Open Access
Add to BibSonomy
Abstract
The effect of Al2O3 content on the structure and properties of phosphate glasses Nd3+: Al2O3-K2O-P2O5 was investigated. Deconvolution Raman spectra were used to elucidate the glassy structures and it was observed that with increased Al2O3 content, the vibration frequencies and relative intensities of Qn (n represents the number of bridging oxygens (BO) per tetrahedra) changed. The Raman derived glass structure units were used as descriptors to build quantitative structure-property relationship (QSPR) models using the partial least squares (PLS) method to investigate the relationship of the glass structure and different physical properties such as glass transition temperature (Tg), density (ρ), and refractive index (nc). The glass properties (Tg, ρ and nc) have positive relationships with the vibration frequency and relative intensity of Q21Al (one bridging oxygen bound to Al, and one bridging oxygen bound to K). In contrast, they have negative relationships with the vibration frequency and relative intensity of Q20Al (two bridging oxygens bound to K) and Q3. The increase in Q21Al leads to the tightening of the phosphate network and a decrease in the molar volume, resulting in improved glass properties (Tg, ρ and nc). The Q21Al tetrahedra that form phosphate chains or rings is one of the main factors that affect the luminescent properties (σems, Δλeff and τrad) of the Nd3+ ions. With increases in vibration frequency and relative intensity of Q21Al, the emission cross section (σems) decreases, but both the effective line-width (Δλeff) and fluorescence lifetime (τrad) increase.
© 2018 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Nd-doped phosphate glasses have been used in high energy/ high power laser applications because of their low nonlinear refractive index and high stimulated emission cross section [1,2]. Poor chemical durability of phosphate glasses can be substantially improved by the addition of aluminum. Kamitsos et al. indicated the phosphate network connectivity by P–O–Al and Te–O–Al cross-links caused a drastic enhancement in Tg of AgI-AgPO3 and Al2O3-TeO2 glasses [3–5]. Upon further investigation, Kamitsos et al. indicated that increasing field strength or force constant significantly influenced Tg [6]. Brow indicated that for xAl(PO3)-(1-x)NaPO3, the addition of Al(PO3)3 increases the Tg, n, and ρ as Al(OP)6 cross-linking groups replace the PO-Na+ bonds. For xAlPO4-(1-x)NaPO3, substitution of AlPO4 for NaPO3 consumes both POP and PO-Na bonds to form POAl(6) bonds, which cross-link and shorten the POP chains [7,8]. Schneider et al. proposed that aluminum ions, with a coordination number of 6, restrict the mobility of phosphate chains by forming interchain linkages causing an increased in Tg [9].
Introducing Al2O3 into Nd: phosphate glass decreases the emission cross section (σems) and broadens the effective line-width (Δλeff) of the emission band. Toratani et al. reported that the Δλeff increases and the σems decreases with increasing Al2O3 content [10]. More recent studies have investigated the effect of Al2O3 content and suggested that P-O-Al bonds affect the luminescent properties of Nd3+ ions [11–14]. However, the relationships between the luminescent properties of Nd3+ and glass structure changes induced by Al remains unclear.
Raman spectroscopy is useful technique for studying the structure of glassy materials and their structure evolution with respect to changes in compositions, which in return correlate to changes in properties [15–17]. The raw data in Raman spectroscopy are often broad bands that envelop many over-lapping components. Deconvolution or curve-fitting is a helpful tool to extract information from these overlapping components. By deconvoluting and assigning the vibrational bands to reported values in the literature, the structural units of the host glass network can be identified. Quantitative structure-property relationship (QSPR) is a widely used method which is used to extract the often-complex relationships between the microscopic (usually molecular) structure and the macroscopic properties (mechanical, thermal, optical, etc.) of materials [18–20]. QSPR modeling has been used to characterize a wide range of materials such as nanomaterials, catalysts, biomaterials, ionic liquids, ceramics and so on [18–20]. Guo et al. used the compositions of a set of BaTiO3 - based dielectric ceramic samples as descriptors to build QSPR models to estimate the values for different properties such as room-temperature dielectric loss [21,22]. Partial least squares (PLS) is one of the QSPR modeling methods [23]. PLS is a statistical method, which is somewhat related to principal component regression and is used to find the fundamental relations between dependent variables y and independent variables X. Michael Gonzalez-Durruthy et al. used Raman spectra and PLS to build QSPR models to predict Young’s modulus and Poisson’s ratio in test sets of carbon nanotubes with large variation across radii, and number and type of surface defects [24].
