Mn4 + Ions for Solid State Lighting

BRIK Mikhail G MA Chong-gengSRIVASTAVA Alok M PIASECKI Michal

(1. College of Sciences & CQUPT-BUL Innovation Institute,Chongqing University of Posts and Telecommunications, Chongqing 400065, China；

2. Institute of Physics, University of Tartu, Tartu 50411, Estonia；

3. Institute of Physics, Jan Długosz University, PL-42200 Częstochowa, Poland；

4. Consultant to GE Current, A Daintree Company, Ohio 44110, USA；

5. Department of Solid State Physics, Eastern European National University, Lutsk 43025, Ukraine)

Abstract： Phosphors activated with Mn4+ ion are gaining prominence in the field of solid-state lighting as generators of red photon. In the present review,we focus on several important points that are fundamentally important to produce a commercially useful phosphor. This includes an understanding of the Mn4+ energy levels in the free state and in the crystal fields and of the host dependent variations in the Mn4+ emission wavelength. Additionally, we formulate several practical recommendations on how to tune the emission wavelength and emission intensity of Mn4+-doped phosphors. The main spectroscopic parameters of the Mn4+ ion in more than 100 phosphor materials are collected and discussed.

Key words： Mn4+ ions； red phosphors； white LED

1 Introduction

1.1 Review Outline

The phosphor converted LEDs are gaining commercial importance in general and specialty lighting because of their higher efficiency and longer lifetime relative to fluorescent lighting. Active research in this field is being performed in many industrial and academic research groups worldwide[1-14]and references therein. Numerous phosphor materials based on different crystalline solids have been developed so far. One common feature of all these phosphors is that their performance relies on additionally introduced impurities(dopants), which after absorption of a properly chosen excitation radiation can emit light in a certain spectral range. The transition metal and rare-earth ions are those widely used impurities.Their unfilled d- and f-electron shells are characterized by a large number of electronic states, which can be split by crystal field[15], thus giving rise to a variety of the absorption and emission transitions in different spectra ranges.In the present review, we summarize the most important information related to the Mn4+ions(their electronic structure, energy levels, crystal field effects) and crystalline solids doped with these ions for applications as the red phosphors for white LEDs. We start with the properties of the 3d3electron configuration, introduce the main spectroscopic parameters that are needed to describe the optical spectra of these materials, give a list of more than 100 phosphor materials doped with the Mn4+ions with a description of the overall trend that related the emission spectra to the nephelauxetic effect. Very important questions of tunability of the Mn4+emission peak position and emission intensity are also addressed. Finally, we explain several misconceptions about the Mn4+spectra in crystalline solids.

1.2 Basic Properties of Considered d3 Electron Configuration

In this review paper, we focus our attention on some representatives of the transition metal group with unfilled 3d electron shell. Comprehensive reviews of crystal field properties of the entire series of 3dnions series were published earlier[16]. In this paper, attention is focused on ions with 3d3electron configuration. These ions are V2+, Cr3+, Mn4+,Fe5+, where Cr3+and Mn4+are widely used for the solid-state lighting due to their energy level structures that is optimum for the required excitation and emission wavelengths.

Deep understanding of the optical properties of these impurities in solids and functionality features of the corresponding phosphors bearing these impurities is possible only if a clear picture of the origin of all electronic states and their interaction with the crystalline environment is formed. Recently[17],we published a detailed tutorial review on the electronic properties of the d3electron configuration. Here we remind the reader of the salient points that should be kept in mind when working with such ions and which will be frequently used in this review：

(1)There are five 3d orbitals, which have the same principal quantum numbern= 3 and orbital quantum numberl=2, but differ by the magnetic quantum numbersml= -2,-1,0,1,2.

(2) Each orbital can accommodate not more than two electrons, whose spin momenta in such a case should be opposite,as is stated by the Pauli exclusion principle. This gives 5 ×2 =10 single-electron states in total. There are 120 ways (different permutations) to distribute three electrons through these single electron states that is calculated as follows：10! /(3! ×(10 -3)!) =120.

