During the period of development and testing of the present Kuehl et (2) described an automatic vent system utilizing a similar principle but consisting O f all-glass construction. We have not examined their system in the chemical ionization mass spectrometer and thus can offer no comparative data.
LITERATURE CITED (1) W. H. McFadden, R. Terinishi. D. R. Black, and J. C. Day, J. Food Sci., 28, 316 (1963). (2) D. W. Kuel, G. E. Glass, and F. A. Puglisi, Anal. Chem., 48, 804 (1974).
RECEIVEDfor review May 27,1975. Accepted July 3 , 1975.
Inexpensive Device for the Determination of Photomultiplier Current Gains N. W. Bower and J. D. Ingle, Jr.’ Department of Chemistry, Oregon State University, Corvallis, Ore. 9733 1
Since its advent, the photomultiplier tube (PMT) has been an unsurpassed transducer for UV-visible radiation with relatively noiseless amplification. If characteristics of the P M T such as the current gain, collection efficiency, and photocathodic responsivity are known, it is possible to perform absolute calculations so that signals can be reported in units of watts or photons per second instead of the somewhat less meaningful photoanodic current or voltage. This would greatly aid the intercomparison of work of different researchers. Knowledge of the current gain also makes it possible to calculate the fundamental noise in the photocurrent signal and ascertain if other noise sources are present. Although typical values of PMT parameters such as the gain are given by the manufacturer, there is considerable variability among individual tubes, and calibration by the user is mandatory for exacting work. Previously, procedures for measurement of the P M T gain have been somewhat tedious or require special instrumentation. In this paper, we describe the construction of a simple apparatus for evaluation of the P M T gain. The materials to construct the apparatus cost less than fifteen dollars and the procedure for current gain determination takes less than five minutes.
BACKGROUND AND THEORY The basic equations that relate the measurable photoanodic current, photocathodic current, and photoanodic pulse rate in the P M T to the radiant power incident on the photocathode are shown in Table I. The equations and variables in Table I are discussed in more detail in another paper ( I ) . The incident radiant power ( P ) can be calculated from the current gain ( m ) ,the collection efficiency of the first dynode ( T ) , and the cathodic responsivity ( Q A ) . Equations are also presented for the anodic dark current (iad) and the shot noise in the dark current and photocurrent. Note that the fundamental shot noise can be determined if the P M T gain, collection efficiency, and the bandwidth constant is evaluated from knowledge of /If (from the electronic time constant) and an estimate of (Y (from the dynode statistics and dynode gains, 6, ( 1 - 3 ) ) . For these equations, it is assumed that photoemission occurs only from the photocathode and not the dynodes, and that photoelectrons that are not collected by the first dynode cause negligible secondary emission a t other dynodes. It is also assumed that the collection efficiency for secondary electrons between dynodes is one or is incorporated into the dynode gains (&), and that m and Q A are independent of the incident radiant power and constant over the incident wavelength interval. Finally, the dark current is taken to be independent of the incident radiant power and Author to whom correspondence should be addressed.
