Laser-induced time-resolved luminescence in analysis of rare earth elements in apatite and calcite

Laser-induced time-resolved luminescence was used to study rare earth element (REE) containing natural apatite and calcite minerals. The luminescence from 400 nm to 700 nm in the minerals was analyzed with excitation ranges 210–340 nm and 405–535 nm. As an outcome, several useful excitation wavelengths to detect one or more REE from apatite and calcite are reported. The feasibility of selected excitations in e.g. avoiding the disturbance of intense Mn2+ luminescence band, results was demonstrated with a non-gated detector.


Introduction
Rare earth elements (REEs) are representing a group of elements consisting of lanthanoids (La-Lu), yttrium (Y), and scandium (Sc). REEs are needed for modern technology [1] and a rapid analysis technique would be a benefit at many stages of their supply chain. REEs can act as luminescence centres in host minerals and therefore luminescence spectroscopy is efficient in detecting them. Natural minerals typically host several REEs simultaneously [2]. Although luminescence emission of REEs is generally observed as relatively sharp lines and only some ions e.g. Ce 3+ and Eu 2+ have broad bands [3], in natural minerals, the luminescence spectrum can be rather complex with several overlapping bands. For example, the crystal field of the host mineral lattice can split the energy levels of REE 3+ ions [3] and as a consequence, more luminescence emission lines may appear. In addition, the intensity of luminescence emission is dependent on both the mineral's crystallinity and on the surroundings of the luminescence centre, which may change the energy transfer, quenching, reabsorption and sensitizing processes [4][5][6].
Although one of the most common techniques to luminescent mineral research is cathodoluminescence [7][8][9][10][11], laser-induced time-resolved luminescence spectroscopy has several benefits, e.g. high laser power excites also weak luminescence centres, time-resolving enables separation of different luminescence centres from each other and the method could be easily combined with other laser-spectroscopic method like Raman spectroscopy or laser-induced breakdown spectroscopy (LIBS), as reviewed in the literature [4,6,[12][13][14]. However, the occurrence of luminescence is strongly dependent on the selection of the laser wavelength used for the excitation. Therefore, the aim of this research was to survey, which excitation wavelengths would be optimal for detecting luminescence of certain or multiple REEs. A tunable ns-pulse laser was used at excitation regions 210-340 nm and 405-535 nm to study luminescent minerals, apatite and calcite, from the same geological site. The emission was detected in the visible range, with a view to the possible on-line and on-site applications in the future. Based on the results, suggestions for optimal excitations for the detection of REEs were produced and the concept was tested using the laser with a basic non-gated spectrometer.

Materials & methods
Two natural mineral samples, apatite and calcite, were chosen from the Siilinjärvi ultramafic alkaline-carbonatite complex in Eastern Central Finland. At the Siilinjärvi complex REE-hosting minerals are occurring in the "glimmerite" (phlogopite rock) and carbonatite rock types [15,16]. REEs are known to be situated in Ca-bearing phases in carbonatites like carbonates and apatites and in previous research overall REE content is analyzed to be smaller in calcite (509 ppm in Mg-rich calcite in apatite sövite i.e carbonatite proper) than in apatite (2986 ppm in glimmerite and 3820 ppm in apatite sövite) [17]. The results enable the general assumption of the occurrence of the same REEs in the host lattice of the chosen samples. At the moment, apatite is only exploited as a phosphate source [15]. Greenish apatite with a length about 2 cm representing medium-grained carbonate glimmerite and a 2 cm piece of reddish calcite were selected because they contained rather large areas of the same mineral macroscopically. Under a 254 nm ultraviolet (UV) lamp, both showed intense luminescence, especially calcite had a very strong red colour. The samples were sawed and polished.
