PII:

SO300-5712(96)00024-3

Journal of Dentistry, Vol. 25, Nos 34, pp. 263-210,1997 Copyright 0 1997ElsevierScience Ltd. All rights reserved Printed in Great Britain 0300-5712/97 $17.00+0.00

ELSEVIER

/II vi&o peel/shear bond strength orthodontic adhesives

G. Willems*, C. E. L. Carels* and G. Verbeket
*Department of Orthodontics and tBiostatistica/ Centre, Department Universiteit te Leuven, Kapucijnenvoer 7, B-3000, Leuven, Belgium of Epidemiology

of

Faculty of Medicine, Katholieke

ABSTRACT Objectives: The purpose of this study was to evaluate the in vitro peel/shear bond strength of a selection

of orthodontic bracket adhesives to human premolar teeth. Methods: Twenty-two commercially available bracket adhesiveswere used to bond the samebracket type (Miniature Twin, 3M Unitek, Monrovia, CA, USA) on 264 intact human premolar teeth and then adhesively tested to failure. Peel/shearbond strength values were calculated in newtons and megapascals. The site of bond failure was scored according to the Adhesive Remnant Index. Statistics included one-way analysis of variance and Tukeys Studentized Range test together with Weibull analysis. The latter is a survival analysis able to describe the performance of a material. Results: The mean bond strengths varied from 9.9 MPa for Concise to 4.1 MPa for Heliosit Orthodontic. The overall F-test showed a significant difference (P<O.OOOl).No significant differences in bond strength were found between Concise, AccuBond, Imperva Dual, Transbond XT, Kurasper and Spectrum. Conclusion: Concise and AccuBond are among the materials of choice for bonding fixed orthodontic appliances to teeth. These materials combine high bond strength with a reliable bond that is easily and quickly debonded. 0 1997 Elsevier Science Ltd. All rights reserved
KEY WORDS: Shear bond strength, Adhesive (Received 31 July 1995; accepted 30 November 1995)

J. Denf. 1997; 25: 263-270

INTRODUCTION
Various papers have been published on in vitro bond

strength tests of bracket adhesives in orthodontics. Comparisons between these different in vitro methods are often laborious or even impossible because of the variety of methods employed. Bond strength values also depend on the quality of the enamel, storage conditions prior to testing, quality of the material and how it is handled. This has led to confusion, since the variation in the results obtained by different researchers is large. Standardization of test methods would lead to comparable results and benefit the development of more effective systems2. Recently, Fox et ~1.~ wrote a critique on bond strength testing in orthodontics, based on an extensive review of the literature and proposed a standard
Correspondence should be addressed to: Dr G. Willems, Department of Orthodontics, KU Leuven, Kapucijnenvoer 7, B-3000 Leuven, Belgium. Tel.: +32 16 332439. Fax: +32 16 332435.

protocol for future bond strength testing in orthodontics. Their criteria were taken into account when the protocol for the present in vitro study was developed. The aim of the present study was to evaluate and compare the in vitro peel/shear bond strength of 22 currently available bracket adhesives on intact human premolars and to investigate the site of bond failure.

MATERIALS

AND METHODS

Two hundred and sixty-four intact human upper and lower premolar teeth, extracted from adolescent patients for orthodontic reasons, were stored in an aqueous solution of chloramine (0.1% wt/vol). All teeth were free of any restorations or carious lesions that might influence the strength of the enamel and were used after 1 month, but within 6 months after extraction3. Each tooth was embedded in a block of dental stone and assigned at random into one of 22 sets of 12 teeth, corresponding to the number of adhesives tested