In this study, Raman spectroscopy and PLS method were used to investigate the relationship between the glass structure and its properties (Tg, ρ and nc). The Tg, ρ and nc have positive relationships with the vibration frequency and relative intensity of Q21Al, but negative relationships with the vibration frequency and relative intensity Q20Al and Q3.The increase in the vibration frequency and relative intensity of Q21Al leads to the tightening of the phosphate network and a decrease of the molar volume, resultings in increased Tg, ρ and nc. The Q21Al tetrahedra that form the phosphate chains or rings is one of the main factors affecting the luminescent properties (σems, Δλeff and τrad) of Nd3+ ions.
2. Experimental and method
2.1 Experimental
KPO3, Al(PO3)3 and a series of xAl2O3-(40-x) K2O-60P2O5 (x = 3, 6, 9, 12, 15) glasses were prepared by a conventional melting procedure. Stoichiometric amounts of Al(PO3)3, Al2O3, KPO3 and P2O5 raw materials were mixed with 2wt% of Nd2O3. The mixtures were then melted in quartz crucibles for 30 minutes at different temperatures in the range 900-1400°C depending on the components. To minimize the concentration of hydroxyl in the glass, dry CCl4 was bubbled along with oxygen during glass melting. After stirring and refining, the melts were cast into preheated steel molds and annealed at the corresponding glass transition temperatures in a muffle furnace. The annealed glass samples were cut, ground, and polished before carrying out various property and Raman spectral characterization.
The glass transition temperature (Tg) of the glasses was measured using a NetzschSTA449/C differential scanning calorimeter (DSC) at a heating rate of 10 °C/min. The error of the Tg value is less than ± 0.2 °C. The densities of the glasses were measured using the Archimedes method with distilled water as the immersion liquid. The error of the density is ± 0.001 g/cm3. The refractive index (nc) were measured using the waveguide prism coupling method. The error of the refractive index is ± 0.001. The content of Nd3+, P5+, K+ and Al3+ were determined using a Thermo iCAP 6300 radial view inductively coupled plasmaoptical emission spectrometer (ICP-OES), the results are listed in Table 1. Absorption spectra of the glasses were recorded using a Lambda 950 UV–VIS–NIR spectrophotometer in the range of 300 - 900 nm. The emission spectra were obtained by excitation at 808 nm for Nd3+ ions using a Xe lamp, and fluorescence lifetimes were measured by a high resolution spectrofluorometer (Edinburgh Instruments FLS 920) following excitation at 808 nm using a microsecond-pulsed Xe flash lamp.
Table 1. Compositions of xAl2O3-(40-x)K2O-60P2O5 (x = 3, 6, 9, 12, 15) glasses with respect to P2O5, K2O, Al2O3 in mol% and Nd2O3 in wt% obtained from ICP-OES.
Raman spectra were recorded at room temperature using a Renishaw inVia Raman microscope in the range of 200-1400 cm−1 using the 488 nm laser line as the excitation wavelength. The measurement parameters were as follows: aperture: 2400l/mm, detector: CCD, exposure time: 10s, number of acquisitions:1, and scan type: SynchroScan.
2.2 Partial least squares (PLS)
PLS is a multivariate method (Dayal and MacGregor, 1997) that yields a stepwise solution for a regression model [23]. The method extracts linear combinations of the predictors, known as factors (also called components or latent vectors). The factors optimally address the combined goals of explaining dependent variables (y) and independent variables (X). The PLS model is built on nonlinear iterative partial least squares (NIPALS) algorithms.
X represents an independent variation (X is a matrix), y represents the dependent variation (y is a vector) and i is the ith component.
where superscript T denotes the transpose of a matrix. Then, the regression coefficients for the PLS model are obtained from the relationThe regression equation of the original variables is given as (omitting residuals)When the regression is linear of the form (y = bX + a), the regression coefficient is a constant (b), which represents the rate of change of the dependent variable (y) as a function of the changes in the independent variables (X). A positive regression coefficient indicates that y increases with increase in X, and a negative regression coefficient indicates that y decreases when X is increased.The square of the correlation coefficient of determination (R2), adjusted R2 (R2adj), and root mean square error (RMSE) are used to evaluate the fit of the model to experimental date [25–27]. The larger the value of R2, the higher value of R2adj, and the smaller the value of the RMSE, indicating a better fit and more suitable model.