Apatite often acts as a host to luminescent ions [18][19][20][21] and for example, manganese and REEs, can replace calcium in the mineral lattice. As the majority of apatite in Siilinjärvi complex is fluorapatite Ca 5 (PO 4 ) 3 F [16], it is assumed that our sample is this phosphate mineral, having a hexagonal structure (P6 3 /m) [22,23]. There are two Ca sites in the apatite's crystal structure: Ca (I) and Ca (II), which differ in both size and site-symmetry [9,18,[22][23][24]. It is presented, how LREEs (light REEs: La-Sm) prefer the Ca (II) site in apatite's structure [25][26][27][28]. However, the REE content in apatite typically corresponds to the whole-rock composition, due to the poor REE selectivity [26,27,29]. Calcite's (CaCO 3 ) structure consists of planar carbonate (CO 3 ) 2groups within the centre of an equilateral oxygen triangle occupied by a carbon ion [18]. Rare earth elements can substitute Ca in calcite lattice and become luminescence centres, e.g. Refs. [4,6,12,[30][31][32][33][34]. Also different divalent  The strongest luminescence peaks from Tb 3+ (marked with orange triangles) can be observed at the laser excitation of 218 nm for apatite (a) and of 226 nm for calcite (b), measured at two different delay and gate times. Unmarked luminescence peaks are caused by Sm 3+ and Eu 3+ .  [18]. Luminescence was measured with a home-built laser-induced timeresolved spectroscopy setup having a tunable laser wavelength system based on Ekspla OPO (optical parametric oscillator). The laser had a pulse duration of 3.3 ns and a repetition rate of 100 Hz. The signal was collected via a plano-convex lens with 45 • geometry to the laser beam. The reflective collimator directed the light to fused silica fiber connected to the spectrograph. To block the laser light from the detector, a longpass filter was used between the lens and reflective collimator. The selected filters had cut-on wavelengths at 266 nm, 400 nm, 450 nm, and 550 nm to match the excitation wavelengths between 210-340 nm and 405-535 nm. Each emission spectrum was accumulated from 400 laser pulses to average the influence of possible pulse-to-pulse laser energy fluctuations. All measurements were carried out at room temperature.
The light was directed to the Czerny-Turner type imaging spectrograph (Acton, SP-150) equipped with a grating having 300 grooves per mm (blazed to 500 nm, resolution >1 nm) and spectra were recorded with ICCD (Instaspec V, Oriel). The spectral range was about 350 nm and the central wavelength was changed by turning the grating. The calibration was carried out with HgAr lamp. A delay generator (model DG 535, Stanford Research System, Inc.) was used to control the time parameters of ICCD. To exclude the broad, strong, and very short-living luminescence emission bands of Ce 3+ and Eu 2+ , which are known to dominate steady-state luminescence spectra [6,35,36], a delay of 50 μs was used in the measurement of excitation-emission maps (EEMs) of calcite and apatite. The luminescence lifetimes (τ) of Ce 3+ and Eu 2+ bands are presented to be < 1 μs in literature, e.g. Ref. [6] and the chosen delay was proven also experimentally to be suitable for exclude these. Spectra were measured with a step of 1 nm (laser wavelength). Gate width was 500 μs to avoid the dominance of the long-living luminescence peaks. The benefit of the time-gated detection is presented on supplementary material ( Fig. A1) with excitation wavelength of 266 nm. To further analyse the observed REE 3+ emissions, certain wavelengths were re-measured with different delay times (10 μs, 50 μs, and 1 ms) with varying gate widths (25 μs, 250 μs, 1 ms, and 5 ms).
To test the predicted optimal excitation wavelength areas, the luminescence signal was directed to a non-gated Czerny-Turner type CCD spectrometer (Thorlabs, model CCS200, grating 600 lines/mm; blazed to 800 nm and FWHM spectral accuracy < 2 nm). The spectral range was 200-1000 nm, and the filters were applied to block the excitation laser light.