(3)The total wave functions of such configuration are antisymmetric linear combinations of products of three single electron wave functions built following the quantum-mechanical rules of the orbital momenta addition. Thorough consideration of all 120 microstates reveals that they are grouped in a way to give rise to eight highly degenerated terms with different values of the total orbital momentumLand total spin momentumS.

(4)These terms are as follows(in the2S+1Lnotation)：4F(the ground term as specified by the Hund's rule),4P,2P,2D(1),2D(2),2F,2G,2H.Here the superscripts “4” and “2” denote the spinquartet states(S=3/2) and the spin-doublet states(S=1/2), respectively. The letters P, D, F, G,H stand for the values ofL=1, 2, 3, 4, 5, correspondingly. The subscripts “(1)” and “(2)” distinguish between two2D terms. The degeneracy degree of each term is found as (2S+1) ×(2L+1).

Tab.1 Free ion values(in cm -1) of the Racah parame ters for some 3d3 ions

(6)Free ions'LSterms are split into a number of states in a crystal field. The splitting patterns depend on the crystal field symmetry and can be obtained by using group theory. In the cubic crystal field, all possible split states are listed in Tab. 2.Orbital singlet, doublet, triplet states are denoted by letters A, E, T, correspondingly, which also stand for the irreducible representations of the Ohpoint group.

In the crystal fields of lower symmetries further reduction of the degree of degeneracy can bepossible, until all states become orbital singlets.

Tab.2 Splitting of the S, P, D, G, G, H terms in the cubic crystal field

(7)Three interactions are mainly responsible for the formation of energy levels of the d ions in solids. In the order of the decreasing magnitude they are as follows： (ⅰ)the Coulomb interaction between electrons in the unfilled d shell, which produces theLSterms of free ions； (ⅱ)the crystal field splitting of the free ion'sLSterms,and (ⅲ) spin-orbit interaction, which produces the fine structure of the split crystal field levels.

(8)Quantitative calculations of the crystal field splittings are possible within the crystal field theory.The Tanabe-Sugano diagrams show variation of the split energy levels depending on the crystal field strength denoted byDq. In these diagrams, the energy of the ground state is always taken as zero； the crystal field symmetry is cubic (which corresponds either to the octahedral or tetrahedral coordination of the impurity ions). The horizontal axis is theDq/Bratio, the vertical axis is theE/Bratio, whereEis the energy of a particular state, andBis the Racah parameter. For the d3ions in the octahedral environment the separation between the4A2and4T2states is equal to 10Dq. All notations in Fig. 1 come from Tab.2.

Fig.1 Tanabe-Sugano diagram for the d3 electron configuration in the octahedral crystal field. C/B =4.5. The spin-quartet and spin-doublet states are shown by the solid and dashed lines, respectively.

(9)It is important to note that at the value ofDq/B≈2 there is a change of the first excited state.WhenDq/B＞2, the absorption spectra are dominated by two broad spin-allowed transitions4A2→4T2and4A2→4T1(blue upward arrows in Fig.1). After non-radiative relaxation shown by the dashed arrows,sharp spin-forbidden transition ascribed to the2E →4A2transition (red downward arrow) takes place. IfDq/B＜2, the absorption spectra are still formed by the same above-listed transitions, but in the emission spectra a broad band appears, which is due to the4T2→4A2transition. In the vicinity of the point of intersection of the2E and4T2states both types of the emission transition occur, which can give rise to interesting phenomena discussed below.

This information presented in the form of the bullet points creates the minimal necessary background for the discussion, clarification and explanation of the main spectroscopic properties of the Cr3+and Mn4+ions doped phosphors.

2 Mn4 + Ions for White LEDs

2.1 General Concepts

Discovery of blue light emitting diodes(LED)based on GaN or InGaN[20-21]allowed to produce bright, efficient and durable sources of white light.The first white LEDs were based on GaN blue LED chip emitting at 450 nm combined with the yellow phosphor Y3Al5O12∶Ce3+(Fig. 2). The latter partially absorbs blue light and converts it into broad yellow emission that is due to the Ce3+5d-4f transition. This emission when blended with the blue light of LED produces white light.

