CALIBRATION OF A SILICON PHOTODIODE
The synchrotron radiation is almost an unique radiation source with a wide continuous spectral range extending over the VUV region and is known as a primary standard radiation source. However, in practice, it is very difficult to obtain accurate spectral quantum efficiencies of VUV detectors by using synchrotron radiation as a primary standard. On the contrary, if a primary standard detector is available, the calibration of the spectral response of a detector is accomplished more easily and more precisely. As Samson demonstrated, in the photon energy range from 12 eV (the ionization potential of Xe) to 49 eV (twice the ionization potential of He) an ion chamber
filled with such a rare gas may work as a primary standard detector with unit quantum efficiency in the case of full absorption. Consequently, assuming identical monochromatized photon flux, the spectral quantum efficiency of any detector may be, by comparison, calibrated. The available spectral range of the’ rare gas ion chamber can be extended to shorter wavelengths by taking into account of secondary and multiple ionization phenomena. Following this method the National Bureau of Standards (renamed the National Institute of Standard and Technology) had, in the middle of 1970’s, established the detector standards using synchrotron radiation in the 20 to 60nm spectral range; eventually this range was ex-
tended to 5 nm 141. Since then, to our knowledge, there has been no announcement of another establishment of the national detector standards in this spectral region. Aim of our work at ETL (Electrotechnical Laboratory) is to establish, in Japan, such standards (spectral responsivity or spectral quantum efficiency) in the 10 to 60 nm spectral range by using synchrotron radiation from the electron storage ring TERAS 151. Since rare gas ion chamber is, at present, considered to he an almost unique and most accurate absolute detector which can he realized easily in this spectral region, we also adopt it as a primary standard. We attempt to make uncertainties in the calibration as small as possible by a proper design of the ion chamber.
A schematic diagram of the VUV detector calibration station constructed at ETL is shown in figure. Synchrotron radiation from TERAS is focused by a toroidal mirror onto an entrance slit of a toroidal grating monochromator. The entrance aperture of an ion chamber iq located at the exit slit of the onochromator. Detectors to he calibrated are located behind the ion chamber. Photon flux impinging a detector can he then measured by the ion chamber. Quantum efficiency of the detector can be derived by the ratio of the detector photocurrent response to the photon flux (both the photocurrent and the photon flux are normalized to the same stored beam current).
Since size of the photon beam, used in the calibration, increases in function of the distance from the entrance of the ion chamber, a silicon photodiode can be moved along the optical axis to the position just behind the entrance aperture to collect all the photons passing through the ion chamber entrance aperture. Once the silicon photodiode is calibrated as a secondary standard detector, detectors, even those having small size photocathode, may he calibrated by using another small sized photon heam through the comparison with the secondary standard detector. As shown in figure I, the ion chamber has four stage ion collectors (i, is the ion current of the m-th stage) and a cylindrical electron collector. The axis of the ion collectors and the axis of the electron collector lie 40 mm and 17.5 mm apart from the optical axis, respectively.
Because of this electrode geometry, the electric field strength around the optical axis is much smaller than that near the ion collectors. This nonuniform electric field has an advantage in that a photoelectron undergoes smaller acceleration than in the case of uniform electric field strength of which must be strong enough to retard all the photoelectrons. Consequently, probability of secondary ionization, which causes errors on photon flux or on gas absorption coefficient measurements can he minimized. Then the four stage ion collectors enable us to evaluate their end effects and to improve consequently the overall accuracy. At the entrance of the ion chamber a glass capillary array (GCA) which has about 100 capillaries with inner diameter 100 pm and length 2 mm each, is
CHARGE COLLECTION CHARACTERISTlCS
Figure 2 shows the photocurrent versus the applied voltage (I-V) characteristics of each stage for argon pressure of 0.05 Torr at the photon energy of 21.2 eV. Since the ion potential (‘P,,,) of argon is 15.76 eV, photoelectrons produced have an initial kinetic energy of 5.4 eV. All the curves except for the first stage curve have good plateaus above about 5 V which coincides well with the kinetic energy of the photoelectrons. As for the first stage, a photoelectric current of about 0.25 pA is observed in a reversed bias voltage without gas, which must be originated either from the GCA or the chamber wall. The I-V curve of i, current seems to he affected by this photoelectric current. Figure 3 illustrates the attenuation of the photon flux through argon gas. It is found that the i, current is much lower than the expected value in each case (probably some ions are escaping from the region determined by the length of the ion collector). The first stage ion currents are also smaller than those expected by about 2 7% almost, this decrease being independent of the gas pressure. This suggests that some ions produced in the first stage might he lost by passing through the glass capillary array. Therefore, we usually use only the second and the third stage ion currents to determine the incident photon flux or the absorption coefficient of the used gas.
where L is the periodic length along the optical axis of each stage. In this case, the wavelength of the incident radiation is 40 nm and the gas used in the ion chamber is neon. The experimental results agree well to the calculation within 3 5% for both of the second and the third stage ion currents. This shows first that the parameters, used in the calculation such as the lengths of the ion collection regions and the absorption coefficient of the gas (independently determined by transmission measurement), are correct and second, that the pressure dependent phenomena such as secondary ionization do not occur in this case. Similar comparison at 15 nm (photon energy 82.7 eV) is shown on figure 5. In contrast to the previous case both second and third stage ion currents gradually exceed, as the pressure increases, the predicted values.
