Self-Diffusion of Lithium in LiAlSi2O6 Glasses Studied Using Mass

ARTICLE pubs.acs.org/JPCA

Self-Diffusion of Lithium in LiAlSi2O6 Glasses Studied Using Mass Spectrometry A.-M. Welsch,*,† H. Behrens,†,‡ I. Horn,† S. Roß,† and P. Heitjans‡,§ †

Institut f€ur Mineralogie, ‡ZFM - Zentrum f€ur Festk€orperchemie und Neue Materialien, and §Institut f€ur Physikalische Chemie und Elektrochemie, Leibniz Universit€at Hannover, Hannover, Germany D-30167 ABSTRACT: In order to improve our understanding of the transport mechanisms of lithium in glasses, we have performed diffusion and ionic conductivity studies on spodumene composition (LiAlSi2O6) glasses. In diffusion couple experiments pairs of chemically identical glasses with different lithium isotopy (natural Li vs pure 7Li) were processed at temperatures between 482 and 732 K. Profiles of lithium isotopes were measured after the diffusion runs innovatively applying femtosecond UV laser ablation combined with inductively coupled plasma mass spectrometry (LA ICP-MS). Self-diffusion coefficients of lithium in the glasses were determined by fitting the isotope profiles. During some of the diffusion experiments the electrical conductivity of the samples was intermittently measured by impedance spectrometry. Combining ionic conductivity and self-diffusivity yields a temperature-independent correlation factor of ∼0.50, indicating that motions of Li ions are strongly correlated in this type of glasses. Lithium self-diffusivity in LiAlSi2O6 glass was found to be very similar to that in lithium silicate glasses although Raman spectroscopy demonstrates structural differences between these glasses; that is, the aluminosilicate is completely polymerized while the lithium silicate glasses contain large fractions of nonbridging oxygen.

1. INTRODUCTION Lithium aluminosilicate glasses belong to the group of fast ionic conducting glasses, which makes them possible materials for various technological applications including solid electrolytes in high performance batteries. In order to optimize the efficiency of such ion conductors, it is of considerable importance to identify the mechanisms which control the ionic transport in glass.1 Ionic mobility in different oxide glass systems has been intensively studied in the past, resulting in different, although often complementary, mechanistic models proposed.2
5 Among these, a model to describe alkali-ionic transport through silicate glass network has been proposed by Greaves and Ngai6 according to which mobile alkali ions move through the structure by network hopping, i.e., places with lower Coulomb barrier facilitate alkali ion diffusion. To which extent such a model could realistically be applied to aluminosilicate glasses where Li is the only alkali present remains to be experimentally established. So far the nature of glass-network pathways and possible channel-like ionic transport in such materials are matters of an ongoing debate.7,8 In the past, most of the experimental information of Li dynamics in silicate glasses, ceramic, and crystal media was acquired using impedance spectroscopy (IS)9
11 and nuclear magnetic resonance (NMR).12
17 Studies of the lithium diffusivity by classical self-diffusion experiments are rare due to the lack of suitable radioactive tracer isotope of lithium. Hence, stable isotopes (6Li and 7Li) have to be used to monitor ionic propagation through solid media and mass spectrometry has been the analytical method of choice. So far this method of studying self-diffusion of lithium in silicate glasses has been applied only in several studies by Frischat and co-workers.18,19 More recently, molecular dynamics simulations have been used as a basis for theoretical modeling of lithium transport through vitreous silicate networks, supporting the experimental results.20
24 r 2011 American Chemical Society

In order to improve our understanding on dynamics and mobility of lithium in glasses, we performed an experimental study on Li self-diffusion in spodumene-composition glass (LiAlSi2O6). In these glasses AlO4
and SiO4 tetrahedra build an aperiodic three-dimensional network while the charge imbalance is compensated by Li ions, the only mobile particles in this system. Due to the absence of any other mobile cation the lithium mobility is unobstructed by mixed-alkali effects.25
27 Dynamics of Li in LiAlSi2O6 glass has been studied already by impedance and NMR spectroscopy. The activation energy Ea of Li-ion conductivity in LiAlSi2O6 glass9,10,28
30 agrees well with Ea values calculated from the results of NMR spin
lattice relaxation14,16 and NMR spin-alignment echo measurements.12 In the present study we performed diffusion couple experiments with the same chemical composition but different Li isotopy in both halves. Li isotope profiles were measured by femtosecond pulsed UV laser ablation combined with inductively coupled plasma mass spectroscopy (ICP-MS).31 Using short pulses as in the femtosecond laser (compared to nanosecond lasers) avoids the problem of sample surface melting or boiling, and isotope fractionation during evaporation.31
34 As shown recently, the UV fs laser ablation ICP-MS technique is suitable for reliable concentrations analyses of light isotopes such as lithium and boron.35 In some of the diffusion experiments, we have intermittently measured the electric conductivity using IS in order to analyze correlation effects during the motion of lithium ions. Raman spectroscopy was used for local structural characterization of pre- and postexperimental glasses. Received: September 27, 2011 Revised: November 21, 2011 Published: November 21, 2011 309