The main parameters that characterize white LEDs are the color correlated temperature(CCT)and the color rendering index(CRI). The CCT is the temperature of the ideal black body radiator that emits light with the spectral distribution approximating a given white LED spectrum in the best way. The higher CCT values(5 000 -6 000 K) are characteristic of somewhat “bluish” white light, which is called “cold white light”. The lower CCT values(3 000 -4 000 K) correspond to the “yellowish”,or “warm white light”. The warm white light LEDs are more popular for living interiors, while the cold white light LEDs are more often used for the office spaces. The CRI indicates ability of the light source to reproduce true or natural colors of the objects. By definition, the sun light has the highest CRI of 100.

However, due to the lack of red emission in the original white LEDs spectra(compare the relative intensities of the blue, yellow, and red emissions in Fig.2),the white light generated has rather low CRI( ～76) and rather high color correlated temperature( ～6 200 K)[10], and is perceived by the human eye as the “cold” white light.

Addition of some red phosphor to such white LED would significantly improve the white light characteristics. One of the best red phosphors,which has already found numerous commercial applications, is K2SiF6∶Mn4+[22-27]. Recently, a large number of other Mn4+-doped hosts have been reported, many of them are listed in Refs.[14,28-30].

Fig.2 Emission spectrum of a white LED based on the In-GaN blue LED and Y3Al5O12∶Ce3+ yellow phosphor

Why are the Mn4+-doped phosphors so attractive? Some reasons are listed here：

(1)Mn is cheaper than the rare-earth activators, therefore, the cost of the LEDs mass production with the Mn4+-based red phosphors can be significantly reduced if compared with the rare-earth ions based phosphors.

(2)The energies of the Mn4+4A2→4T2and4A2→4T1transitions correspond to the blue light(～450 nm) and ultraviolet(UV) light (～330 -380 nm).This allows to use the blue and UV LEDs as the direct excitation sources for this activator.

(3)Due to the direct excitation to the Mn4+states, the energy losses are minimized(if the impurity is excited by the energy transfer from the host after over-band-gap excitation, efficiency of such energy transfer is the crucial factor).

(4) Mn4+ions are always described by the strong crystal field case,i.e. emission is always due to the sharp spin-forbidden2E→4A2transition.

(5)The emission energies of the2E→4A2transition can be tuned in a very wide range, from ～620 nm to ～720 nm[7], by changing the crystalline host, which opens numerous possibilities for tuning emission color from red to deep-red.

(6)The2E→4A2emission is very sharp, with the full width at the half maximum of a few nm only,which ensures color purity—a very important parameter for the display applications.

(7)Plenty of host materials can accommodate the Mn4+ions, and with partial cation and/or anion substitution the tunability of the Mn4+red emission is greatly enhanced.

2.2 Host Materials

All host materials that can be doped with the Mn4+ions can conditionally be divided into three groups： oxides, fluorides and oxyfluorides. This classification may not be necessarily correct from the pure chemical point of view. It is based rather on the type of ligands surrounding impurity, which can be either oxygen or fluorine anions. An interesting intermediate case of oxyfluorides is also existing, but the number of such compounds is not yet as large as the number of the conventional oxide or fluoride phosphors.

Tab.3 -5 list a large number of solids(divided into the groups conditionally referred to as fluorides,oxides, oxyfluorides) doped with the Mn4+ions that have been reported in recent publications. The values of the crystal field strengthDq, Racah parametersBandCand the energetic position of the emitting2E energy level are all given.

Tab.3 Spectroscopic parameters of Mn4+ ions in fluoride phosphors. Dq is the crystal field splitting, B and C are the Racah parameters, E(2E) is the ZPL position, all these values are in cm -1

Tab.3(Continue)

Tab.4 Spectroscopic parameters of Mn4+ ions in oxide phosphors, all notations are the same as in Tab.3, the hosts with questionable Racah parameters B and C are highlighted

Tab.4(Continue)

Tab.5 Spectroscopic parameters of Mn4+ ions in oxyfluoride phosphors, all notations are the same as in Tab.3

