luminescent rise times of inorganic phosphors excited by high intensity ultraviolet light

Luminescent Rise Times of Inorganic Phosphors Excitedby High Intensity Ultraviolet Light

Robert J. Anderson, and Sam G. Ricchio

The relative delay between excitation and luminescence was measured for a number of common inorganicphosphors using short, high intensity excitation pulses. The delays were found to be much shorter thananticipated; on the basis of low intensity pulsed luminescence measurements and were found to be ex-tremely intensity-dependent. Both the luminescence rise and decay times were found to be intensity-dependent as well, with the luminescence pulse waveform tending to approach the exciting pulse at thehigher excitation intensities.

1. Introduction

The characteristics of the luminescence of inorgan-ic crystals have been studied extensively. 1 -5 Initial-ly, luminescence was of interest primarily to miner-alogists engaged in the classification and study ofnaturally occurring crystals. By the 1930's, how-ever, technical applications such as television andfluorescent lamps aroused a tremendous interest inthe study of cathode luminescence and photolumi-nescence.

More recently, exhaustive studies of solid-state lu-minescence have been carried out,3-5 with resultantapplications to a number of devices including displaydevices such as electroluminescent panels, light-emitting diodes, and cathode-ray tubes. In addi-tion, luminescence from crystal phosphors such asZnS and CdS has been applied to such diverse areasas atmospheric mapping by fluorescent particle trac-er analysis 6 ' 7 and fluorescence microscopy. 8

Despite the development of crystal phosphors (II-VI compounds such as ZnS and CdS) and their usein applications such as television and fluorescentlamps, the details of the luminescent processes inthese materials are not yet thoroughly understood.Hence, quantitative predictions of their perfor-mance in a given application can be obtained onlyfrom empirical data. This is particularly true of therise time of the luminescence in these phosphors,since historically the decay characteristics have beenof more interest. Moreover, applications of these

The authors are with Beckman Instruments, Inc., Fullerton,California 92634.

Received 9 April 1973.

phosphors in the past has been primarily energy-lim-ited. That is, both the excitation power densityavailable and the emission output power have beenso low that relatively long time periods have been re-quired both for activation of the phosphor and obser-vation of the luminescence output.

The recent development of ultraviolet emitting la-sers and laserlike sources has made it possible to useII-VI phosphors in applications that are not energy-limited. The time necessary both to excite thephosphor and to observe the luminescence is thuspotentially extremely short. 9 10" 1 Empirical data donot exist for the evaluation of phosphor performancein this area of high intensity, short-time-period excita-tion, and subsequent luminescence, however, and itis for that reason that the current measurementswere undertaken. In particular, it was desired toobtain an indication of the rise-time behavior of thephosphor luminescence using fast, high intensity ex-citation pulses.

Pringsheim2 has discussed the qualitative andquantitative aspects of the photoinduced recombina-tion luminescence found in Type II-VI compounds asa function of time and excitation intensity. Qualita-tively, over a very short time interval after excitation(on the order of tens of nanoseconds) the majority ofthe luminescence output is a direct recombinationfrom the conduction band to an acceptor level, or alattice vacancy, such a process being essentiallybimolecular. After an appreciable time interval(microseconds), electrons in the conduction bandhave had time to interact extensively with trap levelsin the crystal, thereby reducing the luminescentdecay process to a monomolecular process dependenton the probability of an electron's being releasedfrom a trap. Thus, the luminescent decay assumes ahyperbolic form over time intervals on the order of

November 1973 / Vol. 12, No. 11 / APPLIED OPTICS 2751

nanoseconds, but is reduced to a more nearly expo-nential decay over a time interval longer than a mi-crosecond.

