Part I Fundamentals and Instrumentation 24 4 Detector Systems The only development noted in the field of photographic recording was that of gamma compensating filters. The principle practical application and methods of preparation of such filters have been described by Nagy (807 808). 25 Part I Fundamentals and Instrumentation 4.1 PHOTO-ELECTRIC DEVICES Trassy and Robin ( 1 0 7 0 ) have discussed t h e principles and technical solutions in building a n electronic detection system for AAS with particular reference t o its use with a n h.f. plasma torch. 4.1.1 Photomultipliers Photomultipliers with a spectral response covering t h e range 180-600 nm are widely available with q u a n t u m efficiencies rising as high as 25%.Some tubes have a n extended red response stretching to 900 nm. The q u a n t u m efficiency o f "solar blind" photo- multipliers is 7% a t 2 5 0 nm and 0.02% a t 350 nm (243) while that of t h e RCA C31000E photomultiplier at 632.8 nm is 6.1% (237). Young and Schild (249) found t h e photo-electron collection efficiency of t h e EM1 9 5 5 8 t u b e t o be 75%. The dark current due t o thermal emission f r o m t h e photocathode tends t o be greater with cathodes of extended red response. T o minimise dark current t h e cathode area should be as small as possible compatible with receiving all t h e signal from t h e instrument.Owing t o inhomogeneities i n t h e photocathode (278) it should be fully illuminated during measurements. T o overcome t h e effect of cathode inhomogeneity d e Galan and Wagenaar (368) when studying line profiles introduced a field lens behind t h e exit slit t o project a n image o f t h e grating o n t o a larger area of the photomultiplier surface. Several a t t e m p t s have been made t o improve photocathode quantum efficiency b y improving t h e optical capture o f photons (282). In o n e arrangement (43 1 ) t h e p h o t e cathode was illuminated obliquely b y shining t h e incident light i n t o t h e side of t h e lens o n t h e cathode end of t h e photomultiplier this gave a q u a n t u m efficiency enhancement o f 50- 100%.A comparable improvement ( 4 3 0 ) was obtained b y passing the incident light through a quartz prism optically coupled t o t h e face of the photo- multiplier; t h e angle of incidence was such t h a t multiple reflections occurred within t h e envelope of t h e tube. As t h e photomultiplier is n o t a storage device t o measure low light levels some form of integrating electronics must b e incorporated i n t o t h e system. Image dissector tubes (image orthicon and vidicon tubes) store photo-images in t h e phosphor until released b y t h e scanning electron beam. The image of a spectrum can b e formed o n t h e photocathode and it can be used f o r multi-element analysis.It has been claimed ( 2 3 2 ) that t h e advantages o f this system are high q u a n t u m efficiency simplicity and cost when compared with a n array of photomultiplier tubes and true multiplex operation by storage of the information f r o m short pulses. The resolution of the system appeared t o be 1 6 line pairs rnm-l. The disadvantages of t h e system (428) included a small region of linearity a poor signal-to-noise ratio some lag and "sticking" of images in pulsed studies and t h e danger of burn-out at high light levels. The resolution was 1 nm and t h e storage time 1 sec. The S / N ratio varied from 1O:l to 3 0 l and t h e dynamic range was limited t o a factor of 5.The use of a commercial Russian television set (PTY-101) with a television camera tube (LI-17) for detection of emission spectra has been described (1091 1092). Qualitative and quantitative analyses were possible in t h e range 400-750 nm b u t detection limits were inferior t o those obtained by photoelectric detection. 4.1.2 Solid State Detectors The sensitivity of solid state detectors is still significantly less than that of photo- multiplier systems. Where light levels are n o t low however such devices have some Part I Fundamentals and Instrumentation 26 advantages and surface barrier diodes developed for nuclear counting have been evalu- ated as photodiodes for detection of radiation from 4 0 0 nni t o t h e near infra-red (87).Results indicated excellent performance in comparison with t h e more conventional diffused surface junction types of photodiodes. Bournans ( 5 6 9 ) has made a detailed study of t h e use of photodiodes and phototransistors as detection devices for multi- channel emission spectrometry. The detector consisted of a n array of equidistant photodiodes o r phototransistors each 3 mm long and 25 p m wide t h e interspace between t h e diodes being 5 pm. An array of 2 0 diodes was assembled. This system offers t h e possibility of t r u e multiplex operation with simultaneous subtraction of background signals f r o m t h e emission signal.Studies of t h e performance of t h e system are not yet complete. 4.2 SIGNAL PROCESSING 4.2.1 Photon Counting Photon counting as a means of measuring low light levels has been used for a t least 2 0 years. Only recently however has its use become widespread due principally to t h e newly available high-speed counting systems. Most photomultipliers may b e used as p h o t o n counters. However owing t o the limitations of t h e associated electronics non-linearities may arise d u e t o pulse pile up. The principal advantages of photon counting lie in t h e measurement of low light intensities and in t h e production of digital signals with consequent simple integration. The detection and measurement of very low light levels were discussed b y Flint (1).By careful choice of t h e photo- multiplier operating conditions and detection system light levels of about 1 0 photons s-l could be detected in t h e ultra-violet t o blue spectral region (more were required in t h e red). Unless extreme precautions were taken stray light reaching t h e photo- multiplier could influence t h e o u t p u t . Even when lock-in amplification was used Flint considered it pointless spending time and money t o reduce t h e dark current t o t h e pA range when stray light produced a photocurrent greater t h a n this. Many commercial spectrometers slit assemblies and photomultiplier