It can easily be seen that in the fluorides the2E level is generally located higher than in the oxides.Such an observation has been made earlier[7], and it is attributed to a weaker nephelauxetic effect(weaker covalency) in fluorides. Fig.1 shows that the emitting2E level is practically independent of the crystal field strengthDq. On the other hand, the energetic separation between the2E and4A2states in the crystal field is very close to the energy interval between the2G and4F terms of a free Mn4+ion(these are the terms, where the2E and4A2levels come from-see Fig.1), which depends on the Racah parametersBandConly. Since-due to the nephelauxetic effectthe Racah parameters for the impurity ions in solids are reduced(and often quite considerably), when compared to their free ion's values, and the magnitude of such reduction varies to a great extent in different solids due to peculiarities of the chemical bonds formation, the2E level position depends mainly on theBandCvalues. As a consequence,in highly ionic fluorides,the values of the Racah parametersBandCare generally greater than in more covalent oxides(Tab. 3 and 4), leading to a blue shift of the2E level. An interesting case is formed by the solids with mixed anion composition, like oxyfluorides(Tab. 5). Since the oxyfluorides combine the fluorides ionicity and the oxide covalency,their emission maxima are positioned between those corresponding to these two large groups of materials.

As a quantitative measure of the covalent effects in the Mn4+-doped crystalline solids, we introduced a non-dimensional parameterβ1,which indicates degree of reduction of the Racah parametersBandC[28-29,100]：

whereB,C(B0,C0) are the Racah parameters of the corresponding ions in a crystal(free state), respectively. It appears that the position of the2E energy level is a linear function of this parameterβ1.This linearity follows from the properties of the Tanabe-Sugano matrices for the d3electron configuration, as was derived in Ref. [151]. Fig.3 illustrates this linear trend. All data points shown in this figure are taken from Tab. 3 - 5(several entries,which are shown in bold, are omitted, because the reported values ofBfor those materials exceed the free ion value, thus indicating some problems with the experimental spectra treatment). All data points shown in Fig. 1 were fitted to the linear equation,which was obtained as follows：

the root-mean-squared deviationσbetween the calculated from Eq. (5) and experimental positions of the2E level was found to be 359 cm-1. Two dashed straight lines in Fig.3 represent the linear function from Eq.(5) shifted upward and downward by this value ofσ. Then nearly all data points are located within this “corridor” marked by the dashed lines.Deviation from the fitting line of Eq.(5) shifted by±σcan be explained by difficulties in assigning the zero-phonon line(ZPL) of the2E→4A2transition,which is often masked by the vibronic transitions.

Fig.3 Dependence of energy of the Mn4+ 2 E level on the new nephelauxetic ratio β1

2.3 Tunability of The Mn4+ Emission in Crystalline Solids

Analysis of data from Tab. 3 - 5 and Fig. 3 shows that the position of the2E emitting level varies from 13 498 cm-1(741 nm) in La4Ti3O12to 16 920 cm-1(591 nm) in K3HF2WO2F4. In other words,by choosing a proper host, it is possible to tune the Mn4+emission from nearly orange color through deep red to the infrared spectral range, which creates a lot of opportunities for various applications.

The Mn4+-doped phosphors with emission at around 620 nm are very good for the solid-state lighting applications, since such emission is characterized by a greater spectral overlap with the human eye sensitivity curve[152], whereas the deep-red emitting phosphors are more suitable for the agricultural applications as a good light source for effective and fast plant growth[153].

Based on the data presented in this review,it is possible to formulate the following rules that can help in choosing proper hosts for the selected applications：

(1)If red photons in the wavelength range of 600 -630 nm are needed, then highly ionic fluorides should be considered for doping with the Mn4+ions.

(2)If red photons in the spectral range between 630 nm and 700 nm are required, then highly covalent oxides should be chosen to accommodate the Mn4+ions.

(3)If the local symmetry of the Mn4+site is perfectly octahedral(which means the p-states of ligands and d-states of the Mn4+ions are directed along the same axis), lowering of the2E level is anticipated due to increased overlap of the above-mentioned wave functions, that leads to a more considerable decrease of the Racah parametersBandC.

(4)If the local symmetry of the Mn4+site is not perfectly octahedral, then the overlap of the p-states of ligands and d-states of the Mn4+ions is decreased, thus leading to the weaker nephelauxetic effect, greater values of the Racah parametersBandCand, as a result, upward shift of the emitting Mn4+2E level.

