Ionics (2006) 12:315–322 DOI 10.1007/s11581-006-0054-y
ORIGINAL PAPER
Conductivity studies on LiX–Li2S–Sb2S3–P2S5 (X=LiI or Li3PO4) glassy system Z. Nagamedianova & A. Hernández & E. Sánchez
Received: 21 June 2006 / Revised: 6 October 2006 / Accepted: 9 October 2006 / Published online: 7 November 2006 # Springer-Verlag 2006
Abstract New sulfide glasses in the Li2S–Sb2S3–P2S5 system have been prepared by classical quenching technique where glassy domain remains up to 50% molar addition of Li2S and electrical conductivities have been determined by impedance spectroscopy. Room temperature DC conductivity vs Li2S content exhibits two regions implying different conductivity mechanisms. The compositions of low lithium content presented low electronic conductivities close to 0.01 μS/cm at room temperature (due to Sb2S3 semiconducting properties). The compositions of medium lithium content could result to mixed ionic–electronic conductors with predominant ionic conductivity with a maximum close to 1 μS/cm; Arrhenius behavior is found between 25 °C and Tg for all glasses, but activation energy is found to be somehow above most similar systems. A comparative study with glasses belonging to the other chalcogenide systems has been undertaken and values of the decoupling index are reported, and in order to validate conductivity data, a circuit equivalent circuit was proposed and fitted parameters were calculated with good agreement. Keywords Ionic conductivity . Electronic conductivity . Sulfide glasses . Impedance spectroscopy . Circuit equivalent models
Z. Nagamedianova : A. Hernández : E. Sánchez (*) Laboratorio de Investigación del Vidrio, Facultad de Ciencias Químicas, Universidad Autónoma de Nuevo León, Cd. Universitaria, San Nicolás de los Garza, N.L 66450, Mexico e-mail:
[email protected]
Introduction Sulfide glasses have been widely studied in recent times due to their great potential in lithium rechargeable batteries [1] and optoelectronic applications [2]. Single ion-conductor glasses consist [3] of a network former (mostly oxides such as SiO2, B2O3, GeO2, P2O5 or sulfides like SiS2, B2S3, P2S5, etc.); a network modifier as M2X (where M=Li, Ag, or Tl and X=O or S), and in some cases a doping salt (lithium halides, LiX, or lithium oxy-salts such as Li3PO4, Li2SO4, etc.). Several compositions had been found to be excellent lithium-ion conductors, for example, the classic system Li2S–P2S5–LiI [4] and some of the newest as Li3PO4–Li2S–SiS2 [5], which are both used in commercially marketed solid state batteries [6]. We can also mention systems like Li2S–GeS2 [7–9], Li2S–P2S5 [10– 12], Li2S–B2S3–LiI [13], etc. with very high ionic conductivities (σ=10−4–10−3 S/cm) at room temperature. Until now, however, there are several drawbacks in processing lithium sulfide glasses like the extremely hygroscopic nature of Li2S, an undesirable devitrification, or poor electrochemical stability towards Li-metal anode [6]. For these reasons, search for new lithium sulfide glasses is appealing and we present recent results of conductivities in the Li2S–Sb2S3–P2S5 sulfide glassy system doped with LiI and Li3PO4. Also, a comparative study and an equivalent circuit model are proposed to validate conductivity information.
