Experimental search for the low-energy nuclear transition in ²²⁹Th with undulator radiation

In this study, about $1\;\times$ 10^{14 229}Th nuclei are prepared on a surface of a transparent crystalline substrate at an areal density of $\approx 1\;\times$ 10¹³ mm⁻². Since the estimated energy of the investigated nuclear transition of ²²⁹Th is 7.8 eV, the substrate material must have a bandgap larger than that. Among a number of candidate materials, we choose CaF₂ due to its large bandgap energy (12 eV) and the availability of highly pure material mainly developed for photolithography optics in the semiconductor industry. Purity of the substrate is one of the important requirements because luminescence from impurities may obscure the searched signal [21]. In addition, to reduce the luminescence intensity caused by radioactivity (Cherenkov radiation and scintillation light), the thickness of the sample should be as small as possible [22]. Diameter and thickness of our substrate (from Hellma Materials GmbH) are 10 and 0.5 mm.

Thorium atoms are prepared on a surface of CaF₂ via adsorption from a solution. Thorium is known for its strong tendency to be adsorbed on the surface of many kinds of materials such as amorphous silica and alumina [23, 24]. The major factor which determines the adsorption efficiency is the acidity of the solution. We first investigate the adsorption efficiency of thorium on a surface of CaF₂ as a function of pH of the solution. The long-lived isotope ²³²Th that is easier to handle due to its low radioactivity (halflife 1.4 × 10¹⁰ years) is used because we expect no difference of chemical properties between ²²⁹Th and ²³²Th.

²³²Th is available as ${\rm T}{{{\rm h}}^{4+}}$ ions in nitric acid solution (Th(NO₃)₄ in 5% HNO₃, Alfa Aesar). First, we take one drop of solution which contains 10^{14 232}Th atoms and put it into a vessel made of perfluoroalkoxy (PFA). PFA is known for its high chemical inertness, which is helpful to avoid contamination during sample preparation. The PFA container is then heated to 80 $^{\circ }{\rm C}$ to evaporate the liquid. After evaporation, the PFA container is cooled down to room temperature and then dilute nitric acid, whose pH is tuned to the desired value, is added to dissolve the ²³²Th nitrate.

The surface of CaF₂ is cleaned by UV ozone cleaning in order to remove adsorbed hydrocarbons and to improve the wettability before placing a drop of the solution on it. This cleaning procedure is essential to obtain stable and efficient adsorption. The surface of CaF₂ is first masked with an aluminum plate which leaves an aperture of 4 mm diameter. Then, it is irradiated with UV light from a low-pressure Mercury lamp for 20 min in air. The distance between the sample and the lamp is a few millimeters. This UV ozone cleaning is efficient only in the area which is exposed to the light and the border between cleaned and none-cleaned area is used to constrain the region that is wetted by the solution. After irradiation with the UV light, the sample is rinsed with distilled water and subsequently dried. Next, 20 μl of solution is put on the surface and the sample is kept in a closed container to prevent evaporation. We find that, at room temperature, it takes three days for adsorption to reach equilibrium. The quantity of adsorbed ²³²Th is measured by comparing that of ²³²Th in the solution before and after adsorption. To measure the quantity of ²³²Th in the solution, we use the dye Arsenazo-III [25]. The color of the Arsenazo-III is originally red and becomes green when thorium ions are added, depending on the quantity of added Th. By observing the change of the transmission spectrum with an optical spectrometer, we can measure the thorium concentration.

Figure 1 shows the adsorption efficiency as a function of pH of the solution. No measurements have been performed at pH > 4 because precipitation starts in this pH range [24]. Stable complete adsorption is realized between pH = 2 and 3.5, and we use pH = 3 for the preparation of the samples described below. In comparison to [23, 24], where complete adsorption of Th on silica and alumina is observed between pH = 3 and 4, the results in figure 1 indicate a stronger binding of Th to the CaF₂ surface, that may be related to the stronger ionic character of CaF₂.

Zoom In Zoom Out Reset image size

To prepare the ²²⁹Th sample, we repeat the same procedure described above with ²²⁹Th instead of ²³²Th. To characterize the sample, we first compare the UV transmission curve of a pure CaF₂ substrate with one with adsorbed ²²⁹Th. Transmission spectra between 120 and 220 nm are shown in figure 2, which are measured using monochromatized synchrotron radiation at Metrology Light Source (MLS) [26]. Although a slight decrease of transmission is observed for the sample with adsorbed ²²⁹Th over the whole wavelength range, no strong absorption at a certain wavelength is noticeable. We also note that the small degradation of VUV transmission shown in figure 2 ensures that the reduction of a VUV fluorescence signal from ²²⁹Th due to absorption in the surface layer is less than 10%.