This behavior is interpreted as secondary ionization and then multiple ionization should occur since the photon energy is larger than the twice of the ionization potential of argon (21.6 eV) and than the double ionization potential (62.5 eV). To extend the spectral range of such absolute detector to shorter wavelengths, we have to know the amount of the above two effects. For the multiple ionization, we used published data of the mean charge (y value)  to correct its contribution. According to the data, the mean charge of neon is,for example, 1.05 at the photon energy of 100 eV. To evaluate the secondary ionization contribution, we adopted the pressure variation method which has been already developed by Samson . The correction factors, in function of the gas pressure, can be obtained by dividing the experimental ion current by the calculated one derived from known gas absorption coefficient (the calculated scale is set equal to the experimental one at a near-zero gas pressure). In the case of neon ion chamber, we determined the correction factors for secondary ionization in the wavelength range from 10 to 32 nm as a function of gas pressure.
Therefore, we can determine spectral photon flux and/or spectral responsivity in the wavelength range above IO nm. Figure 6 shows the spectral photon flux determined by the neon ion chamber at the exit of the monochromator, the storage ring electron energies being successively of 230, 300 and 750 MeV. Unit of the ordinate is the number of the photons per second per I mA storage current when both widths of the entrance slit and the exit slit are 1 mm, which corresponds to a bandwith width of about 0.3 nm. In order to minimize the unwanted effects of high orders (generally present in such diffraction grating monochromators) we usually lower the electron beam energy as low as 230 MeV or less when the measurements in the wavelength longer than about 34 nm are needed. We evaluated the contribution of the second and the third order radiation quantitatively ; indeed, when the electron beam energy is 230MeV and the monochromator
tuned up, for example, at 50 nm, the ratio of the photon flux of the second and the third order to the one of the tuned first order radiation is 1.4 % and 1.0 %respectively. In determining the photon flux of radiation and/or the quantum efficiency of a detector, we took then account for the contribution of the high orders using the result of this measurements. The incident photon flux and the gas absorption
coefficient reduced at the standard condition can be derived from Eqs. (1) to (3). Since both of them are unique, they, ideally, must be independent of the gas pressure at which the measurement is made. In a typical experimental condition, it was observed that both of the reduced absorption coefficient and the photon flux which were determined using the second and the third ion currents are constant within 1.0 %to
the gas pressure above about 0.07 Torr. Absorption coefficients of neon at STP obtained by this ion chamber are shown in figure 7 by dots ; these measurements are well fitted with previous data (continuous line) from Marr & West characteristics such as long-term stability and spatial uniformity of their responsivities. As reported [IO], we also found that an inversion layer type silicon photodiode well satisfies the above requirements. Calibration result of such UDT XUV-100 photodiode is shown on figure 8 (dotted line). Also shown (continuous line) on this figure a comparison with our model .In this latter, we first derive transmittance of the silicon dioxide layer on the silicon substrate using their complex.refractive index and we assume that the mean energy (3.7 eV) to produce an electron-hole pair in silicon is independent of the
UNCERTAINTIES OF THE CALIBRATION
The uncertainties in the detector calibration depend on the spectral responsivity and/or responsivity uniformity of the detector to be calibrated. As an example, a summary of uncertainties estimated in the calibration of a silicqn photodiode using the neon ion chamber is shown in table I. Sources of uncertainty in the evaluation of the photon flux are fluctuations in the ion chamber currents, errors in correction required for secondary or multiple ionization or spectral impurity (high orders and stray light). At 15 nm, the correction errors for secondary ionization are the most important. Sources of uncertainty in the measurement of the silicon photodiode photocurrent are fluctuations in the photocurrent, errors in the impurity radiation correction, and the nonuniformity of the detector responsivity. Since the quantum efficiency of the silicon photodiode becomes large as the wavelength becomes short, the errors in the correction for the high orders are rather large at long wavelengths. Due to instabilities in the position andlor the size of the electron beam in the storage ring, the photon flux normalized by the electron beam current is not necessarily constant. At present, variation of the normalized photon flux amounts to 4 %. As a total, uncertainties at the wavelengths of and 50 nm are estimated to be approximately 9 %, 5 % and 5 %, respectively.
Author: Umeiono, Tsukuba-shi, Iboraki,