Experimental description Glasses of the Li2S–Sb2S3–P2S5 ternary system were obtained by a classical quenching technique with the use of evacuated sealed silica glass tubes, the details of which
316
are described previously [14, 15]. They were prepared by mixing, inside a glove box (Omnilab, VAC) with controlled atmosphere (0.30 relative to ternary and ternary-doped series. Nyquist spectra of a single semicircle (Fig. 2a) correspond to samples whose conduction is believed to be electronic because of the absence or small amounts of ion mobile species. The presence of an inclined spike (Fig. 2b) on the low frequency side of the semicircle Fig. 2 Opposite examples of nyquist plots at several temperatures showing different situations according to lithium content: a 0–20% Li2S and b above Li2S 20%
317
is characteristic of blocking charges at the Au electrode– electrolyte interface that may be due to interfacial resistance of electronic species, e.g., a Schottky barrier. However, the presence of higher amounts of Li+ ion suggests an interpretation in terms of a barrier of ion-mobile species. In Fig. 2a, a single arc is attributed to electron transport through the Sb conduction band and accumulation of electronic charges at the solid electrolyte/electrode interface. The analysis of conductivity data (Table 1) for Sb2S3 (both crystalline and amorphous) show low electronic conductivity values but similar to reported data [16]. The addition of 10% of Li2S decreases the specific conductivity of Sb2S3 to almost two orders of magnitude and that could be explained in terms of band gap increment (Eg=1.63 eV for pure crystalline Sb2S3). At this point, we were discouraged on the binary system because of their rather low σ at room temperature, and we started to explore the effect of a third component (P2S5) to add randomness to the system and increase the possibility of available spaces or channels to ion conduction mechanism. The introduction of phosphorus sulfide in the binary system allowed us to increase progressively Li2S content up to 50%, and we observe two different types of Nyquist plot (and hence a change of conductivity behavior) and an increasing conductivity was observed. The temperaturedependent DC conductivity data for this ternary system is displayed in Fig. 3. The composition dependence of conductivity at 25 °C, activation energies and pre-exponential factor are presented in Fig. 3 as inserts and those data suggest the presence of two conductivity regions dependent of lithium content suggesting that the conductivity mechanisms on these two regions are different. The compositions with low lithium content (below 20 mol%) have lower conductivities (which could be attributed to a preferred electronic mechanism). The compositions of medium lithium content (30–50 mol%) could be mixed ionic–electronic conductors with preferred ionic conductivities with a maximum close to 10−6 S/cm for sample with
318
Ionics (2006) 12:315–322
Table 1 Summary of results of electrical characterization of xLi2S−(1−x)[yP2S5 +(1−y)Sb2S3] system
Binary
Ternary
Ternary−doped
Composition
C s 25 per S per cm DC
Ea/eV
Sb2S3, crystalline Sb2S3, glassy 0.10Li2S−0.90Sb2S3 0.1Li2S−0.9(0.4P2S5−0.6Sb2S3) 0.2Li2S−0.8(0.4P2S5−0.6Sb2S3) 0.3Li2S−0.7(0.4P2S5−0.6Sb2S3) 0.4Li2S−0.6(0.4P2S5−0.6Sb2S3) 0.5Li2S−0.5(0.4P2S5−0.6Sb2S3) 0.05LiI−0.95[0.5Li2S−0.5(0.4P2S5−0.6Sb2S3)] 0.10LiI−0.90[0.5Li2S−0.5(0.4P2S5−0.6Sb2S3)] 0.05Li3PO4−0.95[0.5Li2S−0.5(0.4P2S5−0.6Sb2S3)] 0.10Li3PO4−0.90[0.5Li2S−0.5(0.4P2S5−0.6Sb2S3)]
(5.8±0.2)×10−9 (2.7±0.1)×10−9 (7.9±0.2)×10−11 (1.3±0.1)×10−9 (7.6±0.3)×10−9 (6.1±0.2)×10−8 (1.6±0.1)×10−7 (4.2±0.1)×10−7 (4.1±0.1)×10−7 (5.2±0.3)×10−7 (3.7±0.2)×10−7 (8.0±0.4)×10−8
0.66±0.01 0.54±0.02 0.73±0.02 0.64±0.03 0.69±0.02 0.61±0.02 0.61±0.03 0.58±0.02 0.59±0.02 0.55±0.01 0.58±0.02 0.63±0.03