Pringsheim has shown that for photoinduced re-combination luminescence, the number of lumines-cent photons emitted per unit time I is given by

I(t) = A(t)[tanh(flA) 2 t]2 (1)

during excitation and by

I(t) = .I + tI) 1/2]-2 (2)during luminescent decay, where A represents thenumber of exciting photons absorbed per unit time,Io is the peak value of I, t is time, and 03 is a factordependent upon the phosphor. Two limiting casesmay be explored for the excitation expression, Eq.(1); these are

I(t) A(t) for (flAt2 )/2 > 1.5. (3)or

1(t) f3 fA2 (t)t 2 for (At 2)"/2 << 1. (4)

Equations (3) and (4) show that two extremely. dif-ferent forms of luminescent time development maybe anticipated during excitation, depending both onthe intensity of excitation and the time interval.

Since the excitation pulses used in this study wereshort, Eq. (4) may be expected to apply for low (lessthan 100 W/cm2) excitation intensities. In this casethe luminescence rise curve will have a form depen-dent on the square of the excitation intensity andthe square of the time. At a moderate excitation in-tensity (1-10 kW/cm2) the luminescence rise curvewill initially have a time development given by Eq.(4) but will later tend toward the saturation valuepredicted by Eq. (3); that is, I(t). will become pro-portional to A(t). At high excitation intensity(greater than 1 MW/cm2 ), it may be anticipatedthat the saturation behavior will approach that pre-dicted by Eq. (3) and the luminescent pulse risecurve will accurately follow the rise of the excitationpulse.

Thus we may expect that the luminescent risecurve will in general be delayed in time with respectto the exciting pulse. Moreover, this delay will bedifferent at the start of the luminescence [I(t) pro-portional to A2t2] than later, when it will tendtoward proportionality with A. The luminescentrise curve will be extremely dependent on the inten-sity of excitation and at high excitation intensity willclosely approach the excitation rise curve. Finally,the initial slope of the luminescence decay curve isproportional to /Io and therefore /A, so that at highexcitation intensity, the decay curve will also tend tobe very fast, approaching the decay of the excitationpulse.

In this study the luminescent rise curve was char-acterized by two independent delay times: (1) thedelay represented by the time interval between theonset of the excitation pulse and the onset of lumi-nescence, and (2) the time interval between the peakof the excitation pulse and the peak luminescence.

These time intervals were measured with an accu-racy of better than 5 nsec.

II. Instrumentation

The basic experiment involved the observation of aluminescent pulse excited by an ultraviolet pulse ofrelatively high energy and power. Simultaneous re-cording of both pulses permits a number of conclu-sions to be drawn concerning both the rise and decaytimes of the luminescent emission. The very fasttime scale involved (nanoseconds) required carefulconsideration of the electronics. The high poweravailable from the pulsed excitation source, however,eased the requirements for high sensitivity in the de-tector. Since recording of the incident excitingpulse at 266 nm would require a change of filters atthe photomultiplier, the pump laser pulse at 532 nmwas recorded instead. The 532-nm pulse was closeenough to the luminescent wavelength to passthrough the photomultiplier filters, hence it was pos-sible to record both the luminescent and the 532-nmpulses without moving the sample, photomultiplier,or filters, assuring that all experimental parametersremained -the same. This technique also assuredthat sample luminescence would not interfere withthe recording of the exciting pulse.

The measurement system used in this investiga-tion consisted of a source, sample, detector, andreadout instrumentation. Figure 1 shows the generallayout and relationship of the major components.

The ultraviolet source in the above system was acrystal of ammonium dihydrogen phosphate (ADP)used to frequency-double the visible output of apulsed laser. The visible input to the ADP crystalwas obtained by frequency-doubling the output of apulsed neodymium:YAG laser operating in the near-infrared. The laser was a Chromatix model OO0D,Nd+ + :YAG (neodymium-doped, yttriun-alumi-num-garnet) laser operating at 1.06 m. An inter-nal second harmonic-generating crystal (lithium iod-ate) produces output in the visible at 532 nm. Cou-pled to the output of this laser was a Chromatixmodel 1030 ultraviolet doubler that converted ap-proximately 50% of the laser output in the visiblerange to second harmonic output in the ultraviolet at266 nm. Output of the model 1030 doubler consist-ed of both the 266-nm ultraviolet beam and the re-mainder of the unconverted 532-nm laser output.