Zoom In Zoom Out Reset image size

Next, we investigate the spatial distribution of adsorbed ²²⁹Th using the emitted ionising radiation. Figure 3 (left) shows an autoradiograph picture of the ²²⁹Th sample. The ²²⁹Th solution used for the preparation of the samples was prepared several years ago and contains ²²⁹Th in radioactive equilibrium with all its progenies [27]. The distance between the sample and the film is 0.5 mm and the exposure time is 10 min. This picture shows that the distribution of adsorbed ²²⁹Th and its radioactive progenies is homogeneous over the area of the excitation beam (beam radius: 1 mm).

Zoom In Zoom Out Reset image size

For the quantitative estimation of ²²⁹Th, α-spectroscopy of the sample is performed. The spectrum obtained in a measurement time of two days is shown in figure 3 (right). The activity of ²²⁹Th is estimated to be 269(7) Bq based on the solid angle of the detector and the observed count rate, which is consistent with the value estimated from the ²²⁹Th content of the original solution. In repeated measurements we could quantitatively confirm that the adsorbed ²²⁹Th remains stably on the surface under heating up to 110 °C and irradiation with VUV radiation during this study.

In order to investigate the effect of a possible contamination layer around the ²²⁹Th nuclei via the lineshape of the α-spectrum, the data in figure 3 (right) is fitted with a combination of 16 Gaussian functions, corresponding to the dominant α-transitions [28]. The peak positions are fixed to the values listed in [28] while the amplitude of each Gaussian function and a common spectral width are taken as fitting parameters. As a result of the fit, the full width at half maximum of the spectrum is found to be 10.5(1) keV, which is consistent with the resolution of the silicon detector used for the measurement. We therefore conclude that the adsorbed ²²⁹Th is contained in a surface layer whose thickness is so small that its effect on the shape of the α-spectrum as tails or broadening of the line is not detectable here.

In order to excite the nuclear transition at about 160 nm, where a powerful and widely tunable laser is not easily available, we use undulator radiation from the MLS of the Physikalisch-Technische Bundesanstalt in Berlin [29]. The MLS is an electron storage ring dedicated to metrology with radiation in the spectral range from the THz- to the extreme ultraviolet regime. In one of its straight sections, the storage ring comprises a U125 undulator with N = 30 full magnetic field periods of 125 mm length each. The undulator spectrum basically consists of a set of quasi-discrete harmonics (with harmonic numbers n) with wavelengths following $1/n$-proportionality. The wavelength of the harmonics can be tuned by change of the magnetic field strength, i.e. change of the gap distance between the pole shoes. Thus, the MLS U125 undulator's first harmonic can be continuously tuned from about 65 nm through the VUV, UV, and visible spectral range up to infrared wavelengths of about 1200 nm. The spectral width of the first harmonic is given by $\lambda /N$, thus it is about 5 nm (full width) with the first harmonic at 160 nm. For the experiments, the undulator beam is focused by a toroidal mirror resulting in a beam size of about 1 mm radius at the sample position.

Figure 4 schematically shows the experimental setup used in this study, which includes a movable sample holder and a PMT. The sample holder can hold four samples and is equipped with CaF₂ with adsorbates of ²²⁹Th and ²³²Th and with pure CaF₂. The latter two samples and an empty position are used for reference measurements. Since VUV light around 160 nm is strongly absorbed by water, all samples are heated to above 100 °C in vacuum before measurement. Background pressure is always kept at 10⁻⁵ Pa or lower during all measurements described here. To detect photons in the VUV and UV wavelength ranges we use PMTs of the types (Hamamatsu Photonics GmbH) R7639 (diamond photocathode, sensitive from 115–230 nm) and R6836 (Cs-Te photocathode, sensitive from 115–320 nm). The quantum efficiency of R7639 is 40% at 155 nm, which is higher than that of R6836 (10% at 155 nm). The distance between the sample and the PMT is 8 mm, which leads to a solid angle collection efficiency of 0.1 for R7639 and 0.36 for R6836. As a result, the total photon collection efficiency is 0.03 for both R7639 and R6836, which includes the electron collection efficiency of 0.8. Both PMTs are operated in photon counting mode.

