High and low temperature syntheses of AgI–AgPO3 glasses: Structural and thermal studies

Journal of Non-Crystalline Solids 415 (2015) 51–56

Contents lists available at ScienceDirect

Journal of Non-Crystalline Solids journal homepage: www.elsevier.com/ locate/ jnoncrysol

High and low temperature syntheses of AgI–AgPO3 glasses: Structural and thermal studies P. Grabowski ⁎, J.L. Nowinski, K. Kwatek Faculty of Physics, Warsaw University of Technology, Koszykowa 75, 00-662 Warsaw, Poland

a r t i c l e

i n f o

Article history: Received 6 November 2014 Received in revised form 22 January 2015 Accepted 24 January 2015 Available online 2 March 2015 Keywords: Silver phosphate; Silver iodide; Glass transition; Mechanosynthesis; MDSC

a b s t r a c t In this work we studied the AgI–AgPO3 and AgPO3 glasses obtained in high and low temperature syntheses. The mechanosynthesis and melt-quenching methods were employed to prepare the amorphous materials. Samples were investigated by means of X-ray diffraction, modulated differential scanning calorimetry and Raman spectroscopy. Despite the identical overall chemical composition of the glasses, we noticed different values of glass transition, non-reversing enthalpy and unequal phase composition in elevated temperatures. The work discusses differences in the glass network and distribution of its components induced by different energy transfer mechanisms of each synthesis method. © 2015 Elsevier B.V. All rights reserved.

1. Introduction The phenomenon of superionic conductivity of glasses attracted the attention of researchers for several decades but, despite the intensive studies, still some issues require clarifications. For instance, a fundamental question whether the method of preparation affects properties of formed glass and possibly to what extent, appears in the light of the recent reports rather complex. Superionic glasses, like silver ion conducting ones, are synthesized mainly in high temperature routes (MQ), which involve melting of the constituents at high temperatures [1–4]. A different way of preparation is offered by the low temperature synthesis routes performed by the mechanosynthesis method (MS) during which the starting materials are agitated by ball-milling to react. The process is usually carried out at room temperature [5–7]. In literature, two materials of the same chemical overall composition, first prepared by the MQ and the second one by the MS method, usually are regarded as the same. Such attitude is based on experimental evidences, developed mainly by X-ray and ionic conductivity which show no significant differences between properties of these materials. For instance, in Ref. [8], the case of the glass of the overall formula 50AgI–30Ag2O–20V2O5 was studied. The material was prepared by the MS and MQ methods. In the latter one, the two different cooling rates were applied: a slow one and an ultra-fast one. The X-ray, differential scanning calorimetry (DSC), impedance spectroscopy, FTiR and Raman spectroscopy methods were used in the investigations. The study concluded that the application of substantially different preparation ⁎ Corresponding author. E-mail address: [email protected] (P. Grabowski).

http://dx.doi.org/10.1016/j.jnoncrysol.2015.01.018 0022-3093/© 2015 Elsevier B.V. All rights reserved.

methods resulted in the formation of materials of very similar properties. However, on the other hand, there are reports pointing, that the method of preparation of the silver ion conducting glasses influences the properties of the final product. Mustarelli, Konidakis and Novita studied properties of the amorphous silver metaphosphate (AgPO3) prepared by MQ method, using constituents of various wetness and considering other process parameters [9–11]. They observed the significant differences of a glass transition temperature Tg among the investigated materials and related those to the modifications of the glass structure by implemented water molecules; namely formation of short (PO4)n chains and rings at the expense of the long chains. Although a direct comparison of Refs. [8] and [11] leads to ambiguous or even opposite conclusions, one should take into account the fundamental differences of the glass structure between these two materials. Whereas, in the 50AgI–30Ag2O–20V2O5 glass short (VO4)n chains dominate, the AgPO3 glass structure consists of long, almost infinite (PO4)n chains, when the material is prepared at a very dry environment [11]. Therefore in fact, the comparison suggests that the glasses which structure is composed of long chains of (MeO4)n tetrahedrons should be more susceptible for structural modifications than those with the short ones. Moreover, one could expect that not only the level of wetness of the starting materials is responsible for these modifications but also the way in which a final glass is formed. The work aims to give more experimental evidences supporting that hypothesis with respect to preparation routes. It presents the case of the materials formed within the AgI–Ag2O–P2O5, [Ag2O/P2O5] = 1 glass system which were synthesized in various ways. Generally three groups of the materials are studied: i) prepared by high temperature method (MQ), ii) prepared by low temperature method, i.e., mechanosynthesis

