1
Structure and Mechanical Properties of Compressed Sodium
Aluminosilicate Glasses: Role of Non-Bridging Oxygens
Tobias K. Bechgaard
1
, Ashutosh Goel
2
, Randall E. Youngman
3
, John C. Mauro
3
, Sylwester J.
Rzoska
4
, Michal Bockowski
4
, Lars R. Jensen
5
, Morten M. Smedskjaer
1,*
1
Department of Chemistry and Bioscience, Aalborg University, Aalborg, Denmark
2
Department of Materials Science and Engineering, Rutgers, The State University of New Jersey,
Piscataway, NJ, USA
3
Science and Technology Division, Corning Incorporated, Corning, USA
4
Institute of High-Pressure Physics, Polish Academy of Sciences, Warsaw, Poland
5
Department of Mechanical and Manufacturing Engineering, Aalborg University, Aalborg, Denmark
* Corresponding author. e-mail: mos@bio.aau.dk
Abstract: Clarifying the effect of pressure on the structure of aluminosilicate glasses is important for
understanding the densification mechanism of these materials under pressure and the corresponding
changes in macroscopic properties. In this study, we examine changes in density, network structure,
indentation hardness, and crack resistance of sodium aluminosilicate glasses with varying Al/Si ratio
and thus non-bridging oxygen (NBO) content before and after 1 GPa isostatic compression at elevated
temperature. With increasing NBO content, the silicate network depolymerizes, resulting in higher
atomic packing density, lower hardness, and higher crack resistance. The ability of the glasses to
densify under isostatic compression is higher in the high-NBO glasses and these glasses also exhibit
more pronounced pressure-induced changes in mechanical properties. The
27
Al NMR data show a
surprising presence of five-fold aluminum in the as-made high-NBO glasses, with additional
formation upon compression. Our study therefore provides new insights into the complicated
relationship between Al coordination and NBO content in aluminosilicate glasses and how it affects
their densification behavior.
Keywords: aluminosilicate glass, structure, non-bridging oxygen, pressure, indentation
© 2016. This manuscript version is made available under the Elsevier user license
http://www.elsevier.com/open-access/userlicense/1.0/
2
1. Introduction
The relationship between structure and properties for sodium aluminosilicate (Na
2
O-Al
2
O
3
-SiO
2
)
glasses and glass-forming liquids with varying thermodynamic variables (e.g., composition,
temperature, and pressure) are important for both industrial and geological processes. These glasses
are commercially used for various products, such as flat panel display glass, scratch resistant cover
glass, and nuclear waste glass. The three oxides constitute more than 80% of andesitic and granitic
magmatic systems and therefore also have important implications for magma dynamics and
properties.
Several structural models have proposed that for Al/Na ratio ≤ 1, i.e., excess Na
+
ions, all Al
3+
is
found in tetrahedral configuration (Al
IV
) [1-4]. Addition of Al
2
O
3
to an alkali silicate glass ideally
leads to the removal of the network-modifying Na
+
ions from their original role in the network until
no more non-bridging oxygen (NBO) atoms remain [5]. For Al/Na ratio ≥ 1, some Al
3+
ions can no
longer be charge balanced in tetrahedral configuration and some excess Al
3+
is forced into higher
coordination number (five-fold Al
V
or six-fold Al
VI
) as a means of charge-balancing additional Al
tetrahedra [6-8]. An alternative hypothesis is that Al
IV
can be incorporated, even in peraluminous
compositions, without the need for a charge-balancing cation through association with a three-
coordinated oxygen (oxygen tricluster) [9-11]. In any case, it is well accepted that a range of
macroscopic properties (e.g., transport and mechanical) depend on the Al/Na ratio and thus the
network connectivity [12-15], i.e., the fraction of NBOs.
Clarifying the effect of pressure on the structure of aluminosilicate glasses is important for
understanding the densification mechanism of these materials under pressure and the corresponding
changes in macroscopic properties. However, high-pressure experiments are challenging, partly due to
the typical small sample volumes [16], prohibiting characterization of post-compression properties. In
this work, we investigate sodium aluminosilicate glasses quenched under isostatic pressure from the
glass transition temperature in a nitrogen gas pressure chamber. Although this approach is relatively
modest in both temperature and pressure (~T
g
and ≤1 GPa), it permits permanent densification of
relatively large glass pieces (cm
2
) that are suitable for characterization of, e.g., mechanical properties
3
[17,18]. This is because permanent densification of glass occurs at significantly lower pressures for
compression at elevated temperature compared to that at room temperature [19,20].
Pressure-induced structural changes of aluminosilicates are manifested by changes in both short-
and intermediate-range structure, including increase in the local coordination numbers of the network
forming Al and Si cations from 4 to 5 and 6, enabling closer packing of the structural units. This
coordination number change has been reported to involve conversion of NBO to bridging oxygen
(BO) [21-24], but in the absence of NBOs, it can also occur through the formation of oxygen
triclusters [25,26]. The densification mechanism of sodium aluminosilicate has also been suggested to
include decrease of Na-O bond distances [23,27,28], decrease of inter-tetrahedral bond angles [29,30],
decrease of average ring size [31], and increase in distribution of Si-O and Al-O bond lengths [30].