3. Results and discussion
3.1 Raman spectra
Raman spectra of KPO3, Al(PO3)3 glasses and KPO3, Al(PO3)3 crystals are depicted in Fig. 1(a). Meyer et al. compared the Raman and IR spectra of Zn-phosphate materials and revealed that crystalline and glassy phosphates with similar composition have similar phosphate anions [28]. The band in the vibration spectra of KPO3 and Al(PO3)3 crystals in the range 1250-1400 cm−1 is identified as vas(Q2), 1000-1250 cm−1 is identified as vs(Q2), 600-850 cm−1 is identified as v(P-O-P) (Q2), and 100-600 cm−1 is attributed to deformation vibration. By comparing the Raman spectra of crystalline KPO3 and Al(PO3)3, it is noted that the bands of KPO3, Al(PO3)3 glasses in the range 1250-1400 cm−1 is assigned to vas(Q2), 1000-1250 cm−1 is assigned to vs(Q2), 600-850 cm−1 is assigned to v(P-O-P) (Q2), and 100-600 cm−1 is assigned to deformation vibration. Figure 2(b) compares the Raman spectra of PKAi (i = 3, 6, 9, 12, 15) glasses to those of KPO3, Al(PO3)3 glasses. PKAi (i = 3, 6, 9, 12, 15) glasses have additional bands in the range of 1300-1400 cm−1, and this vibration band is assigned to v(P = O) (Q3) in accordance with published literature [1]. Table 2 summarizes the Raman band assignments by matching them with literature data covering a wide composition range of phosphate glasses.
Fig. 1 (a) Raman spectra of KPO3, Al(PO3)3 glasses and KPO3, Al(PO3)3 crystals;(b) Raman spectra of KPO3, Al(PO3)3 and PKAi (i = 3, 6, 9, 12, 15) glasses.
Fig. 2 Deconvolution Raman spectra of KPO3, Al(PO3)3 and PKAi (i = 3, 6, 9, 12, 15) glasses in the range 850-1400 cm−1.
Table 2. Raman band positions identified from this study and band assignments taken from literatures.
Figure 2 shows the deconvolution Raman spectra of the PKAi (i = 3, 6, 9, 12, 15) and KPO3, Al(PO3)3 glasses. Owing to the presence of a small amount of –OH, the deconvoluted Raman spectra of KPO3 and Al(PO3)3 glasses contain vs(Q1) bands [30]. The band in the range 1300-1400 cm−1 is assigned to v(P = O) (Q3). The band in the range 1250-1300 cm−1 is assigned to vas(Q2). The band in the range 1000-1250 cm−1 containing two components is assigned to vs(Q2). Gwenn Le Saout et al. assigned these two bands of (PbO)0.5-(P2O5)0.5 glass as vs(Q2) (1150-1200 cm−1, the high frequency band) and vs(Q11) (two Q1 connected) (110-1150 cm−1), the low frequency band) [33]. PKAi (i = 3, 6, 9, 12, 15) glasses, O/P≈3, and therefore, the Q11 content cannot be larger than the Q2 content, which means that this Raman band assignment cannot be applied to PKAi (i = 3, 6, 9, 12, 15) glasses. According to Tsuchida et al. it is possible to resolve different types of Q2 phosphate species according to the number m of non-bridging oxygens (NBOs) bound to Al (m = 0, 1, 2) [34]. In KPO3 glass, two NBOs of are linked to K, and therefore the Q2 phosphate tetrahedra of KPO3 is Q20Al. Two NBOs in Al(PO3)3 glass are bound to Al, and hence the Q2 phosphate tetrahedra of Al(PO3)3 is Q22Al (two NBOs bound to Al). Tsuchida et al indicated that the association of P with Al occurs preferentially through one NBO per phosphate tetrahedra (forming Q21Al tetrahedra) and the distribution is close to binary (Q20Al, Q21Al) for K-Al glasses [34]. With the increase in aluminum content, more Al-O-P linkages are created, which causes the relative area in the range of 1150-1170 cm−1 decreasing, and in the range of 1170-1190 cm−1 increasing. Consequently, it is reasonable consider that the vibration in the range 1150-1170 cm-1 is vs(Q20Al) and that in the range of 1170- 1190 cm−1 is vs(Q21Al).
The original Raman spectra (Fig. 1(b)) in the range of 1000-1150 cm−1 (vs (Q1)) are flat. The preferred arrangement of Al ions and Qn tetrahedra in phosphate glass follows the guidelines: (1) avoids phosphates sharing of both NBOs with Al; and (2) allocates Al ions with maximum NBO coordination. The preference of Qn involves sharing one NBO with Al and the other with K. The possible Q1 has one NBO connecting to Al and two NBOs connecting to K. More K is required to connect to NBOs if Q1 form in the glass. However, the K2O content decreases with the substitution of Al2O3 for K2O. So, the preference of Qn is Q21Al (one NBO connecting to Al and one to K), rather than Q1 (two NBOs connecting to K and one to Al). There is likely a minor presence of Q1, but its content is too small to be represented in the deconvolution Raman spectra.