52

P. Grabowski et al. / Journal of Non-Crystalline Solids 415 (2015) 51–56

(MS) and iii) those formed first by the MQ and then ball-milled (MQ + MS). All technology operations were executed in an open air without any special precautions against moisture. Such approach was adopted to provide the same wetness conditions during preparation and to expose the role of the used methods themselves. It is worthy to add that, according to our knowledge, the materials of the AgI–Ag2O–P2 O5, [Ag2 O/P2O5 ] = 1 system prepared by mechanosynthesis were not presented in the literature. 2. Experimental Investigated materials were formed in high (MQ) and/or low temperature processes (MS). The high temperature synthesis was applied for preparation of both: amorphous AgPO3 (material labeled as AMQ) and AgI–AgPO3 glasses. The former material was synthesized from NH4H2PO4 and AgNO3 substrates. The AgI–AgPO3 glasses were prepared using AgI, NH4H2PO4 and AgNO3 reagents (BMQ sample) or AgI and the amorphous AgPO3 prepared earlier — CMQ material. The next steps of the preparation procedure were the same for all MQ materials. The starting materials taken directly from the commercially packed containers, after weighting, were mixed in appropriate molar proportions and ground with a mortar and a pestle. The reagents were not pre-heated or specially dried. All operations were executed on laboratory benches in an open air without any special precautions against moisture. The melting of the substrates was carried out in an open vertical furnace and it consisted of three stages. First, the batch was annealed from room temperature to about 300 °C and kept until the ingredients melted and volatile products of the first stage of synthesis were released to the atmosphere. Then, the temperature was raised up to about 750 °C. During this annealing some weak bubbling of the melt was still observed, indicating that the synthesis process was still in progress. After the bubbling had stopped, which marked the completion of the synthesis, the melt was additionally annealed for 15–20 min and then it was poured out between two stainless steel plates and quickly quenched. As a result 0.5–0.9 mm thick glassy plates were formed. The low temperature syntheses were carried out by means of Fritsch Pulverisette P7 planetary ball mill. The reagents were located in 45 ml Si3N4 vials. Three Si3N4 grinding balls were used and the total mass of the reagents in a vial was to keep the ball-to-powder mass ratio equal to 10:1. The mill operated at 600 rpm rotation speed for 6 h in a constant work mode. Applying the procedures described above, the amorphous AgPO3 and the materials of an overall formula x · AgI − (1 − x) · AgPO3 for x = 33, 40, 50, 55, 60, 66 and 80 were prepared. To distinguish materials of the same overall formula but prepared by different methods, the given labels will be used. Table 1 presents the list of the materials, labels and selected preparation details. The as-received materials were examined by means of X-ray powder diffraction (XRD) employing Phillips X'Pert Pro diffractometer set up in a Bragg–Brentano configuration with filtered Cu Kα radiation. High temperature X-ray measurements (HT XRD) were carried out with assistance of an Anton Paar oven. Thermal properties were investigated by means of modulated differential scanning calorimetry method (MDSC)

using Thermal Analysis TA Q2000 calorimeter. The device operated in a heat-flow mode within a temperature range from 0 °C to 220 °C. Modulated component of temperature oscillated with 1 °C amplitude and 3 °C period. For Raman measurements, samples were characterized by the Renishaw InVia System, using 633 nm circularly polarized laser: the power was adjusted to avoid heating of the samples. The error values provided for the measured quantities were determined according to the manufacturer's guide of the TA Q2000 and Renishaw InVia System including the random errors measured as the standard deviation of the average value of determined quantities relevant to the: glass transition temperature determination onset points; enthalpy integration area bounded by inflection points of heat flow signal; scattering strength ratios of Raman modes as related to the integrated area; and modes' detectability. 3. Results 3.1. XRD The room temperature XRD investigations of the DMS materials (x · AgI − (1 − x) · AgPO3, x = 33, 40, 50, 55, 60, 66, 80) revealed, that the materials x b 55 were entirely amorphous, whereas for x ≥ 55, the final products contained, apart from the amorphous phase, a crystalline one, identified as the β/γ-AgI (Fig. 1). A similar dependence between a phase composition and an overall concentration of silver iodide added to the material has been reported in the literature for the x · AgI − (1 − x) · AgPO3 glasses obtained by the MQ method [1,2]. Therefore, results of our room temperature XRD investigations, confronted with the literature data did not indicate any substantial differences between the materials prepared by the low and high temperature methods. For further, more detailed study, to search for the plausible differences among the materials synthesized by various routes, the glasses of the 50AgI–25Ag2O–25P2O5 overall chemical composition were selected. These materials (BMQ, BMQ + MS, CMQ, CMQ + MS and DMS) were prepared by low or high temperature processes and combination of the both procedures, as the labels indicate. All of them were entirely amorphous, without presence of any crystalline phases, despite, for related samples, “dissolving” high amounts of AgI in the AgPO3 matrix. The HT XRD investigations performed in a 25–200 °C temperature range for the materials from the A, B and C groups revealed no diffraction peaks on the patterns both during heating and cooling runs. A different observation was recorded for the DMS material — Fig. 2 collects the relevant XRD patterns. During heating, when temperature did not exceed 80 °C, the material was amorphous, as the X-ray patterns indicated. Above, near 85 °C weak intensities attributed to the γ-AgI crystalline phase were recorded. At about 125 °C they were accompanied by the lines of the β-AgI phase. At the temperature of 140 °C, the β/γ-AgI crystalline precipitates transformed to the α-AgI phase. During cooling at 90 °C the γ-AgI lines reappeared again. The recorded temperature of phase transition is slightly lower than the one with characteristic of pure AgI which, in normal conditions for pure AgI, occurs at 147 °C [12]. The high temperature α-AgI phase is a superionic material exhibiting high conductivity: 1.3 S/cm. This value,

