The fluorescent plate is heated; then, the air above the heated glass is layer, which transforms the ultraviolet radiation gener- 3 inhaled through many tiny air holes in a desired-shape ated from the mercury excited species (6 P1) into visible mold right over the heated glass. Spontaneously, the light, is optimized for 253.7 nm wavelength ultraviolet heated glass is brought into the final shape of the inte- light. The fluorescent layer includes red, green and blue rior of the mold to form the upper substrate with chan- phosphors. The red, green and blue phosphors trans- nels. The upper glass substrate includes a plurality of form the ultraviolet light into red, green and blue light, channels (33 channels in the case of a 4000 FFL). The respectively. Substantially same amounts of red light, channels extend substantially parallel to an edge of the green light and the blue light generate white light. A -758- Journal of the Korean Physical Society, Vol. 58, No. 4, April 2011 light-reflecting layer having a high optical reflectivity is generation source. The detailed simulation procedure formed on a lower substrate and is made with aluminum and results will be published in another paper in the fu- oxide (Al2O3). Metal powder or liquid metal is sprayed ture. and dried to form the light-reflecting layer. The FFL schematic including inside layer structures is shown in Fig. 1(b). IV. RESULTS AND DISCUSSION

The fact that the buffer gas pressure and its composi- III. EXPERIMENT tion have a salient influence on the lamp parameters is well known from the work of many authors [1–6]. The An experimental investigation was made to find a new buffer gas in a fluorescent lamp serves a number of func- efficient buffer gas pressure and composition ratio for tions. It acts as a buffer to reduce ion bombardment on a flat fluorescent lamp including a constant amount of electrodes or the phosphor coating and reflects and re- mercury. It is shown that there exists an optimum oper- distributes sputtered material back on the cathode. It ating buffer gas pressure and composition ratio that can also helps the lamp ignite by reducing the voltage re- make the performance of the 4000 flat fluorescent lamp, quired to initiate the discharge, and the most important specifically the brightness, reach a maximum. In this in- function, which is focused on in this article is that it vestigation, the pressure is varied from 30 Torr to 5 Torr controls the ambipolar diffusion rate, which in turn es- with 100%, 10%, and 20% Ar composition ratios for each tablishes the required balance between the ionization of pressure condition. The lamp is driven with a sinusoidal mercury atoms to maintain the discharge and the excita- wave of 50 kHz. The lamp’s operating voltage VL and tion of mercury to provide UV radiation of 253.7 nm by electric current IL was observed by using a digital os- limiting the charged particle’s mean free path. In other cilloscope (Tektronix TDS3014C) and the lamp wattage, words, the balance of diffusion losses and ionization may which is the total power dissipated in FFL itself, was cal- be influenced by the choice of buffer gas, gas pressure, culated from the measured values of VL and IL.VL and and gas composition ratio. IL were measured by using a high voltage probe (Tek- The diffusion losses increase with decreasing gas pres- tronix P6015A) and a current probe (Tektronix P6022), sure, decreasing molecular weight of the , and respectively. The brightness was measured with ProMet- so on. John F. Waymouth said in his book “Electric Dis- ric 4200 (Radiant Imaging) after 90 minutes of aging at charge Lamps” that in a 1.5-inch-diameter tubular lamp an ambient temperature of 25 ◦C. operated at constant current, at low buffer gas pressure, The average value of the measured data for 5 differ- the efficiency increases with increasing buffer gas pres- ent samples (FFLs) under the same condition was used sure because the reduction of the ionization loss rate, and as a representative value of that condition. The RMS therefore of electron temperature, increases the efficiency 3 values of the currents, voltages and wattages for all the of excitation of the 6 P1 state. Above approximately 1 measurements presented were used. An IP (integrated to 1.5 Torr, however, the efficiency decreases with in- power) board was used to drive the lamp. Constant creasing argon pressure because the elastic collision loss power was supplied from the IP board, which served as a increases and the total radiation output decreases with power supply plus inverter, but the supplied real power decreasing electron temperature [5]. Like Waymouth’s to the flat fluorescent lamp itself was not the same for tubular lamp case, in the FFL, there also must be the op- each lamp condition, i.e., in this research, for each buffer timal buffer gas condition for the maximum brightness, gas filling condition. Actually, the supplied lamp power although the FFL has external electrodes unlike Way- was dissipated in the glass and in the anode/cathode fall mouth’s tubular lamp, and in this study, experiments too. In this investigation, the power dissipated in the were performed at constant power. In this investigation, glass and in the cathode/anode fall was not measured, so the buffer gas pressure and composition were optimized the total power, including the glass and cathode/anode for brightness improvement of the 4000 FFL, but after fall power consumption, as well as the positive column 30 Torr was chosen as a reference, one problem was to power consumption was used as the input power. fix the pressure range for the experiment for pressures Also, using commercial package (CFD-ACE+, ESI higher and lower than 30 Torr. The range lower than 30 corp.), the behaviors of the Ne and the Ar plasmas, in- Torr was chosen because, as in the work of Barnes [2] and cluding the Hg gas, were investigated using a volume- Read [4], the intensities of the 253.7 nm line increase with averaged (global) model. The volume-averaged value of a reduction in either the pressure or the atomic weight the electron temperature and all species densities were of the filling gas. As is well known, this effect is caused calculated by solving the energy and the particle balance by the increase in the electron temperature required to equations. Input parameters, i.e., the buffer gas pressure compensate for the more rapid loss of ions and electrons and composition, were varied over the same range as in to the wall. Generally, as the electron temperature of 3 the experiment to achieve the maximum Hg6 P1 species a fluorescent lamp that is operated at constant power density as a 253.7 nm wavelength ultraviolet light (UV) is increased, the efficiency of resonance radiation is also Optimization of the Buffer Gas in a Flat Fluorescent Lamp ··· – Taewon Kim and Kangwhan Lee -759-

