1. Introduction

Persistent luminescence can be defined as emission obtained after the removal of an excitation source [1] which can be visible light, UV radiation, electron beam, plasma beam, X-rays, or even γ-rays. This much is agreed upon at present. Generally, though not universally, it has also been accepted that the excitation (irradiation) energy is stored in intrinsic traps or in those intentionally introduced or in both [2]. The nature and origin of the traps are contested [3], they may be assumed to result from the preparation methods [4] – usually reducing ones – or from the aliovalent co-doping of the material [5] or from unidentified lattice defects and distortions [6]. Too often, the more detailed description and effect of the defects are just omitted although it is evident that the defects play an important role both as an emitting center and the defect generator, as is the case with the persistent luminescence of “non-doped” zirconia – in fact a small titanium impurity partly in the trivalent form is the origin [7].

The general disagreement extends – or, it should be said: starts – from the very name of the phenomenon. In the literature (e.g. [8]), there is practically an endless number of permutations of the adjectives as long, persistent, long-duration and long-lasting as well as nouns as afterglow, fluorescence, phosphorescence and luminescence – in pairs of two or in bigger clusters – the record being perhaps “Long-Persistent Phosphorescence”. It is not clear if this name was given intently but it is doubted when knowing more closely the father of this name, prof. William (Bill) Yen. In practice, the name is afterglow – and this it should still be in the future if the duration of the persistent luminescence is short, up to a limit when it is still a nuisance but not yet long enough to be exploited to something useful. Scientifically, persistent luminescence is a special case of thermally stimulated luminescence (TSL) at a given temperature. Because of this additional restriction, this name should be avoided, though, since it may easily be confused with thermoluminescence [9] which is a broader phenomenon.

Persistent luminescence is definitely not phosphorescence since the relatively long life time of phosphorescence is due to the forbidden transitions within the luminescence center [10]. However, in persistent luminescence materials, the excitation energy is stored in defects – the nature of which is usually totally different from the luminescence center. As an example, rather intense persistent luminescence – though not good enough for practical applications – can be obtained from Ce³⁺ doped aluminates [11]. The origin of the luminescence is the 5d¹(²D) → 4f¹(²F_J; J: 5/2 and 7/2) transitions which are both parity and spin allowed (ΔS = 0) and cannot be named as phosphorescence as the short (order of some tens of ns) conventional lifetime of Ce³⁺ implies. The long lifetime is not as such a reason to name persistent luminescence as phosphorescence. This is, however, only one consequence whereto the ignorance about the real mechanisms of persistent luminescence can lead. Finally, the classification of some authors are trying to introduce – based once again – on the lifetime of persistent luminescence – leads to nothing since the limits are not clear and change when a new top score in the duration is achieved. The reader might wonder the utility of e.g. the following classification: very short persistent, short persistent, persistent, long persistent, and super long persistent phosphorescence [12].

It has been already a few generations ago when – at least some kind of – agreement was achieved to use luminescence instead of – usually theoretically inaccurate – fluorescence and phosphorescence [10]. Maybe it is the time to make the same with persistent luminescence? At least the some 50+ participants of the first meeting on persistent luminescence were overwhelmingly for the use of a single term [13]. Maybe the common sense finally gains over the temptation of using too much imagination.

The imagination of the authors has thus achieved the maximum value when the intensity of persistent luminescence is concerned. Nevertheless, there exist clear rules concerning the intensity (and its evolution vs. time): the luminous intensity (luminance) should exceed the value of 0.32 mcdm⁻² when measured at a defined (one steradian (sr), i.e. 65.54 degrees) viewer angle. The threshold value corresponds to approx. 100 times the lower limit of the perception of the human eye in the scotopic (dark adapted) vision range [14]. This value – combined with the requirement that the persistent luminescence should last at least a few hours to be of any practical use – limits significantly the number of potential phosphors. However, only very infrequently this rule is remembered and even less frequently the measurements have been carried out to validate it with the phosphor studied.