Table 1 Composition, starting materials, preparation method, glass transition temperature (Tg), Rc and Rr modes and (ΔH)exo and (ΔH)nr enthalpies of investigated A–D materials. The N/A stands for “not applicable”, MQ for “melt-quenching” and MQ + MS for “melt-quenching and mechanosynthesis”. Label

Composition

Starting materials

Preparation method

Tg/°C

Rc

Rr

(ΔH)exo/J/g

(ΔH)nr/J/g

AMQ AMQ + MS BMQ BMQ + MS CMQ CMQ + MS DMS

AgPO3 AgPO3 50AgI–50AgPO3 50AgI–50AgPO3 50AgI–25Ag2O–25P2O5 50AgI–25Ag2O–25P2O5 50AgI–50AgPO3

NH4H2PO4, AgNO3 NH4H2PO4, AgNO3 AgI, AgPO3 AgI, AgPO3 AgI, NH4H2PO4, AgNO3 AgI, NH4H2PO4, AgNO3 AgI, AgPO3

MQ MQ + MS MQ MQ + MS MQ MQ + MS MS

177 (2) 153 (2) 88 (2) 85 (2) 88 (2) 85 (2) 75 (2)

1.23 (2) 1.77 (2) 1.75 (2) 1.83 (2) 1.75 (2) 1.83 (2) 1.90 (2)

N/A 6.7 (1) N/A 6.3 (1) N/A 6.3 (1) 1.8 (1)

0.14 (3) 0.12 (3) 0.55 (3) 0.41 (3) 0.55 (3) 0.41 (3) 0.30 (3)

0.76 (3) 0.61 (3) 0.33 (3) 0.13 (3) 0.33 (3) 0.13 (3) 0.24 (3)

P. Grabowski et al. / Journal of Non-Crystalline Solids 415 (2015) 51–56

53

Fig. 1. X-ray patterns of x · AgI − (1 − x) · AgPO3 (where x = 33, 40, 50, 55, 60, 66, 80 mol%) materials formed by means of mechanosynthesis.

is five to six orders of magnitude greater when compared to low temperature β/γ-AgI phases [13]. The significant conductivity increase is a result of structure change: the low temperature phases of AgI show a close packing of iodide ions in zinc blende or wurtzite arrangement for β and γ phases respectively whereas for the high temperature α phase, the iodide ions form a bcc structure allowing the mobile silver ions to occupy 42 available sites [14,15].

Fig. 3. MDSC traces of a) AMQ, b) AMQ + MS, c) BMQ/CMQ, d) BMQ + MS/CMQ + MS, and e) DMS materials. For details see text and Table 1. Black line presents the reversible signal whereas the magenta colored one is for non-reversing one.

3.2. MDSC In the analysis of the MDSC data, a recorded total heat flow signal was separated into reversing and nonreversing heat flows to facilitate detection and distinction of thermal events induced by the heat energy delivered to the material. A typical MDSC trace is seen in Fig. 3a. The shape of the reversing signal curve near 180 °C is characteristic of a glass transition. On the relevant nonreversing signal curve, the glass transition manifests itself as an enthalpy recovery endothermic peak. Generally, both peaks appear together and they hallmark the occurrence of a glass transition event. The nonreversing curves reveal also exothermic processes preceding the (ΔH)nr [16,17]. The presence of similar exotherms was reported by Tomasi et al. for the fast-quenched

Fig. 2. X-ray patterns of 50AgI–50AgPO3 mechanosynthesized material (DMS) recorded in heating and cooling cycles.

