Nano-optical antennas (NOAs) are becoming pivotal tools in modern photonics owing to their unprecedented ability to control light-matter interaction in ultracompact architectures [1]. Various NOA concepts, inherited from low-frequency antenna theory, have, for instance, demonstrated unprecedented ability to tailor emission rate and directionality from luminescent elemental sources, such as fluorescent molecules and quantum dots [2–7]. We address here the use of a NOA concept for controlling the x-ray-excited luminescence (XEL) from scintillators, thus providing a nano-optically driven approach in the development of novel architectures for x-ray imaging and real-time dosimetry.

The development of miniaturized x-ray sensors and dosimeters is hindered by the difficulty in achieving efficient conversion of x-ray photons to electrons in electronic devices, which imposes large detection volumes [8]. Indirect detection, which combines luminescent materials with visible optical detectors, has demonstrated acceptable performances in terms of image contrast and signal dynamics, and is now widely exploited by a large variety of detectors in the scientific, medical, and industrial domains. In such a technique, semiconductor materials used under various forms, such as crystals or powders (called scintillators or phosphors), convert high-energy impinging radiations into light that is detected with silicon-based photodiodes and cameras or with photosensitive films. The resulting optical devices are often of modest compactness, which may represent limitations from practical point of view. However, the technique offers great promises in low x-ray flux detection, since one can detect optical luminescent signals down to a single photon.

Accessing direct in-fiber XEL detection offers the prospect of a wide range of x-ray sensors and dosimeters free from bulky optics. The integration of x-ray detection functionality at the end of an optical fiber is highly desirable, as it would lead to ultracompact, plug-and-play, and flexible architectures, establishing a completely new versatility in x-ray imaging and dosimetry. Reaching efficient optical coupling between scintillators and conventional step index optical fibers remains a real challenge owing to the strong mismatch between the almost omnidirectional dipolar emission of luminescent particles and the noticeably low numerical aperture and weak guiding properties of optical fibers. The approach followed so far consists of compensating for this low-coupling efficiency by employing a large scintillating volume coupled to large core multimode fibers [9–14]. These techniques lead to modest resolution on the mm to cm range and a compactness weakly compatible with endoscopy. Recently, sub-millimeter architecture has been demonstrated by covering the cleaved end facet of a 600-μm core diameter multimode fiber with a thin layer of scintillators [15]. Scaling dosimeter architecture with 125 μm down to an 80-μm outer diameter, single-mode fibers with a direct scintillator-to-fiber coupling, for instance, by improving optical detection with photon counters, remains questionable and has yet to be addressed.

We recently theoretically proposed an alternative approach relying on the engineering of an ultracompact interface between a luminescent source and a fiber, aimed at optimizing the in-fiber photon outcoupling [16]. This is essentially the application of the concept of a horn antenna to optical frequencies, with the capability of collecting and transferring up to 70% of the luminescence photons to the fiber-guided mode. In this Letter, we use this NOA approach to show experimentally direct imaging and dose detection of an x-ray-focused beam with a spatial resolution of a few micrometers. By optimizing the coupling channel between a tiny scintillation cluster and an optical fiber, we achieve highly miniaturized and spatially isotropic x-ray detection systems: x-ray detection volume of a few tens of ${\mathrm{\mu m}}^3$ is obtained at the end of a 125-μm-diameter, single-mode fiber, thus demonstrating both system overall compactness and local detection capability. This holds promise of ultracompact, flexible, plug-and-play, and high-resolution devices for monitoring, mapping, and controlling the dose of high-energy radiations in a wide panel of applications covering scientific, medical, and industrial domains.

From antenna theory [17], two methods are possible for impedance matching the dipolar emission at the end of a coaxial cable with vacuum. The first technique relies on the generation of localized resonances over subwavelength assemblies of metallic elements of specific shapes and sizes, a concept that has recently been demonstrated optically [1]. The second method is based on the efficient coupling of the point-like source to a flaring waveguide, leading to larger antenna structures, such as the horn antenna, which also has been recently successfully extended to optical frequencies [16,18,19].