The history of persistent luminescence extends to some 1000 years ago making persistent luminescence probably the first kind of documented luminescence [15]. Unfortunately, the materials used in the Chinese painting presented to the Japanese emperor are not known. In the beginning of the 17th century, the persistent luminescence certainly became the best documented kind of luminescence because of the discovery – or better manufacture – of the famous Bologna Stone [16]. This was apparently BaS made by the reduction of the mineral Barite containing a selection of impurities to give a good persistent luminescence material – though not yet good enough for practical applications. The Bologna Stone aroused the interest of the great Galileo though he never published anything about it – maybe better for him since, at that time, he was already in house arrest proclaimed by the Roman Catholic pope. He definitely needed no extra trouble by using too much imagination since the Bologna Stone was suggested to have some celestial properties and to be closely related to alchemy. The final explanation of the origin of persistent luminescence from the Bologna Stone had to wait until the present day, however [17]. It was revealed that the Bologna Stone is indeed BaS and not, as incorrectly though frequently believed, BaSO₄ (barite). The persistent luminescence originates from the monovalent copper impurity. The mechanism suggested [17] is – again – similar to the one for Eu²⁺ to be discussed in detail below.

The first commercial applications (self-lit watch dials etc.) came with ZnS:Cu about 100 years ago. Similar sulphide materials – CaS:Eu²⁺, (Ca,Sr)S:Bi³⁺ etc. appeared later – the famous Lenard’s phosphors [18] – but their utility is hampered by their extreme sensitivity to moisture. The final boom in the research and applications came in mid 1990s with the introduction of the efficient Eu²⁺,R³⁺ co-doped aluminates – CaAl₂O₄:Eu²⁺,Nd³⁺ [19] and SrAl₂O₄:Eu²⁺,Dy³⁺ [20] emitting in the blue and green, respectively. The disilicate, Sr₂MgSi₂O₇:Eu²⁺,Dy³⁺ [21] and another aluminate (Sr₄Al₁₄O₂₅:Eu²⁺,Dy³⁺) [22] with improved moisture stability followed in the early 2000s. All these phosphors extend the persistent luminescence beyond the limit of 24+ hours enabling the use of natural excitation by sun light – if this is at all possible, however, since the soft UVA-blue light excitation characteristics of the phosphors may vary significantly [23]. However, only a few phosphors – in addition to those mentioned previous only Y₂O₂S:Eu³⁺,Mg²⁺,Ti – have gained commercial success – and the red emitting oxysulfide only because of the lack of really good persistent phosphors emitting in red [23].

The use of the persistent luminescence phosphor finally determines the properties, the following three being the most important: excitation, colour, and duration. The first one is usually widely ignored though very recently the excitation spectroscopy of persistent luminescence phosphors has proved invaluable to fill this gap [25]. In principle, at present it is possible to have persistent luminescence phosphors to cover the entire visible spectral range (Fig. 1 ) though the weakness of luminescence and the relative insensitivity of the human eye in red is still a problem. This is discussed in some detail in the end of this paper.

 

Fig. 1 A three-color persistent luminescence panel based on the Sr₂MgSi₂O₇:Eu²⁺,Dy³⁺ (blue), SrAl₂O₄:Eu²⁺,Dy³⁺ (green) and Y₂O₂S:Eu³⁺,Mg²⁺,Ti (red) phosphors (modified from [24]).

Download Full Size | PDF

The future trends involved in the persistent luminescence research are not difficult to predict – in the short term, of course. First, in order to design new phosphor materials – or at least to tailor the present ones – the exact knowledge of the mechanisms behind the phenomenon is required. The present development is based on the trial and error methods and on the use of experimental techniques which are either very difficult or nearly impossible to carry out or require sophisticated apparatus with enormous costs involved as the synchrotron radiation sources. Secondly, the development of new phosphors covering the entire visible spectrum – especially in the red and NIR ranges – are acutely needed e.g. for biomedical applications [26]. Thirdly, the new methods of manufacture of these phosphors need to be studied carefully since there are indications that e.g. the nanophophors may have different defect structure from the powders and single crystals. This paper tries to address to the two former trends – both stating the status quo and the possible future trends – not forgetting the milestones of the human imagination in the past.

