High-dynamic-range transmission-mode detection of synchrotron radiation using X-ray excited optical luminescence in diamond
aa r X i v : . [ phy s i c s . i n s - d e t ] F e b High-dynamic-range transmission-mode detection of synchrotron radiation using X-ray excitedoptical luminescence in diamond ∗ Stanislav Stoupin † Cornell High Energy Synchrotron Source, Cornell University, Ithaca, NY 14853, USA
Sergey Antipov
Euclid Techlabs LLC, Solon, OH 44139, USA
Alexander M. Zaitsev
College of Staten Island and Graduate School of the City University of New York, Staten Island,NY 10314, USA and Gemological Institute of America, New York, NY 10036, USA
We demonstrate enhancement of X-ray excited optical luminescence in a 100-micron-thick diamond plateby introduction of defect states via electron beam irradiation and subsequent high-temperature annealing. Theresulting X-ray transmission-mode scintillator features a linear response to incident photon flux in the range of7.6 × to 1.26 × photons/s/mm for hard X-rays (15.9 keV) using exposure times from 0.01 to 5 s. Thesecharacteristics enable a real-time transmission-mode imaging of X-ray photon flux density without disruptionof X-ray instrument operation. I. INTRODUCTION
Noninvasive (or minimally invasive) visualization of X-raybeam profiles under in-operando conditions is of high prac-tical importance at experimental stations of large-scale syn-chrotron and XFEL user facilities. The desired generalizeddevice functionality can be described as real-time imaging offlux density distribution in a chosen observation plane placedacross the direction of propagation of X-ray beams. Com-mon metrics, which reflect performance of the experimen-tal station, such as, average beam position, intensity/photonflux over a chosen region of interest as well as position andintensity fluctuations can be derived from the real-time pro-files of flux density. A recently demonstrated, quantitative ap-proach for flux density monitoring features detection of elec-trical charge in a lithographically patterned (pixelated) dia-mond plate [1]. X-ray transmission-mode scintillators (pro-ducing X-ray excited optical luminescence) with low X-rayabsorption, coupled to a visible-light area detectors is an alter-native, more straightforward strategy towards implementationof the imaging functionality. Diamond is a preferable choicefor the X-ray transmission-mode scintillator due to its lowX-ray absorption, high radiation hardness, remarkable ther-mal and mechanical properties. Video-monitoring of X-rayexcited optical luminescence in thin diamond plates is com-monly used for this purpose, however, predominantly in asemi-quantitative manner, where the observed profile is eval-uated based on visual appearance characteristics (e.g., spa-tial resolution, relative brightness) often without documentedknowledge on response linearity and dynamic detection range.Such advanced characterization is perhaps unnecessary formonitoring profiles of intense synchrotron beams upstream ofthe beamline monochromator (beamline front-end) [2–4]. In ∗ accepted for publication in J. Synchrotron Rad. † Electronic address: [email protected] other cases (e.g., X-ray beams downstream of the monochro-mator), a more quantitative approach is required. Park et al.[5] recently demonstrated ≃ µ m beam position stability fora monochromatic beam using a commercial diamond screenas a real-time imaging detector. However, no information onthe response linearity is provided, therefore the dynamic de-tection range of their study remains unclear. In this work, weachieved enhancement of X-ray optical luminescence in a thindiamond plate by introduction of additional defects via elec-tron irradiation and subsequent high-temperature annealing.The resulting responsivity and the dynamic range of the lu-minescence detection was increased by more than one orderof magnitude compared to commercially available diamondscreens of the same thickness. Contrary to the commercial di-amond screens, the response to the incident X-ray photon fluxwas found to be linear in the range of 7.6 × to 1.26 × photons/s/mm . II. SAMPLES