264

J. Dent 1997; 25: Nos 3-4

Table /. List of bracket adhesives with method of cure, manufacturer (LC, light-cure; NM, no-mix; CC, chemical-cure; DC, dual-cure) Bracket adhesives Transbond XT Unite Concise APC Miniature Twin Precoated Spectrum No-mix: 30 Force II Ideal Ottho-Lot AccuBond Kurasper F Insta-Bond VL Insta-Bond NM Prestige Sequence System l+ Pulpdent OBA lmperva Dual UltraLight Right-on Extend-A-Bond Heliosit Orthodontic LC NM cc LC LC NM cc LC NM cc LC LC NM cc DC NM LC DC LC NM cc LC Manufacturer 3M 3M 3M 3M American Orthodontics American Orthodontics American Orthodontics GAC International GAC International GAC International Kuraray Co Lee Pharmaceuticals Lee Pharmaceuticals Lee Pharamaceuticals Ormco Ormco Pulpdent Corporation Shofu TP Orthodontics, Inc. TP Orthodontics, Inc. TP Orthodontics, Inc. Vivadent and batch number specified

Batch number 100493 024 940126 93HOl Lot 401725 SI 2248 AJ 2246 SJ 2442 SJ2486 111993 079273 039293 089243 11118 31451Vl DO 3328lBl DO 402OOPl DO 3LOl 4802 701293 0693121 069303 3193000 2793000 3403000 560458

(Table I). Among this selection of commercially available systems,there were four types of bracket adhesives: chemical-cured, light-cured, dual-cured and no-mix. The Miniature Twin 018 bracket (3M Unitek, Monrovia, CA, USA) was selectedto evaluate the bond strength of the different adhesive systems. The bracket base projected surface area in this system is 7.90 mm2. Before bonding, the enamel surfaces were carefully polished with pumice and water, rinsed and air dried. After conditioning the enamel surface with a 37% phosphoric acid solution for 60 s, the manufacturers instructions for bonding were followed for each material evaluated. Each bracket was bonded separately, using a fresh mix of material, as appropriate. Each bracket was positioned over the mid-point of the long axis of the clinical crown on the buccal of the prepared premolar tooth. Excess bonding material was carefully removed with a probe from around the bracket base. A light source (Visilux 2, 3M, Monrovia, CA, USA) with a controlled light intensity of more than 400 mWtcm2 was used to activate the light-cure materials from each side of the bracket and by transillumination using exposures of 30 s each. After bonding, the samples were immersed in distilled water for 24 h at 37C. All bonding procedures were carried out by the same operator. Each prepared specimen was mounted in an Instron universal testing machine (Instron 4505, Instron Corporation, Canton, MA, USA) in such a way that the bracket base was at an angle of 30 to the direction of the applied dislodging force. A round steel wire was placed directly underneath the cervical bracket wings as close as possible to the bracket wings in order to

minimize differences in the nature of the applied force. In this way, the point of application and the direction of the debonding force was similar for all specimens3.The bonds were tested to failure at a crosshead speed of 0.1 mm/min with the tests being performed at a room temperature of 22C and a relative humidity of 45%. After peel/shear testing the predominant site of bond failure was assessedaccording to the Adhesive Remnant Index (ARI). This index was designed as a clinical tool to assessthe amount of bonding material remaining on the enamel after bond failure. The index has four grades, ranging from 0 to 3 (0, no adhesive left on tooth; 1, less than half of the adhesive left on tooth; 2, more than half of the adhesive left on tooth; 3, all adhesive left on tooth with distinct impression of the bracket base). In this study, a light-optical stereomicroscope (Wild M5A, Wild, Heerbrugg, Switzerland) was used to evaluate fracture sites. Statistical analysis was performed with the SAS statistical software package (SAS Institute, Cary, NC, USA). The significance level was set at 0.01 for all statistical tests. The residuals of the data were calculated and plotted and were found to be normally distributed. Therefore, the comparison of sample means was carried out using an F-test. One-way analysis of variance and Tukeys Studentized Range test were used to determine which of the means were significantly different from the others. According to McCabe and Carrick4, and Fox et al.? statistical analysis of bond strength data should include survival analysis, since these data tend to be skewed. It is known that materials perform well in most casesbut occasionally fail for no apparent reason. Additionally,

Willems et al. Bond strength of orthodontic adhesives

265

Concise AccuBond Imperva Dual

19.9 18.7 18.6

4 MPa

10

Fig. 1, In

vitro peel/shear bond strength means for the different materials.