Data processing was carried out with MathWorks MATLAB R2015a program. As the laser energy varies in OPO as a function of wavelength, it was compensated for each spectrum by dividing the spectral intensities by the average energy value measured with Ophir (model PE10-SH-V2) laser power meter. All presented spectra have slight smoothing for enhanced visualization.  3. In apatite (a) Sm 3+ luminescence is observed only from substitution at Ca (I) site (dark blue triangles) or also from Ca (II) site (light blue triangles), depending on the excitation wavelength, here 210 nm (black) and 412 nm (violet), respectively. In calcite (b) the strongest Sm 3+ signal is seen with excitation at 210 nm (black), but for other peaks of Sm 3+ , the excitation at 221 nm (violet) is better. Note the scaling of the violet spectra in calcite. Peaks related to Tb 3+ (orange triangles) and Eu 3+ (red triangle) are also marked. All spectra were measured with a delay of 1 ms and a gate of 1 ms.

Table 2
Transitions and luminescence peaks of Sm 3+ observed in apatite at two Ca sites and in calcite.

Laser-induced time-resolved luminescence of apatite and calcite
Time-resolved luminescence spectra from apatite and calcite were measured at a range of 400-700 nm using excitations 210-340 nm and 405-535 nm. The delay time was 50 μs and the gate width 500 μs.
The chosen delay and gate times seem to be suitable for observing several narrow REE 3+ luminescence peaks simultaneously. Shape of spectrum varies as the excitation wavelength changes. Intensities of the luminescence peaks and bands in calcite were lower than in apatite, which is most likely caused by the lower REE content in calcite in the Siilinjärvi [17].
In apatite (Fig. 1a) more REE 3+ peaks can be observed than in calcite (Fig. 1b), where a broad luminescence band around 620 nm is dominating spectra at wide excitation regions. This band is caused by Mn 2+ and is thoroughly studied, e.g. Refs. [8,30,[37][38][39][40][41][42][43]. Also, the broad Mn 2+ band is peaking at 573 nm in apatite (Fig. 1a) note that the change in the emission wavelength is caused by different mineral structures [8]. In literature Mn 2+ band in apatite in this spectral region has been observed earlier [21,44] and Mn 2+ can accommodate both Ca (I) and Ca (II) sites in apatite's structure [4,19,45]. The absolute intensity of Mn 2+ band in the measured excitation wavelength regions has no considerable difference between the apatite and calcite, even though the relative intensities of Mn 2+ and REE 3+ luminescence between the minerals differ ( Fig. 1). Calcite from Siilinjärvi contains typically 0.10-0.32 wt % of MnO, whereas apatite has only 80-300 ppm of Mn [16]. In order to study how REEs contribute to the EEMs in Fig. 1, in the following selected excitation wavelengths are discussed in detail.
Several strong Sm 3+ luminescence peaks can be seen in both minerals. In apatite Sm 3+ can occupy Ca (I) and Ca (II) sites in the mineral structure [4,6,9,12,19,36,61,62]. For example, Sm 3+ luminescence at excitation wavelength of 210 nm originates from the substitution to Ca (I) site (Fig. 3a, black). When excitation is changed to 412 nm (Fig. 3a,   Fig. 4. In apatite (a) Eu 3+ luminescence peaks from Ca (I) and Ca (II) sites can be observed with the same excitation wavelength of 236 nm by changing the delay time. Emission of Eu 3+ in Ca (II) site can be seen with the shorter delay (10 μs, black) and emission of Eu 3+ in Ca (I) site with the longer delay (1 ms, violet). In calcite (b) either the strong background caused by the Mn 2+ band at excitation of 268 nm (black) or emission peaks from other REEs disturb the detection of Eu 3+ luminescence (violet). Both spectra of calcite were measured with a delay of 1 ms and a gate of 1 ms. The peaks of Sm 3+ (blue triangle) and Tb 3+ (orange triangles) are marked in spectra and the peak with a green star in calcite is possibly related to Pr 3+ . violet), also luminescence from Ca (II) site is detected. The band ~700 nm in apatite is a combination of Eu 3+ and Sm 3+ transition 4 G 5/2 → 6 H 11/2 (Fig. 3a, black).