AgPO3 glasses, however for the slow-cooled ones no such peak was recorded [18]. In addition, Avramov et al. described these exothermic processes as the initial part of the transition region where the heat capacity deviates measurably from its glassy state frozen value [19]. The integrated peak area under the curve will be labeled in the text as (ΔH)exo. Fig. 3 collects the heat flow traces recorded from all investigated materials. The black lines mark reversing heat flows whereas the colored ones represent the non-reversing heat flows, both plotted as functions of temperature. The MDSC reversible heat flow traces recorded for the AMQ and AMQ + MS materials reveal only a glass transition event. However, for the as-prepared AMQ material, the glass transition occurs at Tg = 177 °C, whereas for the additionally ball-milled material (AMQ + MS) Tg is lower by 22 °C. The signals of BMQ and CMQ or BMQ + MS and CMQ + MS materials are alike, thus in Fig. 3c and d, only one curve is shown. The glass transition for the BMQ and CMQ occurred at Tg = 88 °C, regardless of different starting materials used in their preparation. When the materials were additionally ball-milled (BMQ + MS and CMQ + MS) the values of glass transition temperature were detected at lower temperatures only by 3 °C. The non-reversing signal shows exothermic reaction for AMQ sample occurring at around 135 °C with the enthalpy of the process (ΔH)exo = 0.14 J/g. For subsequent processes, the nonreversing enthalpy, which assists the glass transition, (ΔH)nr value equals 0.48 J/g. For the ballmilled material, the AMQ + MS sample, the exotherm appears at 126 °C with enthalpy of (ΔH)exo = 0.12 J/g while the (ΔH)nr of later endothermic event is 0.44 J/g. In the case of materials with silver iodide (B, C samples) the exothermal reaction occurred more intense and appeared in lower temperatures around 54 °C. Further we recorded values for BMQ/CMQ material, where the (ΔH)exo = 0.55 J/g whereas for BMQ + MS/CMQ + MS the (ΔH)exo equals 0.41 J/g. For endothermic reaction the (ΔH)nr values equal 0.29 J/g and 0.14 J/g respectively.

54

P. Grabowski et al. / Journal of Non-Crystalline Solids 415 (2015) 51–56

The MDSC traces for the DMS material are more complex. The reversing curve indicates only a glass transition event at 75 °C. On the nonreversing signal, the endothermic peak attributed to the glass transition appeared, with enthalpy of (ΔH)nr = 0.31 J/g. Other thermal events were also noticed: prior to (ΔH)nr exothermic one around 53 °C (with (ΔH)exo = 0.3 J/g), very broad exothermic one peaking at 110 °C, and an endothermic with minimum at 130 °C. To maintain the clarity of results, Table 1 collects values of investigated observables for all samples. Confrontation of the MDSC results with the HT XRD ones facilitates attribution of the observed thermal events to changes of a microstructure of the material while in thermal treatment. The investigations of A, B and C materials (MQ or MQ + MS) did not detail any phase changes within the 70–80 °C temperature range, supporting the interpretation postulated by Tomasi and Avramov. In the case of DMS material, the exothermic peak at 110 °C was identified as the γ-AgI crystallization, the endothermic one at 130 °C was ascribed to the γ-AgI to β-AgI phase transition and subsequent high temperature phase transition namely β-AgI → α-AgI.

3.3. Raman The Raman spectra of the investigated materials are shown in Fig. 4. For the amorphous AgPO3 (AMQ) the strong vibration modes at 681 cm−1 and 1140 cm−1 are observed accompanied with two broad and weak modes located at 916 cm−1 and 1231 cm−1 (Fig. 4a). The Raman spectrum of the ball-milled amorphous AgPO3 (AMQ + MS) reveals additional vibration modes at 766 cm−1 and 1023 cm− 1 (Fig. 4b). The Raman spectra of the BMQ and CMQ, shown in Fig. 4c, are both practically the same and they are similar to the AMQ one showing only the vibration modes at 681 cm−1, 911 cm−1, 1135 cm− 1 and 1230 cm−1 respectively (Fig. 4c). A noticeable shift of the 1140 cm−1 peak position to 1135 cm−1 was recorded (Fig. 4c) [20]. The ball-milled BMQ + MS and CMQ + MS materials exhibit, like the AMQ + MS, modes at 681 cm−1, 766 cm−1, 916 cm−1, 1023 cm−1, 1135 cm−1 and