survival analysis might give a prediction of the performance of the material at any level of stresswhich may be related to the clinical situation5. The Weibull analysis is such an approach that has been used to evaluate the fracture processesin other dental materials4. This distribution may be more applicable to the evaluation of data from mechanical tests. As discussed by Fox et aL3, mean and standard deviation may not be the best indicator for the performance of a material. Data from mechanical tests such as bond strength tests tend to be abnormally distributed. These data are more or less survival type data and survival analysis can be used to better describe the performance of each material. The Weibull distribution is a parametric test. It takes into account the tail values of a distribution. It is characterized by the Weibull modulus, which expressesthe spread of the data, and the normalizing parameter or characteristic level, in this case the characteristic bond strength, which more or less corresponds with the mean bond strength for a Gaussian distribution. As reported by Fox et aL3 the Weibull distribution can be characterized by the following equation which relates probability of failure (Pr) to applied stress (CT). The equation is:

n=I++1\

where I is the gamma distribution and m is the Weibull modulus6 The Weibull analysis was performed to calculate characteristic strength and Weibull modulus5. Material performance was compared by calculating the 99% confidence limits and the probability of failure at a given value of force. A Weibull distribution with a Weibull modulus of 3.64 approximates the bell-shape of a Gaussian distribution. The higher the Weibull modulus, the higher and the narrower the shape of the distribution and the smaller the variation among the data, that is for a constant characteristic level. It is clear that both parameters influence the shape of the distribution.

RESULTS
The means of the in vitro peel/shear bond strengths for the different materials tested are displayed in Fig. 1 and listed in Table II together with part of the outcome of the descriptive statistics. The results are displayed in newtons (N) and in megapascals (MPa). The mean bond strengths vary from 9.9 MPa for Concise to 4.1 MPa for Heliosit Orthodontic. The overall F-test showed a significant difference (P<O.OOOl). The output of the Tukeys Studentized Range test is summarized in Table III. Statistical analysis reveals that there are no significant differences in bond strength between Concise, AccuBond, Imperva Dual, Transbond XT, Kurasper and Spectrum. These six systems display

o, and o0 are constants. o, is the lowest level of stress at which Pf approaches zero and is normally assumedto be equal to zero. o0 is the normalizing parameter or characteristic level and is described by Fox et aL3, as the characteristic strength. It can be related to the mean of the Gaussian distribution according to the following equation:

266

J. Dent

1997; 25: Nos 3-4

Tab/e II. List of peel/sheer bond strengths (in N and in MPa) for the different bracket adhesives evaluated. All tests were performed using Miniature Twin brackets Bracket adhesives Mean (N) SD. (N) 9.3 5.0 12.2 5.0 8.9 6.1 4.4 il.0 6.9 4.3 12.1 12.4 12.7 12.3 5.7 11.0 11.1 13.4 9.3 6.5 14.2 5.9 Mean (MPa) 8.7 5.4 9.9 7.0 6.4 4.1 6.1 8.6 5.0 5.4 8.3 6.3 6.1 5.3 6.1 6.2 6.5 8.2 5.8 8.4 7.5 7.2 SD. (MPa) 1.2 0.6 1.5 0.6 1.1 0.8 0.6 1.4 0.9 0.5 1.6 1.6 1.6 1.5 0.7 1.4 1.4 1.7 1.2 0.8 1.8 0.7 Min (MPa) 5.4 4.2 7.6 6.0 4.8 3.1 5.3 5.7 4.3 4.4 6.0 3.3 3.8 3.7 5.3 4.2 4.1 6.3 3.9 7.0 5.5 6.1 Max (MPa) 9.9 6.6 12.2 8.4 8.6 5.7 7.3 11.2 6.8 6.3 10.1 8.2 8.9 7.9 7.8 8.6 8.0 11.9 7.8 9.4 10.1 8.5