to originate solely from thermionic emission at the photocathode. Equations 8 and 9 show the relationship between the current gain and dynode gains (Si), and in Equation 9, all dynode gains are assumed equal. Usually, the PMT specification sheets give only typical and maximum and minimum values of the cathodic responsivity (sensitivity) ( Q A ) and the anodic responsivity (mQ,J. Since the anodic responsivity depends on the gain, it is specified a t one P M T voltage, or a plot of responsivity vs. P M T voltage is reported. The cathodic and anodic responsivities are evaluated with a calibrated light source such that the radiant power incident on the photocathodic surface is known ( I , 2). Usually the responsivities are measured at relatively high radiant powers and are assumed to be independent of the radiant power. Manufacturer’s data indicate that cathodic responsivity can vary approximately f100% from the typical value. The standard procedure ( I , 2, 4 ) for calculating the gain involves taking the ratio of the photoanodic current to the cathodic current measured under equivalent radiant power and biasing conditions. The photoanodic current is first measured with the P M T wired in its typical configuration. Then the dynode chain is rewired or the P M T is transferred to another housing and socket so that all the dynodes are tied together to the anode and used as the “new anode”. Then the voltage is adjusted until the cathodeanode potenti’al is the same as the cathode-first dynode potential in the normal PMT configuration. There are a number of disadvantages and assumptions to this standard technique ( 4 ) . First, it assumes a collection efficiency of one. Second, the incident radiant power cannot be too large or the photoanodic current will be in the nonlinear region (i.e,, the photoanodic current will not be proportional to the incident radiant power). Under conditions of high P M T gain, the photocathodic current may be too small and difficult to measure because of noise limitations. Third, elaborate and time consuming changes to the biasing network are often required. Fourth, if neutral density filters are used to attenuate the radiant power by a known amount for photoanodic current measurements so that the photocathodic current will be larger and easier to measure, then the accuracy depends upon the accuracy of the filter characteristics. Fifth, the two currents are not measured under equivalent conditions because of a time factor (Le., drifts in the calibration light source) or because the photocathode may be moved (differences in cathode responsivity with respect to position). One assumption of the standard procedure which was stated above is that the collection efficiency is one. Note that the ratio of photoanodic to photocathodic current as defined by Equations 1 and 2 is f m and, hence, the calcu-
ANALYTICAL CHEMISTRY, VOL. 47, NO. 12, OCTOBER 1 9 7 5
2069
Table I. Basic Equations i,, = PQ,q)n
(5)
i,, = photoanodic current, A Qi = cathode radiant responsivity or sensitivity, A/W 17 = collection efficiency of first dynode, dimensionless m = current gain of photomultiplier, dimensionless P = radiant power incident on photocathode, W i,, = photocathodic current, A Y,, = observed photoanodic pulse rate, sec-' d l = discriminator coefficient or fraction of photoelec trons that reach the active part of the first dynode and a r e seen a s a photoanodic pulse (0 < A l < l ) , dimensionless iad= anodic dark current, A ice = effective cathodic dark current, A (u&* = r m s shot noise in dark current, A K = bandwidth constant, sec-' e = charge of an electron, C Af = noise equivalent bandwidth, sec" CY = secondary emission factor of photomultiplier, di mensionless ( u , ) ~= ~ r m s shot noise in photocurrent, A 6 , = gain of ith dynode stage, dimensionless k = number of dynode stages, dimensionless
cedure has most of the limitations of the standard procedure, but no special wiring is required. In addition, it must be assumed that the two cathodic surfaces, or a t least that the parts of the cathode illuminated, are identical. Robben (11) and Klobuchar e t al. (4)developed photon counting procedures in which the gain is taken as iaplerap or the ratio of the photoanodic current to the product of the photoanodic pulse rate and the charge of an electron. Note that from Equations 1 and 3 this ratio is mlA1. Unlike the standard procedure, the gain determination is not affected by the collection efficiency. However the gain is the average charge of the photoanodic pulses seen and does not take into account pulses not passed by the discriminator or photoelectrons that are collected by the first dynode which do not result in photoanodic pulses. The gain determined by this procedure may be higher than the gain determined by the standard procedure (if the discriminator coefficient is less than one) because the usual gain takes into account the finite probability of no secondary emission. The advantages of photon counting techniques ( 4 ) are that both the current and pulse rate measurements may be made simultaneously (hence, under exactly equivalent conditions) and, a t any reasonable incident radiant power, there is no need for attenuation filters, and no modification of the dynode chain is required. The main disadvantages are that photon counting equipment is required and, a t larger incident radiant power, dead time corrections must be made (which requires some measurements and graphical extrapolation). The manufacturer's specifications indicate that the anodic responsivity can vary about f500% from the typical value. This is due primarily to the variability of the gain between individual tubes rather than to differences in the cathodic responsivity. Hence, the most variable of the P M T parameters discussed, the gain, must be individually determined for each P M T if reasonable absolute signal and noise calculations are to be made. Below we describe a simple apparatus and procedure for P M T gain calibration which is based on the standard method but which does not suffer from many of its limitations.