The samples procured for this study were circular polycrys-talline diamond plates prepared using chemical vapor deposi-tion (CVD) with diameter of 10 mm and thickness of 100 mi-cron. These were of the nominal tool, thermal and opticalgrades (supplier specification), presumably of different im-purity concentration. An additional polycrystalline diamondplate of the ”optical” grade of the same shape was acquiredfrom the same supplier. It was subjected to irradiation andsubsequent annealing to generate luminescent centers (irradi-ated and annealed plate). To create vacancies in the diamondlattice irradiation was performed with an energetic electronbeam. Practically any diamond contains impurities. Both di-amond plates of the ”optical” grade had completely transpar-ent visual appearance. The main impurity was N with ex-pected concentration on the order of 0.1-10 ppm. Anneal-ing was performed in vacuum to promote formation of NV(nitrogen-vacancy) defect centers. In this work we made no
400 500 600 700 800 900Wavelength [nm]1000020000300004000050000 I n t e n s i t y [ a . u . ] optical × 5irradiated and annealed FIG. 1: Spectra of UV-excited luminescence in the irradiated andannealed diamond plate (red solid line) and diamond plate of opti-cal grade (yellow dashed line; data collected using 5 times longerexposure time per point). attempts to study boron doped diamond samples for the fol-lowing reasons. While it is known that some natural and syn-thetic boron-doped single crystals produce enhanced lumines-cence the characteristic luminescence lifetimes are on the or-der of seconds [6, 7], which can result in a nonlinear responsefor real-time detection. Prior studies indicate that boron is notan efficient luminescent center in boron-doped polycrystallineCVD diamond [8, 9].Luminescence spectra under UV excitation at wavelengthof 360 nm were measured for all samples. Selected spec-tra are shown in Fig. 1. The luminescence spectrum of theunannealed ”optical” grade sample (from here on referred toas ”optical”) is shown with yellow dashed line. It can bedescribed as a broad band with a maximum response in thespectral range from 500 to 600 nm. This band is due to char-acteristic optically active defects, which usually dominate lu-minescence of as-grown nitrogen-doped CVD diamonds. Inluminescence measured at low temperature, these optical cen-ters produce zero-phonon line at wavelength 468 nm (notclearly seen in the present spectrum taken at room tempera-ture) [10, 11]. Although the 486 nm center is very typical fornitrogen-doped CVD diamonds, its origin has not been iden-tified yet. The luminesence spectrum of the ”thermal” gradesample was found to be relatively weak, having similar spec-tral shape (not shown in the figure). The sample of ”tool”grade did not produce any reasonably measurable lumines-cence. The spectral response of the irradiated and annealedsample (shown with red solid line) was much stronger com-pared to that of the ”optical” grade sample. Its appearance ischaracteristic to the NV0 center [10], which includes the zero-phonon line (narrow peak at 575 nm) and the broad phononside band above it.
FIG. 2: Experimental setup (top view, see text for details).
III. EXPERIMENT, RESULTS AND DISCUSSION
The characterization experiment was conducted at 2Bbeamline of Cornell Synchrotron Radiation Source (CornellUniversity, USA). The beamline features CHESS compact un-dulator source and a side-bounce monochromator operating ata set of fixed photon energies (9.7, 15.9, 18.65, 22.5 keV).Diamond screens were placed in light-tight environment at45 degrees with respect to incident X-ray beam, which wasshaped to 1 × size using X-ray slits. X-ray excited lu-minescence was measured using Mako G319C camera (Al-lied Vision) equipped with an objective lens. The camerawas placed at a distance of about 100 mm from the sample.An ionization chamber (IC0) was placed upstream of the dia-mond screen to monitor the incident photon flux and anotherionization chamber (IC1) was set downstream the diamond tomonitor photon flux transmitted through the sample. The ex-perimental setup is shown schematically in Fig. 1 (top view).During preliminary tests with X-ray beam, luminescenceintensity of the tool grade sample was found negligible com-pared to that of other samples. Therefore, results for this sam-ple were excluded from analysis. This can be easily explainedby dark appearance of the sample (high self-absorption of lu-minescence). The X-ray transmissivity was measured at 15.9keV. The observed values were 0.965, 0.969 and 0.973 for thethermal, optical and the irradiated plates, respectively. For a100-micron-thick diamond plate oriented at 45 degrees