1231 cm−1. The same vibration modes are visible on the Raman spectrum recorded for the DMS material (Fig. 4d). The observed modes were assigned to various vibrations of phosphorous and oxygen atoms inside (PO4)n chains [21,22]. The strong 681 cm−1 scattering peak was related to a P–Obr symmetric vibration mode of a long (P–Obr–P)n chain. The second, very intensive mode at 1140 cm−1 or at 1135 cm−1 for the materials containing AgI, was attributed to a P–Ot symmetric vibration of a P and a terminal oxygen Ot in a PO4 tetrahedron. The remaining detected modes were connected respectively to: the 766 cm−1 peak with a symmetric vibration of the P–Obr occurring for small (PO4)n rings, the 916 cm−1 with asymmetric P–Obr vibrations of long chains, the 1023 cm− 1 with an asymmetric P–Obr vibration of small (PO4)n rings and the 1231 cm−1 with an asymmetric P–Ot vibration of long chains. A visualization of tetrahedron long chain and small ring structures is shown in Fig. 4f and g respectively. One can notice that the relative intensity of selected vibration modes varies from material to material. To compare the obtained Raman spectra, we introduced the scattering strength ratios Rc and Rr. The first one, Rc, was defined as an area of the 1140 cm−1 (or 1135 cm−1) peak to an area of the 681 cm− 1 mode, whereas the Rr as the 1140 cm−1 to 1023 cm−1. Thus, the Rc can be interpreted as a quantity which characterizes a length of a (PO4)n chain. For an infinite length, the Rc value approaches unity. Such value determined Novita et al. for the AgPO3 glass to be prepared in a very dry environment [11]. When shorter chains are comprising a glass structure the Rc value becomes higher, more and more with increasing number of these chains. The second introduced quantity Rr relates to the number of small rings in a glass structure. When long chains are dominating the Rr value is very high, practically infinite. It adopts lower and lower value with increasing number of the short rings existing in the glass. For the melt-quenched AgPO3 glass the Rc = 1.23. After additional ball-milling of the glass, the Rc value increased to 1.77. The Rr for the as-quenched AgPO3 glass is practically non-determinable because the 1023 cm−1 mode is not visible on Raman scattering. However, for the ball-milled glass the scattering strength ratio is Rr = 6.7. The BMQ and CMQ materials show, in some respect, similar characteristics like the AMQ glass, i.e., non-determinable Rr value. However, their Rc values equal 1.77 are considerably higher than that for the AMQ. Ball-milling of these materials resulted in the increase of the value of scattering strength ratios to Rc = 1.83 and decrease of Rr to 6.3. Therefore, it is clearly seen that ball-milling was applied to the as-quenched glasses: the AgPO3 or 50AgI–50AgPO3, regardless of the starting materials used in preparation, generally causes the increase of Rc accompanied with the decrease of Rr values. It is worthy to emphasize, that the DMS material demonstrates the highest Rc and the lowest Rr values: 1.9 and 1.8 respectively. The determined values of the scattering strength ratio for all materials are collected and shown in Table 1. 4. Discussion

Fig. 4. Raman spectra of a) AMQ, b) AMQ + MS, c) BMQ/CMQ, d) BMQ + MS/CMQ + MS, and e) DMS materials. Vertical dashed lines represent positions of long chains (black lines) and short rings (magenta lines) with the respective vibrational modes labeled beside each line. For details see text and Table 1. Below are the provided visualizations of tetrahedron structures of f) long chains and g) small rings.

All collected experimental evidences, presented above, clearly indicate that the low and high temperature syntheses both produce the AgI–Ag2O–P2O5, [P2O5]/[Ag2O] = 1 amorphous, i.e., glassy, materials, however, each of different properties. These manifest as different values of glass transition temperature, (ΔH)nr, and also are reflected in the Raman spectra. To explain the cause of the differences it is necessary to refer to the conclusions withdrawn from the analysis of the Raman spectra. They univocally point out to a structure of a glass network as the key factor responsible for the observed differences. In the glasses prepared by means of the high temperature route, long (PO4)n chains compose predominantly a structure of a glass network, whereas short (PO4)n chains and (PO4)n rings form the glass network of the glasses synthesized at low temperature. Because both processes were conducted at different temperatures, the value of temperature, low or high, at the first look, could be regarded as the factor determining the structure of the glass network. It is obvious

