A Comprehensive Laboratory Screening of Three-Step Etch-and-Rinse Adhesives

The use of 85% phosphoric acid to improve the infiltration of an acrylic resin into enamel was the genesis of all dental adhesive systems used in dentistry.¹  The interaction of dental adhesives with enamel and dentin results from resin monomers permeating the microporosities created by acidic agents and polymerized monomers subsequently becoming wrapped around the exposed hydroxyapatite crystals and/or collagen fibers.² 

In spite of the different classifications that have appeared within the last two decades, dental adhesives are currently classified according to how they interact with the dental substrate and thus are divided into etch-and-rinse (ER) and self-etch (SE) adhesives.²  The number of application steps categorizes adhesive systems within each of the two strategies. For example, ER adhesive systems can be available in a three-step and a two-step protocol, depending on whether the primer and bonding resin are separate or combined in one bottle.

Although three-step ER adhesives have been considered the clinical gold standard in dental bonding,^3-6  this issue seems to be controversial, according to recent systematic reviews of the literature.^3,5,7-9  The retention rates of three-step ER systems are quite variable, with an annual failure rate varying from 0% to 16%.³  Furthermore, Peumans and others³  in their systematic review reported that at least three out of 10 three-step ER adhesives did not meet the requirements of the American Dental Association guidelines for provisional and full acceptance of the restorations, and in some of the clinical studies reviewed, they did not gain full acceptance.

It is claimed that the use of a hydrophobic resin coating in the three-step ER is responsible for better in vitro and in vivo performance than that of their 2-step counterparts¹⁰ ; however, one cannot rule out the role of the chemistry of three-step ER systems, which varies considerably depending on the brand.

For instance, the type of solvent presented in the formulation of the primer has been mentioned as a factor affecting the performance of the materials, with the ethanol-based systems performing better than the acetone-based types.⁴  Moreover, the bonding resin usually claimed to be a hydrophobic resin coating can contain hydrophilic monomers, such as hydroxyl-ethyl-methacrylate (HEMA), which may ultimately influence the bond performance.¹¹  Other differences such as filler loading may also impact the material's performance.

Based on the foregoing, the aim of this study was to compare several bonding and mechanical properties of three-step ER adhesives available on the market. The microtensile bond strengths (μTBS), nanoleakage (NL), in situ degree of conversion (ISDC) on dentin and ultimate tensile strength (UTS), degree of conversion (DC), water sorption (WS), and solubility in water (SL) for all the adhesives were also evaluated.

A total of 28 extracted, caries-free human third molars were used. The teeth were collected after obtaining the patients' informed consent under a protocol approved by the local Ethics Committee Review Board. The teeth were disinfected in 0.5% chloramine, stored in distilled water, and used within six months after extraction.

A flat occlusal dentin surface was exposed after wet grinding the occlusal enamel with 180-grit silicone-carbide (SiC) paper for 60 seconds. The exposed dentin surfaces were further polished with wet 600-grit SiC paper for 60 seconds to standardize the smear layer. These teeth were used for testing resin-dentin μTBS, NL, and measurement of ISDC.

Teeth were randomly assigned for bonding with four different materials (n=7): All-Bond 3 (ALB3; Bisco Inc, Schaumburg, IL, USA); Fusion Duralink (FSDL; Angelus, Londrina, Brazil); Optibond FL (OBFL; Kerr, Orange, CA, USA), and Scotchbond Multi-Purpose (SBMP; 3M ESPE, St Paul, MN, USA). Each adhesive system was applied on the flat dentin surfaces, according to the manufacturers' instructions (Table 1).

On each prepared surface a composite restoration (Filtek Z350, 3M ESPE) was built up in two increments of 2 mm and was light-polymerized for 40 seconds using a LED light-curing unit set at 1200 mW/cm² (Radii-cal, SDI Limited, Bayswater, Australia). The same light-curing unit was used throughout this study.

After the restored teeth had been stored in distilled water at 37°C for 24 hours, they were longitudinally sectioned in the mesio-to-distal and buccal-to-lingual directions across the bonded interface, using a low-speed diamond saw (Isomet, Buehler Ltd, Lake Bluff, IL, USA) to obtain resin-dentin sticks with a cross-sectional area of approximately 0.8 mm², measured with a digital caliper (Digimatic caliper, Mitutoyo, Tokyo, Japan). Four resin-dentin bonded sticks from each tooth were not tested in the tensile mode, because they were used for the ISDC and NL evaluations.