2. Earlier persistent luminescence mechanisms

Not much was known about the mechanisms of persistent luminescence prior to the introduction of the aluminate materials in the mid 1990s [19,20]. This is no wonder despite the ZnS:Cu materials had been already at that time commercially available for nearly a hundred years. In fact, the mechanism of the conventional luminescence from copper is still contested [10]. The luminescence center as well as the co-dopant in the Eu²⁺,R³⁺ co-doped aluminates (mainly SrAl₂O₄:Eu²⁺,Dy³⁺) offered a seemingly – or deceivingly – simple system: the electron from the valence band is trapped to Eu²⁺ converting it to Eu⁺ (?) whilst Dy³⁺ traps the hole becoming Dy⁴⁺. It is implied that both the electrons and holes act as charge carriers. The schematic mechanism follows this reasoning (Fig. 2 [27]).

 

Fig. 2 One of the earliest proposed persistent luminescence mechanism for SrAl₂O₄:Eu²⁺,Dy³⁺. Note the unrealistic monovalent Eu⁺ species (redrawn from [27]).

Download Full Size | PDF

Old habits die hard since the latest paper suggesting the presence of Eu⁺ was published in 2010 [28]. It was thus not really too fast, unfortunately, when it has been realized that the Eu⁺ species is very improbable and would require huge amounts of energy to be formed. This energy is just not available in ambient conditions. More rational researchers soon suggested a reverse process, i.e. the oxidation of Eu²⁺ and reduction of Dy³⁺ [29]. The charge carriers are again the same. Although being much more realistic, in general, the problem with the “redox” mechanisms is that several materials – including both Sr₂MgSi₂O₇:Eu²⁺,Dy³⁺ (blue) and SrAl₂O₄:Eu²⁺,Dy³⁺ (green emission) – show rather intense persistent luminescence without the co-dopant, albeit weaker and shorter than with the co-doping. Moreover, except for the Eu²⁺ → Eu³⁺ oxidation [30,31], no change in the co-dopants’ valence has been observed experimentally by EPR, XPS, or XANES [30–32] techniques. The first lesson is that maybe the existence of a luminescence center – co-dopant redox couple is not at all necessary. Sometimes the luminescence center and co-dopant are the same, as with CdSiO₃:Tb³⁺ [33]. The role of the co-dopant may thus be completely different. From this idea there was only a short way to recognize that the charge carriers may be trapped to defects created by the co-dopants. The existence of defects is more or less obvious as a result of charge compensation since, in most cases, the co-dopant has the valence incompatible with the host cation. Instead, it should be observed that the luminescence center has usually the same valence as the host cation it replaces. This is evident since the luminescence center must 1) yield efficient luminescence by its very nature (as Eu²⁺, Ce³⁺, Pr³⁺, Eu³⁺, and Tb³⁺ show) and 2) this luminescence should not be quenched by uncontrolled lattice defects (as color centers).

It is interesting to note, in retrospective of course, that all the efficient luminescence centers involved in different persistent luminescence materials form chemically (and physically, too) meaningful redox couples: Eu²⁺, Ce³⁺, Pr³⁺, Eu³⁺, Tb³⁺, Mn²⁺, Ti³⁺, Cr³⁺, Bi³⁺, Pb²⁺, and Cu⁺ with Eu³⁺, Ce^IV, Pr^IV, Eu²⁺, Tb^IV, Mn³⁺,^{TiIV, Cr2+/CrIV, BiV, PbIV, and Cu2+, respectively. The second lesson to be learnt is that efficient persistent luminescence seems to involve mainly the oxidation of the luminescence center, much less frequently – but not completely excluded – the reduction. This means that the charge carriers are electrons and the process should include the conduction band where the mobility of the electrons is far higher than that of holes in the valence band. Thus the electron finding a trap below the conduction band is more probable. The third lesson is that the oxidation/reduction of the luminescence center and the co-dopant is perhaps only a virtual one; instead of the actual oxidized and reduced species, maybe Rn+-h+ and Rn+-e- pairs are formed, respectively. The actual oxidized species have been observed only for the Eu2+-Eu3+ pair [30,31]. The pair formation may be a conceivable possibility since in a M2+ lattice the small and highly charged R3+ ion with high charge density attracts easily electrons.}

As a conclusion concerning the early mechanisms proposed for persistent luminescence, it is evident that at least a basic knowledge of the general chemical behavior of the species in the materials studied could have helped to reject the most improbable mechanisms. Restricting the imagination – so useful in some cases – could have done the same, as well. As an extreme case of the use of imagination and to those who long for exact numbers, the following mechanism was suggested [34] for the Eu²⁺ and Dy³⁺ doped SrAl_2−xB_xO₄ and Sr₄Al_14−xB_xO₂₅ (Fig. 3 ). However, the mechanism – apart from the participation of the borate – resembles the presently accepted mechanisms. If the authors had known – as shown in the next section – the correct position of the 4f⁷(⁸S_7/2) ground level of Eu²⁺ in the host band gap and omitted the complications…

 

Fig. 3 An extreme example of the use of imagination in the construction of persistent luminescence mechanism: the Eu²⁺ and Dy³⁺ doped SrAl_2−xB_xO₄ and Sr₄Al_14−xB_xO₂₅ materials (redrawn from [34]).