AccuBond 69.1 APC Miniature Twin 43.0 Precoated Concise 78.1 Extend-A-Bond 55.3 Force II 50.3 Heliosit Orthodontic 32.5 Ideal 48.6 lmperva Dual 67.8 Insta-Bond NM 39.7 Insta-Bond VL 42.6 Kurasper F 66.0 No-mix: 30 49.6 Ortho-Lot 48.0 Prestige 41.7 Pulpdent OBA 48.4 Right-on 48.7 Sequence 51.5 Spectrum 64.4 System 1+ 46.2 Transbond XT 66.3 UltraLight 59.3 Unite 57.5

Tab/e 111.Statistical analysis: the F-test displayed a significant difference (PcO.0001). The significant pairwise comparisons resulting from Tukeys Studentized Range test are marked with an asterisk

Concise AccuBond lmperva Dual Transbond XT Kurasper Spectrum UltraLight Unite Extend-A-Bond Sequence Force II No mix:30 Right-on Ideal Pulpdent OBA Ortho-Lot System 1+ APC Insta-Bond VLC Prestige Insta-Bond no mix Heliosit

* *

* *

* *

* *

* *

* *

* *

* * * 1

* *

* *

* *

l * *

* * * * *
* *

* * * * *
* *

* * * * *
* *

* * * * *
* *

* * * * *
* i

* * * * * *
* * * * * * * * * *

Willems et al, Bond strength of orthodontic adhesives

Tab/e IV, Statistical analysis: output of the Weibull analysis (Weibull modulus, characteristic strength and its 99% confidence limits, calculated bond strength for 1% chance of failure) Stress for 1% chance of failure (MPa) 6.5 3.6 5.9 4.8 3.3 2.0 4.3 4.7 2.5 3.8 4.5 2.9 2.3 1.7 3.7 2.6 3.3 3.6 2.8 6.2 3.2 5.1

267

Bracket adhesives AccuBond APC Miniature Twin Precoated Concise Extend-A-Bond Force II Heliosit Orthodontic Ideal Imperva Dual Insta-Bond NM Insta-Bond VL Kurasper F No-mix:30 Ortho-Lot Prestige Pulpdent OBA Right-on Sequence Spectrum System 1+ Transbond XT UltraLight Unite

Weibull modulus 13.7 9.8 8.1 10.8 6.2 5.6 11.5 6.9 5.9 11.7 6.6 5.3 4.4 3.8 8.3 4.9 6.1 5.0 5.7 13.5 4.9 11.3

Characteristic strength (MPa) 9.1 5.7 10.5 7.3 6.8 4.4 6.4 9.1 5.4 5.6 8.9 6.8 6.7 5.9 6.5 6.7 7.1 8.8 6.3 8.7 8.2 7.6

99% Confidence limits (MPa) 8.6 5.2 9.5 6.8 6.0 3.8 6.0 8.2 4.8 5.3 7.9 5.9 5.6 4.7 5.9 5.7 6.2 7.5 5.5 8.2 7.0 7.1 9.7 6.2 11.6 7.9 7.8 5.1 6.8 10.2 6.2 6.0 10.1 8.0 8.0 7.3 7.1 7.9 8.0 10.4 7.3 9.3 9.6 8.2

comparable bond strength. The outcome of the Weibull analysis is shown in Table IV. The Weibull distribution for each of the materials tested is graphically displayed in Fig. 2. X- and Y-axes of all these graphs are the same, which enables comparisons between them. These distributions were calculated based on the two parameters obtained after statistical analysis, Weibull modulus and characteristic strength. The site of failure results are listed in Table V. Based on these results of the six materials mentioned above only Concise and AccuBond may be found to give a debonded tooth surface less than half covered with composite remnants. More than half of the debonded surface was covered with remnants when using the other four materials, and in particular Imperva Dual had a tendency towards an adhesive fracture at the resin-bracket interface. The only material that showed a complete adhesive fracture at the resin-bracket interface was APC Miniature Twin Precoated, closely followed by Ideal and Pulpdent OBA. Brackets bonded with Prestige showed a clear failure at the enamel-resin interface when debonded.