In this work, we study the structure and micromechanical properties of a series of Na
2
O-Al
2
O
3
-
SiO
2
glasses before and after isostatic compression at elevated temperature. Specifically we study the
influence of NBOs on the changes in structure and properties by varying the Al/Si ratio at constant
Na
2
O concentration. Our results provide insight into the composition-dependent structural changes
that facilitate densification of the aluminosilicate network during compression and the consequence
for the micromechanical properties. Improved understanding of the link among structure and
mechanical properties is important due to the need for more scratch-resistant and mechanically
durable glasses to enable new advanced glass applications.
2. Experimental Section
2.1 Sample preparation
We have prepared six glasses in the (75-x)SiO
2
-xAl
2
O
3
-25Na
2
O system with x = 0, 5, 10, 15, 20, and
25. In this series, the [Al
2
O
3
]/[Na
2
O] ratio is ≤ 1 and therefore prevailing models of network structure
in peralkaline glasses would indicate sufficient charge-compensating Na
+
ions to keep all Al
3+
in
tetrahedral configuration. High purity powders of SiO
2
(Alfa Aesar; >99.5%), Na
2
CO
3
(Sigma
Aldrich; >99%), and Al
2
O
3
(Sigma Aldrich; ≥99%) were used for glass melting. Homogeneous
mixtures of batches (corresponding to ~70 g of oxides), obtained by ball milling, were melted in Pt-
4
Rh crucibles at 1650 ºC for 2 h in air. The melts were poured on a metallic table and were initially
annealed at 600 ºC for 1 h. The chemical compositions of the glasses were determined using flame
emission spectroscopy and inductively coupled plasma mass spectroscopy. The results are given in
Table 1. To ensure uniform thermal history, the glasses were annealed at their respective glass
transition temperature (T
g
) for ~2 h. T
g
was determined using differential scanning calorimetry (DSC
449C, Netzsch) at 10 K/min (Table 1). The glasses were cut to dimensions of about 10 × 10 × 8 mm
3
and polished to an optical finish.
The six glass compositions were isostatically compressed at 0.5 and 1.0 GPa at their respective
ambient pressure T
g
value (see Table 1) in a gas pressure reactor with nitrogen as the compression
medium. The system was kept at the high-pressure/high-temperature condition for 30 min before
cooling to room temperature at 60 K/min, followed by decompression at room temperature at 30
MPa/min. The setup used for this pressure treatment has been described in detail in Ref. [17]. X-ray
diffraction analyses showed no evidence of crystallization following the pressure treatment.
2.2 Density
The density values of the as-prepared and compressed glass samples were determined using
Archimedes’ principle with ethanol as the immersion medium. The weight of each glass sample in
both air and ethanol was measured ten times.
2.3 Indentation
Vickers hardness (H
V
) and crack resistance (CR) of as-prepared and isostatically compressed glasses
were measured using a Vickers micro-indenter (Duramin 5, Struers A/S). The measurements were
performed in air at room temperature with a dwell time of 15 s. Thirty indentations at each load (0.49,
0.98, 1.96, 2.94, 4.91, 9.81, and 19.6 N) were performed. H
V
was calculated at 0.98 N from the length
of the indentation diagonals. CR was determined as the load leading to an average of two radial cracks
per indent [35].
5
2.4 Raman Spectroscopy
Raman scattering spectra were measured in backscattering geometry with a Renishaw Invia Raman
microscope on freshly polished samples. A diode laser with a wavelength of 532 nm was used as the
excitation source. The collected Raman spectra were baseline-corrected using an asymmetric least
square algorithm [32]. Afterwards the processed spectra were deconvoluted using Fityk software with
Gaussian and Voigt lineshapes.
2.5
27
Al NMR Spectroscopy
27
Al magic-angle spinning (MAS) nuclear magnetic resonance (NMR) and triple quantum magic-
angle spinning (3QMAS) NMR experiments on both as-prepared and compressed (1.0 GPa)
aluminosilicate glasses were conducted at 16.4T using a commercial spectrometer (VNMRs, Agilent)
and a 1.6mm MAS NMR probe (Agilent) with spinning speeds of 25 kHz. MAS NMR data were
acquired using radio frequency pulses of 0.6 µs (equivalent to a π/12 tip angle), relaxation delays of 2
s, and signal averaging of 1000 acquisitions. MAS NMR data were processed using commercial
software, without additional apodization and referenced to aqueous aluminum nitrate at 0.0 ppm. A
weak background signal from the zirconia MAS rotors was detected by
27
Al MAS NMR of an empty
rotor and subsequently subtracted from the MAS NMR data of the glass samples. This signal, at
approximately 16 ppm, is clearly distinct from the Al peaks in the glasses, but nonetheless has been
removed to ensure higher accuracy in the
27
Al MAS NMR experiments. Unfortunately, this weak
zirconia signal cannot be removed from
27
Al 3QMAS NMR data, and appears in some of the spectra
as a weak set of contours around 16 ppm in the MAS NMR dimension.
MQMAS NMR spectra were measured using the three pulse, zero quantum filtering method
[33].
The hard 3π/2 and π/2 pulse widths were calibrated to 1.8 and 0.7 s, and the soft reading pulse of the
z-filter was optimized to 10 μs. 48 scans were collected for each of 88 t
1
points, using a recycle delay
of 1 s. Spectra were processed using commercial software (VNMRJ, Agilent) and modest line
broadening (100 Hz) was used in processing the
27
Al 3QMAS NMR data. For each resonance in the
3QMAS NMR spectra, the centers of gravity in the MAS and isotropic dimensions,
CG
2
and
CG
iso
,