Download Full Size | PDF

3. Advanced persistent luminescence mechanisms

The early persistent luminescence mechanisms proposed were usually nothing but qualitative sketches on what could be if there were some experimental data available to support them. The quantitative understanding of the said materials and processes therein and the design of new, even better materials could hardly be done with these mechanisms. However, it must be said that all the elements of the present more advanced mechanisms were there but not in the same one or two mechanisms and there were also things which were evidently not correct at all. For example, the mechanism of persistent luminescence in Eu²⁺ and R³⁺ doped SrAl₂O₄ [35] considered [8] as one of the fundamental ones is purely qualitative and contains highly suspicious processes as charge transfer emission of Eu³⁺ at 445 nm. In addition, the authors have considered only direct electron transfer between traps and the 5d levels of Eu²⁺. The electron transfer via the conduction band was excluded based on inconsistent photoconductivity measurements. Taken these deficiencies, this work cannot be considered as a model.

Much of the progress in understanding the persistent luminescence mechanisms owes to the work on combining the energy level systems of the host, the luminescence centers and the co-dopants [36]. Basically only now it was made possible to predict at least the following: 1) the nature – an electron or hole transfer – of the charging process of persistent luminescence, 2) identification of the most probable centers (dopants) for persistent luminescence, 3) the role of (rare earth) co-dopants, 4) to predict the energy sufficient to charge persistent luminescence (i.e. persistent luminescence excitation). Despite the model probably failed to explain the role of the co-dopants, i.e. the capture of an electron in the conduction band by an R³⁺ ion to form R²⁺, the energy level schemes were a major breakthrough. It allows, among other things, to design an efficient persistent luminescence phosphor by a careful choice of the band gap energy and the luminescence center, i.e. extending the persistent luminescence from predominantly Eu²⁺ doped materials to e.g. Tb³⁺ doped ones [37,38]. The delicate interplay between several energy level systems can now be controlled though some experimental difficulties still exist e.g. in the exact determination of the band gap energy in VUV by synchrotron radiation sources (e.g. [39]).

The ground energy level positions (Fig. 4 ) for the R²⁺ and R³⁺ ions in Sr₂MgSi₂O₇ – as in most wide band gap materials – indicate that the best candidates for the persistent luminescence centers are (in this order) Eu²⁺, Ce³⁺, Tb³⁺ and Pr³⁺. First of all, all these ions are known to be strong emitters with weak tendency to lose the excitation energy in cross-relaxation or multiphonon de-excitation processes. As for persistent luminescence, most present day applications prefer the excitation (daylight or artificial lighting) with blue light or, if not possible, with UV radiation having energy as low as possible. The Eu²⁺, Ce³⁺, Tb³⁺ and Pr³⁺ have the lowest 5d energies known (Fig. 4) in the rare earth series ensuring the low excitation energies. Further, the overlap between the excited 5d levels and the host’s CB is required and achieved by these ions for efficient electron transfer through CB to traps.

 

Fig. 4 The experimental R²⁺ and R³⁺ (R: La-Lu) 4fⁿ and 4f^n-15d¹ ground level energy positions in Sr₂MgSi₂O₇ showing the possible electron (A) and hole (B) trapping mechanisms [42].

Download Full Size | PDF

The proximity of the host’s conduction band (for electron trapping materials, of course) becomes then important as will be discussed in detail with the Eu²⁺ persistent luminescence mechanism. Though the excitation of the trivalent ions (Ce³⁺, Tb³⁺ and Pr³⁺) usually requires the use of UV radiation, Eu²⁺ can be excited in many hosts by the blue light, as well. The lowering of the band gap energy to e.g. 5-6 eV, i.e. changing the host to e.g. Lu₂O₃ or CdSiO₃, would enable the persistent luminescence of Tb³⁺ as was observed experimentally [40] and [41], respectively. The persistent luminescence of the “typical” R³⁺ ions (e.g. Er³⁺, Ho³⁺, Nd³⁺) can still be observed only in very specific cases.