The laser was operated in the Q-switched mode,with a peak output power of approximately 2.5 kWat 532 nm and a pulse repetition rate of about 20pulses per second. Pulse width at 532 nm was ap-proximately 75 nsec, with the pulse peak occurringapproximately 50 nsec after the start of the pulse.The output of the doubler was approximately 1-kWpeak pulse power at 266 nm.

The use of two types of filters, a Corning 7-54 thattransmits uv radiation and a Schott GO-21 that ab-sorbs uv, allowed measurement of both the 532-nmlaser pulse and the luminescent radiation pulse

2752 APPLIED OPTICS / Vol. 12, No. 11 / November 1973

Fig. 1. Schematic diagram of the apparatusused to measure luminescence rise curves.

*CORNING 7-54 UV PASSSCHOTT GG - 21 UV BLOCK

** FILTER STACK CONSISTS OFSCHOTT GG - 13SCHOTT OG - 5

546.lnm BAND PASS

through identical optical paths. By alternatingthese filters in and out of the laser beam, indepen-dent traces of each signal could be photographed.

Luminescent energy from the phosphors was de-tected by an RCA C31034 photomultiplier with agallium arsenide photocathode. This photomultipli-er has an extremely uniform sensitivity from 200 nmto 800 nm and has a rise time on the order of a nano-second. The photomultiplier anode was coupled di-rectly to a Tektronix model 585A oscilloscope using a50-ohm coaxial cable terminated in a 50-ohm load atthe input. Thus the full bandwidth of the detectionsystem was utilized, permitting rise times of a fewnanoseconds to be accurately measured.

To reduce error, both the laser and the lumines-cent rise times were measured by identical electricalpaths. Signal amplitudes were adjusted by means ofthe variable attenuator between the laser and the ul-traviolet doubler. All measurements were made atpeak anode currents of less than 1 mA from the pho-tomultiplier, thus avoiding saturation of the photo-multiplier. The 1-mA current level corresponded todetected signals of less than 50 mV full scale into a50-ohm load.

The oscilloscope was triggered through the use ofabeam splitter that deflected a portion of the 532-nmlaser output onto a fast photodiode (Hewlett-Pack-ard PIN photodiode). The longer signal path fromthe photomultiplier assured that the signal would bedelayed approximately 50 nsec after the trigger, sothat reliable observation of the rise time was ob-tained.

To facilitate data evaluation, each photograph wasdouble-exposed to show both the laser and the lumi-nescent pulses. Luminescent emission was moni-

tored at a wavelength of 546.1 nm by using a SchottGG-13 glass filter, followed by a Schott OG-5 glassfilter, with an Infrared Industries 546.1-nm bandpassfilter next to the photomultiplier. No signal was ob-served due to luminescence of this filter sandwich.

11. Experiment

A. Luminescence Spectra

Prior to measurement of the luminescence risetime, an experiment was performed to determine theeffect on the luminescent output of very high intensi-ty laser emission at 266 nm as compared to a low in-tensity source. Low intensity luminescence spectraexcited by 365-nm light were first obtained using anAminco Bowman spectrofluorometer and are shownin Fig. 2. The same spectra were then run us-ing thepulsed ultraviolet laser as a high intensity excitationsource at 266 nm. The apparatus used to obtain thelaser-excited luminescence spectra was similar tothat shown in Fig. 1 except that the photomultiplierfilters were replaced by a Beckman DU monochro-mator, arranged to scan automatically from 350 nmto 800 nm. The photomultiplier (RCA C31034) out-put was amplified by a boxcar integrator and record-ed on a strip-chart recorder.

The spectra obtained corresponded closely to thoseshown in Fig. 2, with the exception of two samples,nos. 2267 and 2225, in which the peaks of the lumi-nescence curve have been shifted to shorter wave-lengths. We can offer no explanation for this wave-length shift at the present time, particularly since itappeared in only two samples. It was reproducible,however, and it did appear to be power-dependent.


100

ILLnzn

Fig. 2. Luminescent particle emission spectra (nor-malized) using low intensity excitation at 365 nm.