DISCUSSION
While a strong and durable bond is required, the problem of removing the bracket without damaging the labial enamel must not be overlooked. The ideal bonding material should fail during debonding by the clean separation of the resin from the etched enamel. This adhesive failure at the enamel-resin interface is most

desirable in terms of debonding and final polishing. In the present study, the failure pattern of the bond was evaluated using the Adhesive Remnant Index. Adhesives with score 0 for AR1 leave a clean enamel surface after debonding (Table v). Only Prestige showed this characteristic. Failure at the resin-bracket interface (score 3) was found for the materials APC Miniature Twin Precoated, Ideal and Pulpdent OBA. Microscopic examination of the enamel surface after debonding showed a clear impression of the bracket base for these samples. Scores 1 and 2 correspond with adhesive and cohesive failures at the enamel-resin and the resin-bracket interface, respectively. Generally, an even distribution of all four scores of the AR1 index was found. Fox et aL3 suggestedthat at least 20 and preferably 30 specimens should be used per test. This number should be modified in view of the very limited availability of intact human premolar teeth, certainly if one aims to evaluate many materials. A sample size of 10 is commonly used, without jeopardizing statistical analysis. In this study, sample size was set at 12. By evaluating a large variety of adhesives in the same laboratory conditions, more valid comparisons of materials can be produced and problems with comparing the results of different studies are omitted. This is exactly why test procedures have to be standardized. Statistical analysis included an F-test to outline general statistically significant effects and one-way analysis of variance and Tukeys Studentized Range test to describe significant differences between the materials

268

J. Dent. 1997; 25: Nos 3-4

AccuBond

APC Precoated

Concise

Extend-A-Bond

Force II

Concise

Insta-Bond NM

Ins&Bond

VL

Kurasper F

No-Mix:30

Ortho-Lot

Prestige

Pulpdent OBA

Ripht-On

Spectrum

Transbond XI

UltraLight I

Unite

Fig. 2. Weibull distribution

of all bracket adhesives

evaluated.

All graphs are produced on an intercomparable

scale.

tested. Data that are subjected to these tests have to be drawn from a normal population. Therefore, the distribution of the residuals of these data was analysed

to see whether it matched the Gaussian distribution. Descriptive statistical analysis showed a nearly perfect Gaussian distribution.

Willems et a/. Bond strength of orthodontic

Tab/e V. Site of failure: scores of the Adhesive adhesive Bracket adhesives AccuBond APC Miniature Twin Precoated Concise Extend-A-Bond Force II Heliosit Orthodontic Ideal lmperva Dual Insta-Bond NM Insta-Bond VL Kurasper F No-mix:30 Ortho-Lot Prestige Pulpdent OBA Right-on Sequence Spectrum System I+ Transbond XT UltraLight Unite I 2 3 Remnant Index for each of the 12 samples per

adhesives

IO

11 11 3 11 13 1 2 3 33 2 11 12 11 13 11 3 13 2 2 12 12 11 2

12

1113111111 3333333333 1111011120 3301111103 1 1 1 2222222212 3331333333 2111331332 3 2 2 2331233213 2221222111 1121111111 2331131121 0000000000 2332233133 1211121111 1112122112 2333322211 2223301211 2323232131 1332323122 2212221222

1 2 1 3

322

3 1 1

The Weibull distribution of each material is shown in

Fig. 2. The X- and Y-scale of all graphs is the same,

which enables visual comparisons. A material with a Weibull distribution similar to that for APC Miniature Twin Precoated or for Insta-Bond VL combined with the characteristic strength of AccuBond or Concise would be the material of choice. Narrow distributions equal smaller variations of data and thus more dependable bond strengths. On the other hand, high bond strength values as produced by Concise, AccuBond or Transbond XT have advantages. However, bond reliability and dependability rather than high mean bond strength is required. As indicated by the conventional statistics, Concise displayed the highest bond strength in this study, but it was not significantly different from Accubond and four other materials. Survival analysis (Table Iv> indicates that Concise has the highest characteristic strength of all the adhesives tested (Fig. 2). However, it seemsthat AccuBond can produce somewhat more reliable bond strengths compared with Concise, but the latter will produce somewhat higher bond strengths with slightly more variation. Calculation of the 99% confidence limits for the characteristic strength of both materials reveals overlapping intervals indicating that there is no statistical difference between them at the 0.01 level. Weibull analysis and calculation of the confidence limits for characteristic strength together with the 1% chance of failure value (Table IV) indicated that the adhesive Ultralight was not significantly different from the six materials listed above. Taking 30 N or calculated for this study - 3.8 MPa as representative of stresses applied clinically7, 13 of the materials