Zoom In Zoom Out Reset image size

The main chamber has two mechanical shutters (see figure 4). One is a beam shutter in front of the sample, the other is mounted between the sample and the PMT. To excite the sample, we open the beam shutter and close the PMT shutter. To detect the fluorescence from the sample after the excitation, we close the beam shutter and open the PMT shutter. To make dead time due to shutter operation as short as possible, both shutters are attached to the same pneumatic linear slide. As a result, the dead time between excitation and detection phases is only 10 ms. Great care is taken to prevent scattered light at the beam shutter.

It is well known that under irradiation with intense VUV light optical elements may get rapidly contaminated with surface layers from cracked hydrocarbons [30]. We found that the fundamental undulator radiation used in this measurement (wavelengths 130–350 nm) does not cause severe carbon contamination under the ultra-high vacuum conditions of our experiment. But the second and higher harmonics in the beam create the carbon contamination very rapidly. After irradiation for 3 min with the full undulator spectrum, transmission of the input window of our chamber decreases to less than 40%. To prevent the harmonics with wavelengths shorter than 120 nm from hitting the thorium sample, an additional filter made of MgF₂ is installed in front of the chamber, protecting all elements behind the filter from the carbon contamination. Since the filter itself is now subject to darkening within the irradiated area, it is continuously moved during the measurements in order to obtain stable excitation power.

For the very narrow nuclear transition the undulator device is effectively a broadband continuous radiation source and one may estimate the excitation rate based on the Einstein B coefficient and the spectral density of radiation intensity:

$$\begin{eqnarray}&&R=({{c}^{2}}P)/(8\pi h\nu _{0}^{3}\tau A\Delta \nu ).\end{eqnarray} \tag{ 1 }$$

Here, P is the light power available in bandwidth $\Delta \nu$ over the focus area A, τ is the lifetime of the ^229mTh isomeric state and ${{\nu }_{0}}=1.9\times {{10}^{15}}$ Hz the resonant frequency at 7.8 eV. This reasoning is independent of details of the electrodynamics of the transition and the only relevant nuclear parameter is the lifetime τ for spontaneous emission. So far, τ has not been measured experimentally. Theoretical predictions for an isolated nucleus at 7.8 eV transition energy are in the range of 1000 s to 1 h [31, 32]. In a solid, the decay can be considerably accelerated due to coupling with the surrounding electrons, either in radiative (so-called electronic bridge) or non-radiative processes [33, 34].

Taking the presented experimental parameters into account, figure 5 shows the predicted initial photon count rate after excitation as a function of the isomer lifetime. The photon flux of the undulator radiation is 100 photons s⁻¹ per 1 Hz of spectral bandwidth at 160 nm. The number of atoms is 2 × 10¹³ within an excitation beam cross section of 3 × 10⁻⁶ m². The selected excitation time is 500 s and the total photon collection efficiency is 0.03 as mentioned above. We take into account not only radiative decay but also non-radiative relaxation processes which are not detected by our measurement set-up. Both channels are modeled as exponential decay with time constants τ for the radiative lifetime (isomer lifetime) and ${{t}_{{\rm nr}}}$ for the non-radiative decay. Curves plotted in figure 5 are for ${{t}_{{\rm nr}}}=1$ s (dotted), 100 s (dashed) and infinity (solid). The estimate shows that with a signal detection threshold of a few photons per second (determined by PMT dark counts and luminescence background) it should be possible in this experiment to observe the ²²⁹Th fluorescence for the range of predicted values of the isomer lifetime, unless non-radiative decay dominates, i.e. ${{t}_{{\rm nr}}}\ll \tau$.

Zoom In Zoom Out Reset image size

Targeting a nuclear lifetime of ^229mTh in the predicted range, we excite the ²²⁹Th sample for 500 s. Figure 6 (left) shows typical observed fluorescence signals from the sample. The origin of the horizontal axis corresponds to the beginning of the shutter operation to stop excitation and start detection. The data is averaged over five measurements. Gate time of the counting unit is 1 s. The lower (black) curve shows the signal obtained with the PMT R7639 which is sensitive in the wavelength range 115–230 nm. To investigate a wider wavelength range, we also use the PMT R6836 which is sensitive in the wavelength range 115–320 nm (upper, red curve). The constant level of PMT signals (65 counts s⁻¹ for R7639 and 220 counts s⁻¹ for R6836) is due to radioluminescence from the ²²⁹Th sample, i.e. Cherenkov light and scintillation light. The rates of dark counts of the PMTs are 5 s⁻¹ (R7639) and 60 s⁻¹ (R6836), respectively.

