Aspects of powder design affecting reactivity and cement composition are discussed.
Two formulations of commercial cement showed bimodal distribution of micro-pores.
X-ray computed tomography revealed that a gel-like phase precipitates in the pores.
Porosity develops by liquid segregation and not by air-entrapment during mixing.
Strength is higher for the cement obtained with less liquid used as filling material.
To characterize the microstructure of two zinc phosphate cement formulations in order to investigate the role of liquid/solid ratio and composition of powder component, on the developed porosity and, consequently, on compressive strength.
X-ray powder diffraction with the Rietveld method was used to study the phase composition of zinc oxide powder and cements. Powder component and cement microstructure were investigated with scanning electron microscopy. Small angle neutron scattering (SANS) and microfocus X-ray computed tomography (XmCT) were together employed to characterize porosity and microstructure of dental cements. Compressive strength tests were performed to evaluate their mechanical performance.
The beneficial effects obtained by the addition of Al, Mg and B to modulate powder reactivity were mitigated by the crystallization of a Zn aluminate phase not involved in the cement setting reaction. Both cements showed spherical pores with a bimodal distribution at the micro/nano-scale. Pores, containing a low density gel-like phase, developed through segregation of liquid during setting. Increasing liquid/solid ratio from 0.378 to 0.571, increased both SANS and XmCT-derived specific surface area (by 56% and 22%, respectively), porosity (XmCT-derived porosity increased from 3.8% to 5.2%), the relative fraction of large pores ≥50 μm, decreased compressive strength from 50 ± 3 MPa to 39 ± 3 MPa, and favored microstructural and compositional inhomogeneities.
Explain aspects of powder design affecting the setting reaction and, in turn, cement performance, to help in optimizing cement formulation. The mechanism behind development of porosity and specific surface area explains mechanical performance, and processes such as erosion and fluoride release/uptake.
Zinc phosphate cements are a class of acid-base cements which find application in dentistry largely as luting or lining agents . ZPC is one of the oldest dental cements, being introduced in the 1880s. Because of the inferior mechanical and biological properties, compared to some of the more recent bioactive restorative materials , the use of ZPC has significantly declined, although it possesses a successful track record, supported by clinical evidences . However, a huge resident population of cemented restorations still exists, making the understanding of structure-limited properties important for the interpretation of their clinical longevity. ZPC is supplied as a liquid component, consisting in a solution of phosphoric acid (45–65%) containing up to about 3% of aluminum and possibly zinc (up to 10%), and a zinc oxide powder. Additions to the solution are aimed at gaining control over the reaction rate and the heat of reaction. Reactivity of the powder is reduced by annealing ZnO at temperatures between 1000 °C and 1400 °C, after mixing it with about 10 wt% of MgO. Sometimes, fluxing agents, like borax, are added to improve the degree of sintering . In analogy with MgO employed in magnesium phosphate cements , the thermal treatment results in ZnO of higher mean crystallite size and particles with smooth surface . With these compositions, the first product of the setting reaction is completely amorphous, and the presence of aluminum has been recognized to play a crucial role in modifying the chemistry of the reaction and stabilize the amorphous fraction .
The study of the setting reaction showed that this first product is an amorphous zinc phosphate hydrate which forms at the surface of the ZnO particles . Crystallization of zinc phosphate is thought to be hindered by the disorder introduced by the polymeric complexes formed when Al is introduced in the phosphoric acid solution or by the entanglement of chains of aluminum phosphate hydrogel embedding the particles . In the latter case, the experimental evidences pointed to an amorphous zinc phosphate with composition Zn 2 P 4 O 12 ·8H 2 O, a cyclophosphate in which some of the water molecules are loosely bound. This was found consistent with the crystallization of hopeite (Zn 3 (PO 4 ) 2 ·2H 2 O) from the amorphous, an occurrence observed in aged commercial cements . The compressive strength of the cement was shown to increase with the increase in the powder fraction up to values above which probably the wetting of ZnO particles is compromised, in analogy with other dental cements . In the clinical practice, the liquid to solid weight ratio ( l/s ) is optimized for the specific applications, therefore, variable amounts of unreacted ZnO can be found in the final cement . It follows that the ZPC can be considered as a composite material and the interaction between ZnO grains and the amorphous matrix has important implications for the longevity of the restoration. In this respect, crack deflection around composite particles, a well-documented toughening mechanism , has been observed in ZPCs .