P. Grabowski et al. / Journal of Non-Crystalline Solids 415 (2015) 51–56

that the formation of a glass from substrates requires energy delivered to the reactants when a synthesis is conducted at high temperature that is heat energy. However, at low temperature, the amount of the heat energy is insufficient to proceed the reaction, therefore the process has to be driven by another form of energy — a mechanical one. Therefore, the discussed syntheses are supported by various mechanisms of energy transfer. Therefore not the temperature but rather the mechanisms of energy transfer should be regarded as the true ‘spiritus movens’ of the structural differences of the glass networks, and in consequence, of the different properties of the glasses, despite their identical overall chemical composition. During the low temperature synthesis, the mechanical energy is transferred to the reactants via the ball-milling. Such process involves crushing, crumbling and grinding of the processed materials and finally leads to a chemical reaction — the mechanosynthesis. The question is: how these side effects accompanying and simultaneously facilitating the mechanosynthesis affect a structure of a glass network of the formed glass? Each process puts its own specific constrains on the substrates which could (or not) be used in the synthesis of the glass. The high temperature one allows to use various starting materials. They can be a direct source of required groups of atoms and ions for the building of the glass network or an indirect source after earlier decomposition during annealing of the melt. However, the high temperature rules out application of the compounds which during annealing could decompose forming some unwanted chemicals. Therefore, in the case of synthesis of the AgI–Ag2O–P2O5 glasses, the AgI, AgNO3 and NH4H2PO4 can be used (ternary system) or alternatively the AgI and AgPO3 (pseudobinary system). The AgNO3 and NH4H2PO4 decompose during annealing supplying structural units for formation of the (PO4)n chains. Thoughtfully possible substitution of the AgNO3 by Ag2O and NH4H2PO4 by P2O5 is rather not recommended: the former one decomposes producing metallic silver whereas the latter one is highly hygroscopic and its application requires a rigorous protection against moisture during synthesis. On the contrary, the low temperature process takes place in a solid state. Therefore, the reagents should be a direct source of the structural units. Therefore, in the case of the AgI–Ag2O–P2O5 glasses, the AgI, Ag2O and P2O5 can be used for the synthesis. However, taking into account the mentioned complications and troubles caused by phosphorous pentoxide, it is more reasonable to use AgPO3 instead of the mixture of Ag2O and P2O5. Such way was adopted in this work. Additionally, apart from these reasons, the approach offered opportunity for comparative study of the high and low temperature synthesized glasses both prepared from the same reagents. Therefore, considering the reasons causing the formation of different glass network structures during synthesis, one should take into account two aspects: i) the impact of the side effects of the ball-milling and ii) the influence of chemical composition of the substrates on the formed amorphous materials. Therefore, referring to the first aspect, the as-prepared AgPO3 and 50AgI–25Ag2O–25P2O5 glasses were subjected to the additional ballmilling. It was expected, that such treatment of material in which the reagents already have reacted, could expose the impact of the ball-milling itself. Indeed, all ball-milled glasses show different properties with respect to the unprocessed materials, those demonstrated as lowering of the Tg, increase of the Rc and decrease of the Rr vibration mode ratios (Table 1). Such results indicate that the structure of the glasses network was modified by the milling. Part of the long (PO4)n chains was broken forming rings and shorter chains. The milling affected the AgPO3 glasses the most (A materials) for which the Tg value decreased by 25 °C. Surprisingly, the changes of the glass network for the glasses containing silver and iodine ions (BMQ and CMQ glasses) were relatively modest. The Rc ratio increased from 1.75 to 1.83 and Tg decreased only by 3 °C. In addition, results of MDSC investigations for both: A, B or C materials expose differences for as-received and ball-milled materials. At first, the value of (ΔH)exo refers to the rigidness of the glasses. The observed

55

lowering of the enthalpy of the exothermal process preceding glass transition (see Table 1), points, that the as-received quenched glasses, AMQ and BMQ/CMQ, are highly strained as opposed to the ball-milled materials [18]. Onwards, comparing the values of (ΔH)nr between the MQ and MQ + MS samples, the lower ones were recorded for additionally processed samples. This observation – in the light of connectivity of the network reflected by (ΔH)nr [23] – underpins the conclusions withdrawn from Raman investigations. A number of works characterize the material formation in the mechanosynthesis process. It is described as a constant interplay between welding and fracturing to yield a powder with a refined internal structure [24–26]. The impact forces occurring in collisions plastically deform the powder particles and lead to hardening and fracturing; the shape of grain develops, while new surfaces are being created. These enable mechanosynthesized particles to weld together and to form refined products. One may expect, that fracturing of the AgPO3 or AgI–AgPO3 glass – a mechanical separation of particles – contributes to splitting and shortening of the structural chains. The second aspect refers to the influence of chemical composition of the starting products on formed amorphous materials. In light of this, obtained results revealed other interesting correlations. Although the BMQ and CMQ glasses were prepared from different substrates, AgI, AgPO3 and AgI, and NH4H2PO4, and AgNO3 respectively, our investigation did not indicate any significant differences between these materials. On the contrary, vibration mode ratios Rr and Rc, also Tg adopted the same values. Even after milling – although, as it was mentioned above, the values slightly changed – they were the same for both glasses. This observation emphasizes the role of energy transfer mechanism as fundamental one for discussed system, subjecting the sequence of chemical reagents synthesis. Further, the influence of silver iodide incorporation should be considered. Comparison of the Rc values for the AMQ and BMQ or CMQ, 1.23 and 1.75 respectively, suggested that the glass network of meltquenched glasses without AgI contained a lower fraction of the (PO4)n rings and short chains than those with AgI. All these glasses were formed directly from the melt. Therefore, the silver and iodide ions, in fact, impede formation of the long (PO4)n chains favoring instead formation of the shorter ones or rings. Such behavior is a helpful hint to explain the observed modest change of the Tg and vibration mode ratios for the BMQ or CMQ glasses after milling. Their glass networks formed in melt quenching process contain already high concentration of the rings and short chains, so the additional milling practically does not cause further increase of these structural units in the glass network. The glass prepared via the mechanosynthesis method (DMS) distinguishes itself from the other ones studied in our work. For DMS, the measurable quantities Tg and Rr adopt the lowest values, and the Rc the highest one. It means, that the fraction of the (PO4)n ring and short chains inside the glass network is the highest when compared with the remaining investigated glasses, no matter the MQ or MQ + BM processed. Moreover, the (ΔH)nr value was lower than for all MQ materials, yet higher than for MQ + MS B or C samples. Lastly, the (ΔH)exo adopted the lowest value relating to other samples containing AgI. To explain such results it is worthy to refer to the discussion above and its preliminary conclusions. According to it, formation of the rings and short chains is promoted by the two factors: the ball-milling itself and Ag+ and I− ions incorporation into the glass network structure. In the case of the process applied for fabrication of the mechanosynthesized glasses, silver and iodine ions were inserted into the AgPO3 network composed of long (PO4)n chains. The insertion was conducted by the ball-milling process — mechanoinsertion [27]. Therefore, both factors were involved in the formation of the glass. It is very likely that some interferences between both factors – the ball milling and AgI dissolvement – play some role, enhancing the breakage degree of the long chains, more than it happens, when the insertion and milling are executed consecutively.