All the remaining bonded sticks were attached to a Geraldeli jig¹²  (Odeme Biotechnology, Joaçaba, Brazil) with cyanoacrylate adhesive and tested under tensile forces (Model 5565, Instron, Canton, OH, USA) at 0.5 mm/min until failure. The μTBS values were calculated by dividing the load at failure by the cross-sectional bonding area. The failure mode was classified as cohesive (failure exclusively within dentin or resin composite), adhesive (failure at the resin/dentin interface), or mixed (failure at the resin/dentin interface that included cohesive failure of the neighboring substrates). The failure mode was analyzed under a stereomicroscope at 100× magnification (Olympus SZ40, Tokyo, Japan). Because few specimens presented premature failures, they were included in the bond-strength mean.

Two resin-dentin bonded sticks were placed in ammoniacal silver nitrate¹³  in darkness (24 hours), rinsed thoroughly in distilled water, and immersed in photo-developing solution (8 hours) under a fluorescent light to reduce silver ions to metallic silver grains. Specimens were polished down with 600-, 1000-, 1200-, 1500-, 2000-, and 2500-grit SiC paper and 1 and 0.25 μm diamond paste (Buehler Ltd) using a polishing cloth. They were ultrasonically cleaned, air-dried, mounted on stubs, and sputter-coated with carbon-gold (MED 010, Balzers Union, Balzers, Liechtenstein). Resin-dentin interfaces were analyzed in a scanning electron microscope operated in the backscattered mode (LEO 435 VP, LEO Electron Microscopy Ltd, Cambridge, UK) (Figure 1).

Three pictures from each bonded stick were taken.¹⁴  The relative percentage of NL in each image was measured by a blinded researcher, using the UTHSCSA ImageTool 3.0 software program (Department of Dental Diagnostic Science at The University of Texas Health Science Center, San Antonio, TX, USA).

The adhesive interfaces of the two other resin-dentin bonded sticks were wet polished with 1500-, 2000-, and 2500-grit SiC paper for 15 seconds each. They were ultrasonically cleaned for 20 minutes in distilled water and stored in water at 37°C for 24 hours prior to performing the DC readings. The micro-Raman spectrophotometer (Bruker Optik GmbH, Ettlingen, Baden-Württemberg, Germany) was first calibrated for zero and then for coefficient values using a silicon specimen. Specimens were analyzed using the following micro-Raman parameters: 20 mW neon laser with 532 nm wavelength, spatial resolution of ≈3 μm, spectral resolution ≈5 cm⁻¹, accumulation time of 30 seconds with six coadditions, and magnification of 100× (Olympus UK, London, UK) to a ≈1 μm beam diameter. Spectra were taken at the dentin-adhesive interface, at three different sites for each specimen.

Spectra of unpolymerized adhesives were taken as reference. Postprocessing of spectra was performed using the Opus Spectroscopy Software Program version 6.5 (Bruker Optik GmbH, Ettlingen, Baden-Württemberg, Germany). The ratio of the double-bond content of the monomer to the polymer in the adhesive was calculated according to the following formula:

where R is the ratio of aliphatic and aromatic peak areas at 1639 cm⁻¹ and 1609 cm⁻¹ in polymerized and unpolymerized adhesives.

An hourglass-shaped metal matrix (10 mm long, 2 mm wide, and 1 mm deep with a cross-sectional area of 0.8 mm²) was isolated with petroleum jelly. To simulate a clinical application, the primer was first inserted into a metal matrix and gently air-dried in accordance with the manufacturer's directions (Table 1). Then, the bonding resin was put into the metal matrix and slightly mixed with the primer using a microbrush. The mixture was again air-dried in accordance with the manufacturer's recommendations (Table 1), and all visible air bubbles trapped in the bonding resin were carefully removed. The bonding resin and the primer were added in a ratio of 1:1. A thin strip of polyester sheeting was placed over the bonding resin, and the surface was light-polymerized for 40 seconds. Eight specimens were prepared for each group, and were tested under tensile forces as described for the μTBS test.