Based on the classical theory of electromagnetism, the Raman scattering intensity I(v) of a molecule at a certain frequency v can be expressed as:
where c is the speed of light in vacuum, h is the Planck constant, IL is the intensity of the exciting radiation, N is the number of molecules, v and v0 respectively are the molecule vibration frequency and exciting light frequency, μ is the atom reduced mass, K is the Boltzmann constant, T is the absolute temperature, αa'is the average of susceptibility tensor, γa'is the anisotropic polarization tensor, and I(v) is the integrated intensity of Raman bands. According to this formula, for a given material, the Raman scattering intensity is directly proportional to molecular concentration if the exciting light intensity and temperature of measurement are constant, which means that under these conditions, the Raman spectra can be used in quantitative analysis.Table 3 lists the vibration frequencies and relative intensities of Raman deconvolution bands of PKAi (i = 3, 6, 9, 12, 15) glasses. The change in vibration frequencies and relative intensities of these bands as a function of the Al2O3 content are depicted in Fig. 3(a) and 3(b). The frequency of vs(Q21Al) increases with increase in the Al2O3 content. Velli et al. suggested that an up-shift of the vs(Q2) vibration frequency occurs with increasing ion field strength in alkali phosphate glasses [30]. Schneider et al. indicated that the vs(Q2) vibration of (1-x)NaPO3-xAl2O3 glasses shifts to higher frequencies with increasing Al(PO3)3 content [9]. Our results are consistent with those of Schneider et al. With increase in Al2O3 content, the relative intensity of vs(Q21Al) increases, while that of vs(Q20Al) decreases. The decrease in vas(Q2) relative intensities indicate that the Q2 tetrahedra become less asymmetric with increasing Al2O3 content. The vibration frequency of v(Q3) also decreases with increasing Al2O3 content because the P = O bond of Q3 lengthens [15]. The decrease in v(Q3) relative intensity with increasing Al2O3 content is caused by detachment of P = O to form P-O-Al.
Table 3. Vibration frequencies (v cm−1) and relative intensities (I %) of Raman deconvolution bands of PKAi (i = 3, 6, 9, 12, 15) glasses.
Fig. 3 (a) Raman deconvolution band shift; and (b) change in relative intensities as a function of Al2O3 content.
3.2 Physical and fluorescence properties
Figures 4(a) 4(b) and 4(c) respectively show glass transition temperature (Tg), density (ρ) and refractive index (nc) of PKAi (i = 3, 6, 9, 12, 15) glasses. It is clear that Tg, ρ and nc of the glasses monotonically increase with increase in Al2O3 content.
Fig. 4 Effect of Al2O3 content on (a) glass transition temperature (Tg); (b) glass density (ρ); (c) refractive index (nc) of PKAi (i = 3, 6, 9, 12, 15) glasses.
The absorption spectra of all samples were tested. The spectrum of PKA3 is shown in Fig. 5(a) as a typical example. For the Nd3+ ion, absorption bands correspond to transitions from 4I9/2 to other excited levels. The eleven absorption bands are assigned to transitions 4I9/2→4D11/2, (4D5/2, 4D3/2), (2D5/2, 2P1/2), (4G11/2, 2G9/2, 2D3/2, 2K15/2), (4G9/2, 4G7/2, 2K13/2), (4G5/2, 4G7/2), 2H11/2, 4F9/2, (4F7/2, 4S3/2), (2H9/2, 4F5/2), 4F3/2 at 329, 353, 429, 477, 523, 582, 626, 684, 746, 801, 870 nm, respectively [1].
Fig. 5 (a) Room absorption spectrum of PKA3; (b) Emission spectrum of PKA3 with excitation wavelength is 808nm.
Based on the absorption spectra, spectroscopic intensity parameters (Ω2, Ω4 and Ω6) of Nd3+ ion were calculated according to the Judd-Ofelt theory [35,36]. The calculated values of Ω2,4,6 are listed in Table 4 and depicted in Fig. 6. The value of Ω2 increases with increase in Al2O3 content, while Ω4 and Ω6 show the opposite trend. The parameter Ω2 is most sensitive to the glass composition and local structure, which reflects the extent of covalent bonding and the asymmetry of the local environment in the vicinity of the rare-earth site. The increase in Ω2 with the increase in aluminum content suggests the presence of a less centrosymmetric coordination environment and a greater extent of covalent bonding around the Nd3+ ions [37,38].
Table 4. Judde - Ofelt parameters (Ω2, Ω4 and Ω6), fluorescence lifetime (τrad), effective line-width (Δλeff) and emission cross section (σems) of PKAi (i = 3, 6, 9, 12, 15) glasses.
The fluorescent spectra of glasses pumped at 808nm are shown in Fig. 5(b). Four emission bands at 895 nm, 1053 nm, 1324 nm and 1800 nm correspond to the 4F3/2 → 4I9/2, 4I11/2, 4I13/2, 4I15/2 transitions. The fluorescence lifetime (τrad), effective line-width (Δλeff) and emission cross section (σems) values are listed in Table 4.
