Three-Dimensional Nonlinear Finite Element Analysis

and Microcomputed Tomography Evaluation of Microgap

Formation in a Dental Implant Under Oblique Loading
Daniela Jörn, Dipl-Ing1/Philipp Kohorst, Prof Dr Med Dent2/
Silke Besdo, Dr-Ing 3/Lothar Borchers, Dr-Ing 1/Meike Stiesch, Prof Dr Med Dent4

Purpose: Since bacterial leakage along the implant-abutment interface may be responsible for peri-implant
infections, a realistic estimation of the interface gap width during function is important for risk assessment.
The purpose of this study was to compare two methods for investigating microgap formation in a loaded
dental implant, namely, microcomputed tomography (micro-CT) and three-dimensional (3D) nonlinear
finite element analysis (FEA); additionally, stresses to be expected during loading were also evaluated by
FEA. Materials and Methods: An implant-abutment complex was inspected for microgaps between the
abutment and implant in a micro-CT scanner under an oblique load of 200 N. A numerical model of the
situation was constructed; boundary conditions and external load were defined according to the experiment.
The model was refined stepwise until its load-displacement behavior corresponded sufficiently to data
from previous load experiments. FEA of the final, validated model was used to determine microgap widths.
These were compared with the widths as measured in micro-CT inspection. Finally, stress distributions were
evaluated in selected regions. Results: No microgaps wider than 13 µm could be detected by micro-CT
for the loaded implant. FEA revealed gap widths up to 10 µm between the implant and abutment at the
side of load application. Furthermore, FEA predicted plastic deformation in a limited area at the implant
collar. Conclusion: FEA proved to be an adequate method for studying microgap formation in dental
implant-abutment complexes. FEA is not limited in gap width resolution as are radiologic techniques and
can also provide insight into stress distributions within the loaded complex. Int J Oral Maxillofac Implants
2016;31:e32–e42. doi: 10.11607/jomi.4179

Keywords: dental implant, finite element analysis, microcomputed tomography, microgap formation, preload,
tightening torque

S ince the first description of osseointegration by

Brånemark et al,1 dental implants have been an
important element in restorative dentistry, with current
long-term success rates of at least 94%.2 Nevertheless,
1Research Associate, Department of Prosthetic Dentistry and
Biomedical Materials Science, Hannover Medical School,
Hannover, Germany.
2 Associate Professor, Department of Prosthetic Dentistry and
Biomedical Materials Science, Hannover Medical School,
Hannover, Germany.
3Research Associate, Institute of Continuum Mechanics,
Leibniz University Hannover, Hannover, Germany.
4Head, Professor, Department of Prosthetic Dentistry and
Biomedical Materials Science, Hannover Medical School,
Hannover, Germany.
Correspondence to: Dr Lothar Borchers, Department
of Prosthetic Dentistry and Biomedical Materials
Science, Hannover Medical School, Carl-Neuberg-Str. 1,
30625 Hannover, Germany. Fax: +49 511 532 4790.
Email: Borchers.Lothar@mh-hannover.de
©2016 by Quintessence Publishing Co Inc.
failures and complications still occur, often with unclear
causes. The principal complications, occurring in up to
32% of implant placements,3 are acute inflammatory
processes of the tissue around the implant-abutment
interface, and this can result in bone loss.4–6 Inflamma-
tion may be due to bacterial leakage from microgaps
at the interface between the implant and abutment,
which might serve as a bacterial reservoir.7 Microgaps
can be induced during manufacture by different fabrica-
tion tolerances, by misfit, relative micromovement and
plastic deformation of the connecting parts, or by screw
loosening.8,9 Several studies using different techniques
have been dedicated to this problem. The possibility
of bacteria penetration through the interface between
the implant and abutment has been investigated using
dyes10–12 or bacteria suspensions5,12,13 as penetrants. Gap
widths have also been quantified by scanning electron
microscopy (SEM),5,10,11 microcomputed tomography
(micro-CT),12,14,15 and microfocus x-ray videos.16 For
example, in the study of Jansen et al,5 microbial leakage
occurred in 69% of the specimens tested, even though