Fig.3.Luminescence spectrum of sample no. 2225 excited at

266 nm with a power density of 100 W/cm 2 .

An additional affect of high power density excita-tion can be seen in Figs. 3 and 4. Figure 3 is thespectrum of sample no. 2225 at an excitation intensi-ty of about 100 W/cm 2 , while Fig. 4 is the samesample excited at a level of 1 MS/cm 2 . At low exci-tation levels, no luminescence is seen in the 450-HPT-nm region while at the higher excitation inten-sity a peak appears in this region. Luminescence inthe region from 450 nm to 480 nm was also observedin all the other samples when excited at intensitiesgreater than 1 MW/cm2. The silicone grease sub-strate on which the samples were placed was also ob-served to fluoresce slightly at an excitation powerdensity of 1 kW/cm2, with a fluorescent peak at 415nm. This spectrum is shown in Fig. 5.

Table I summarizes the wavelengths of peak lumi-nescent emission as measured using the high intensi-ty laser excitation as well as the low intensity mercu-ry lamp.

B. Comparison of 266-nm and 532-nm Pulses

Since the luminescent pulses were to be comparedwith the 532-nm laser pulse rather than the 266-nmultraviolet exciting pulse, it was necessary to estab-lish the relationship between these two pulses in thetime domain.

Figure 6 is a representative photograph of the os-cilloscope traces obtained and shows both the 266-


Fig. 4. Luminescence spectrum of sample no. 2225 excited at 266nm with a power density of 1 kW/cm 2 .

Fig. 5.Luminescence spectrum of silicone grease excited at

266 nm with a power density of 1 kW/cm 2 .

Table I. Characteristics of Phosphor Samples

Peak Luminescence (nm)

Low excitation High excitationintensity of intensity of

No. Color Mfr. Lot no. Compound 365 nm 266 nm

Blue Sylvania HBSB-1 ZnS: Ag 463 4603206 Blue, green U.S. Radium H447 ZnS: CdS 493 4952210 Green U.S. Radium H1096 ZnS 545 5302267 Yellow U.S. Radium H779 ZnS: CdS 588 5103336 Orange, pink U.S. Radium H1102 ZnS 590 5852225 Red Sylvania PR-2 ZnS: Eu 650 600

nm and 532-nm pulses on a time base of 50 nsec perdivision. It is apparent from these waveforms thatthe 266-nm pulse is appreciably shorter (FWHM =40 nsec) than the 532-nm pulse (FWHM = 70 nsec).It is also apparent, however, that this difference isdue almost entirely to the pulse decay time, and thatthe rise times are almost identical, the peak ampli-tudes occurring simultaneously within the experi-mental accuracy of the system (3 nsec).

Fig. 6. Comparison of pulse waveforms due to 266-nm and 532-nm laser output pulses. Time base is 50 nsec per division. Thewider pulse is the 532-nm output; the narrower pulse, the 266-nm

output.

C. Measurement of Quinine Sulfate Decay Time

The experimental decay time of the fluorescence ina solution of quinine sulfate has been frequently


studied and may be considered to be a standard offluorescence decay (cf Refs. 12-15). For this reasonwe performed a measurement of this decay time toestablish the over-all system accuracy over the timescale involved.

Approximating both the excitation pulse decayand fluorescence decay by exponentials, we calculatethe mean lifetime of the fluorescent state as 7 =

20.5 nsec. This result corresponds well to previousmeasurements of the fluorescence lifetime in quininesulfate (cf Ref. 12-14), confirming the validity of theexperimental technique.

D. Direct Measurement of System Delay Resolution

The purpose of the experiment was to measure thedelay between excitation of the luminescent particlesand emission of the luminescence. Thus, it was nec-essary to perform a direct measurement of the over-all system response to a calibrated optical delay-timestandard. This calibrated optical delay was ob-tained in the form of a Beckman ten-meter gas cell,in which the optical path through the cell may bevaried in increments from 10 cm to 10 m. The ex-perimental apparatus used for this measurementduplicated the arrangement used in the final lumi-nescence measurements, with the exception of theinsertion of the 10-m gas cell in the optical path andthe substitution of a diffuse reflector for the fluo-rescent sample.