The change in τrad with Al2O3 content is shown in Fig. 7(a). The fluorescence lifetime increases from 367 μs to 441 μs with an increase in of Al2O3 content. The corresponding change in Δλeff is shown in Fig. 7(b), where it is seen that Δλeff increases from 18.15 nm to 22.58 nm with increase in Al2O3 content. In contrast, as seen in Fig. 7(c), the emission cross section decreases from 1.46 × 10-20 cm2 to 1.22 × 10-20 cm2 with increasing Al2O3 content.
Fig. 7 Effect of Al2O3 content on (a) fluorescence lifetime (τrad); (b) effective line-width (Δλeff); and (c) emission cross section (σems) of PKAi (i = 3, 6, 9, 12, 15) glasses.
3.3 Quantitative structure-property relationship
We used JMP Statistical Discovery Software version 9.0 (SAS Institute, Cary, NC) to establish the QSPR models. PLS technique was applied to model the glass structures – properties relationships of PKAi (i = 3, 6, 9, 12, 15) glasses. During models building, cross - validation was completed by using the leave-one-out technique to ensure the accuracy of the models [23]. The predicted residual sum of squares (PRESS) value was used to determine the number of factors, where the number of factors is determined by the minimum PRESS value [26]. PRESS can be calculated by the following equation.
where yp,ij are the predicted values, yij are the known values, n is the number of samples, and d is the number of factors.QSPR models were built to extract the complex relationships between properties, Tg, ρ, nc and the composition of the glasses in terms of components (P2O5, Al2O3, K2O) using the PLS method. When PLS regression and leave-one-out prediction were applied, the optimal number of PLS principal components was found to be 2. The results of Tg (as example) can be expressed by the following linear equation:
The regression coefficients (RC) are listed in Table 5 and R2, R2adj, RMSE, PRESS for the three QSPR models are listed in Table 6.
Table 5. Regression Coefficients (RC) of QSPR models for the different components in the glass, and Tg, ρ, and nc values obtained using the PLS method.
Table 6. Coefficients of determination (R2), adjust R2 (R2adj), root mean square error (RMSE) and predicted residual sum of squares (PRESS) of QSPR models for the different components in the glass and the physical properties, Tg, ρ, and nc, by PLS method.
The R2 value (≥ 0.97), R2adj value (≥0.96), and the small RMSE value allow to conclude that the three models are reasonable.
The regression coefficient represents the rate of change of the dependent variables (y) as a function of the independent variables (X). The larger the regression coefficients the greater is the effect of X on y. A positive regression coefficient indicates that y increases with increase in X, and conversely, a negative regression coefficient indicates that y decreases with increase in X. The regression coefficient of Al2O3 is positive, whereas the regression coefficients of the other components (P2O5 and K2O) are negative. This means that the increase in Al2O3 content results in increasing the Tg, ρ, nc of PKAi (i = 3, 6, 9, 12, 15) glasses.
In order to find out the relationships between Tg, ρ, nc properties and the glass structures of PKAi (i = 3, 6, 9, 12, 15) glasses, the structure units derived from Raman spectral analysis were used as descriptors to build QSPR models using the partial least squares (PLS) method. When PLS regression and leave-one-out prediction were applied, the optimal number of PLS principal components was found to 2. The result for Tg (as example) is given by the following linear equation:
The regression coefficients (RC) are listed in Table 7 and R2, R2adj, RMSE, PRESS values for the three QSPR models are listed in Table 8.
Table 7. Regression Coefficients of QSPR models for the vibration frequencies and the relative intensities of Qn tetrahedra, and values of Tg, ρ, nc of the glasses obtained using the PLS method.
Table 8. Coefficients of determination (R2)and adjust R2 (R2adj), root mean square error (RMSE), and predicted residual sum of squares (PRESS) obtained from QSPR models of the vibration frequency and relative intensities of Qn tetrahedra, and the values for Tg, ρ, nc of glasses by PLS method.
The R2 value (≥ 0.98), R2adj value (≥0.97), and the low value of RMSE value allow to conclude that the three models are reasonable.
The regression coefficients of vs(Q21Al) and I(vs(Q21Al)) are positive, but the regression coefficients of the other parameters (Q20Al and Q3 tetrahedra) are negative. This indicates that the glass properties (Tg, ρ and nc) have positive relationships with the vibration frequency and relative intensity Q21Al, but negative relationships with the vibration frequency and relative intensity of Q20Al and Q3. Upon the substitution of Al2O3 for K2O, both P = O and PO-K are consumed to form PO-Al, and the Q21Al content increases as a result. The increase in Tg occurs because the interconnections between Q21Al and other Qn tetrahedra harden the network. The density increases as the phosphate network becomes tighter and the molar volume decreases with increasing Q21Al content. The refractive index has a negative relationship with molecular volume and positive relationship with the ionic polarization. Despite the ionic polarization of K+ is larger than that of Al3+, the increase in nc occurs because molar volume decrease with the increase in Q21Al content.