The bonding resin was mixed with the primer in a ratio of 1:1 and placed on a thin strip of polyester sheeting in the same way as reported for the UTS. Then, the adhesive was covered with another thin strip of polyester sheeting and light-polymerized for 40 seconds. Each thin adhesive layer was carefully removed with a blade and stored in a dark and dry environment for 24 hours, after which the DC was determined by Fourier transformed infrared analysis (Spectrum 100, Perkin Elmer, Waltham, MA, USA).

The spectra of the polymerized and unpolymerized bonding resins were obtained with 32 scans at a resolution of 4 cm⁻¹ in transmission mode. The percentage of unreacted carbon-carbon double bonds (% C=C) was determined from the ratio of absorbance intensities of aliphatic C=C (peak height at 1640 cm⁻¹) against an internal standard before and after specimen polymerization. The aromatic C=C bond (peak height at 1610 cm⁻¹) absorbance was used as an internal standard. The DC was determined by subtracting the % C=C from 100%. Five specimens were tested for each group.

The bonding resin was mixed with the primer in a ratio of 1:1 and placed on a thin strip of polyester sheeting in the same way as reported for the UTS. Ten resin disks of each adhesive were produced in a circular metal mold (5.8 mm diameter, 0.8 mm thick) after isolation with petroleum jelly. The primer and bonding resin were applied in a ratio of 1:1 as previously reported for the UTS test. All visible air bubbles trapped in the adhesives were carefully removed. Next, the solvent evaporation was performed with an air syringe at a distance of 10 cm for 40 seconds. A thin strip of polyester sheeting was placed on top of the adhesive, which was light-polymerized for 40 seconds to allow the specimen to be removed without permanent deformation.

WS and SL were determined according to the International Organization for Standardization (ISO) specification 4049.¹⁵  Immediately after polymerization, the specimens were placed in a desiccator, transferred to a preconditioning oven at 37°C, and left undisturbed for 10 days. After this period, specimens were repeatedly weighed at 24-hour intervals until a constant mass (m₁) was obtained (ie, variation was less than 0.2 mg in any 24-hour period). Thickness and diameter of the specimens were measured using a digital caliper, rounded to the nearest 0.01 mm, and these measurements were used to calculate the volume of each specimen (in mm³).

After this, specimens were then individually placed in sealed vials containing 10 mL of distilled water (pH 7.2) at 37°C. After fixed time intervals of 1, 2, 3, 4, 5, 6, 7, 14, and 28 days of storage, the vials (15 mL; Eppendorf of Brazil, São Paulo, Brazil) were removed from the oven and left at room temperature for 30 minutes. The specimens were washed in running water, gently wiped with a soft absorbent paper, weighed on an analytical balance (m₂), and returned to the vials containing 10 mL of fresh distilled water. After 28 days of storage, the specimens were dried in a desiccator containing fresh silica gel in an oven at 37°C and left undisturbed for 10 days. They were weighed daily until a constant mass (m₃) was obtained (as previously described). The initial mass determined after the first desiccation process (m₁) was used to calculate the change in mass after each fixed time interval during the 28 days of storage in water. Changes in mass were plotted against the storage time in order to obtain the kinetics of WS during the entire period of water storage.

WS and SL over the 28 days of water storage were calculated using the following formulas: WS = (m₂ – m₃) / V and SL = (m₁ – m₃) / V.

The values originating from the same specimen for μTBS, NL, and ISDC were averaged for statistical purposes. The μTBS, NL and ISDC means for every group were expressed as the average of the seven teeth used per group. Data from μTBS, NL, ISDC, UTS, and DC of the adhesive pellicle were individually analyzed using one-way analysis of variance (ANOVA) and the Tukey post hoc test (α=0.05). Data from WS and SL were calculated for each experimental condition, and the data after 28 days were analyzed by one-way ANOVA (adhesive) and the Tukey post hoc test at α = 0.05.

The failure modes of all experimental groups are shown in Table 2. The majority of the specimens (84.4%) presented adhesive/mixed failures. Dentin and resin cohesive failures were observed in 7.0% of the specimens. A number of premature failures (8.6%) were observed.

One-way ANOVA detected significant differences between the μTBS values (Table 2; p<0.001). OBFL resulted in the highest mean μTBS (Table 2). All other adhesives had statistically similar mean μTBS (p>0.05).