INSTRUMENTATION lated gain will be too low if 7 is significantly less than one. In other words, some of the photoelectrons emitted from the photocathode do not cause secondary emission a t the first dynode. Collection efficiencies are not usually reported in specification sheets and are difficult to measure as illustrated by the recent controversy in the literature (5-8). Foord e t al. ( 5 ) have determined 7 to be 25% for a particular PMT. Conversely, Young and Schild (8) have measured a collection efficiency of 86% for the same PMT, which is more in keeping with past experience. The exact definition of collection efficiency can vary. Here we define 7 as the fraction of the photoelectrons from the photocathode which strike the active part of the first dynode's surface, even though some of the electrons have a finite probability of not causing secondary emission. Some of the photoelectrons may strike an inactive part of the first dynode (e.g., the dynode support) or may even be collected by dynodes further down the chain. Sometimes 7 is taken as the fraction of photoelectrons which strike the first dynode and result in secondary emission (a smaller fraction). Another procedure (referred to as the independent PT procedure) which we have used to calculate the gain in the past (9, 10) is to take the ratio of the photoanodic current from a P M T to the photoanodic current of a vacuum phototube (PT). The PT is chosen to have the same cathodic responsivity a t the wavelength of measurement. This pro2070
*
ANALYTICAL CHEMISTRY, VOL. 47,
NO. 12,
The apparatus is illustrated in Figures 1 and 2. Figure 1 shows the bias box (Part A) which consists of a P M T socket, dynode resistors, and a rotary switch in a metal box. The wiring schematic is shown in Figure 2. The bias box is connected directly to the P M T with the exposed socket on the box. The bias box switch has two positions: the PMT configuration (closed switch position shown in Figure 2) and the PT configuration. In the PT position, the cathode is biased a t I/~O of the voltage it has in the PMT configuration (so that the cathode-first dynode potential is the same as for the PMT) and the first dynode and anode (as well as the other dynodes through the 100 K resistors) are connected to the measurement circuit. Note that resistor R1 can be replaced by another resistor or a zener diode and the cathode-first dynode biasing will be the same for both configurations. A PVC sleeve (part B) is machined to fit snugly around the PMT (part C ) base. Some care must be taken in making the sleeve, as it should be light tight, yet there is some latitude in the diameters of the PMT bases, even from the same manufacturer. The basic housing (part D)is a bored PVC rod with one outside surface milled flat. A hole or window is milled in the flat side to which a metal plate (part E) with a corresponding hole is cemented. Holes are drilled in the metal plate to attach the housing to the monochromator. The sleeve plus PMT assembly is inserted into the housing and secured with a thumb screw. The P M T housing could be modified to accept different size side-on PMT's or end-on PMT's. For photocurrent measurements, the housing is attached to a monochromator (Heath EU-70) and tungsten light source (Health EU-701-50) combination. The output of the biasing box (anode or anode plus first dynode) is connected to an operational amplifier
OCTOBER 1975
Li
,W!
E
,100
100
U
100
100
-
100 100
100 100
I
r To current measurement system
A
To high voltage source
Flgure 2. Bias box schematic
Flgure 1. PMT gain measuring apparatus Part A: Bias box: aluminurn BLidd box dimensions: 2.25 X 2.25 X 4 in.; BNC plugs. Part B: Sleeve; i.d. = 7.279 in.: 0.d. = 1.575 in.; height = 1.325 in. Part C: PMT; RCA 1P28. Part C): PMT housing: lower i.d. = 1.575 in.; lower inside height = 1.325 in.; upper i.d. = 1.279 in.; upper inside height = 1.5 in.; 0.d. = 2 in.; outside height = 3.25 in.; window side milled flat = 1.0 X 3.25 in.: window = 0.4 X 1.0 in.: center of window is 2.0 in. above base of PVC housing. Part E: Mounting plate; y8 X 5 X 7 in. aluminum; holes drilled to fit monochromator: window dimensions the same as on PVC housing
All resistors are in kohms. The 100 K resistors are 1 % metal film. The 900 K resistor is a combination of 10% carbon resistors good to 1 % of the specified value. The switch (S)is a 4-position, 2-pole rotary switch
Table 11. Currents a n d Calculated Gains iCp x 1011“
P h r r Yo.