Fig. 4. (Color online) Measured total power consumption Fig. 2. (Color online) Measured lamp current as a function in the lamp as a function of the buffer gas pressure in a flat of the buffer gas pressure in a flat fluorescent lamp discharge fluorescent lamp including a constant Hg mass (80 mg) with including constant Hg mass (80 mg) with different Ne/Ar different Ne/Ar composition ratios (0:100, 90:10, 80:20). composition ratios (0:100, 90:10, 80:20).

Fig. 5. (Color online) Measured lamp brightness as a func- tion of the buffer gas pressure in a flat fluorescent lamp in- Fig. 3. (Color online) Measured lamp operating voltage as cluding a constant Hg mass (80 mg) with different Ne/Ar a function of the buffer gas pressure in a flat fluorescent lamp composition ratios (0:100, 90:10, 80:20). discharge including constant Hg mass (80 mg) with different Ne/Ar composition ratios (0:100, 90:10, 80:20). Figs. 2, 3, and 4 for the lamp current, for lamp operat- ing voltage and the lamp power consumption as functions increased. Therefore, much effort has been devoted to of the buffer gas pressure for different Ne/Ar composi- increasing the electron temperature [5]. tion ratios (0:100, 90:10, 80:20). Cases of 100% Ar at 25 In the same way as reduction of the gas pressure the Torr and 30 Torr, whose discharge is very unstable, were substitution of Ne for Ar results in increased ambipolar removed from the measurement range. The positive col- diffusion to walls because of the lighter weight of Ne, umn potential drop was not investigated, but the lamp i.e., higher mobility, this necessitates a higher electron operating voltage was measured at the lamp terminals temperature to maintain the ionization at the same level and consisted of two glass voltages applied to glass, the as the charged particle loss. Therefore, in this experi- cathode/anode fall voltage and the positive column volt- ment to increase the brightness of the FFL that is, to age. Therefore, instead of positive column power con- increase the electron temperature, we reduced the pres- sumption, the total lamp power consumption was calcu- sure of buffer gas from the reference of 30 Torr, and we lated by using the integral of the product of the measured increased the neon composition ratio from 80% to 90%. lamp current and the lamp operating voltage. Because The case of 100% argon was also added as a conventional the negative resistance characteristic is well known in lamp buffer gas case. the positive column, as the lamp current increases, the According to the measurement results of this study at positive column voltage must decrease [7,8]. This means a constant power of 170 W supplied by the IP board, that the positive column voltage, i.e., the electric field as the buffer gas pressure is decreased, the lamp current across the positive column of the FFL, decreases with and the total power dissipated in the lamp, including the decreasing buffer gas pressure. glass and cathode/anode fall, increases while the lamp Generally, the electron density increases linearly with operating voltage decreases. These are well shown in the discharge current, and at higher electron density, the -760- Journal of the Korean Physical Society, Vol. 58, No. 4, April 2011