Significant differences were detected in the NL values (Table 2; p<0.0001). OBFL showed the lowest mean NL (Table 2). SBMP showed intermediate values, but only statistically different from FSDL (p<0.01). The mean NL associated with ALB3 was similar to those of SBMP and FSDL (p>0.05). Examples of the NL pattern can be seen in Figure 1.

One-way ANOVA detected significant differences in the ISDC means (Table 2; p<0.00001). OBFL showed the highest mean ISDC and FSDL showed the lowest (Table 2). FSDL and SBMP showed intermediate values, with the former showing the lower mean ISDC (Table 2).

Statistically significant differences were detected in the UTS means (Table 3; p<0.001). OBFL showed the highest mean UTS (Table 3). All other materials had statistically similar mean UTS (p>0.05).

One-way ANOVA detected statistically significant differences (Table 3; p<0.00001): OBFL showed the highest mean DC, whereas ALB3 resulted in the lowest (Table 3). FSDL and SBMP showed intermediate values, with the latter showing the lowest mean DC (Table 3).

Significant differences were also detected in terms of WS and SL values (Table 3; p<0.001 and p<0.01, respectively). FSDL and OBFL showed the lowest mean WS; SBMP resulted in the highest mean WS (Table 3; p<0.001). FSDL showed an intermediate mean WS, which was similar to that of OBFL (Table 3; p>0.05). FSDL resulted in the highest mean SL in comparison with all the other materials (Table 3; p<0.001). All other adhesives had statistically similar mean SL (p>0.05).

The results of this study confirmed that there were several differences between the three-step ER adhesive systems tested. Among the adhesive evaluated, only OBFL and SBMP have been extensively evaluated. When the immediate bond strengths of OBFL and SBMP were compared with each other, controversial results were obtained.^16-23  This lack of consensus among studies can be attributed to methodological differences. A systematic review of bond-strength tests²⁴  reported that SBMP and OBFL are usually reported as being similar when assessed by shear or microshear bond-strength tests. However, OBFL usually presents significantly higher bond strength values when tested in tensile and microtensile bond-strength tests such as the type observed in the present study.

An earlier study²⁵  reported that OBFL exhibited a very uniform hybrid layer, whereas the hybrid layer formed with SBMP showed a varying electron density with a less sharply demarcated transition to the unaffected dentin. This observation suggested a less-than-optimal resin infiltration, as has been reported by other authors,^26-29  and this is corroborated by our NL results.

SBMP contains a specific polyalkenoic acid copolymer (PAC) that was primarily added to the composition of the resin-modified glass ionomer Vitrebond (3M ESPE). PAC bonds chemically and spontaneously to hydroxyapatite in glass ionomer materials,³⁰  which may explain why an ER with PAC showed more enhanced bond strength than a PAC-free adhesive with the same composition.³¹  On the other hand, it has been demonstrated that because PAC is a compound with a high molecular weight, it does not dissolve in the adhesive solution, which may lead to phase separation and formation of resin globules within the polymer.²⁸  In addition, the collagen network can filter it out and the PAC can be deposited as a distinct gel on the exposed collagen network surface.^28,32 

In a more extreme case, the gel can hinder adequate monomer infiltration, and the hybrid layer produced would be constituted of collagen and linear polymer chains, with entrapped residual water and insufficiently removed solvent. The presence of water may reduce the degree of conversion of the material, which is shown by the presence of silver nitrate grains, thus explaining why a higher NL and a lower DC (in situ and from the adhesive pellicle) was observed for SBMP when compared with OBFL.

The primer of SBMP contains approximately 50 volume percent (vol%) of water, whereas the OBFL primer contains 23 vol% of water and 29 vol% of ethanol.²⁵  The high boiling temperature and low vapor pressure of water, when compared with that of ethanol, makes this solvent more difficult to remove in comparison with water-ethanol solutions.¹¹  The excess water in the adhesive film reduces the degree of conversion of SBMP and diminishes the UTS of the material by preventing molecule approximation during polymerization. A poorly polymerized adhesive is more prone to WS, as could be detected in this study.^33,34 