On the other hand, pores are fracture initiation sites under load , although, in glass materials they are considered more stress concentrators rather than flaws, as for polycrystalline or glass-ceramics . In general, since ZPCs are highly defect-limited, the measure of the porosity is a significant parameter for their characterization. The final porosity of the cement can be strongly affected by the mixing technique adopted and the viscosity of the paste (dependent on l/s ) during preparation . The imperfect homogenization of the powder was found to yield clusters of ZnO particles which develop large irregular pores with diameter D > 5 μm of deleterious nature . In this respect, a distinction should be made between the large pores (>1 μm in size) and the smaller meso-micro-pores (as defined in Ref. ). In fact, the amount of the former has been directly related to the cases of failure of the restorations and, in general, was found to correlate negatively with the mechanical properties . Much less information is available about the latter. Few examples of characterization of porosity in ZPCs are reported , in some of them optical techniques were used , which show obvious limits of representativeness and accessible range in pore size. A technique usually employed to retrieve the pore-size distribution in solids is mercury intrusion porosimetry (MIP). However, MIP possess several drawbacks that can sometimes lead to misleading results ; it is destructive, it relies on the assumption of cylindrical pore shape and it is limited to the detection of open porosity . Reliability of MIP results might be also impaired by microstructural changes occurring in the sample under the vacuum conditions required for the analysis. Substantial loss of chemically bound water in the porosimeter has been observed in magnesium phosphate cements . Furthermore, availability of MIP for the next future is threatened by the progressive mercury phase out ratified by many countries in 2013 .
In any porous material, a further critical descriptive parameter of its microstructure is the S V specific surface area. Being related to the degree of subdivision, it is most sensitive to changes in the smallest microstructural details and controls to a large extent the rate of the chemical and/or physicochemical reactions, therefore the behavior of the material in applications. Dissolution/erosion, take up or release of elements from solution (such as fluoride) are of interest for restorative materials, and they have been shown to be more surface-dependent rather than volume-dependent .
SANS has been recognized to be a valuable tool for the evaluation and quantification of a statistically representative microstructure (and nano-structure) of heterogeneous materials. It has been successfully applied to porous solids, such as ceramics , and, recently, acid-base cements . The microstructure can be proved for a volume defined by the beam spot size (of the order of 5–10 mm) and the sample thickness (up to few mm). The advantages of using neutrons are that, thanks to their high depth of penetration, they can probe the microstructure in the entire volume of sample (e.g. open and closed porosity), the measurement is not perturbing the sample, and no special sample preparation is required.
XmCT is a powerful non-destructive 3-D imaging technique for characterizing the microstructure and morphology of materials. The high contrast and spatial resolution of the obtained images allow for visualization and quantification of porosity and pore-size distribution without any hypothesis on pore geometry. XmCT has been widely used in dental research , however there are few examples in literature concerning the investigation of restorative dental materials . Depending on the experimental setup, SANS allows for gaining access to the porosity from the nano-scale up to about 1 μm, whereas with XmCT, this range is typically shifted towards larger pore sizes, from around 1 μm up to several mm. It follows that the combination of the two techniques, by making accessible a wide range in porosity, greatly enhances our capability of describing sample microstructure.
In this work, the microstructure of two ZPC formulations of the same product, prepared according to manufacturer’s recommendations, has been investigated in order to provide insights into the role of l/s and composition of powder component on the developed porosity and, consequently, on compressive strength. The morphology of ZnO grains and the phase composition of zinc oxide powder and ZPC have been investigated with a combination of analytical techniques, including XRPD with the Rietveld method and SEM. Pore-size distribution, porosity and S V have been derived from the analysis of SANS data. The volume-weighted size distribution of pores has been calculated through a procedure of profile fitting of the curves. Pore structure was visualized, and from the analysis of reconstructed 3-D images of the two ZPC samples, produced from the tomographic projections collected with XmCT, porosity and S V have been calculated. Compressive strength was determined according to the standard EN ISO 9917-1:2007 .