56

P. Grabowski et al. / Journal of Non-Crystalline Solids 415 (2015) 51–56

One may attempt to explain this phenomenon by the distribution manner of AgI in the material. The issue was previously discussed in the literature with regard to the high ionic conductivities of AgI-based glasses, for which a cluster level inhomogeneity has been pointed out [28–30]. The model suggests that Ag+ and I− ions are introduced into interstices of the glass network to form α-AgI microclusters. Moreover, the AgI rich parts in such glasses were pointed out to be expected to grow with increasing AgI concentration [31]. While the AgI amount reaches up to 75 mol%, the cluster level AgI-rich parts most likely grow in order to form AgI-rich amorphous particles about 40–60 nm in diameter. It is very likely that, these regions crystallize as the α-AgI in the rapidly quenched glasses [31–34]. However, only for the DMS that we observed a different phase composition in elevated temperatures. The wide temperature range of β-AgI crystallization (ΔT = 45 °C) peaked at 110 °C whereas, at the same temperatures, no thermal events were recorded for other materials — neither for the MQ samples nor for MQ + MS samples. This observation unanimously points to AgI inhomogeneous distribution in the material, though, a different one than presented in Refs. [28–30]. To bring more details, we will refer to the work of Peng et al. [35], which reports AgI–Ag2PO3.5 systems formed by means of mechanosynthesis. For all synthesized products the XRD method revealed only Ag4P2O7 or Ag5IP2O7 phases while no lines ascribed to AgI were detected. The value of ionic conductivity at 300 K appeared to be a function of milling time. The highest values were observed for samples milled for a longer time, up to 20 h. The authors correlated the recorded conductivity value with the amount of dissolved AgI and sketched the chemical reactions progress at the interfaces of reacting grains. Therefore, the conclusions of the studies showed, that despite the AgI remaining undetected in XRD measurements, it was not entirely incorporated to the glassy matrix and the amount of dissolved AgI was found to be a function of milling time. In our case the DMS material was obtained in a 12 h ball-milling process. In the literature [36,37], there are some other reports supporting the idea of surface reaction model of AgI in the ball-milling process given in Ref. [35]. Hence, the differences in the values of Tg, Rr and Rc or even (ΔH)nr revealed from comparing the DMS to A or B, C (MQ or MQ + MS) samples, emphasize contrasting structure of the mechanosynthesized material. Relating the above to the reaction processes described in Ref. [35], it is likely that the DMS sample, despite no XRD evidence, contains AgI in weakly bonded amorphous volumes or microclusters on the surface of synthesized, in some extent, AgI–AgPO3 amorphous phase. Moreover, the crystallization occurring at 110 °C, also votes for this interpretation: after providing sufficient heat energy, the AgI agglomerates detach and form crystalline phase recorded in XRD measurements. 5. Conclusions We investigated structural and thermal properties of AgPO3 based glasses formed in high and low temperature synthesis processes: melt-quenching and mechanosynthesis. Both methods lead to formation of amorphous products. However, obtained materials differed in temperatures of glass transition, values of nonreversing enthalpy and intensities of detected on Raman spectra vibrational modes. Contrasting properties of the glasses – despite their identical overall chemical composition – are caused by different energy transfer mechanisms, which play a fundamental role in determination of the glass network. As

observed for the melt-quenched AgPO3 or AgI–AgPO3 glasses, the process of ball-milling itself induces structure disordering: mechanical treatment resulted in shortening of long (PO4)n chains and initiated production of small (PO4)n rings. Likewise, dissolving AgI in AgPO3 glass lowers the glass connectivity. In the case of mechanosynthesized AgI–AgPO3 material both factors are involved in formation of the glass. Moreover, the sample exhibited the lowest Tg value and crystallized after heating up. Discussed evidences, point to a non-uniform distribution of AgI in the mechanosynthesized glass, distinctly variant than the one adopted for melt-quenched AgI–AgPO3 materials. Acknowledgments This work is partially sponsored by the European Social Fund in Human Capital Programme. It is a pleasure to thank Mariusz Miśkiewicz for having prepared the tetrahedron structures' visualization. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37]