3.4 Relationships between structure and Nd3+ luminescent properties
Nd3+-emission properties (σems and Δλeff) and fluorescence lifetime (τrad) are related to the glass structure. Q21Al has an important effect on σems, Δλeff and τrad of PKAi (i = 3, 6, 9, 12, 15) glasses. Figure 8 (a-1 ─ b-2) shows the effects of Q21Al on σems, Δλeff and τrad of PKAi (i = 3, 6, 9, 12, 15) glasses. With the increase in vibration frequency and the relative intensity of the Q21Al, the emission cross section (σems) decreases, but the effective line-width (Δλeff) and the fluorescence lifetime (τrad) increase.
Fig. 8 Effects of Q21Al tetrahedra on the (a-1, a-2) fluorescence lifetime (τrad), (b-1, b-2) effective line-width (Δλeff), and (c-1, c-2) emission cross section (σems) of PKAi (i = 3, 6, 9, 12, 15) glasses.
The Q2 possess two NBOs bond to modifies and two BOs bond to neighboring tetrahedra [15,30]. The two NBOs of Q20Al bond to K, but the Q21Al bond to K and Al [9,34]. These Q2 are linked to form phosphate chains or rings. The Q3 possess three BOs bond to neighboring tetrahedra, and one NBO bonds to a terminal oxygen. The Q3 does not form phosphate chains as these Q3 are linked to form a three-dimensional network [15]. Phosphate chains based on Q2 terminated by Q1 as the Q1 possess only one BO bonds to neighboring tetrahedra, and three NBOs bond to modifiers [15]. The structure of the phosphate chains affects the luminescent properties of Nd3+ ions, which could be certified by commercial glasses that have the approximate molar composition near the metaphosphate glass [1]. Thus the Q21Al that forms the phosphate chains or rings is one of the main factors affecting the luminescent properties (σems, Δλeff and τrad) of Nd3+ ions. With the increase in Q21Al content, the coordination environment around the Nd3+ ions becomes more covalent and less symmetrical, as evidenced by the increase in Ω2 value. The increase in Q21Al vibration frequency and relative intensity influence the crystalline field of Nd3+ ions which has a great effect on luminescence properties. The increase in the Q21Al vibration frequency and relative intensity results in a decreased oscillator strength of the 4I9/2→4F5/2 transition and emission band broadening. The decrease in oscillator strength lead to the σems decrease, but Δλeff and τrad increase.
4. Conclusion
Effect of Al2O3 content on the physical (Tg, ρ), optical properties (nc, σems, Δλeff) and τrad of Nd3+: xAl2O3-(40-x) K2O-60P2O5 (x = 3, 6, 9, 12, 15) glasses were systematically investigated. Overall, within the concentration range, the addition of Al2O3 was shown to systematically improve the physical properties of the glass (Tg, ρ) as well as the optical properties (nc, Δλeff) and also τrad, but decreased σems. Raman spectroscopic were utilized to elucidate the changes in the phosphate network structure. A standard deconvolution procedure was adopted to simulate Raman spectra, and the component vibrational bands were resolved and subsequently assigned by matching them with literature data covering phosphate glasses in a wide composition range. Our study shows that, with an increase in the Al2O3 content, more NBOs connect to Al and the Q21Al content increase. Vibrational frequencies and relative intensities of Qn tetrahedra were used as descriptors to build QSPR models using PLS method to study the relationship of the glass structures to its physical properties such as glass transition temperature (Tg), density (ρ) and refractive index (nc). The glass properties (Tg, ρ and nc) have positive relationships with the vibration frequency and relative intensity of Q21Al, but they have negative relationships with the vibration frequency and relative intensity of Q20Al and Q3. Increasing Q21Al content leads to phosphate network tightening and a molar volume’s decrease, which results in an increasing in glass properties (Tg, ρ and nc). The Q21Al tetrahedra that form the phosphate chains or rings is one of the main factors affecting luminescent properties (σems, Δλeff and τrad) of Nd3+ ions. The σems has negative relationships with the vibration frequency and relative intensity of the Q21Al, whereas the Δλeff and τrad have positive relationships with the vibration frequency and relative intensity of the Q21Al.