1 (Analog Devices Model 415) wired in the current to voltage configuration. Rotary switches allow selection of feedback resistors (105-108R) and capacit.ors to vary the time constant. The output of the OA is connected to an integrating voltmeter (Heath EU-805)(a 1-second integration time was always used) and the high voltage supply (Heath EU-4211is set to the desired voltage (600 V for all measurements). To evaluate the independent PT gain procedure, measurements of PM’T and PT photoanodic current were made using commercial housings (McKee-Pedersen MP-1016 and M P 1021). The housings were attached to the monochromator with a plate to Cannon connector combination.
RESULTS AND DISCUSSION The photoanodic and photocathodic current of 7 different PMT’s (all RCA lP28’s) were measured with the new assembly. In the PT configuration, the photocurrent which reaches any of the dyn’odesor anode is actually measured. This current should be equivalent to the photocathodic current (Le., the current which leaves the photocathode) and this equivalence was confirmed by other measurements. The currents and calculated gains are shown in Table 11. In addition, the photoanodic current of 7 different PT’s (RCA 929) was measured to obtain the average A) to provide an alternate calcuphotocurrent (2.1 X lation of the P M T gains as shown in Table 11. The calculated relative standard deviation in the PT photoanodic currents was 34% and the mean photoanodic current was not significantly different from the mean photocathodic currents of the PMT’s. which indicates similar cathodic surfaces for the PT’s and PMT’s. Measurements were :made with a 2-A spectral bandpass (100-wm slit) a t 450 nm (the cathodic responsivities of the PT’s and PMT’s are the same a t this wavelength). With the new apparatus, ten measurements of each current were made to establish a good mean, and ten measurements with the shutter closed were made for every current measurement to establish the zero. After switching from one configuration to the other, it takes about two minutes for the signal to stabilize, possibly because of charging effects of the tube wall. This period is short enough that the lamp radiance does not change significantly. I t also takes only a minute to switch from one P M T to another so that relative or anodic radiant responsivities can be obtained and absolute responsivities if a calibrated light source is available. The geometric mean (rather than the arithmetic) of the photoanodic current and P M T gains were calculated be-
2 3
4 5 6 7
(A
3.5 2.3 2.2 1.2 2.6 3.5 1.2
iaP x 106’ (A)
9.2 1.1 0.73 1.1 2.3 5.2 3.8
10-4‘
26 4.8 3.4 9.0 9.0 15 33
Arithmetic mean 2.4 x lo-” ... .. Geometric mean ... 6.37 5.03 Standard deviation 0.91 x lo-“ ... ... Standard deviation 0.19 0.41 0.36 of logarithm a Ri = los Q , R/ = lo5 a. Gain from new procedure
10-4‘
44 5.4 3.6 5.8 9.9 27 14
... 5.04
... 0.39 ( L ~ , ~ / L ~ ~ ~ ) .