withrespect to the incident beam the transmission factors calcu-lated using tabulated values for the mass attenuation coef-ficient and the mass energy-absorption coefficient [12] are0.965 and 0.977, respectively. Thus, X-ray attenuation in thediamond plates was small as expected. In the first experiment,linearity of the system was explored by performing measure-ments at a fixed incident photon flux at 15.9 keV using dif-ferent exposure times. The range of exposure times was from0.01 up to 5 s (except a few 10 s exposures for the weaklyluminescent ”thermal” grade sample). This range could bedescribed as real-time conditions for most experiments per-formed at synchrotron sources (excluding fast time-resolvedexperiments). The response was measured as a sum of pixelintensity across the image of the beam footprint, normalizedby the number of pixels. An offset representing camera darkcurrent (sum of pixel intensity for a region of the same size, R e s p o n s e [ c t ] (a) thermalopticalirradiated and annealed10 −2 −1 E posure time [s]10 R e s p o n s e [ c t ] (b) thermalopticalirradiated and annealed FIG. 3: X-ray excited optical luminescence (average response overthe 1 × X-ray beam footprint) as a function of exposure timefor the as-received diamond plates (thermal and optical grades) andfor the irradiated and annealed plate. The linear-linear plot (a) issupplemented with log-log plot (b). The solid lines are fits to thelinear function. not exposed to X-rays) was subtracted. The fluctuation of theresponse was found to be less than one percent. In the figuresthat follow the related uncertainties are less than the size ofthe figure markers. The values of the photon flux were evalu-ated using ion chamber flux calculator for N2 filled chambersof length 6 cm [13]. The response as a function of exposuretime was found to be linear to within experimental uncertain-ties for all studied diamond plates. The results are shown inFig. 3.This observation enabled further characterization of the lu-minescence as a function of the incident photon flux regard-less of exposure time (normalization of the response by theexposure time was performed). Since the dynamic range ofthe camera was only 12 bit (up to 4096 counts per pixel), vari-able exposure time enabled greater total dynamic range forquantitative evaluation of luminescence intensity. In the sec-ond experiment the luminescence was evaluated at variableincident flux levels at a photon energy of 15.9 keV. The time-normalized response (response from here on) as a functionof the incident photon flux for the different diamond plates R e p o n e [ c t / ] (a) thermalopticalirradiated and annealed10 Incident flux [photon / ]10 −2 −1 R e p o n e [ c t / ] (b) thermalopticalirradiated and annealed FIG. 4: X-ray excited optical luminescence (average time-normalized response over the 1 × X-ray beam footprint) asa function of the incident photon flux for the as-received diamondplates (thermal and optical grades) and for the irradiated and an-nealed plate. The linear-linear plot (a) is supplemented with log-logplot (b). The dashed line connecting data for the optical grade sam-ple in (b) illustrates nonlinear behavior. The solid lines are fits to thelinear function. is shown in Fig. 4. Fits with a linear function are shownby the solid lines. The proportionality coefficients in the lin-ear fits were 3.7 × − , 2.9 × − , and 3.2 × − for thethermal, optical, and irradiated and annealed samples, respec-tively. For the sample of thermal grade, only two data pointswere measured, which precludes analysis of response linear-ity. The response of the optical grade sample deviates fromthe linear behavior (as shown by the dashed line), which re-duces the dynamic range for detection of photon flux, whilethe response of the irradiated and annealed sample is linear toa good approximation. The response of the irradiated sampleis greater by more than one order of magnitude. The result-ing measured dynamic detection range is more than 3 ordersof magnitude. The detected levels of the photon flux are from7.6 × to 1.26 × photons/s.Figure 5 shows color images (RGB) of the X-ray beam foot-print taken at 15.9 keV. Different exposure times and flux lev-els were used to optimize image statistics. FIG. 5: Color images of X-ray excited optical luminescence in di-amond plates of different grades: thermal (a), optical (b), irradiatedand annealed sample (c).