References and links
1. J. H. Campbell and T. I. Suratwala, “Nd-doped phosphate glasses for high-energy/high-peak-power lasers,” J. Non-Cryst. Solids 263(23), 318–341 (2000). [CrossRef]
2. C. Wang, H. Wei, J. Wang, D. Huang, W. Fan, and X. Li, “1 J, 1 Hz lamp-pumped high-gain Nd: phosphate glass laser amplifier,” Chin. Opt. Lett. 15(1), 011401 (2017). [CrossRef]
3. I. Konidakis, C. P. E. Varsamis, and E. I. Kamitsos, “Effect of synthesis method on the structure and properties of AgPO3-based glasses,” J. Non-Cryst. Solids 357(14), 2684–2689 (2011). [CrossRef]
4. D. Palles, I. Konidakis, C. P. E. Varsamis, and E. I. Kamitsos, “Vibrational spectroscopic and bond valence study of structure and bonding in Al2O3-containing Agl-AgPO3 glasses,” Rsc. Adv 6(20), 16697–16710 (2016). [CrossRef]
5. N. S. Tagiara, D. Palles, E. D. Simandiras, and E. I. Kamitsos, “Synthesis, thermal and structural properties of pure TeO2 glass and zinc-tellurite glasses,” J. Non-Cryst. Solids 457, 116–125 (2017). [CrossRef]
6. D. Möncke, E. I. Kamitsos, D. Palles, R. Limbach, A. Winterstein-Beckmann, T. Honma, Z. Yao, T. Rouxel, and L. Wondraczek, “Transition and post-transition metal ions in borate glasses: Borate ligand speciation, cluster formation, and their effect on glass transition and mechanical properties,” J. Chem. Phys. 145(12), 124501 (2016). [CrossRef] [PubMed]
7. R. K. Brow, “Nature of Alumina in Phosphate Glass: I, Properties of Sodium Aluminophosphate Glass,” J. Am. Ceram. Soc. 76(4), 913–918 (1993). [CrossRef]
8. R. K. Brow, R. J. Kirkpatrick, and G. L. Turner, “Nature of Alumina in Phosphate Glass: II, Structure of Sodium Alurninophosphate Glass,” J. Am. Ceram. Soc. 76(4), 919–928 (1993). [CrossRef]
9. J. Schneider, S. L. Oliveira, L. A. O. Nunes, and H. Panepucci, “Local structure of sodium aluminum metaphosphate glasses,” J. Am. Ceram. Soc. 86(2), 317–324 (2003). [CrossRef]
10. H. Toratani, PhD thesis, Kyoto University, Kyoto,Japan, 1989.
11. A. R. Kuznetsov, S. G. Lunter, S. I. Nikitina, A. G. Plyokhin, and Y. K. Fedorov, “Properties of neodymium-doped aluminophosphate laser glasses,” J. Appl. Spectrosc. 56(1), 68–72 (1992). [CrossRef]
12. Y. S. Jiang, S. B. Jiang, and Y. Y. Jiang, “Spectral properfiles of Nd3+ in aluminophosphate glassws,” J. Non-Cryst. Solids 112(1-3), 286–290 (1989). [CrossRef]
13. J. S. Hayden and S. A. Payne, “Laser and thermophysical properties of Nd-doped phosphate glasses,” Proc. SPIE 1761, 162–173 (1993). [CrossRef]
14. Y. N. Ch, Ȧ. Ravibabu, and S. V. G. V. A. PrasadḂ., “Effect of Al2O3 on Optical Absorption Properties of Heavy Metal Oxide Glass System doped with Nd3+ Ion,” Int. J. Eng. Sci. 4, 2558–2594 (2014).
15. R. K. Brow, “Review: the structure of simple phosphate glasses,” J. Non-Cryst. Solids 263(28), 1–28 (2000). [CrossRef]
16. L. Zhang and R. K. Brow, “A Raman study of iron-phosphate crystalline compounds and glasses,” J. Am. Ceram. Soc. 94(9), 3123–3130 (2011). [CrossRef]
17. A. K. Yadav and P. Singh, “A review of the structures of oxide glasses by Raman spectroscopy,” RSC Advances 5(26), 67583–67609 (2015). [CrossRef]
18. T. Le, V. C. Epa, F. R. Burden, and D. A. Winkler, “Quantitative structure-property relationship modeling of diverse materials properties,” Chem. Rev. 112(5), 2889–2919 (2012). [CrossRef] [PubMed]
19. M. Karelson, V. S. Lobanov, and A. R. Katritzky, “Quantum-chemical descriptors in QSAR/QSPR studies,” Chem. Rev. 96(3), 1027–1044 (1996). [CrossRef] [PubMed]
20. A. R. Katritzky, M. Kuanar, S. Slavov, C. D. Hall, M. Karelson, I. Kahn, and D. A. Dobchev, “Quantitative correlation of physical and chemical properties with chemical structure: utility for prediction,” Chem. Rev. 110(10), 5714–5789 (2010). [CrossRef] [PubMed]