The International Journal of Oral & Maxillofacial Implants e32

© 2016 BY QUINTESSENCE PUBLISHING CO, INC. PRINTING OF THIS DOCUMENT IS RESTRICTED TO PERSONAL USE ONLY.
NO PART MAY BE REPRODUCED OR TRANSMITTED IN ANY FORM WITHOUT WRITTEN PERMISSION FROM THE PUBLISHER.
Jörn et al
microgap widths were found to be in the range of only
1 to 2 µm. Abdelhamed et al12 demonstrated leakage
of toluidine blue solution through microgaps in the
implant assemblies but could only estimate that the
gap widths were less than approximately 10 µm, the
resolution limit of the micro-CT device used. While in
these two studies, the implants (Osseo Speed, Astra Tech)
were inspected unloaded, Zipprich et al16 assessed the
gap widths of these and other implants under dynamic
loading. However, up to an oblique load of 200 N, the
gap widths were less than the resolution limit of the
analytical method (approximately 4 µm).
Because of the limited resolution of the aforemen-
tioned methods, no reliable conclusions are possible
about the tightness of the interface between the implant
and abutment against leakage of typical oral bacteria
with diameters down to 0.1 µm.14 As an alternative
method, finite element analysis (FEA) can be applied,
an approximate numerical method for solving problems
in structural mechanics that is capable of calculating
displacements of structures under simulated load with
unlimited resolution. In recent decades, FEA has been
used in implant dentistry to study the biomechanical
performance of dental implants,17 but to the authors’
knowledge, only two studies18,19 have used FEA to
investigate microgap formation in an implant-abutment
complex. As these latter studies neglected the mechanical
effects of abutment screw tightening, their results can
only be regarded as approximate.
Thus, it was the primary goal of this study to construct
and validate a realistic finite element model of an implant-
abutment assembly in an in vitro test set-up, and to
assess gap widths after loading by FEA and, additionally,
by micro-CT for cross-checking of results. Furthermore,
stresses in selected regions were evaluated by FEA, and
how far the results for microgap widths and stresses
were influenced by typical simplifications made in FEA
regarding screw preload18–22 and contact definitions17,23
at interfaces was investigated.
MATERIALS AND METHODS
Both the micro-CT experiment and three-dimensional (3D)
FEA were set up according to the load test described in
ISO 14801.24 The implant-abutment complex investigated
consisted of a titanium implant (Osseo Speed 4.5, conical
neck design, length 13 mm) and a titanium abutment
(TiDesign 4.5/5.0; diameter 5.5, 1.5 mm, both Dentsply
Implants, formerly Astra Tech; for material properties,
see Table 1).25–29
Micro-CT Examination
The implant was embedded with reinforced polyurethane
resin (PUR, Alpha Die Top, Schütz Dental) in a stainless
e33 Volume 31, Number 3, 2016
© 2016 BY QUINTESSENCE PUBLISHING CO, INC. PRINTING OF THIS DOCUMENT IS RESTRICTED TO PERSONAL USE ONLY.
NO PART MAY BE REPRODUCED OR TRANSMITTED IN ANY FORM WITHOUT WRITTEN PERMISSION FROM THE PUBLISHER.
steel cylinder (inner diameter 12 mm, wall thickness
1 mm, height 15 mm) in such a manner that, according
to the requirements of ISO 14801,24 an assumed bone
loss of 3 mm from the implant collar was simulated. Sub-
sequently, the abutment was connected to the implant
with the abutment screw and tightened with a torque
of 25 Ncm, as recommended by the manufacturer. A
hemispherical load cap made of cobalt chromium alloy
(coron, Etkon) was then cemented to the abutment,
giving a distance of 11 mm between the implant collar
and the site of force application.
For the examination by micro-CT, the implant assembly
was inserted into a special loading jig, as shown in Fig 1.
The loading jig basically consisted of a hollow aluminum
cylinder (inner diameter 14 mm, wall thickness 3 mm,
length 120 mm) surrounding the assembly, a calibrated
coil spring (spring constant 26 N/mm) for exerting an
axial force to the bottom of the steel cylinder with
implant assembly, and a loading plate, inclined by 30
degrees, in contact with the hemispherical load cap. As
the loading plate rested on a linear ball bearing, only
forces normal to the contact area could be transferred
to the load cap, which is, due to the inclination of the
plate, under an angle of 30 degrees with respect to the
implant axis. Microgap formation between the implant
and abutment was analyzed with an oblique load of
200 N resting on the specimen after adjusting the spring
force via a set screw accordingly. In the micro-CT device
(type X8011, Viscom), the specimen was inspected at 80-kV
acceleration voltage with 720 exposures per revolution
and 2-second duration per exposure. A 1024 × 1024
charge-coupled device (CCD) detector recorded the
transmitted radiation. By means of the software XVR-CT
(Viscom) and using a cone beam algorithm, the images
were reconstructed with a voxel edge length of 6.5 µm.
The latter was set by the program according to the
detector pixel size and the source-object and source-
detector distances, so neither over- nor undersampling
of the voxel size took place except for the systematic
variation due to the cone beam geometry. The effect
of beam hardening was reduced by the software with
the corresponding beam-hardening factor set to 6.5.
Finite Element Analysis
The geometrical data of implant body, abutment, and
abutment screw, as supplied by the manufacturer in
International Graphics Exchange Specification (IGES)
format, were imported into the computer-aided design
(CAD) part (Design Modeler) of the finite element program
ANSYS (Workbench 14.0, ANSYS), which was subsequently
used for FEA.
Analogously to the micro-CT experiment, the implant
was virtually embedded in polyurethane resin, and
a load cap was placed onto the abutment. To reduce
computational time, the abutment screw was modeled
 Jörn et al