(5) External hydrostatic pressure causes decrease of the interionic distances. As a result, this will be accompanied by an increase of the overlap integrals and enhanced nephelauxetic effect. The values of the Racah parametersBandCwill be further reduced leading to a slight red shift of the2E→4A2emission transition.

(6) The so-called “chemical” pressure can lead to a similar result. “Chemical” pressure can be produced by partial cation or anion substitution(especially with greater ionic radii), after which the neighboring chemical bonds will shrink-similar to the hydrostatic pressure. The only difference between these two phenomena is that the “chemical” pressure affects only nearest neighbors around the ion with a greater ionic radius, whereas the hydrostatic pressure acts upon the whole sample.

(7)Fabrication of the Mn-containing nano-sized crystallites gives opportunities to design an exciting class of red luminescence materials showing quantum confinement. The emission color can be tuned by choosing proper size of the crystallites.

2.4 Enhancement of The Mn4+ Emission Intensity in Crystalline Solids

An efficient phosphor for the solid-state lighting should produce bright emission after suitable excitation. When it comes to the Mn4+emission spectra,they typically consist of several closely located sharp lines ascribed to the2E →4A2transition. One of these individual peaks corresponds to the zero-phonon line, the peaks at the higher/lower energy side of it-to the anti-Stokes and Stokes vibronic transitions, respectively. As an empirical rule, intensity of the Stokes peaks is higher than the ZPL and anti-Stokes peaks(relative intensity of these peaks depends on the strength of the electron-vibrational interaction, normal mode energiesetc). An important question is how to identify the ZPL position. The following example explains this in detail.

Fig.4 shows the emission spectrum of NaKSiF6∶Mn4+[152].

Fig.4 Emission spectrum of NaKSiF6 ∶Mn4+[152]. The ZPL position and the Stokes/anti-Stokes peaks are indicated.

To find the ZPL location, it is necessary to plot the emission spectrum in the energy scale. Then by looking at the spectrum, one has to identify the peak(it may have a very low intensity, sometimes really almost zero), whose position serves as a symmetry point, about which positions of all other peaks in the spectrum become the mirror images of each other. It should be emphasized that the “mirror symmetry”should not be understood literally in this case： the matched peaks intensities are not supposed to match each other,but their positions must differ by approximately the same amount of energy, taken with different signs. The ZPL in Fig.4 is at about 16 120 cm-1(620 nm). The two nearest peaks located to the left and to the right from it differ by 219 cm-1.This difference corresponds to one of the normal mode energies of the octahedral MnF6cluster. The high-energy peak is the anti-Stokes peaks, which is formed by the pure electronic2E →4A2transition emission transition with absorption of one vibrational quantum(it would increase the energy of the emitted photon), whereas the low-energy peak corresponds to the partial loss of the pure electronic2E →4A2emission transition energy(one phonon would be produced then). Another pair of emission peaks is mirrored about the ZPL position with the energy difference of about 345 - 348 cm-1. They correspond to the vibronic transitions, which are formed with participation of the pure electronic2E →4A2transition and another normal mode with the abovegiven energy.

Fig.5 Human eye sensitivity curve and emission spectra of K2SiF6 ∶Mn4+ and NaKSiF6 ∶Mn4+ phosphors. The figure below is a zoomed view of the area shown by a rectangle in the top figure.

Fig.5 shows comparison of the human eye sensitivity curve peaked at about 550 nm with emission spectra of K2SiF6∶Mn4+and NaKSiF6∶Mn4+phosphors.

It is clear that the greater the overlap of the emission spectrum with the human eye sensitivity curve is, the brighter the phosphor to the human eye is. From this point of view, the NaKSiF6∶Mn4+phosphor is brighter than K2SiF6∶Mn4+phosphor,as evidenced by the lower diagram in Fig.5. Since the Stokes peaks in the emission spectra are more intensive than the ZPL and the anti-Stokes ones(Fig.4),increase of the ZPL intensity is a very important task to be achieved for getting efficient bright phosphors.In addition, the ZPL for the ideal red phosphors based on the Mn4+ions should be located at about 620 nm-this would ensure a large overlap with the human eye sensitivity curve, still staying in the red part of the visible spectrum.