In calcite the strongest Sm 3+ luminescence was observed with excitation at 210 nm, but also a broad Mn 2+ band occurs, nearly hindering the detection of other Sm 3+ peaks than the double peak at 599/604 nm (Fig. 3b, black). At excitation 221 nm, also peaks at 563/568 nm and 650 nm can be detected clearly (Fig. 3b, violet). The occurrence of Sm 3+ luminescence as two double peaks have also been reported for synthetic calcites [30,43,63,64]. Due to the strong luminescence of the overlapping Mn 2+ band, the used visible excitations are not relevant for successful detection of Sm 3+ in the measured calcite. All observed Sm 3+ luminescence peaks are collected in Table 2 with the respective transitions recognized via literature [53,58,[65][66][67][68][69][70][71][72][73].
Analogous to samarium, also luminescence of Eu 3+ has been observed from different Ca sites in apatite, e.g. Refs. [9,19,36,[74][75][76][77][78][79][80][81][82][83][84][85][86]. This is also the case in our study and as an example, spectra measured with excitation 236 nm are presented (Fig. 4a). The appearance of the luminescence from two Ca sites can be controlled by changing the delay time of the measurement. At the shorter delay of 10 μs, luminescence peaks of Eu 3+ in Ca (II) site can be observed at 577 nm, 595 nm, and 623 nm (Fig. 4a, black). With the longer measurement having the delay of 1 ms, the luminescence from Eu 3+ in Ca (I) site can be observed in 579 nm, 591 nm, 618 nm, 655 nm and 700 nm (Fig. 4a, violet). The peak ~620 nm is rather wide regardless of the occupied Ca site, which suggests further fine structure within this peak. It can be noted that we observe peaks at 636 nm and 720 nm in apatite (pink stars in Fig. 4a), always when also the peaks of Eu 3+ from Ca (II) site are seen. Our assumption is, that both are luminescence of Eu 3+ in Ca (II) site, somewhat supported by previous research, where the peak observed at ~636 nm was explained as a vibronic transition of Eu 3+ in Ca (II) site [76,86].
In calcite, the detection of Eu 3+ peaks is challenging due to the broad and strong Mn 2+ band, as they are excited simultaneously. With excitation wavelength 268 nm, luminescence of Eu 3+ is observed at 577 nm, 591 nm, 610 nm, 620 nm, 655 nm, and 702 nm although there is also Mn 2+ band underneath (Fig. 4b, black). It can be remarked, that the transition 5 D 0 → 7 F 2 here is observed as a double peak at 610/620 nm, also reported for doped calcite samples [43,63,87,88]. When luminescence from calcite is measured with excitation 229 nm, the contribution of Mn 2+ is also relatively low and only the Eu 3+ peak at 620 nm is seen along the adjacent Tb 3+ peak (Fig. 4b, violet), whereas other Eu 3+ peaks are not resolved. Table 3 summarizes the observed Eu 3+ luminescence peaks in both minerals and the corresponding transitions were recognized based on e.g. Refs. [49,67,[89][90][91][92].
We assume, that peak at 641 nm in calcite (Fig. 4b, green star) might be related to Pr 3+ , which is observed around this spectral region [61,62,71,[93][94][95]. Also, the suitability of deep UV excitation for Pr 3+ in minerals has been proved very recently [96]. However, in the scope of this study, the occurrence of Pr 3+ in our sample cannot be proven, because the other intensive Pr 3+ peaks in this spectral region are located around 600 nm and most likely be overwhelmed by strong Sm 3+ luminescence and by Mn 2+ in calcite.