50% Li2S at room temperature. As the Li2S content increases, the Sb–S–Sb bonds are broken and non-bridging sulfur anions appear in the neighboring of the Li+ cations, resulting in an ionic conductivity of lithium cations. An analogous result has been previously observed for the Li2O–TeO2–V2O5 glassy system [17, 18] where a marked Fig. 3 Arrhenius plot for xLi2S−(1−x)[0.4P2S5−0.6Sb2S3], x=0–0.50 glassy system. Inserts shows composition dependence of room temperature dc conductivity (log σDC), activation energy (Ea) and exponential factor ðlog 0 Þ
log s 0 1.89 0.74 1.19 1.37 2.75 2.53 2.61 3.10 3.05 2.10 3.14 3.06
Nyquist Single arc
Single arc Mix Arc+spike Arc+spike arc+spike arc+spike arc+spike arc+spike arc+spike
electronic-to-ionic conductivity switchover was found and explained in terms of optical basicity increase according to Li2O content and favoring the oxidation of V+4 to V+5 and diminishing the amount of polaron conductivity mechanism and favoring ionic conductivity as Li2O went up to higher concentration (and eventually “killing” the electronic
Ionics (2006) 12:315–322
conductivity). In this Sb2S3–P2S5 system, we do not account for a similar effect on the redox glassy state on Li2S addition (being Li2O more basic than Li2S), but instead we analyze our present information displayed in Fig. 3 insets: a simple arc to arc + spike Nyquist plot change, a drastic drop on activation energy and a mild diminution on exponential factor. It is suggested that electronic conductivity remains low (because band gap enhancement, if any, is not easy to detect one on glassy samples) and ionic conductivity becomes the dominant conduction mechanism as Li+ is the most mobile specie present in this ternary system. A comparison study of composition with predominant ionic conductivities (30–50% Li2S) with glasses belonging to the other chalcogenide lithium [4–13] conductors (Fig. 4) shows that the Li2S–Sb2S3–P2S5 glassy system presents the following features: first, DC room temperature conductivities ðs 25 C Þ are comparable to those of the Li2S–As2S3 system (following extrapolation line); second, activation energies (Ea) are situated on the 0.3–0.6 eV interval characteristic of the fast ion conduction; however, the Li2S–Sb2S3–P2S5 glassy system presents the largest values and it is reflected on lower conductivity values. Extrapolating values to higher Li2S content could overcome this situation. However, a large number of attempts were
319
performed but we systematically obtained partially crystallized amorphous samples and their thermal characteristics are discussed on detail [15] and their conductivity characteristics would be a subject of a separate communication in the near future. Third, despite pre-exponential factors, (log s 0 ) are slightly increased with Li2S content as most of the systems do if we consider an ion hopping mechanism explained in terms of weak electrolyte model in these glasses [19]. Analyzing the introduction of dopand on the ternary system, we observe almost no effect of the introduction of LiI on ionic conductivity (Table 1). If we compare our results with other sulfide systems [4–13] doped with LiI (Fig. 5), we observe that the dopant amount is very important to enhance ionic conductivity. We believe that the small amount of introduced LiI in the Li2S–P2S5–Sb2S3 system is the principal factor of small change on conductivity for doped glasses (attempts to introduce higher concentration keeping the glassy nature of samples failed). In the case of glasses doped with Li3PO4 (up to 10 mol%), the ionic conductivity started to decrease slowly (Table 1), which could be explained by the higher tendency of crystallization in this system corroborated by DSC analysis [15]. Decoupling index For new glass conductors, we were interested in the freedom of mobile ions species from glassy network matrix. Angell [20] has quantified the separation of ion mobility
Fig. 4 Comparison of DC room temperature conductivity, activation energy, and exponential factor (log DC , Ea y log σ0 ) for different [4–13] sulfide glassy lithium–ion conductors