A horn antenna normally directs emission from a dipolar source in free space by connecting it to a coax-to-waveguide adapter, resulting in a simple architecture in which a point-like emitter is placed in between a reflector and the flaring waveguide (dipole-to-mirror distance of $\lambda /4$) [16]. In our case, the x-ray-to-light conversion from a scintillation cluster is promoted by its distribution of crystalline defect centers [20]), which can be considered as an ensemble of randomly oriented dipolar optical sources coupled to the antenna. To assess the ability of the horn NOA to direct the XEL from a tiny scintillation cluster, we modeled with commercial finite difference time domain (FDTD) software a horn NOA coupled to one of its crystalline defects (i.e., a single dipolar emitter). The point-like source is embedded in a high-refractive-index host matrix (2-μm-diameter sphere) covering the apex of a dielectric horn [see Fig. 1(a)]. Its refractive index $n=1.85$ is the average refractive index of the barium platino-cyanide [21]. The dielectric horn is 38-μm long, has a rounded apex with a 0.5-μm radius of curvature, and is terminated with a 6-μm radius circular end-facet. The overall structure is covered with a 150-nanometer-thick aluminum layer. Figures 1(b)–1(d) show the antenna’s angular emission for the three dipole positions noted A, B, and C in Fig. 1(a), respectively. We see that whatever the dipole position, the directivity [17] of the horn NOA (up to 273) is high enough to efficiently outcouple the luminescence into a low-acceptance angle, single-mode fiber. The antenna gain is given for the three dipole positions in the inset of Figs. 1(b)–1(d). The antenna gain [17] is defined as the product of the maximum antenna directivity and the quantum yield of the emitter coupled to the antenna [22,23]. The maximum gain of the horn NOA is reached for point-like sources placed close to the antenna symmetry axis ($0z$), for which up to 40% of the radiated photons are coupled to the directional antenna emission in free space. The lower gain observed at points B and C is mainly due to the excitation of surface plasmons at the dielectric-metal interface, which dissipates energy. These results predict that the horn NOA can be used as an efficient interface between a tiny scintillation cluster and a single-mode optical fiber.

 

Fig. 1. Directivity for an x-oriented emitter coupled to the horn NOA, in the two major planes (xz) (blue curves) and (yz) (red curves). The emitter is embedded in a high-refractive-index medium (host matrix) covering the dielectric tip apex [see schematic diagram (a)]. (b)–(d) Directivity for an emitter positioned at points A–C, respectively [see (a)]. The gain of the horn NOA is given in the inset of (b)–(d) for the three different emitter positions.

Download Full Size | PDF

To produce the scintillator-coupled horn NOA fiber system, polymer micro-tips are first grown by photopolymerization at the cleaved end-facet of a single-mode fiber (SMF-28) [24]. Tips are 38-μm long, have a radius of curvature of 1 μm at their apex, and are terminated with a 12-μm-diameter circular end-facet in contact to the fiber. Next, a scintillator cluster (barium platino-cyanide) is attached at the micro-tip end under microscope observation, following the approach proposed in Ref. [25]. Isolated micrometer-sized clusters (between 1 and 5 μm) are deposited onto a flat surface and the micro-tip is brought into contact with a single cluster. Owing to adhesion forces between the tip and cluster, the luminescent material remains attached to the micro-tip when it is removed from the surface. Finally, the resulting fiber-integrated structure is metal coated with a few-nanometer-thick titanium adhesion layer followed by an aluminum layer that is 150 nm thick. Aluminum is chosen for its high reflectivity at visible wavelengths and high transparency to x rays. Figure 2(a) displays optical and SEM images of a resulting fibered x-ray dosimetry platform. Because the SEM irradiation tends to burst the polymer micro-tip [see inset of Fig. 2(a)], which irremediably results in a drop of luminescence collection efficiency, our scintillator grafted fiber micro-tips used in the following experiments have not been characterized by SEM.