Table 1  Material Properties Used for the FEA

Young's Yield Ultimate Ultimate Tangent
modulus Poisson's strength strength strain modulus
Part Material (MPa) ratio (MPa) (MPa) (%) (MPa)
Abutment and implant Titanium, 105,000* 0.36* 483* 550* 1525 461
Grade 4
Abutment screw Ti-6Al-4V 120,000* 0.36* 930* 1,095* 1826 958
Embedding resin Reinforced PUR 3.527,29 0.3327,29 – – – –
Load cap CoCr alloy 190,50028 0.2628 – – – –
Data marked by an asterisk were provided by the manufacturer; tangent moduli were calculated according to Equations 1 and 2.

1 2 3 4 5
20 mm

Fig 1   Loading jig with (1) tilted ceramic loading plate with frictionless linear ball bearing
to avoid generation of transverse force components, (2) embedded implant with load
cap, (3) loading piston, (4) loading spring, and (5) adjusting screw.

as a cylinder by infilling the 0.2-mm-deep thread, and

the 0.1-mm microthread at the implant neck was cut F
off (Fig 2).
In general, linearly elastic material behavior was
assumed, but for the implant components consisting
of titanium grade 4 and Ti-6Al-4V, respectively, plastic Load cap
deformation was taken into account by bilinear stress-
strain curves. To describe the bilinear stress-strain curves
of the material behavior, the tangent modulus (Εm) was Frictional
Abutment contacts (blue)
calculated, using the following equation:
Abutment screw
R –R
Εm = ε m – ε p0.2 (1)
m p0.2 2 mm I

with Rm being the ultimate strength, Rp0.2 the yield 2.3 mm S Detaching
contact zones of
strength, and εm the ultimate strain. The yield strain model C (red)
εp0.2 is derived from
Implant
Rp0.2
εp0.2 = Embedding
Ε (2) resin

where Ε is Young’s modulus.

The material properties assumed for the model com-
ponents are specified in Table 1.
To demonstrate the effect of often-made simplifica-
tions concerning interface contact conditions17,23 and
screw preload18,19 on FEA results, three geometrically
identical model variants (A, B, C) with different boundary
conditions were created. All variants had in common
Fig 2  Cross-sectional view of model representing differ-
ences in contact definitions of models A to C. Zones with con-
tact conditions varying between model variants are marked
in red (I = implant collar/embedding resin, contact length
2  mm; S = abutment screw/implant, contact length 2.3  mm);
zones with frictional contact in all variants are marked in blue.
F = force.