In the above, the occurrence of Eu 3+ luminescence was only presented with UV excitation. In apatite, where the Mn 2+ luminescence was not as strong as in calcite, the Eu 3+ peaks can also be efficiently observed using also visible excitation. For example, excitation at ~465 nm is suitable for detecting Eu 3+ in Ca (I) site whereas ~435 nm is good for Eu 3+ in Ca (II) site (Fig. 5). The measurement times are analogous to our previous observations; the luminescence of Eu 3+ from Ca (I) site is strong at the longer delay (1 ms) whereas from Ca (II) it is better observed at the shorter delay time (10 μs). In Fig. 5 those Eu 3+ transitions, which have a strong contribution of nearby Sm 3+ luminescence, are not marked. At excitation 435 nm there are two peaks related to transition 5 D 0 → 7 F 0 at 573 nm and 577 nm (Fig. 5, violet). Therefore, it is assumed that in our sample Eu 3+ is occupying at least three different Ca sites because europium's states 5 D 0 and 7 F 0 are non-degenerated [82]. It can be also remarked that the same peaks at 636 nm and 720 nm (Fig. 5, pink stars), which were seen in Fig. 4a along with the Eu 3+ occurrence in Ca (II) site in apatite's structure are observed here as well.
The two strongest Dy 3+ bands can be seen at ~480 nm and ~575 nm in both apatite and calcite e.g. with excitation 427 nm (Fig. 6). The corresponding transitions are 4 F 9/2 → 6 H 15/2 and 4 F 9/2 → 6 H 13/2 also reported previously for various dysprosium containing samples [69,73,90,[97][98][99][100]. The recognized Dy 3+ peaks are presented in Table 4. The peak at ~480 nm is wide and seems to consist of several peaks located very close together in both minerals. Such fine structure for the Dy 3+ peak has been reported on apatite [61], in doped calcites [43,63] and other materials [7,100]. It seems that in apatite (Fig. 6a) the relative intensities of three peaks at 475 nm, 480 nm, and 484 nm vary when the measurement delay is changed from 10 μs to 1 ms, but such behaviour is not observed in calcite (Fig. 6b). We anticipate that this could be caused by the Dy 3+ occupying different Ca sites in apatite, which would lead to the difference in relative intensities at changed delays, as demonstrated earlier for Eu 3+ .
The Dy 3+ band at ~575 nm is clearly observed with excitation 452 nm for apatite and 454 nm for calcite (Fig. 6, cyan). Also here is the fine structure of closely located peaks and similar patterns have been presented for Dy 3+ at ~575 nm by others, e.g. Refs. [7,61,77]. With the excitation wavelength of 427 nm, the Dy 3+~5 75 nm luminescence is strongly contributed by other luminescence species. In apatite adjacent Eu 3+ (II) peak is dominating at the shorter measurement delay 10 μs (Fig. 6a, black). If the delay is changed to 1 ms, Dy 3+ at 578 nm in apatite nm can be better resolved, but there is a background from the Mn 2+ , Sm 3+ (I), and Sm 3+ (II) (Fig. 6a, violet). In calcite, the Dy 3+ peak at 579 nm can be observed at the shorter delay only, as at the longer  The right selection of the measurement time parameters with excitation 427 nm has importance in discovering the Dy 3+ band at ~575 nm.
In apatite, at the measurement of delay 10 μs, Dy 3+ has a strong contribution of Eu 3+ (II) peak (Fig. 6a, black). Measurement performed with the longer delay (1 ms), the Dy 3+ at 578 nm in apatite nm can be better resolved (Fig. 6a, violet), but a strong contribution of Mn 2+ , as well as Sm 3+ (I) and Sm 3+ (II), can be seen. In calcite (Fig. 6b), the detection of Dy 3+ peak at 579 nm can be observed at the shorter delay   only, as at the longer delays the luminescence is overwhelmed by intensive Mn 2+ band.

Selection of optimized excitation wavelengths for apatite and calcite
The appearance of the luminescence spectrum depends significantly on the excitation wavelength. This is efficiently visualized as animations for excitation in the UV (video B1) and the visible (video C1) range. Normalization offers better comparison of the changed spectral shapes, thus the intensity values of apatite and calcite are normalized to the maximum of respective spectrum. The delay time in the measurements was 50 μs and the gate width 500 μs. For further applications in the luminescence analysis of natural minerals, it is useful to know which excitation wavelengths can be used to separate specific REEs or, on the other hand, to observe multiple REE 3+ peaks simultaneously. Therefore, the laser-induced time-resolved luminescence spectra measured from apatite and calcite (Fig. 1) are illustrated as 2D contour presentations of EEMs (Figs. 8-10). The excitation regions (UV, Vis) are shown separately and for clarity, only the locations of the most prominent REE 3+ peaks are marked as dashed lines. Deviating from Fig. 1, Contour EEM of apatite in the visible excitation region (Fig. 8), shows also spectral area of 750-900 nm presented in Fig. 7.