dx.doi.org/10.1021/jp209319b | J. Phys. Chem. A 2012, 116, 309–318

The Journal of Physical Chemistry A

ARTICLE

Table 1. Experimental Conditions and Diffusion Dataa sample LDI12

Teff (K)

td (s)

teff (s)

DLi,IE (m2/s)

482

250256

263852

3.54  10
14 ( 6.70  10
15

DLi,σ (m2/s)

HR/f

7.54  10
14

0.47

9360

3.06  10


14

6.37  10
13

0.48

50400c

3.26  10
13 ( 4.69  10
14

6.37  10
13

0.51

2.94  10
13 ( 4.25  10
14

6.37  10
13

0.46


13

( 5.26  10

LDI6

553

8220

LD9b

553

50400

LD9b (2)

553

LD7

640

6861

6960

2.71  10
12 ( 3.00  10
13

4.56  10
12

0.59

LDI15

688

3518

10664

6.22  10
12 ( 2.88  10
13

1.09  10
11

0.57

LDI15 (2)

688

6.42  10
12 ( 2.94  10
13

1.09  10
11

0.59

LDI14 LDI14 (2)

693 693

4240

6810

4.66  10
12 ( 1.33  10
13 4.50  10
12 ( 1.24  10
13

1.19  10
11 1.19  10
11

0.38 0.38

LD8b

732

941

1072

1.22  10
11 ( 1.32  10
12

2.20  10
11

0.55

2.20  10
11

0.57

b

LD8 (2)

1.25  10

732


11


12

( 1.35  10

a

(2) refers to second profiling on the same sample. Teff is the average temperature during the high temperature dwell. td is the time the sample spent at maximum Teff ( 5 K. teff is the effective time at Teff calculated by eq 1. DLi,σ is calculated from eq 4 using the DC conductivity data of Kuhn et al.28 HR/f is the correlation factor calculated as DLiσ/DLiIE. b Not tested for conductivity. c Rapid heating and cooling, no time correction made.

2. MATERIAL AND SAMPLE PREPARATION Our starting material were two glasses with spodumene (LiAlSi2O6) composition prepared by the Schott Glaswerke AG through melting of stoichiometric high purity mixtures of Li2CO3, Al2O3 and SiO2 and subsequent roller-quenching. One half of the diffusion couple contained lithium with the natural isotopic composition (natLi: 7.42% 6Li, 92.58% 7Li), while the second half was essentially pure 7Li (>99%). Glass plates with plane- parallel surfaces were prepared ca. 3  3  1 mm3 in size and well polished on the side of contact. For experiments with contemporaneous measurements of electric conductivity a circular gold electrode (diameter ∼2 mm) was sputtered on the outer side of each glass plate, which was only roughly polished, using an Edwards SCANCOAT six sputter device (conditions: voltage of 1.3 kV, current of 35 mA, duration of 180 s). In order to improve the contact between the glass plates in the diffusion couple, a thin layer of lithium nitrate (melting temperature of 533 K36) with natural isotopic composition was inserted between both plates. The LiNO3 interface layer was produced in two different ways. In the first experiments (LDI6 and LDI12, see Table 1) about 0.1 mg of LiNO3 was deposited on one glass plate and the plate was rapidly heated to ∼573 K to melt the nitrate. Next the second plate was placed on the first one, and the assemblage was cooled within a few seconds to room temperature. The total annealing time was less than 0.5 min. For the high-temperature diffusion experiments (LDI14, LDI15) we chose another pretreatment to avoid significant annealing of samples. Crystals of LiNO3 3 3H2O were produced by adding a droplet of distilled water to a couple of milligrams of dry LiNO3. After evaporation of excess water clear crystals of lithium nitrate trihydrate were obtained. A small crystal (,1 mg) was placed on one of the glass plates and melted using a hair dryer (melting point of LiNO3 3 3H2O = 303 K36) before the second plate was pressed onto the first one.

room temperature in a tube furnace Nabertherm R50/500/13 which has excellent temperature stability in the center, where the sample was positioned (gradients