The International Journal of Oral & Maxillofacial Implants e34

© 2016 BY QUINTESSENCE PUBLISHING CO, INC. PRINTING OF THIS DOCUMENT IS RESTRICTED TO PERSONAL USE ONLY.
NO PART MAY BE REPRODUCED OR TRANSMITTED IN ANY FORM WITHOUT WRITTEN PERMISSION FROM THE PUBLISHER.
Jörn et al
Table 2  Configuration of the Three Model
Variants According to Contact and
Loading Conditions (Contact Zone
Designation According to Fig 2)
Contact behavior
Contact zone I Contact zone S
(Implant collar/ (Abutment screw/
Model embedding resin) implant )
A Bonded Bonded
B Bonded Bonded
C Detachment Detachment
enabled enabled
that fixed contact was assumed between the load cap
and abutment, and frictional asymmetric surface-to-
surface contact was chosen at the screw-abutment and
implant-abutment interfaces (marked in blue in Fig 2).
In models A and B, the implant, embedding resin,
and abutment screw were combined to form one com-
ponent assembly with bonded contacting faces. In
contrast, detachment in the contact zones S and I in
Fig 2 was permitted in model C. This corresponded to
the findings of a previous study,29 in which detachment
was observed in the respective areas. An overview of
the contact definitions is given in Table 2. Contacting
faces were always modeled using target and contact
elements (TARGET170 and CONTA174 in ANSYS), with a
friction coefficient of 0.4. For problem solving, the pure
penalty method was applied with the contact stiffness
factor set to 1.0.
In the analysis of model A, the screw preload due to
the tightening torque was neglected, and the implant-
abutment complex was directly loaded with an external
force of 200 N, built up incrementally and acting on the
load cap at an angle of 30 degrees with respect to the
implant axis. To simulate the posterior implant position in
the mandible, the load was directed toward the bevelled
surface of the abutment (Fig 2). In contrast, two load
steps were applied in the analyses of models B and C. In
the first load step, the tightening torque was simulated
by a preload (FP) of 271 N generated by contraction of
the screw shaft. FP was calculated from the tightening
torque TA, as recommended by the manufacturer, using
the following formula30:
TA
FP =
d × μT dH + dB
0,159 • P + S + × μH
α
2 × cos (2 (4
where P is the thread pitch (0.35 mm), dS is the screw
shaft diameter (1.83 mm), µT is the coefficient of friction
at the thread (0.4), a is the flank angle of the thread (60
e35 Volume 31, Number 3, 2016
© 2016 BY QUINTESSENCE PUBLISHING CO, INC. PRINTING OF THIS DOCUMENT IS RESTRICTED TO PERSONAL USE ONLY.
NO PART MAY BE REPRODUCED OR TRANSMITTED IN ANY FORM WITHOUT WRITTEN PERMISSION FROM THE PUBLISHER.
degrees), dH is the head diameter (2.35 mm), dB is the
bore diameter (2.03 mm), and µH is the coefficient of
friction at the screw head (0.4). In the second load step,
the oblique external load of 200 N was applied as in
the analysis of model A. To ensure static equilibrium of
the models, all bottom nodes of the embedding resin
were held fixed. The loading conditions chosen for each
Preload analysis are also summarized in Table 2.
Tetrahedral solid elements with quadratic trial function
Not
applied (element type SOLID187) were used. After appropriate
Applied
refinement at critical locations using semiautomatic mesh-
ing, the meshes consisted of a total of 347,515 (A and B)
Applied
and 399,913 (C) elements. Figure 3 shows cross-sectional
views of the meshed models used for the analyses. Very
small elements (size 0.05 to 0.1 mm) were chosen in the
vicinity of contact areas between the abutment screw
and the abutment, as well as between the abutment
and implant. Due to different contact definitions, mesh
sizes differ between models A/B and C.
For the validation of the finite element models, data
were available from five previous load experiments29 with
identical specimens and a set-up identical to that used
in the numerical analyses. Especially, conditions at the
implant-abutment interface (frictional contact) were the
same as defined in the FEA. The forces and correspond-
ing displacements (in force direction) of the point of
load application at the load cap had been recorded in
the experiments and could be extracted from the FEA
calculations as well. Agreement between the experimental
and numerical load-displacement curves was taken as
evidence for the validity of the respective model.
Furthermore, the stress results of the FEAs of model A
to C were compared at selected areas of the implant and
abutment to reveal differences due to different loading
and contact conditions. Additionally, abutment regions
were examined with high stresses after screw tightening
(at the screw seat) and after external loading (at the
abutment hexagon).
Finally, the postprocessor of the program was used to
evaluate the distribution of the microgap width between
the contacting surfaces of the abutment and implant,
defined as the distance between neighboring nodes of
previously specified contact and target elements in the
direction of the surface normal.
RESULTS
Evaluation of Microgap Formation with
Micro-CT
(3) The results of the micro-CT experiment are illustrated
in Fig 4. The images show sections of the implant-
abutment assembly under the oblique load of 200 N.
Two mutually orthogonal midsections (in the xy and yz
planes) are shown, together with a cross section in the
 Jörn et al