The experiments were performed in a manner sim-ilar to the subsequent luminescence measurements.The 10-cm path through the gas cell was selected,and the laser output at 532 nm was allowed to im-pinge on the diffuse reflector after passage throughthe cell. The trace of these pulses on the oscillo-scope was photographed. The switch on the gas cellwas then rotated to change the pathlength to a pre-viously selected long path, thus introducing a knowndelay. The trace of the pulse waveform on the oscil-loscope was now photographed on the same filmframe as the previous trace, the two waveforms beingdisplaced from one another by a time interval pro-portional to the difference in the pathlengths.

Table II compiles the various pathlengths avail-able using the Beckman gas cell, as well as the corre-sponding pathlength differences and their delay-time

Table II. Calculated Delay Times for Various Pathlengths inthe 10-m Gas Cell

Pathlength Pathlength Delay time(i) difference (m) (nsec)

1.0 0.9 3.02.8 2.7 9.04.6 4.5 15.06.4 6.3 21.08.2 8.1 27.0

10.0 9.9 33.0

increments. It was found that a delay of 3 nseccould just be resolved, thus establishing the systemdelay resolution as approximately 5 nsec.

E. Phosphor Rise-Time Measurement

The experimental apparatus used in the measure-ment of the luminescent particle rise times is shownschematically in Fig. 1. Detection of the lumines-cent signal was made at an angle to the sample suchthat the specularly reflected excitation beam wouldnot be within the acceptance aperture of the detec-tor. The delay time between the excitation pulseand the luminescent pulse was measured for eachsample at a number of different power densities:100 W/cm2 , 1 kW/cm2 , 10 kW/cm2, and 1 MW/cm2 . These various power levels were obtained byfocusing the beam and adjusting a precision attenua-tor. The samples were prepared by dusting a smallportion onto a CRES (corrosion-resistant steel) slidecontaining a film of silicone grease (Dow-Corning).In this manner, a fairly uniform sample layer wasobtained on the slide, which simplified handling dur-ing measurement periods.

Figures 7 through 9 are representative photographsof the oscilloscope traces of the luminescent emissionfrom the samples. Each photograph also includesthe trace of the 532-nm pump laser pulse, whoseleading edge and peak correspond closely to theleading edge and peak of the 266-nm exciting pulse.All photographs were obtained by first blocking the266-nm pulses and photographing the 532-nm lightscattered by the sample. The 532-nm pulses werethen blocked and the 266-nm exciting light was per-mitted to impinge on the sample.

From these data it was possible to discern two rel-atively independent features of the luminescent risecurve: first, the initial delay, that is, the time inter-val between the onset of the exciting pulse and theonset of luminescence; second, the peak delay, thatis, the time interval between the peak of the excita-tion pulse and the peak of the luminescence. TableIII is a summary of these delay times for the varioussamples as a function of excitation power density.

IV. DiscussionSeveral features of the results summarized in

Table III are worth noting. First, in all of the sam-ples measured, the peak delays from excitation to lu-minescence were less than 40 nsec. This result is inmarked contrast to the results of low excitation in-tensity pulsed luminescence, in which case rise anddecay times on the order of milliseconds are com-mon. Moreover, it is very apparent in the two seriesof oscilloscope traces shown in Figs. 7 and 8 that thedelay between excitation and luminescence is highlyintensity-dependent and decreases with increasingexcitation intensity. In addition, the luminescencedecay is seen to be highly intensity-dependent, withthe decay becoming faster at high levels of excitationintensity. The result of these two intensity-depen-dent phenomena is that as the excitation intensity


Fig. 7. Waveforms of the excitation and luminescence pulses from sample no. 2225, Red (PR-2), at various levels of excitation intensity:(a) 100 W/cm 2 , (b) 1 kW/cm 2 , (c) 10 kW/cm 2 , (d) 1 MW/cm 2 . All traces are shown on a time base of 50-nsec per division.