It has been noticed that the ZPL has a higher intensity in the host materials with rather low local symmetry, especially in those ones, which lack an inversion center at the local site occupied by the emitting impurity ion[152]. This is because the inversion center absence removes the parity selection rule for the electric dipole transitions. Therefore, choosing the low-symmetry hosts for doping with the Mn4+ions is one possible way to get bright phosphors.

It is also possible to remove the inversion center artificially, when making the phosphor, by employing the strategy of the cation or anion substitution.The above given example of K2SiF6∶Mn4+and NaKSiF6∶Mn4+phosphors is a good illustration. In both phosphors the ZPL is at about 620 nm, but in the latter phosphor its intensity is increased drastically.This is because of random occupation of the first cation site by K and Na the inversion symmetry at the Si site(occupied by the Mn4+ions) is removed. Another example is the Rb2HfF6, Cs2HfF6, and Rb-CsHfF6phosphors with Mn4+ions. By performing the first-principles calculations, it has been shown that in the “mixed” RbCsHfF6compound the Cs and Rb ions in the second coordination sphere around the Hf ions(occupied by the Mn4+ions after doping) affect the opposite Hf—F chemical bond lengths in the HfF6octahedron,which loses then the inversion center[152,154].

We also mention a recent publication[155]that is focused on the ways of achieving higher efficiency and greater stability of the Mn4+-doped phosphors for white LEDs.

2. 5 Several Misconceptions Related to The Analysis of The Mn4+ Ions Spectra in Crystalline Solids

Quite often it is possible to find examples of improper interpretation and analysis of the Mn4+spectra in solids. Here we shall not give the references to those wrong publications. Instead, we just list those incorrect statements we have seen in the papers and explain what was wrong there.

(1)Sometimes-although not often-the tetrahedral Mn4+centers are mentioned. This is ultimately wrong-the Mn4+ions occupy only octahedral positions.

(2) Sometimes-especially in the oxide hoststhere are certain issues with assignment of the4A2→4T1absorption band. This is because this absorption band overlaps with the O-Mn charge transfer transitions(in fluorides, luckily, this charge transfer band is located at considerably high energy position due to higher electronegativity of the fluoride anion). Then people mistakenly assign the charge transition band to the4A2→4T1transition, which eventually leads to unrealistically high values of the Racah parameterB-even exceeding the free ion value(which is physically impossible!) and unrealistically low values of the Racah parameterC.

(3)Since the Mn4+ions are always located at the strong crystal field sites (Dq/B＞2.2 -2.5 in the Tanabe-Sugano diagram),theDq/Bratios which are either smaller or unrealistically higher( ＞5 -6,for example) immediately indicate some mistakes in determination of theDq,B,Cvalues.

(4)There are often some issues with the ZPL determination；e. g.people attempt to assign the ZPL to the most intensive peak in the group of vibronic lines corresponding to the2E→4A2emission transition. This should not be done, and the careful analysis of the ZPL location, as described in the present review, should be performed.

(5)There are also some attempts to assign the origin of the2E→4A2emission transition to the electron-vibrational interaction(EVI). This is wrong.The EVI can affect the spectral shape of emission peaks, but the physical origin of appearance of this transition is in the spin-orbit interaction, that mixes together the states with the same total angular momentumJin the2G and4F terms even for a free ion[17].

(6)The structure of the2E→4A2emission spectrum is due to the crystal field effects. This is only true partially. The most important reason for the structured2E→4A2emission spectrum is EVI, and a thorough analysis of the ZPL position and associated vibronics should be performed.

3 Conclusion

In the present review, we have described the main spectroscopic properties of the Mn4+ion,which is a very important activator for the solid-state lighting. The origin of the Mn4+energy levels and their splitting in crystal fields were described. The main advantages of these ions over others are also highlighted. The main spectroscopic parameters of the Mn4+ions in more than 100 solids are collected and the main trend across this group of materials is discussed. Several practical recommendations on how to analyze the spectra of Mn4+ions in crystalline solids are given in detail to help the researchers working in the field for the discovery of novel Mn4+red phosphors.