housings were insufficiently light tight f o r critical work.In atomic spectroscopy it is n o t usual t o be measuring such low light intensities. Other applications of p h o t o n counting include t h e determination of t h e relative quantum efficiencies of monochromator and detector systems (96). There has been some dispute ( 2 5 1 2 5 2 ) o n t h e contribution of small noise pulses t o the pulse height spectrum obtained from a photomultiplier. Jones et a1 (250) showed t h a t to obtain t h e same relative variance f o r a signal t h e observation time for capacitative store should be 2.6 times greater t h a n t h a t required f o r p h o t o n counting with digital store. The application of p h o t o n counting t o spectroscopy has been examined b y several workers Aldous a n d Bailey (491) Dagnall e t a1 (193 376 487 and 1041) Dawson (446) a n d King and Williams (950).The latter compared t h e relative sensitivity of p h o t o n counting with photographic detection for t h e measurement of Pb and Ni b y emission spectroscopy. Photon counting has been compared with lock-in amplification detectors f o r measuring transient signals in atomic fluorescence using t h e carbon rod atomiser (491). The transient response o f p h o t o n counting and d.c. systems was com- pared f o r fluorescent and absorption measurements of Cd vaporised from a Pt wire in a flame (1041). The digital system gave a lower detection limit b y factors of between 2 and 5.A detailed examination (446) of t h e comparative performance of analog and digital systems f o r t h e measurement of flame emission absorption and fluorescence 27 Part I Fundamentals and Instrumentation signals using a n EM1 9 5 9 2 A photomultiplier showed t h a t b o t h analog and digital systems gave detection limits close t o t h e ultimate in emission and fluorescence set by t h e randomicity of photoelectron emission. In atomic absorption however t h e limit was set b y t h e instability of t h e light source. When using photon counting systems care must be taken t o ensure t h a t t h e response of t h e system is sufficiently fast t o avoid a n y possibility of pulse overlap as this will lead t o spuriously large pulses; if voltage discrimination is used this should n o t lead t o t h e loss of small signal pulses unless it is demonstrated that these arise from dynode emission.At t h e present time t h e highest convenient counting rate is of t h e order of l o 6 counts s-l. When a low background atomising system is used t h e photomultiplier dark signal may be compar- able with t h e emission from t h e a t o m reservoir. In this instance cooling of t h e photo- cathode can lead t o lower detection limits. When evaluating t h e performance of p h o t o n counting and other methods for t h e measurement of light intensity it is desirable t h a t some measure of t h e linearity of t h e response of t h e system be made.A scheme f o r t h e testing of t h e linearity of absorp- tion spectrophotometers has been outlined b y Hawes (276) and is based o n t h e use of Bouguer's Law and t h e superposition of optical fields. A theoretical analysis o f t h e linearity of the response of a photomultiplier (67) shows t h a t t h e deviation from linearity is less than 3% for all photomultipliers when t h e anode current is less t h a n 10% of t h e current in t h e final resistor of t h e dynode supply chain. 4.2.2 Fast Response Systems With t h e use of impulse atomisation i n a variety of spectroscopic techniques time resolution of t h e resultant signal can be used t o give improved analytical precision nd t o study atomisation and excitation processes.Photomultiplier signal currents have been gated i n t o integrating capacitors with t h e aid of a complementary switched pair of FETs a t t h e rate of 2 0 MHz (305 672). An RCA C70045 photomultiplier has been operated t o give a time resolution of 1 ns (692). Fast gating of t h e photomultiplier signal was used by Treytl et a1 (40 8 7 3 ) t o eliminate t h e continuum emission signal which occurred almost entirely during t h e initial 1 0 0 ns of t h e pulse generated when a Q-spoiled laser beam was used t o excite t h e emission of solid samples. 4.2.3 Modulation and Signal-to-Noise R a t i o It has been shown (771) in o n e instrument t h a t modulation of the light source in atomic absorption did n o t completely remove flame interference effects.If a strong band occurred a t t h e same wavelength as an analytical line a negative absorption reading resulted from a lean flame (i.e. emission was greater t h a n absorption) while in a rich flame positive absorption readings were obtained (i.e. absorption greater than emission). These effects may be attributed t o a frequency in t h e noise spectrum o f t h e flame emission passing through t h e electrical filters. The strong atomic emission from Ca in t h e nitrous oxide flame leads t o interferences in atomic absorption when light source modulation is effected b y a mechanical chopper interposed between the lamp and flame (39).In this system emission from t h e flame was reflected back from t h e hollow cathode lamp window and thus was modulated b y t h e chopper. The inter- ference could be avoided by tilting t h e window of t h e hollow cathode lamp and would n o t occur in systems using electronic source modulation. The effect of sine- and square-wave modulation of t h e light intensity o n t h e photon count distribution obtained for differing integration periods has been evaluated (227). Part I Fu ndam en tals and Ins tru men tu t io n 28 When t h e integration time is greaJer t h a n t h e modulation time t h e photocount distri- bution is Poissonian and 0 2 N = N.S / N ratio measurements have been carried o u t b y several workers (236 273 429 446). It is found that t h e S / N ratios for current measuring and p h o t o n counting systems are very similar and t h a t these d o n o t change significantly with change of photomultiplier supply voltage. A comparison between photomultipliers a n d photodiodes (273) shows t h a t in general photomultipliers have a better S/N a t photo-anodic currents less than 5 X lop9 amps b u t photodiodes are marginally better a t higher currents.