Table Ill. Summary of Delay Time Measurements

Relative Excitation power Delay times (nsec)intensity density at 266 nm

Sample at 546 na (cm-2 ) Initial Peak

3336 Orange, pink w 1 kW 10 103336 Orange, pink w 1 MW 5 52225 Red w 100 W 20 202225 Red w 1 kW 15 152225 Red w 10 kW 10 102225 Red w 1 MW 0 52210 Green s 100 W 25 302210 Green s 1 kW 15 302210 Green s 1 MW 0 152267 Yellow s 100 W 20 302267 Yellow s 1 kW 15 302267 Yellow s 1 MW 10 15Blue-HBSB 1 m 1 kW 15 40Blue-HBSB 1 m 1 MW 0 103206 Blue-green s 1 kW 15 303206 Blue-green s 1 MW 0 15Silicone grease vw 1 kW 5 5

a s = strong, m = moderate, w = weak, vw = very weak.


Fig. 8. Waveforms of the excitation and luminescence pulsesfrom sample no. 2210, Green (H1096), at various levels of excita-tion intensity: (a) 100 W/cm 2 , (b) 1 kW/cm 2 , (c) 1 MW/cm 2 .

All traces are shown on a time base of 50 nsec per division.

Fig. 9. Waveforms of the excitation and luminescence pulses fromthe silicone grease substrate at an excitation intensity of 1 kW/

cm2 . The time base is 50 nsec per division.

increases, the luminescent pulses tend to follow theexciting pulse waveform, with little or no delay.

The degree to which the luminescence pulse fol-lows the excitation pulse is dependent on the compo-sition of the phosphor. In Fig. 7, it is seen that thered phosphor (no. 2225) exhibits a fast decay even at10 kW/cm2 excitation intensity, while sample no.2210 (Fig. 8) is seen to exhibit a relatively long-liveddecay even at an excitation intensity of 1 MW/cm 2.In Fig. 9 it is seen that the weak fluorescence in-duced in the silicone grease substrate closely followsthe exciting pulse, even at an excitation intensity of1 kW/cm 2 .

This work was supported by Contract DAAD 09-72-C-0028 for the Deseret Test Center, Departmentof the Army.

References1. P. Pringsheim, and M. Vogel, Luminescence of Liquids and

Solids and Its Practical Applications (Wiley-Interscience,New York, 1946), pp. 14, 15.

2. P. Pringsheim, Fluorescence and Phosphorescence (Wiley-In-terscience, New York, 1949), pp. 508-669.

3. S. Shinoya, in Luminescence of Inorganic Compounds, P.Goldberg, Ed. (Academic Press, New York, 1966), pp. 206-277.

4. F. E. Williams, in Advances in Electronics and ElectronPhysics (Academic Press, New York, 1953), Vol. 5.

5. D. Curie, Luminescence in Crystals (Methuen and Co., Ltd.,London 1963).

6. P. A. Leighton, W. A. Perkins, S. W. Grinnell, and F. X.Webster, J. Appl. Meteorol. 4, 334 (1965).

7. J. D. Ludwick, and R. W. Perkins, Anal. Chem. 33, 1230(1961).

8. H. P. Mansberg, and J. Kusnetz, J. Histochem. Cytochem.14, 260 (1966).

9. S. A. Pollack, IEEE J. Quant. Electron. QE-4, 703 (1968).10. S. A. Pollack, J. Appl. Phys. 38, 5083 (1967).11. S. A. Pollack, J. Appl. Phys. 41, 3526 (1970).12. M. Ch. Studer, V. P. Wild, and Hs. H. Gunthard, J. Phys. E.

3, 847 (1970).13. J. B. Birks, d I. H. Munro, in Progress in Reaction Kinetics,

G. Porter, Ed. (Pergamon Press, Oxford, 1967), Vol. 4, p. 239.14. J. Zynger and S. R. Crouch, Appl. Spectrosc. 26, 631 (1972).

15. S. S. Brody, Rev. Sci. Instrum. 28, 1021 (1957).


luminescent rise times of inorganic phosphors excited by high intensity ultraviolet light

Documents