For the samples of thermal and optical grades the lumines-cence color is predominantly in the blue range. We note thatthe luminescence spectrum measured under UV excitation(Fig. 1) corresponds to green color. This discrepancy suggestsnon-equivalence of X-ray and UV excitation. A more detailedinterpretation of this observation falls outside the scope of thepresent study. For the irradiated and annealed sample the lu-minescence color is predominantly red due to the character-istic NV0 luminescence (consistent with the UV-excited lu-minescence spectrum). Some structure in luminescence of theirradiated sample was observed (lines of intensity propagatingoutside of the X-ray illuminated region). We attribute this ef-fect to surface quality (luminescence re-scattering on surfacedefects/scratches).In the final experiment the response of the irradiated andannealed sample was measured at the several fixed photonenergies available at the beamline. Under the conditions ofnegligible re-scattering and self-absorption of luminescence,which are applicable to the thin diamond plate of the opticalgrade, the response as a function of the X-ray photon energy E X can be approximated with [14]: R ( E X ) ∝ F E X (1 − exp( − µ ( E X ) t )) , (1)where F is the incident photon flux, µ ( E X ) is the X-ray at-tenuation coefficient of the material (for practical purposesmass-energy attenuation coefficient is often used), and t isthe thickness of the plate. Here it is assumed that the energyconversion efficiency and the quantum yield do not dependon the X-ray photon energy, which is a valid assumption forhard X-rays since the photon energies are substantially abovethe characteristic energies of any electronic transitions of thecarbon atom. Figure 6 shows the measured luminescence re-sponse normalized by F and E X as a function of the photonenergy. This ratio should be proportional to the absorptivityof the diamond plate according to Eq. 1. The absorptivity ofthe sample for a 100- µ m-thick plate, scaled to the experimen-tal data using the optimal proportionality coefficient is plottedwith a solid line. The agreement between the experiment andEq. 1 is good. The observed minor discrepancies can be at-tributed to uncertainties in determination of the incident pho-ton flux. IV. SUMMARY
In summary, we have demonstrated a more than one or-der of magnitude improvement in responsivity of X-ray ex-
10 12 14 16 18 20 22Photon energy [keV]2468 R e s p o n s e [ c t / ( p h / s ) / e V )] FIG. 6: X-ray excited optical luminescence (average energy-normalized response over the 1 × X-ray beam footprint) asa function of the X-ray photon energy for the irradiated and annealeddiamond plate. The solid line represents scaled absorptivity of X-rays in the 100- µ m-thick diamond plate (Eq. 1). cited luminescence in thin (nearly transparent for hard X-rays) diamond scintillator by introduction of NV defect states.The results of synchrotron X-ray measurements show that thenew scintillator has a linear response to the incident hardX-ray flux density in the range from 7.6 × to 1.26 × photons/s/mm . This was demonstrated using a simple imag-ing scheme in the visible range with exposure times for in-dividual frames from 0.01 to 5 s. Thus, an imaging deviceusing the new scintillator can provide minimally invasive real-time transmission-mode imaging of photon flux density with-out disruption of X-ray instrument operation. Further im-provement in responsivity can be achieved by exploring otherhighly luminescent defect centers, reducing influence of com-peting non-radiative processes (e.g., via control of crystal lat-tice quality) [15] as well as by using more advanced imag-ing detectors and image intensifiers. Future work could beextended to detailed studies of spatial resolution and time re-sponse of the new scintillator. Outcomes of such studies maylead to the next-generation technology for beam diagnosticsat large scale X-ray user facilities. Acknowledgments
This work is based upon research conducted at the Cor-nell High Energy Synchrotron Source (CHESS) which issupported by the National Science Foundation under awardDMR-1332208. The work was supported by Department ofEnergy SBIR grant de-sc0019628. [1] T. Zhou, W. Ding, M. Gaowei, G. De Geronimo, J. Bohon,J. Smedley, and E. Muller, J. Synchrotron Rad. , 1396 (2015).[2] M. Degenhardt, G. Aprigliano, H. Schulte-Schrepping,U. Hahn, H.-J. Grabosch, and E. Wrner, J. Phys. C: Solid St.Phys. , 192022 (2013).[3] B. Kosciuk, Y. Hu, J. Keister, and S. Seletskiy, AIP Conf. Proc. , 020030 (2016).[4] S. Takahashi, T. Kudo, M. Sano, A. Watanabe, and H. Tajiri,Rev. Sci. Instrum. , 083111 (2016).[5] J. Y. Park, Y. Kim, S. Lee, and J. Lim, J. Synchrotron Rad. ,869 (2018).[6] S. Eaton-Maga˜na and R. Lu, Dia. Rel. Mat. , 983 (2011).[7] E. Gaillou, J. E. Post, D. Rost, and J. E. Butler, Am. Mineralo-gist , 1 (2012).[8] R. J. Graham, F. Shaapur, Y. Kato, and B. R. Stoner, Appl. Phys.Lett. , 292 (1994).[9] K. Iakoubovskii and G. J. Adriaenssens, Phys. Rev. B , 10174 (2000).[10] A. M. Zaitsev, Optical properties of diamond: a data handbook (Springer, Berlin, 2001).[11] K. Iakoubovskii and G. Adriaenssens, Phys. Stat. Solidi (a)