Contour-EEM of apatite in the visible excitation region shows luminescence from several REEs and no significant contribution of the Mn 2+ band is observed (Fig. 8). The highest intensities for the Dy 3+ peak at 578 nm are obtained with excitations at 427 nm and 452 nm. All the excitation wavelength regions, where Dy 3+ is detected clearly, here 424-429 nm and 444-458 nm, are presented as grey boxes on the left side of the map, and stars show the most suitable excitations in our study. The same notation is used throughout all EEMs. Eu 3+ can be excited from apatite using several spectral regions. With the delay of 50 μs luminescence from both mineral sites is seen and to separate Eu 3+ peaks, those related to Ca (II) site are marked with pink. The good excitation wavelength for Eu 3+ in apatite's Ca (I) site is 465 nm and for Eu 3+ in Ca (II) sites either 435 nm, 473 nm, or 527 nm Sm 3+ peaks have high intensities in apatite at wide excitation range. At Ca (I) site Sm 3+ is well observed at excitations 405 nm and 473 nm. If also Sm 3+ at Ca (II) site should be measured, the highest intensities were obtained with 414 nm excitation, marked with a pink star. The other REEs do not disturb the detection of Nd 3+ peaks and in this research, the excitation 526 nm was the best.
In the apatite's EEM in UV excitation region (Fig. 9), background from some shorter living luminescence is here seen around the Tb 3+ peaks (see also Fig. 1a), but they are clearly resolved and the highest intensities are obtained at 218 nm. The broad and intense Mn 2+ band ~575-650 nm disturbs the detection of the other REEs, especially at excitations from 275 to 320 nm. The strongest intensities for Dy 3+ peaks were observed at excitations of 287 nm and 326 nm. For Eu 3+ the most suitable excitation wavelength at UV was 236 nm and the Sm 3+ peak offered a strong signal when 210 nm was used for the excitation.  In calcite, the visible excitations strongly induce luminescence of Mn 2+ and it readily dominates the spectra, as demonstrated in Fig. 1b. As an exception, the excitation 454 nm is applicable to Dy 3+ and excitation 405 nm for the Sm 3+ , but both have a notable background. For this reason, the contour-EEM for calcite is only presented in the UV excitation range (Fig. 10). For Tb 3+ luminescence, the best excitation is at 226 nm. Because of Mn 2+ , for the other REE 3+ peaks the optimal excitation ranges (grey boxes) in calcite are not as wide as was shown for  apatite. The excitation of 287 nm is applicable to Dy 3+ peaks and for Eu 3+ and Sm 3+ the excitations 268 nm and 221 nm, respectively, were found good.
The results offer possibility to locate those excitation wavelengths, where the emission signal of one or several REE 3+ peaks are intensive and the disturbance of Mn 2+ luminescence is minimal. The applicability was tested by measuring luminescence with a non-gated spectrometer with a few selected excitation wavelengths from the same samples. Without time-gating the long-living Mn 2+ luminescence could readily dominate the spectrum even at low concentrations of Mn. However, the relative intensities of the REE peaks will not be comparable due to the varying lifetimes of the luminescence activators.
Firstly, the optimal excitations of Tb 3+ for calcite and apatite were measured (Fig. 11). At these deep UV excitation wavelengths, both spectra show the luminescence peaks of Tb 3+ and Sm 3+ clearly (orange and blue stars), despite the background caused by the luminescence of Eu 2+ , Ce 3+ , and Mn 2+ . In apatite (Fig. 11, black) also several Eu 3+ related peaks (red star) and the Nd 3+ band are seen.