Materials and methods
A commercially obtained ZPC product (Adhesor, SpofaDental a.s., Jičin, Czech Republic; LOT 5412375) was employed to prepare two cements, recommended as filling and fixed bridge bonding materials. They have been prepared 2 weeks after purchase and within the first year out of the five years expiry period. The compositions, labelled Z1 and Z2, with l/s = 0.348 and 0.571, respectively, have been hand mixed on a chilled glass pad with a stainless steel spatula. Both tools were kept in refrigerator at 4 °C for 1 h, then left to heat up in air to reach temperature above the dew point (15–18 °C) and carefully dried to prevent moisture contamination. In line with manufacturer’s instructions, the powder was added to the liquid in five increments (approx. 200 mg each) while mixing. The entire procedure was accomplished at ambient temperature of 22 ± 2 °C and 60 ± 3% relative humidity. Specimens were casted in form of discs (2 mm thick, 10 mm diameter) and cylinders (3.8 mm diameter, about 20 mm height), sealed in plastic bags, and cured at 37 °C for 3 days at 60 ± 3% relative humidity. Discs were employed for XRPD and SANS measurements, as well as for SEM observations, while cylinders for XmCT. ZPC specimens used for compressive strength tests were prepared in accordance with the standard EN ISO 9917-1:2007 . The number of replicates employed for each test and studied material are summarized in Table 1 . The time interval between samples preparation and testing was 3 days for SANS and XmCT, and 1 week for XRPD and SEM. The latter longer time interval was observed also in order to evaluate the potential crystallization of phases like hopeite in the bulk cement.
Samples of ZPC and zinc oxide powder for XRPD data collection were ground by hand in an agate mortar and mounted on an Al sample holder using the side loading technique to minimize a priori preferred orientation of crystallites. Analytical grade ZnO (Merck) was also analyzed for comparison with zinc oxide powder. XRPD data were collected in the angular range 4–82° 2 θ at 40 kV and 40 mA using a Bragg–Brentano θ–θ diffractometer (Bruker D8 Advance, Cu Kα radiation ( λ = 1.5418 Å)), equipped with a LynxEye 1-D silicon strip detector. Divergence 0.6 mm slits and 2.5° Soller slits were mounted on the incident beam pathway. The pathway of the diffracted beam included a Ni filter and Soller slits (2.5°). A virtual step scan of 0.0102° 2 θ , with 0.4 s/step counting time, was employed. Samples were allowed to spin at 15 rpm to improve particle statistics. Quantitative phase analysis (QPA) including both amorphous and crystalline fraction was performed with the Rietveld method by spiking the sample with 10 wt% of internal standard (NIST SRM 676a). Refinements were accomplished with the TOPAS 4.2 software (Bruker AXS).
SEM observations were performed at 20 kV accelerating voltage on the powder and the freshly exposed internal surface of ZPC samples mounted on aluminum stubs and coated with 5 nm thick gold film, employing a FEI QUANTA FEG 450 instrument equipped with an EDAX Apollo X energy dispersive detector, whose window is capable of detecting all chemical elements down to beryllium. EDS analysis was performed on 12 points for each sample.
The sample discs were mounted between two quartz windows for SANS data collection. SANS curves were recorded at the KWS-2 instrument operated by JCNS at the Heinz Maier–Leibnitz Zentrum MLZ (Garching, Germany). The Q range (0.025–3.15 nm −1 ) was covered by merging data collected at wavelength λ = 5.151 Å with sample-to-detector distance 2.23 m and 7.73 m, and at wavelength λ = 10.308 Å with sample-to-detector distance 19.43 m. Time for data collection was 5 min for each Q range. The scattering intensity was obtained as a function of momentum transfer Q :
where 2 θ is the scattering angle. Data were collected with a 2-D detector and radially averaged in order to obtain 1-D intensity patterns. The 2-D raw data have been corrected for beam attenuation (according to measured sample thickness), the scattering from the empty cell, the electronic and background noise. Intensity was calibrated against a plexiglass standard material to set the data to absolute scale. Sample thickness was chosen in order to minimize multiple scattering effects.