A/B C F
Possible start
of microgap

Y Y

a X b X

Z X

c Y d Z
10.00 mm
1 mm

Fig 3  Cross-sectional view of the meshed models
(left) A and B and (right) C.
Fig 4   Micro-CT sections of implant-abutment assembly ([b, c] midsec-
tions in xy and yz planes, respectively; [a] xy section enlarged 5× [red
box]; [d] cross section in xz plane approximately 65 µm below entrance
level of abutment into implant); implant under 200-N load (30 degrees).
Force components in respective planes indicated on right side. F = force.

xz plane perpendicular to the implant axis. In all sec-

tions, the transition from the abutment to the implant 200 
in the region of the conical seat is just visible as a thin 180 
line slightly darker than its environment and always 160 
with higher grey values than air pixels in other regions. 140 
Even in the magnification of the critical collar region, no 120 
continuous gap between abutment and implant can be
Force (N)

clearly detected. This leads to the conclusion that any
putative gaps would have to be considerably smaller
than two voxel edge lengths, ie, 13 µm.
Comparison of Load-Displacement Curves of
FEA and Experimental Results
The force-displacement curves as determined by exper-
iments and FEAs are compared in Fig 5. All curves run
almost linearly up to a displacement of 0.05 mm, but
then exhibit a transition to smaller slopes, possibly due
to the onset of plastic deformation or beginning relative
motion of contacting parts. The experimental curves,
generated with identical specimens, (Fig 5, experiments
1 to 5) are less steep than the FEA graphs and vary con-
siderably. Of the computed curves, those for models
A and C are very similar and, up to a displacement of
0.05 mm and a force of 80 N, are almost identical to the
graph for experiment 4. With increasing displacement
and force, however, the computed curves deviate more
and more from the experimental curves. The closest
approximation to experimental data was reached with
model C, the model with the most realistic simulation
of contact conditions.
100  Experiment 1
80  A
Experiment 2
B Experiment 3
60  Experiment 4
C
40  Experiment 5
20 
0 
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Displacement (mm)
Fig 5   Force-displacement curves calculated in FEAs for models
A, B, and C and experimental data for five identical specimens of
the same type as used in this study according to Dittmer et al.29
Von Mises Stresses in Abutment and Implant
The plots of von Mises stresses in the abutment (Fig
6) show similar distributions for all three models. The
highest stresses arose in the hexagon region, with the
highest stress levels present in model A. Models B and
C exhibited almost identical distributions, with slight-
ly higher stresses at the screw base than for model A.
A more thorough assessment and better compara-
bility are provided by the stress profiles given in Figs
7 and 8, which were each taken along identical paths
at the abutment surface for all three models. Figure 7
shows a plot of von Mises stresses after loading with

The International Journal of Oral & Maxillofacial Implants e36

© 2016 BY QUINTESSENCE PUBLISHING CO, INC. PRINTING OF THIS DOCUMENT IS RESTRICTED TO PERSONAL USE ONLY.
NO PART MAY BE REPRODUCED OR TRANSMITTED IN ANY FORM WITHOUT WRITTEN PERMISSION FROM THE PUBLISHER.
Jörn et al

Fig 6  Cross-sectional views

F of the abutment with distribu-
tion of von Mises stresses
after loading with 200 N for
models A, B, and C. The force
vector lies in the cross-sec-
tional plane. F = force.

A B C
10.00 mm

Fig 7  Von Mises stresses

at the abutment-screw seat
after loading with 200 N; the
400  Model A path of evaluation is marked
Model B in red in the cross section of
350  Model C F the abutment. The force vec-
tor lies in the cross-sectional
300  plane. F = force.
Von-Mises stress (MPa)

250 

200 

150 
A B
100 

50 

0 
0 1 2 3 4 5 6 7
A Path (mm) B

Fig 8  Von Mises stresses

at the lower edge of the abut-
500  ment hexagon after loading
with 200 N; the path of evalu-
450  ation is marked in red in views
of the abutment. F = force.
400 
F
Von-Mises stress (MPa)

350 
300 
250 
200 
150 
Model A
100  Model B
Model C
50 
0 
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Path (mm)

e37 Volume 31, Number 3, 2016

© 2016 BY QUINTESSENCE PUBLISHING CO, INC. PRINTING OF THIS DOCUMENT IS RESTRICTED TO PERSONAL USE ONLY.
NO PART MAY BE REPRODUCED OR TRANSMITTED IN ANY FORM WITHOUT WRITTEN PERMISSION FROM THE PUBLISHER.