The adhesives ALB3 and FSDL also had a lower DC than SBMP did, and they did not present such a high WS. Thus, other factors apart from lower DC may be responsible for this finding. SBMP is the only material that contains the hydrophilic monomer HEMA in the bonding resin. HEMA can reduce the vapor pressure of water,³⁵  leading to solvent retention. Additionally, poly (HEMA) formed after polymerization behaves as a hydrogel, which swells and rapidly takes up water. Water droplets were found to be located adjacent to the adhesive resin-composite interface formed by HEMA-rich adhesives^36,37 The literature has confirmed that SBMP usually showed poor long-term in vitro and in vivo results when compared with OBFL^22,38  and the clinical results of this material are usually controversial.^3,38 

The bond strength of ALB3 and FSDL was similar to that of SBMP; however, these materials showed the lowest DC (in situ or from the adhesive pellicle) and the highest NL values when compared with OBFL and SBMP. This was expected, given that it was recently demonstrated that the DC is negatively correlated with the NL.³⁹  Conversion of monomer into polymer (percentage of C=C incorporated into polymer chains as C=C) plays an important role in successful dentin bonding.⁴⁰  To some extent it is true that increased permeability of the adhesive layer is observed when there are unreacted monomers within the hybrid layer.^41,42  This results in the formation of a porous hybridoid structure with reduced sealing ability, which is prone to silver nitrate uptake,^43,44  as could be seen in the results of the present study. Unfortunately, the reason why ALB3 and FSDL showed higher amount of NL and lower DC is not clear, and future studies need to be conducted to test these hypothesis.

Earlier studies reported negative solubility values for three-step ER,^45-47  which does not agree with the present findings. Contrary to this study, in which a mixture of primer and bonding resin was used for specimen fabrication, the previous studies only evaluated the bonding resin. The mixture of the primer with the bonding resin affects the mechanical properties of the adhesive interface when compared only with the bonding resin.^48,49 

Materials with a high SL usually present a high level of WS. WS causes polymer swelling, thereby facilitating the elution of unreacted monomers and/or solvents trapped in the polymer network.⁴⁰  However, the adhesive system FSDL showed the highest SL in this study without presenting a high WS. This fact along with the low DC of this material suggests that many unbound oligomers and residual monomers were readily leached out without the need for polymer plasticization by water. The high amount of silver nitrate uptake in the adhesive layers produced by FSDL reinforces this hypothesis. Because the manufacturer does not specify the material composition, it is quite difficult to interpret the study findings based on its chemistry. Moreover, as far as we know, no published article with FSDL was found, and future studies need to be conducted to confirm the present results.

The good performance of OBFL in this study is in agreement with the literature findings. OBFL usually showed the best performance in all tests, confirming that it is the gold standard in terms of ER adhesives^2-7  in both laboratory^18,22  and clinical studies.^38,50,51  At least two hypotheses can be used to explain these good results. Sezinando and others⁵²  speculated that the higher bond-strength durability of this material is due to the presence of glycerol phosphate dimethacrylate, which can interact chemically with hydroxyapatite.^53,54  Additionally, OBFL has a bonding layer with very high filler loading (48 vol%). This filled adhesive, along with the hybrid layer, may also create an artificially elastic cavity wall⁵⁵  and act as a potential elastic shock absorber⁵⁶  during polymerization of the composite resin. ALB3 also has filler in its composition (>40%), but the benefit of its presence was not confirmed in terms of immediate bond strength and NL into the dentin. Of course, further long-term in vitro and clinical studies need to be conducted to prove this hypothesis, mainly because the only clinical study⁵⁷  that was found with ALB3 showed very promising results, as was pointed out in the recent systematic review of clinical studies published by Chee and others.⁷ 

The evaluation of several bonding (μTBS, NL, and ISDC on dentin) and mechanical (UTS, DC, WS, and SL in water) properties in the short term allowed the adhesives available on the market to be more effectively screened, without requiring long-term studies. OBFL showed the best results in the all properties evaluated, and it can be considered the gold standard in the three-step ER adhesive systems.

The authors of this study would like to thank Prof Dr Jorge Perdigão for the critical revision of the manuscript. This study was partially supported by The National Council for Scientific and Technological Development (CNPq) under grants No. 301937/2009-5 and 301891/2010-9.

The authors of this manuscript certify that they have no proprietary, financial, or other personal interest of any nature or kind in any product, service, and/or company that is presented in this article.