Fig. 5 Ionic conductivities of selected [4–13] chalcogenide glasses showing LiI doping effect
320
Ionics (2006) 12:315–322
Table 2 Summary of calculated decoupling index on different sulfide systems Composition
σCD (μS/cm] at 25 °C
Tg/°C
Extrapolated σDC (mS/cm) at Tg
log s DC at Tg
log Rt
Rt
0.3Li2S−0.7[0.4P2S5−0.6Sb2S3] 0.4Li2S−0.6[0.4P2S5−0.6Sb2S3] 0.5Li2S−0.5[0.4P2S5−0.6Sb2S3] 0.05LiI−0.95[0.5Li2S−0.2P2S5−0.3Sb2S3] 0.10LiI−0.90[0.5Li2S−0.2P2S5−0.3Sb2S3] 0.05Li3PO4−0.95[0.5Li2S−0.2P2S5−0.3Sb2S3] 0.10Li3PO4−0.90[0.5Li2S−0.2P2S5−0.3Sb2S3] 0.3Li2S−0.7GeS2 0.4Li2S−0.6 GeS2 0.5Li2S−0.5 GeS2 0.67Li2S−0.33As2S3 0.70Li2S−0.30As2S3 0.75Li2S−0.25As2S3
0.061 0.16 0.42 0.41 0.52 0.37 0.08 0.44 3.2 40.0 4.6 11.0 13.0
221 218 208 207 205 207 214 316 313 310 164 142 124
0.216 0.242 0.524 1.96 0.242 1.07 0.374 79.4 72.1 632.0 1.70 1.58 1.66
−3.666 −3.616 −3.281 −2.707 −3.737 −2.971 −3.427 −1.100 −1.142 −0.199 −2.769 −2.800 −2.781
10.634 10.684 11.019 11.594 10.563 11.329 10.873 13.200 13.158 14.101 11.531 11.500 11.519
4.31×1010 4.83×1010 1.05×1011 3.93×1011 3.66×1010 2.13×1011 7.47×1010 1.58×1013 1.44×1013 1.26×1014 3.39×1011 3.16×1011 3.30×1011
from the network structural movements in terms of a decoupling index, Rt Rt ¼
ts ts
ð2Þ
where τs is the structural relaxation time—which at Tg is approximately 100 s, and t s is the conductivity relaxation time, which refers to the rate of decay of electric field in the solid electrolyte caused by the ion transport process. It is possible to evaluate t s by the expression: ts ¼
"0 "1
ð3Þ
T
g s DC
Parametric adjustment of equivalent circuit model proposed In order to separate bulk conductivity from interfacial response, an equivalent circuit model of three elements in a serial array Relect(RbulkQbulk)(RintQint) was proposed. On this model, Relect represents the internal resistance of the sampler holder and cables. Rbulk corresponds to the electrical resistance of ion conductive paths within glassy bulk and Qbulk to the dispersive energy storage (capacitance) of the bulk. Rint is assigned to the possible redox reactions within the electrode and the solid electrolyte
where ɛ0 is the free space permittivity ð" ¼ 8:854 1014 F =cmÞ, "s is the relative glass permittivity at infinite Tg frequency ("s ~12 for this system), and s DC is the measured DC conductivity at glass transition temperature. This leads to an easier relation: T
g log Rt 14 þ log s DC
ð4Þ
In Table 2, we show values of Angell’s decoupling indexes for Li2S–Sb2S3–P2S5 glassy system (ternary and doped) and some similar systems [4–13]. The studied glasses and most of the common glasses have high Rt values (1010–1014), so the movements of the monovalent cations on these glasses are strongly decoupled from structural relaxations characteristic at glass transition temperatures. The transport numbers of cations in ionic solid glass are usually close to unity (which means a wellconnected glassy network). This is not the case of polymer electrolyte systems (strongly coupled), where cation transport numbers may be as low as 0.2 and Rt 105 to107 [20]. For this reason, interest in the search of new glassy ionic systems remains high. On the other hand, new electrolytes based on ionic liquids [21] are attracting the attention on the development of new lithium secondary batteries.
Fig. 6 Nyquist plot at several temperatures for 0.5Li 2S−0.5 [0.4P2S5−0.6Sb2S3] glassy sample showing fitting data from adjusting of the R(QR)(QR) equivalent circuit model
Ionics (2006) 12:315–322
321
Table 3 Fitting results of circuit model Relect(RbulkQbulk)(RintQint) on the 0.5Li2S−0.2P2S5−0.3Sb2S3 glass T/°C
24 37 63 93 137 a
Relect/Ω
26 26 26 26 26
Bulk properties
Electrode/electrolyte interactions
Fitting parameter
Z0/kΩ
R/kΩ
Q/nS
n
fmax/kHz
C/pF
R/MΩ
Q/μS
n
fmax/mHz
C/μF
χ2
1,715 959 238 44.3 6.5
1,700 950 238 45.5 6.4
0.815 0.655 2.090 3.311 1.023
0.70 0.72 0.68 0.68 0.78
2.0 5.0 15.1 88.0 640.0
50.4 37.5 51.2 44.5 37.7
60 150 10 1.5 0.07
1.572 1.614 3.644 7.31 35.63
0.54 0.55 0.59 0.67 0.55