 

Fig. 2. (a) Optical microscope image of a fiber-integrated NOA-based x-ray dosimeter. Inset: SEM micrograph of a scintillation cluster at the apex the micro-tip before aluminum coating (scalebar: 4 μm). (b) Scheme of the experimental setup.

Download Full Size | PDF

Our concept of a fiber-integrated x-ray dosimeter is first demonstrated with the fully calibrated x-ray focusing test bench shown in Fig. 2 [26]. Polychromatic x-ray radiation from a low-power Rh target lab source (800 μA at 35 kV) is focused with a polycapillary lens into a 25-μm-wide spot (FWHM, manufacturer’s data). The x-ray fiber dosimeter is positioned at the focus with a 3D manual translation stage. The final centering process, together with beam scanning, is realized with a precision-motorized system. During scanning, x-rays cross the transparent tiny dosimeter with minimum absorption by the thin aluminum layer and directly excite the NOA-embedded scintillators. Owing to the high reflectivity of aluminum at visible wavelengths, a large portion of the XEL from the scintillators is collected by the horn NOA and launched towards the fiber core with high directionality. The unique impedance-matching property of horn NOA ensures optimum photon outcoupling into the fiber mode. In-fiber XEL is registered with a room temperature photon counter (LynXea) from Aurea Technology.

Figure 3(a) shows the x-ray beam profile at the focal plane, revealing a full width at half-maximum (FWHM) of 24.6 μm, which is in good agreement with the manufacturer’s spot size value. Note that, because both the beam focus and cluster size are much larger than the incident x-ray wavelengths, one can retrieve the real beam profile with direct deconvolution techniques between the experimental plot and the point-spread function of the fiber probe. Whereas point-spread function can be easily obtained at optical frequencies, it remains challenging with x-rays. Their high penetration depth and the x-ray fluorescence phenomenon impede the achievement of the sharp transmission steps required to unambiguously estimate the point-spread function. The point-spread function could then be measured by probing the focal spot produced by an x-ray Fresnel lens, whose spot size is small enough to be considered as a Dirac function [27]. According to the basic rules related to the convolution of Gaussian functions, the experimentally measured FWHM is found to be very close to the manufacturer’s reference value of 25 μm, which confirms that the luminescent source is on the micrometer scale, consistent with the cluster sizes measured by SEM during preliminary tests of the scintillator-to-tip grafting process [cf. Inset of Fig. 2(a)]. Possible rounding of the manufacturer’s spot size value also may explain a part of the discrepancy observed here. Note that, as in the case of the fluorescence process involving tiny fluorescent beads, the luminescence of small scintillator clusters is almost insensitive to the x-ray incident angle. Our detection system thus led to the same level of signal for two x-ray incidence directions parallel and perpendicular to the fiber axis.

 

Fig. 3. (a) Experimental plot of an x-ray-focused beam ($>8\mathrm{keV}$) at focus. (b) Calibration curve of the dosimeter (detected luminescence versus incident x-ray flux). Minimum detectable x-ray power density is estimated to be of the order of ${10}^3\mathrm{photons}/\mathrm{s}/{\mathrm{\mu m}}^2$. (c) Time trace of the luminescence intensity obtained with the same fiber dosimeter positioned within a collimated x-ray beam emanating from a Cu target source of a commercial diffractometer (Bruker). The source is successively switched on and off.