Secondly, the occurrence of Sm 3+ in both Ca (I) and Ca (II) sites in apatite's structure was measured. Luminescence of Sm 3+ in the Ca (I) site can be detected in many excitation areas and here UV excitation 210 nm was selected (Fig. 12, black). Several peaks of Sm 3+ in Ca (I) site Fig. 11. Non-gated laser-induced luminescence measurements with the best excitations for Tb 3+ for apatite (λ ex 218 nm, black) and calcite (λ ex 226 nm, violet). The Tb 3+ peaks are marked with orange stars, Sm 3+ as blue stars and Eu 3+ as a red star. Note scaling of the violet spectrum.   (blue stars) can be observed in addition to the luminescence peaks of Eu 3+ occupying both sites. If also Sm 3+ in Ca (II) is needed, the excitation at 414 nm (Fig. 12, violet) shows intensive luminescence peaks (light blue stars).
Eu 3+ was also observed in the time-resolved measurements to occupy Ca (I) and Ca (II) sites in apatite, but optimal delay times were varying. In the non-gated measurement, the influence of the excitation wavelength seems to be decisive (Fig. 13). At excitation 236 nm (Fig. 13, black) the luminescence of Eu 3+ is seen from Ca (I) and Ca (II) sites (red star). For comparison, excitation wavelength 435 nm shows Eu 3+ peaks from Ca (II) site (Fig. 13, violet) and excitation 465 nm from Ca (I) site (Fig. 13, cyan). It can be remarked that the peaks at 636 nm and 720 nm (pink stars) are observed also here when the luminescence of Eu 3+ from Ca (II) site is induced. This supports our assumption, that these peaks are related to Eu 3+ and especially to Ca (II) site. Sm 3+ peaks are seen in all spectra of Fig. 13.
For non-gated measurements of calcite, only UV excitations were used, as the luminescence of Mn 2+ was strong already in the timeresolved study. Even with the selected UV excitations 222 nm and 262 nm, the background caused by Mn 2+ in calcite is notable (Fig. 14). With excitation of 222 nm strong peaks of Sm 3+ and Tb 3+ are seen (Fig. 14, black). For non-gated measurement of Eu 3+ in calcite, the excitation 262 nm seems to be the best (Fig. 14, violet).

Conclusion
The aim of the research was to study how REEs could be efficiently detected in natural apatite and calcite by optimizing the laser excitation wavelength. Luminescence from both minerals was systematically measured using excitations of 210-340 nm and 405-535 nm with tunable OPO and a time-resolved ICCD. The complex luminescence spectra of several REEs were studied and multiple suitable excitation wavelengths for detecting the REE 3+ luminescence peaks from both apatite and calcite were presented. A broad band of Mn 2+ was seen as a strong background, especially in calcite. For apatite, the luminescence of Eu 3+ and Sm 3+ occupying different Ca sites was observed with specific excitation regions, which would offer further information about the crystal structure. As a main result, the laser excitation region in deep UV around ~220-230 nm was discovered very promising. Within this range, the intensive luminescence of Tb 3+ and Sm 3+ was observed from both minerals and in the case of apatite peaks of Nd 3+ and Eu 3+ as occupying Ca (I) and Ca (II) sites were detected. In addition, the luminescence of Mn 2+ is not strong with these excitations. Applicability of the optimized excitation wavelengths found here was demonstrated with a non-gated detector, as it would be more feasible for in-situ analysis of minerals.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments
This study was partly supported by the Academy of Finland (Grant 281955 to S.K. and Grant 282240 to H.H.). S.R. wishes to acknowledge K.H. Renlund's Foundation for the grant which offered the possibility to finalize this work.
The apatite sample for the research was chosen from the samples kindly offered by Yara Suomi Oy and Pasi Heino and Aleksi Salo are appreciated for the co-operation. The Mn-bearing calcite sample from Siilinjärvi was from the collection offered by Geological Survey of Finland: Jari Nenonen and Satu Hietala are thanked for it.