investigated were found to have a 1% chance of failure smaller than 3.8 MPa. Among this group is Ultralight. The spread of the data for this material, visualized in the Weibull distribution (Fig. 2), supports this result. Through the calculation of the 99% confidence limits, one is able to assessthe performance of a material. However, it must be remembered that these results are laboratory findings which may not reflect clinical performance. Another phenomenon was noted when comparing Transbond XT with APC Miniature Twin Precoated. Both materials are essentially the same except that the latter is more viscous and is delivered as a precoated bracket system. This is to prevent the material from flowing off the bracket base during storage. Comparison of the Weibull distribution of both materials shows a similar shape, but a significantly lower characteristic strength was found for APC Miniature Twin Precoated. This indicates that the process of precoating and the characteristics of the adhesive used result in a lower adhesive capacity of the bracket system. An AR1 score of 3 for APC Miniature Twin Precoated indicates that the weak link of the precoating process might be the adhesion of the adhesive to the bracket base. It is well known that an adhesive material with a relatively high viscosity is less likely to flow into the undercuts of the bracket base compared with a less viscous adhesive. An important consideration, as concluded by Shiau et al. is that specifying bond strength as breaking load/area may be incorrect. Specimenstested with some shear bond strength methods may fail because of tensile stress induced by the applied bending moment rather than by shear stress, although further tests are

270

J. Dent. 1997; 25: Nos 3-4

required to prove this. Therefore, results in this study are given in both newtons and in megapascals. As can be seen from the results, light-cured and dual-cured systemscan both produce strong bonds, but are time consuming and expensive. However, these systems offer the clinician more time for bracket positioning. The advantage of dual-cured systems over light-cured systemsis that, once initiated, the polymerization reaction proceeds to completion. No-mix adhesives are relatively easy and quick to bond, although they do not produce high bond strengths, except for Unite.

References
1.

Wrtun J, Bergland S. Clinical trials with crystal growth conditioning as an alternative to acid-etch enamelpretreatment.Am J Orthodont 1984; 85: 333-340.
Wile G. Bond strength testing-what
Dent J 1993; 43: 492-498.

2. 3.

does it mean? Int

4.

5.

CONCLUSION
The results suggest that Concise and AccuBond are among the materials of choice for bonding orthodontic brackets to teeth. These materials combine high bond strength with a reliable bond, but allow brackets to be debonded relatively easily and quickly. Statistical analysis of in vitro bond strength tests should include survival analysis. However, as in other fields, clinical performance can only be established by means of a clinical trial.
6.

7.

8.

Fox NC, McCabe JF, Buckley JG. A critique of bond strength testing in orthodontics. Br J Ovthodont 1994; 21: 33-43. McCabe JF, Carrick TE. A statistical approach to the mechanical testing of dental materials. Dent Mater 1986; 2: 139-142. Fox NG, McCabe JF, Gordon PH. Bond strengths of orthodontic bonding materials: an in vitro study. BP J Orthodont 1991; 18: 125-130. Johnson NL, Kotz S. Continuous univariate distributions. In: Distributions in Statistics. Boston: Houghton Mifflin, 1970; pp 250-254. Millet D, McCabe JF, Gordon PH. The role of sandblasting on the retention of metallic brackets applied with glass ionomer cement. Br J Orthodont 1993; 20: 117-122. Shiau JY, Rasmussen ST, Phelps AE, Enlow DH, Wolf GR. Analysis of the shear bond strength of pretreated aged composites used in some indirect bonding techniques. J Dent Res 1993; 72: 1291-1297.