As a conclusion concerning the advanced mechanisms for persistent luminescence, the experimental techniques and the construction of the host independent evolution of the energy levels for both the R²⁺ and R³⁺ ions in any host has enabled the prediction of the doping schemes for efficient persistent luminescence. A minimum amount of experimental work is still required to establish at least one energy level value relating at least one R²⁺ or R³⁺ ion to the host band structure. In addition to the energetics in the system, the kinetics may play an important role to achieve efficient persistent luminescence. This knowledge is, however, easily available – at least for Eu²⁺, Ce³⁺, and Pr³⁺ though excluding Tb³⁺: the lifetimes for the 4f^n-15d¹ levels are ca. 1 µs, 50 ns and 1 µs, respectively, and for Tb³⁺ a value of 1 µs can be used (the exact value is not known, however). The short lifetime of Ce³⁺ means severe competition between emission and the transfer of electrons to the host’s conduction band (and to subsequent trapping). Despite the very favourable energetics, the number of potential persistent luminescence materials amongst the Ce³⁺ doped compounds is serendipitously low. The improvements to the present models require the better description of both the nature and the energy level structure of the defects. Thermoluminescence measurements are now a commonplace practice but the theoretical calculations using the DFT methods have given encouraging results [42], as well.

As a result of the sometimes tedious progress, the mechanism of persistent luminescence of the Eu²⁺ doped materials can now be considered to have reached the maturity – perhaps except for the exact nature of the defects whereto the excitation energy is stored. As an example of the “model” mechanism for persistent luminescence, that for Sr₃SiO₅:Eu²⁺,Nd³⁺ is described here (Fig. 5 ). This model has been used to construct the corresponding mechanisms for the materials based on the persistent luminescence of Tb³⁺ [40] and even Ti³⁺ [43], Cr³⁺ [44] and Cu⁺ [17], the latter three revealing the utility of the model far beyond the rare earths.

 

Fig. 5 The model mechanism of persistent luminescence for the Eu²⁺ doped materials: Sr₃SiO₅:Eu²⁺,Nd³⁺ [45].

Download Full Size | PDF

The Eu²⁺ ion is thus the basic choice, though versatile and unsurpassed in efficiency, as a persistent luminescence center. The band gap value (for Sr₃SiO₅:Eu²⁺: 6.0 eV) is easily found from the synchrotron (or other VUV) excitation spectrum of Eu²⁺, as well as is the charge (electron) transfer (e^-[O(2p)] + Eu³⁺ → Eu²⁺) transition energy (3.2. eV) from the excitation spectrum of Eu³⁺, used to estimate the Eu^{2+ 8}S_7/2 ground state position. The excitation and emission energies can be obtained from the elementary photoluminescence studies. Finally, the trap energies are acquired from the straightforward deconvolution of the TL glow curves.

The resulting mechanism involves, as the charging phase, the excitation of an electron from the 4f⁷(⁸S_7/2) ground state to the excited 4f⁶5d¹(²D) states of Eu²⁺. Due to the proximity of the host’s conduction band (the 4f⁶5d¹ levels definitely overlap, though only partially, with CB), some electrons can easily escape from the excited 4f⁶5d¹ levels, directly or aided by thermal energy (kT), to the conduction band of Sr₃SiO₅. The relatively long lifetime of the 4f⁶5d¹(²D) → 4f⁷(⁸S_7/2) emission of Eu²⁺ (ca. 1 µs) facilitates this charge transfer. The electrons then move further quite freely in the conduction band until they meet a defect and are trapped. The reverse, decharging, process is initiated by the absorption of thermal energy allowing the trapped electrons to escape back to the conduction band and populate the 4f⁶5d¹(²D) levels of Eu²⁺ (partially overlapping the CB of Sr₃SiO₅). The non-radiative (within the 4f⁶5d¹ states) and finally the radiative relaxation of Eu²⁺ (from the lowest 4f⁶5d¹ state) takes care of producing the persistent luminescence. Alternatively, retrapping of the electrons can occur in every step of the decharging process still lengthening the persistent luminescence. Although the mechanism is convincing, it is not entirely quantitative since the energies of the processes are only estimates because the values are based on data consisting of broad band emission, excitation and absorption. The estimated accuracy is thus only of the order of 0.5 eV.