Zoom In Zoom Out Reset image size

In order to investigate whether the obtained signals contain a fluorescence decay signal or not, we fit them by a function consisting of a constant offset and a single exponential function. No statistically significant decay signal is found. To obtain information on the scatter and uncertainty in the amplitude of the exponential function, it is left as the only free fit parameter, while the decay constant τ is fixed. In figure 6 (right), fitting results for both PMTs are shown as a function of excitation wavelength for τ = 100 s. No clear decay signal exceeding the noise level is observed. We also conduct the same analysis for different decay constants such as 50 s and 500 s. However, the results are similar to figure 6 (right).

Targeting a nuclear lifetime shorter than one second, the photomultiplier signal within 500 ms after the excitation (for a time = 500 s) is analyzed. Figure 7 shows a typical signal taken by the R6836 with a gate time of 1 ms. A fluorescence decay with a lifetime of about 100 ms is observed both for ²²⁹Th-coated and pure CaF₂ samples. In order to see the wavelength dependence of this signal, we fit the data by a combination of a Gauss error function and a single exponential function, which describes the shutter operation and decay of the fluorescence, respectively. In figure 7, the amplitude of the exponential function determined by the fitting is shown as a function of the excitation wavelength. At 160 and 240 nm strong fluorescence is observed. But, since they are observed not only for the ²²⁹Th sample but also for the pure CaF₂ sample, we regard that these are emitted from the CaF₂ substrate. The origin of the fluorescence at 160 and 240 nm is not clear so far. It is known that CaF₂ shows emission of a self-trapped exciton around 270 nm under exposure to UV light [21]. But, the lifetime of such fluorescence is of the order of 1 μs. Again, no clear fluorescence related to ²²⁹Th nuclei is observed.

Zoom In Zoom Out Reset image size

Different reasons may be considered why no clear signal from excitation and decay of the ^229mTh isomer is observed in this study. The isomer excitation energy or lifetime may be outside our detection range, or radiationless decay may prevent the observation of fluorescence. We have covered a range of excitation energies from 3.9–9.5 eV, corresponding to $-7\sigma$ to $+3.5\sigma$ around the result from γ spectroscopy 7.8(5) eV [1], that is generally accepted, but has not been confirmed in an independent experiment so far. The commonly accepted nuclear lifetime is around 1000 s, and our experiment is designed for this range. However, the electron bridge mechanism may make the nuclear lifetime much shorter [34]. If it is too short, our apparatus cannot detect the nuclear fluorescence because of the time required for operation of the shutters. There are theoretical predictions about the nuclear lifetime being much longer than 1000 s [35], which is also outside the sensitive range of our apparatus, because the excitation rate will be small and the fluorescence hidden in background noise of the photo detection.

Based on the large bandgap of the CaF₂ substrate and the predominant tetravalent character of thorium compounds we do not expect radiationless decay to significantly reduce the isomer lifetime, because bound electronic resonances and ionization potentials of CaF₂ and ${\rm T}{{{\rm h}}^{4+}}$ are higher than the isomer excitation energy. This is confirmed by the absence of resonances in the VUV transmission spectrum of our sample (see figure 2). Phonon energies, on the other side, are much smaller than the isomer energy so that fluctuating electric charges or magnetic moments from adsorbate molecules or the substrate lattice would couple to the nuclear transition only in a high-order multiphonon process. However, without complete information on the structure of the Th adsorbate and about unexpected chemical compounds relating to, for example, the nitric acid used to optimize pH during sample preparation, we can not rule out radiationless decay with certainty. X-ray photoelectron spectra of Th(NO₃)$_{4}\cdot n$H₂O have shown a broad structure around 5.6 eV [36]. If such compounds are formed, non-radiative decay due to internal conversion may shorten the isomer lifetime. More detailed information about the chemical environment around the adsorbed ²²⁹Th on CaF₂ could be obtained by x-ray surface spectroscopy such as XPS or EXAFS [36, 37] or from electron energy loss spectroscopy.