21. D. Guo, Y. Wang, C. Nan, L. Li, and J. Xia, “Application of artificial neural network technique to the formulation design of dielectric ceramics,” Sens. Actuators A Phys. 102(5), 93–98 (2002). [CrossRef]
22. D. Guo, Y. Wang, C. Nan, J. Xia, and L. Li, “Investigation of BaTiO3 formulation: an artificial neural network (ANN) method,” J. Am. Ceram. Soc. 22(5), 1867–1872 (2002). [CrossRef]
23. Z. Daren, “QSPR studies of PCBs by the combination of genetic algorithms and PLS analysis,” Comput. Chem. 25(2), 197–204 (2001). [CrossRef] [PubMed]
24. M. González-Durruthy, L. C. Alberici, C. Curti, Z. Naal, D. T. Atique-Sawazaki, J. M. Vázquez-Naya, H. González-Díaz, C. R. Munteanu, T. A. S. David, M. V. N. José, G. D. Humberto, and R. M. Cristian, “Experimental–computational study of carbon nanotube effects on mitochondrial respiration: in silico nano-QSPR machine learning models based on new Raman spectra transform with Markov–Shannon entropy invariants,” J. Chem. Inf. Model. 57(5), 1029–1044 (2017). [CrossRef] [PubMed]
25. S. Yin, Z. Shuai, and Y. Wang, “A quantitative structure-property relationship study of the glass transition temperature of OLED materials,” J. Chem. Inf. Comput. Sci. 43(3), 970–977 (2003). [CrossRef] [PubMed]
26. N. Tavassoli, Z. Chen, A. Bain, L. Melo, D. Chen, and E. R. Grant, “Template-oriented genetic algorithm feature selection of analyte wavelets in the Raman spectrum of a complex mixture,” Anal. Chem. 86(21), 10591–10599 (2014). [CrossRef] [PubMed]
27. Y. Zheng, X. Zhu, Z. Wang, Z. Hu, F. Gao, R. Nie, X. Cui, J. She, and B. Peng, “Noninvasive blood glucose detection using a miniature wearable Raman spectroscopy system,” Chin. Opt. Lett. 15(8), 083001 (2017). [CrossRef]
28. K. Meyer, “Characterization of the structure of binary zinc ultraphosphate glasses by infrared and Raman spectroscopy,” J. Non-Cryst. Solids 209(3), 227–239 (1997). [CrossRef]
29. B. N. Nelson and G. J. Exarhos, “Vibrational spectroscopy of cationsite interactions in phosphate glasses,” J. Chem. Phys. 71(7), 2739–2746 (1979). [CrossRef]
30. L. L. Velli, C.-P. E. Varsamis, E. I. Kamitsos, D. Möncke, and D. Ehrt, “Structural investigation of metaphosphate glasses,” Phys. Chem. Glasses 46(3), 178–181 (2005).
31. D. A. Magdas, R. Stefan, D. Toloman, and N. S. Vedeanu, “Copper ions influence on lead-phosphate glass network,” J. Mol. Struct. 1056(4), 314–318 (2014). [CrossRef]
32. E. Metwalli and R. K. Brow, “Modifier effects on the properties and structures of aluminophosphate glasses,” J. Non-Cryst. Solids 289(9), 113–122 (2001). [CrossRef]
33. G. Le Saout, P. Simon, F. Fayon, A. Blin, and Y. Vaills, “Raman and infrared study of (PbO) x (P2O5)(1− x) glasses,” J. Raman Spectrosc. 33(6), 740–746 (2002).
34. J. Tsuchida, J. Schneider, M. T. Rinke, and H. Eckert, “Structure of ternary aluminum metaphosphate glasses,” J. Phys. Chem. C 115(44), 21927–21941 (2011). [CrossRef]
35. S. Tanabe, T. Hanada, T. Ohyagi, and N. Soga, “Correlation between 151Eu Mössbauer isomer shift and Judd-Ofelt Ω6 parameters of Nd 3+ ions in phosphate and silicate laser glasses,” Phys. Rev. B Condens. Matter 48(14), 10591–10594 (1993). [CrossRef] [PubMed]
36. F. Qi, F. Huang, T. Wang, R. Lei, J. Zhang, S. Xu, and L. Zhang, “Influence of Tm 3+ ions on the amplification of Ho3+: 5 I 7→ 5 I 8 transition in fluoride glass modified by Al (PO3)3 for applications in mid-infrared optics,” Chin. Opt. Lett. 15(5), 051604 (2017). [CrossRef]
37. W. F. Krupke, “Optical absorption and fluorescence intensities in several rare-earth-doped Y2O3 and LaF3 single crystals,” Phys. Rev. 145(1), 325–337 (1966). [CrossRef]
38. A. Novais, N. O. Dantas, I. Guedes, and M. Vermelho, “Spectroscopic properties of highly Nd-doped lead phosphate glass,” J. Alloys Compd. 648(7), 338–345 (2015). [CrossRef]