V. CONCLUSION

The optimal buffer gas pressure condition of 25 Torr is proposed for brightness improvement of a 4000 flat flu- orescent lamp. No significant result was obtained from the variation of the buffer gas composition ratio, and the relation between the brightness and the discharge mech- anism of the FFL due to variations in the buffer gas pressure and the composition ratio was described with a view to two main parameters, the electron temperature and the total power dissipated in the lamp’s positive col- umn. Also, the global model was applied to Ne, Ar and 3 constant Hg gas discharges, and the Hg6 P1 density as 3 Fig. 6. (Color online) Hg 6 P1 density as a function of the a 253.7 nm wavelength ultraviolet light (UV) generation buffer gas pressure calculated by using the volume-averaged source was calculated. The simulation result also peaked (global) model with a different Ne/Ar composition ratios at the same pressure and composition, agreeing with ex- (0:100, 90:10, 80:20) and including constant Hg gas. periment very well. In the future, to confirm the relation between the brightness and the discharge mechanism de- scribed in this article, measurements of the electron tem- discharge can be sustained with a lower electric field i.e., perature, a positive column voltage are needed, and the a lower electron temperature [9, 10]. Therefore, in the measurement method should be deliberated. Basically, case of constant power feeding of 170 W by the IP board, the Langmuir probe method will be considered. Also, the from the fact that the discharge current is increased as cathode fall voltage and the glass voltage are so critical the filling gas pressure is reduced, it can be deduced that for understanding the lamp completely. the electron temperature will be decreased. In addition, the power supplied to the positive column is also seen to be increased with increasing lamp current [3]. In other words, a reduction in the buffer gas pressure has two ACKNOWLEDGMENTS marked effects; the power consumption’s increase in the positive column and a decrease in the electron temper- ature. As a result, as Fig. 5 shows, as the buffer gas This paper was partially supported by the Education pressure is decreased from 30 Torr to 5 Torr, that is, un- and Sabbatical Year Research Promotion Program of der a trade-off relationship between a reduction in the KUT. And this research is supported by the collabora- electron temperature and an increase in the power sup- tive research program among industry, academia and re- plied to positive column, the fact that the brightness search institutes through of Korea Industrial Technology intensity peaks at 25 Torr indicates that for maximum Association (KOITA) funded by the Ministry of Educa- resonance radiation of 253.7 nm wavelength, the compro- tion, Science and Technology (KOITA-2011). mise buffer gas pressure point exists at 25 Torr. In other words, at 25 Torr, the effect of increasing energy, which is dissipated in the positive column, is bigger than the REFERENCES efficiency reduction on the absolute amount of generated resonance radiation at a 253.7 nm wavelength. Through this investigation, the optimal pressure of 25 [1] W. Verweij, Physica 25, 980 (1959). Torr was attained for the maximum brightness of FFL. [2] B.T. Barnes, J. Appl. Phys. 31, 852 (1960). Also, Ne and Ar plasma discharges, including a con- [3] M. Koedam, A. A. Kruithof and J. Riemens, Physica 29, stant amount of Hg gas, were simulated with a global 565 (1963). (volume-averaged) model [11]. For a specified discharge [4] T. B. Read, Br. J. Appl. Phys. 15, 837 (1964). size, absorbed power, pressure, and feed gas composition, [5] J. F. Waymouth, Electric discharge lamps (MIT Press, as well as the appropriate reaction rate coefficients and Cambridge, 1971). [6] T. G. Verbeek and P. C. Drop, J. Phys. D: Appl. Phys. surface recombination constants, the energy and particle 7, 1677 (1974). balance equations were solved to determine all species [7] T. B. Read, Br. J. Appl. Phys. 14, 36 (1963). densities. Especially, to find the optimal buffer gas pres- [8] C. Kenty, J. Appl. Phys. 21, 1309 (1950). sure and composition for achieving the maximum bright- [9] M. A. Easley, J. Appl. Phys. 22, 590 (1951). 3 ness, the Hg6 P1 density as a 253.7 nm wavelength ultra- [10] G. M. Petrov and J. L. Giuliani, J. Appl. Phys. 94, 62 violet light (UV) generation source was calculated and is (2003). shown in Fig. 6. This simulation result also agreed with [11] CFD-ACE+ V2006 Modules Manual (ESI corporation, the experiment very well. 2006).