Download Full Size | PDF

The calibration curve of our x-ray dosimeter, shown in Fig. 3(b), reveals a linear response of the system with respect to the incident x-ray power. For a maximum x-ray power of ${210}^9\mathrm{photons}/\mathrm{s}$, the detected luminescence signal reaches 130 kcts/s, which dramatically exceeds the noise level of the photodetector. This curve is used to evaluate the minimum x-ray power density detectable with our approach to a value smaller than ${10}^3\mathrm{photons}/\mathrm{s}/{\mathrm{\mu m}}^2$. Our dosimeter is, therefore, sensitive enough to detect and probe x-rays from diffractometers. To validate this point, we performed a direct x-ray dose measurement of a low-power Cu-target source used in a commercial diffractometer from the Bruker company (D8 Advance). A time trace of the luminescence intensity with the source successively on and off is reported in Fig. 3(c). We see that the detected signal is well beyond the noise level (signal-to-noise ratio equal to 5.5), which validates our approach in low-power x-ray radiation sensing. From Fig. 3(b), the flux density of the diffractometer source is estimated at ${5.110}^3\mathrm{photons}/\mathrm{s}/{\mathrm{\mu m}}^2$. Note that the sensitivity of our detector may decrease while the energy of the impinging x-rays increases, because the absorption cross-section of the cluster is known to be a decreasing function of x-ray energy. However, the correlated detection opportunity provided by photon counters offers the perspective of enhancing detection sensitivity for either investigating higher x-ray energies or increasing resolution ability at low energies [28]. Note that the upper-dose rate leading to luminescence saturation should be larger than ${10}^6–{10}^7$ impinging x-ray photons per second per single point-like emitter. This lower limit of the saturation dose is estimated from the decay time of the emitters (on the range of ${10}^{-7}–{10}^{-6}\mathrm{s}$ [29]) and on the assumption of perfect emitters having maximum absorption rate and an intrinsic quantum yield equal to 1. The dose range considered in this study is well below this lower limit.

In this study, we used barium platino-cyanide as the x-ray-to-light converter, which is known to produce XEL emission at a wavelength of 520 nm due to a scintillation process promoted by its crystalline defect centers [21,29,30]. However, other luminescent materials are available, depending on the targeted application [31–33]. To ensure that the luminescence is induced by the scintillation cluster, we measured the in-fiber collected luminescence spectrum from the scintillator-grafted NOA by connecting the fiber to a spectrometer (Princeton SP2300) (see Fig. 4). We see that the detected signal is spectrally centered at 530 nm, which is consistent with the barium platino-cyanide emission spectrum. The little red shift of the emission peak with respect to the predicted spectrum may be due to the influence of the structured environment upon the cluster’s emission properties. Moreover, we probed the x-ray-focused beam with a fiber NOA free from scintillators. In that case, a detection signal of a few hundreds of luminescence photons per seconds was measured, which is much lower than the signal level obtained in the presence of the scintillation cluster [up to 130 kcts/s, see Figs. 3(a) and 3(b)]. These measurements show that the detected XEL is undoubtedly due to the scintillators embedded within the NOA.

 

Fig. 4. Spectrum of the detected optical signal when the fiber probe is centered with respect to the focus.

Download Full Size | PDF

We proposed a nano-optically driven technological approach for the real-time dose detection of high-energy radiations in ultracompact and flexible architectures. By coupling scintillators to a horn NOA and by exploiting the record emission directionality and impedance matching of this metallo-dielectric antenna, we successfully developed a micrometer-sized x-ray sensor at the end of a 125-μm-diameter, single-mode optical fiber. By leveraging the versatility and ubiquity of fiber-optics technology, this may constitute a key step towards the widespread use of ultracompact x-ray detectors in a wide range of scientific, medical, and industrial domains. For example, high-energy imaging and real-time dosimetry (x-rays, gamma rays, charged particles, etc.) for radiotherapy and brachytherapy would become possible with completely new versatility, accuracy, and control possibilities, and with ultra-low footprint detection devices. More generally, to the best of our knowledge, with our approach, NOAs make a first key contribution to the development of x-ray sensing protocols and architectures. In that context, nano-optics could open the way to radically new concepts for smaller and faster x-ray detectors for scientific, medical, and industrial metrology and characterization. While the capability of plasmonic NOAs to dramatically increase fluorescence decay rate has been intensively studied [3–5], their ability to control light-matter interaction generated under x-ray exposure has not been reported. NOAs may thus also contribute to the development of faster x-ray detectors.