Details of the SANS theory may be found elsewhere , here only few concepts are summarized. SANS can be described as the Fourier transform of the fluctuation in scattering length density (a measure of the interaction of neutrons with matter) within the sample. In a porous solid, when the difference in scattering density originates from the boundary between the pores and the surrounding matrix, the single phase approximation is usually made and the system is treated as biphasic . The complex microstructure is usually described assuming pores of spherical shape . According to the Bragg law, the size of detected features in a SANS experiment is inversely proportional to Q , thus, large structures will be visible at small Q . The upper limit in distance periodicity d is thus defined by the lowest Q value accessible to the experiment according to:
Commonly, in the limit of high Q , in dense systems with sharp and smooth pore surfaces, the scattering curve exhibits a single power law obeying the Porod law . Therefore, the curve can be fitted with the equation:
The exponent in Eq. (3) identifies the slope of the SANS curve, K is the previously mentioned scattering contrast within the sample (that can be calculated from the chemical composition) and C is a constant background. From these considerations it follows that the scattered intensity becomes proportional to S V . A change in slope of the SANS curve, marks the boundary between different regimes, characterized by different length scales. This allows for the identification of different scattering regimes, but, since the scattering profile contains information on the shape and size distribution of the scatterers, it is possible to retrieve the particle-size distribution of the scattering objects. In this paper, this quantity has been obtained with the software McSAS , implementing a fitting procedure based on the Monte Carlo method, applying the sphere model for the pore shape.
XmCT analyses were carried out using the TomoLab instrument custom-developed at the Elettra synchrotron light laboratory in Trieste (Italy). The employed instrument is based on a microfocus Hamamatsu L9181 X-ray source, which guarantees a minimum focal spot size of 5 μm, in a voltage range from 40 up to 130 kV and a maximum current of 300 μA. A 12 bit, water-cooled, Photonic Science VHR CCD camera with a maximum field of view of 49.9 × 33.2 mm 2 and a pixel size of 12.5 μm was used as detector system . Data were collected at a voltage of 130 kV, a current of 61 μA, with a source-to-sample distance D1 = 80 mm and a sample-to-detector distance D2 = 400 mm corresponding to an effective pixel size of 5.0 μm (pixel binning = 2 × 2). Each tomographic scan was performed rotating the sample over 360° and recording 1440 projections with an exposure time/projection of 7 s. The pathway of the incident X-ray beam included a 1.5 mm thick Al filter. The scanning procedure adopted for these XmCT measurements corresponds to phase-contrast modality in edge-detection regime .
Slice reconstruction was obtained by using the commercial software COBRA (Exxim). The COBRA software was also used to reduce beam hardening artefacts from the reconstructed images while for ring artefacts removal a filter, implemented in the Pore3D software library developed at Elettra, was applied directly to the reconstructed slices . The commercial software VG Studio Max 3.0 was employed for the 3-D rendering, the image analysis, and the extraction of numerical data for both samples. The 3-D analysis was performed on Volumes of Interest (VOIs) of the same size (368 × 398 × 929 voxels; isotropic voxel size = 5 μm, volume of 17.00 mm 3 ) cropped from the center of the reconstructed volumes of each sample. This step allowed excluding the regions close to the sample borders of the pseudo-cylindrical sample from the subsequent analysis. The VOI was chosen as the largest possible from the original data set, in order to be representative of the whole sample. The quality of the analysis of porosity depends also on the number of voids included in the VOI. In this case, considering that the voids are small with respect to the VOI size, and their total number of the order of 10 4 , a satisfactory statistical consistency can be assumed. An appropriate threshold Iso value, below which the density of the material is considered virtually zero, i.e. the corresponding volume is considered as pore, was set for each sample. The choice of the optimal value was done with the aid of the automatic definition routine implemented in the software and visually verified on several slices. Low level Gauss adaptive digital filtering, available from the VGDefX algorithm of the software, was applied on VOIs to reduce the noise of the data set before thresholding, in order to preserve the pore shape. The obtained threshold values were 100.5 and 76.0 for the samples Z1 and Z2, respectively. Interactive output of the software, showing the histograms with the threshold Iso values, is reported as Supplementary material Fig. S1. The objects considered as pores in a meaningful way and included in porosity calculations, were those with size larger than 8 voxels (volume of 1.0 × 10 −6 mm 3 ). A total number of 26324 and 25729 pores for Z1 and Z2 samples, respectively, were used in the analysis. The calculation of porosity was performed processing voxel data sets according to the standard VDG P201 . Therefore, pore radius/diameter and the corresponding S V were calculated considering the circumscribed sphere of the pore.
Compressive strength was determined according to the standard EN ISO 9917-1:2007 , averaging the values obtained from 5 samples for each ZPC composition.