Gain calculated from PT data (ratio of PMT photoanodic current measured in cbmmercial housing to mean photoanodic current of
I PT’s). cause of the exponential nature of the gain distribution. Similarly the standard deviations for these parameters are reported in terms of logarithms. The manufacturer’s value (12) for the typical gain is 9 X IO4 and this compares well with the calculated geometric means for both gain procedures. Note that the data indicate that the standard deviation in the anodic current (i.e., anodic responsivity) is due primarily to variations in the gain, rather than in the cathodic responsivity (Le., 0.19 < 0.36). Of further interest is that the manufacturer’s reported maximum deviation from the typical gain is about the same as our calculated standard deviation. T h e apparatus will work except a t high P M T voltages or gains where the photocathodic current will be too small to measure if the incident radiant power is reduced enough so that the photoanodic current will be in the linear region. Here filters can be used or the gain can be determined from the ratio of two photoanodic currents measured for equivalent incident radiant power a t two different P M T voltages where the gain a t the lower voltage has been evaluated. Usually the log-log plot of P M T current gain vs. P M T bias voltage is linear and the slope is the same for equivalent PMT’s although the intercept varies. Hence, the P M T gain a t any P M T voltage can be extrapolated from the gains determined a t two P M T voltages or from the slope and the gain a t one P M T voltage. For squirrel cage PMT’s like the one we used, the collection efficiency is near one ( 1 3 ) so that this procedure gives
ANALYTICAL CHEMISTRY, VOL. 47, NO. 12, OCTOBER 1975
2071
a good estimate of the actual gain and not just of Tm. Equivalent results were obtained over a wide spectral region, even when we used a modulated hollow cathode and lock-in amplifier to make photocurrent measurements. This apparatus and photon counting equipment could be used to evaluate the collection efficiency as previously described (8, 14). From Equations 2 and 3, the ratio of the product of the photocathodic current (measured in the PT configuration) and the charge of an electron to the photoanodic pulse rate (measured in the P M T configuration) is pIA1. By extrapolation, the photoanodic pulse rate a t A1 = 1 can be estimated so that the ratio yields p directly.
(12) RCA-1P28 Specification Sheet. (13) R. M. Schaffer, RCA, personal communication. (14) L. Birenbaum and D. B. Scarl, Appl. Opt., 12, 519 (1973).
RECEIVEDfor review May 2, 1975. Accepted July 7, 1975. Acknowledgment is made to the donors of the Petroleum Research Fund, administered by the American Chemical Society, for partial support of this Research.
ADDENDUM
ACKNOWLEDGMENT We thank Steve Hoyt for machining the P M T housing.
LITERATURE CITED (1) J. D. Ingle, Jr., and S. R. Crouch, Anal. Chem., 44, 785 (1972). (2) RCA Photomultiplier Manual, Technical Series PT-61, RCA Electronic Components, Harrison, N.J., 1970. (3) EM1 Photomultiplier Tubes, Brochure 30M/6-67 (PMT), Gencom Division, Varian/EMI, Plainview, N.Y. (4) R. L. Klobuchar, J. J. Ahumada, J. V. Michael, and P. J. Karol, Rev. Sci. hstrum., 45, 1071 (1974). (5) R. Foord, R. Jones, C. J. Oliver, and E. R. Pike, Appl. Opt., 8, 1975 11969). ---, (6) R. Foord, R . Jones, C. J. Oliver, and E. R. Pike, Appl. Opt., 10, 1683 (19711. (7) A. T. Young, Appl. Opt., 10, 1681 (1971). (8) A. T. Young and R. E. Schild, Appl. Opt., I O , 1668 (1971). Jr.. Anal. Chim. Acta.. 77. 71 (1975). 191 \., .I. -. E. Hawlev - ,and -~-J. D. Inole.,~ (10) L. D. Gthman. S. R. Crouch, and J. D. Ingle. Jr., Ana/. Chem., 47, 1226 (1975). (11) F. Robben, Appl. Opt., 10, 776(1971). \
~I
~
2072
~
I
Carbon Black Adsorbates: Separation and Identification of a Carcinogen and Some Oxygenated Polyaromatics
While the paper by Avram Gold, Anal. Chem., 47, 1469 (1975), describing the identification of polycyclic aromatic hydrocarbons and oxygenated polycyclics adsorbed on carbon black was in press, a paper by L. Wallcave, D. L. Nagel, J. W. Smith, and R. D. Waniska, Enuiron. Sci. Technol., 9, 143 (1975), appeared in which one of the same polycyclics (cyclopenta[cd]pyrene)was identified. The Wallcave paper should be added as a reference. Both papers are in agreement on the physical-chemical properties of the compound; in addition, the Wallcave paper presents PMR data. Also, reference 10 should read: L. Wallcave, Environ. Sci. Technol., 3,948 (1969).
ANALYTICAL CHEMISTRY, VOL. 47, NO. 12, OCTOBER 1975