4. Red emitting persistent luminescence materials

The difficulties to obtain good persistent luminescence materials emitting in red is due to both the lack of stable red emitting phosphors and the insensitivity of the human visual perception (Fig. 6 ). The red emission should be very useful especially for biomedical applications [26,46], as well.

 

Fig. 6 The spectral response of the human eye in bright (photopic) and dark (scotopic) illumination [47].

Download Full Size | PDF

It is evident that – especially in the dark and just after entering into the low level illumination conditions [48] – the perception of red requires a very strong source. Unfortunately, the choices are very few. The standard solution would be to use Eu²⁺ doped materials but the extension of the Eu²⁺ luminescence to red requires an extremely strong crystal field or a shift of the whole 4f⁶5d¹ electron configuration to exceptionally low energies. From the electrostatic point of view, the former seems more or less impossible whilst the latter can be achieved in such sulphide lattices as CaS and SrS. The disadvantage of these hosts is their extremely hygroscopic nature. The excitation can, however, be at relatively low energy, even in the visible, so these phosphors should be preferred if the hygroscopy is not a problem.

The other possibilities include the doping with Eu³⁺ and, indeed, the strong luminescence of the conventional Eu³⁺ doped phosphors as rare earth oxides and oxysulfides can be used at ca. 610 – 625 nm. The problems arise from the high energy required to reach good excitation (to the charge transfer band at 200 – 350 nm) and the difficulty of creating necessary defects by co-doping. When the energy level scheme of Eu³⁺ (Fig. 4) is considered, the ground ⁷F₀ level is well in the valence band (or even lower) and thus the conventional persistent luminescence mechanism is not operational. In addition, the chemistry of the Eu³⁺/Eu²⁺ system does not allow for the Eu^IV species. It has recently been suggested [49] that the persistent luminescence of Y₂O₂S:Eu³⁺,Ti,Mg²⁺ is based on the hole trapping in charge compensation defects close to the top of the valence band. The efficient co-doping seems to include a combination of Mg²⁺ and Ti^IV which both could induce hole traps, e.g. the single negative substitutional defects ($\text{M}{g}_{\text{Y}}^{\text{'}}$) and triply negative yttrium vacancies (${\text{V}}_{\text{Y}}^{\text{'}\text{'}\text{'}}$) though the probability of the latter should be low mainly due to the stable YO⁺ complex cation sublattice in Y₂O₂S. Nevertheless, persistent luminescence times of 5+ h have been obtained [50].

Still other possibilities to obtain red persistent luminescence include the use of the “persistent energy transfer” between Eu²⁺ (absorbing and storing) and Mn²⁺ (emission), as in CaMgSi₂O₆:Eu²⁺,Mn²⁺,Pr³⁺ [46]. The exact mechanisms seem to be largely unknown since, in addition to the sensitizer and luminescence, a third trap forming ion is usually needed. The same holds with the Cr³⁺ doped materials [44] though their efficiency seems to be even less than that of the Mn²⁺ doped systems.

5. Conclusions

The modelling of the processes occurring in persistent luminescence has achieved a stage where the main features are well known but the details concerning the traps and their depths are still somewhat an open question. As a small but interesting and intriguing detail probably worth of a study is the evidently unbroken Sr host cation and Dy³⁺ co-dopant combination yielding the best performance irrespective of the host lattice, i.e. in SrAl₂O₄, Sr₂MgSi₂O₇, Sr₄Al₁₄O₂₅, and even Sr₃SiO₅. The commercial market seems to be saturated with only a few persistent luminescence materials. There is, however, an ever-growing demand for red emitting materials – especially in nanocrystalline form for biomedical applications. The Eu²⁺,Mn²⁺,R³⁺ co-doped materials still offer a challenge to explain in a satisfactory manner the “persistent energy transfer” between Eu²⁺ and Mn²⁺. The effect of the R³⁺ co-doping is of interest, too, since it may affect the properties of the assumed storage (Eu²⁺) and/or the emission (Mn²⁺).