T. Minami, K. Imazawa, M. Tanaka, J. Non-Cryst. Solids 42 (1980) 469–476. T. Minami, Y. Takuma, M. Tanaka, J. Electrochem. Soc. 124 (1977) 1659. T. Minami, M. Tanaka, J. Solid State Chem. 32 (1980) 51–55. J. Kuwano, Solid State Ionics 40/41 (1990) 696–699. J.L. Nowiński, P. Grabowski, J.E. Garbarczyk, M. Wasiucionek, Solid State Ionics 188 (2011) 86–89. A. Dalvi, A.M. Awasthi, S. Bharadwaj, K. Shahi, J. Phys. Chem. Solids 66 (2005) 783–792. A. Dalvi, A.M. Awasthi, S. Bharadwaj, K. Shahi, Mater. Sci. Eng. B103 (2003) 162–169. A. Czajkowska, J.L. Nowiński, M. Mroczkowska, J.E. Garbarczyk, M. Wasiucionek, Solid State Ionics 188 (2011) 94–98. P. Mustarelli, C. Tomasi, A. Magistris, S. Scotti, J. Non-Cryst. Solids 163 (1999) 97–103. I. Konidakis, C.P.E. Varsamis, E.I. Kamitsos, J. Non-Cryst. Solids 357 (2011) 2684–2689. D.I. Novita, P. Boolchand, Phys. Rev. B Condens. Matter 76 (2007) 184205. C. Tubandt, E. Lorenz, Z. Phys. Chem. B24 (1914) 513–543. K.H. Lieser, Z. Phys. Chem. 9 (1956) 302. L.W. Strock, Z. Phys. Chem. B25 (1934) 441. K. Funke, R.D. Banhatti, P. Grabowski, J. Nowinski, W. Wrobel, R. Dinnebier, O. Magdysyuk, Solid State Ionics SOSI-13477 (2014) 1–8. Leonard C. Thomas, Modulated DSC #2, T.A. Instruments Inc. www.TAInstruments. com2007. Leonard C. Thomas, Modulated DSC #5, T.A. Instruments Inc. www.TAInstruments. com2007. C. Tomasi, P. Mustarelli, A. Magistris, Phys. Chem. Glasses 36 (1995) 136. I. Avramov, I. Gutzow, J. Non-Cryst. Solids 104 (1988) 148. J. Malugani, R. Mercier, Solid State Ionics 13 (1984) 293–299. L.L. Velli, C.P.E. Varsamis, E.I. Kamitsos, Phys. Chem. Glasses 46 (2) (2005) 178–181. E.I. Kamitsos, J.A. Kapoutsis, G.D. Chryssikos, J.M. Hutchinson, A.J. Pappin, M.D. Ingram, J.A. Duffy, Phys. Chem. Glasses 36 (1995) 141. D.I. Novita, P. Boolchand, Phys. Rev. Lett. 98 (2007) 195501. B.J.M. Aikin, T.H. Courtney, Metall. Trans. A 24A (1993) 647. D. Maurice, T.H. Courtney, Metall. Mater. Trans. A 21A (1990) 01. C. Suryayanrayana, Prog. Mater. Sci. 46 (2001) 1. J.L. Nowinski, Solid State Ionics 179 (2008) 93. L. Borjesson, W.S. Howells, Solid State Ionics 40–41 (1990) 702. E.I. Kamitsos, J.A. Kapoutsis, G.D. Chryssikos, J.M. Hutchinson, A.J. Pappin, M.D. Ingram, J.A. Duffy, Phys. Chem. Glasses 36 (1995) 141. T. Minami, M. Tanaka, J. Non-Cryst. Solids 38–39 (1980) 289. M. Hanaya, M. Nakayama, A. Hatate, M. Oguni, Phys. Rev. B 52 (1995) 3234. M. Tatsumisago, Y. Shinkuma, T. Minami, Nature 354 (1991) 217. M. Tatsumisago, N. Torata, T. Saito, T. Minami, J. Non-Cryst. Solids 196 (1996). M. Tatsumisago, T. Saito, T. Minami, J. Non-Cryst. Solids 293–295 (2001) 10. H. Peng, N. Machida, S. Nishida, T. Shigematsu, J. Non-Cryst. Solids 318 (2003) 112–120. A. Dalvi, K. Shahi, J. Non-Cryst. Solids 341 (2004) 124–132. N. Machida, S. Nishida, T. Shigematsu, H. Sakai, M. Tatsumisago, T. Minami, Solid State Ionics 136–137 (2000) 381–386.