Fig 9   Cross-sectional views
of the implant with distribu-
tion of von Mises stresses
after loading with 200 N, for
models A, B, and C. The force
vector lies in the cross-sec-
tional plane. F = force.
Fig 10  Von Mises stresses
at the implant collar after
loading with 200 N; path of
evaluation is marked in red
in views of the implant. F =
force.
Von-Mises stress (MPa)
200 N along a circular path at the abutment-screw
seat. Different stress levels were evaluated for all mod-
els. The lowest stress values—between approximately
5 and 80 MPa, with a maximum at the side facing
toward the force (point A in Fig 7)—were determined
for model A. The highest stresses, with strong local
variations but similar courses, occurred with models
B and C. They ranged between 130 and 190 MPa,
were slightly higher for model C than for model B,
and exhibited a slight increase at the side facing away
from the force. These values were much less than the
material’s yield strength (483 MPa).
Figure 8 displays von Mises stresses along a path
at the edge of the abutment hexagon facing toward
the external force. With all three models, the high-
est von Mises stresses, exceeding the yield strength
(483 MPa) of the material, were observed at both ends
of the path of evaluation. In the middle of the path, all
© 2016 BY QUINTESSENCE PUBLISHING CO, INC. PRINTING OF THIS DOCUMENT IS RESTRICTED TO PERSONAL USE ONLY.
NO PART MAY BE REPRODUCED OR TRANSMITTED IN ANY FORM WITHOUT WRITTEN PERMISSION FROM THE PUBLISHER.
Jörn et al
F
A B C
10.00 mm
600 
500  Model A F
Model B
A B
Model C
400 
300 
200 
100 
0 
0 2 4 6 8 10 12
A B
Path (mm)
profiles showed stress minima, the highest of which
(approximately 380 MPa) was evaluated for model
A, whereas models B and C exhibited similar stress
curves, with much lower minima of approximately
280 MPa.
For the implant, all analyses found that the highest
von Mises stresses developed at the implant collar
(Fig 9). Local stress maxima occurred at the side facing
away from the applied force, both at the collar and,
somewhat lower, on the outside of the implant at the
crestal level. At the seat of the abutment screw, the
highest stresses were found for model C.
The von Mises stress profiles along a circular path
at the implant collar demonstrate that yield strength
was reached and plastic deformation could take place
at the side opposite to the external oblique force of
200 N (Fig 10). The area of plastic deformation was
larger for model A than for models B and C.
The International Journal of Oral & Maxillofacial Implants e38
Jörn et al
Model A Model B Model C
2.00 mm
Evaluation of Microgap Formation with FEA
Microgap widths determined by FEA for all three models
are presented in Fig 11. Gaps in all three models emerged
in a semicircular shaped area at the side facing toward
the load. The largest gaps, with widths ranging between
2 µm at the bottom of the implant-abutment interface
and approximately 20 µm at the implant collar, were
calculated for model A, where screw preload had been
neglected. In contrast, gaps determined with both
model B and model C showed the highest widths of
approximately 10 µm only near the implant collar and
were confined to a small semicircle with a radius of
approximately 1.5 mm.
DISCUSSION
The aim of this study was to investigate microgap forma-
tion in an implant-abutment complex by FEA and micro-
CT. Three finite element models of different complexity
were generated and verified by comparing analysis
results with previously presented force-displacement
behavior in real load tests29 and gap measurements in
micro-CT experiments. To the authors’ knowledge, this
method for finite element model validation has not yet
been used in other studies of this kind.
To reduce the complexity of the finite element models
and to significantly save central-processing-unit time,
several simplifications had to be adopted. Furthermore,
specific boundary conditions had to be chosen, without
precise knowledge of relevant parameters. The possible
consequences are discussed in the following.
Cutting off the 0.1-mm-deep microthread at the
implant neck created a slightly more resilient support
of the abutment and may have led to slightly overes-
timated implant displacements, gap widths, and local
stresses in the analyses. In contrast, modeling the abut-
ment screw thread as a cylinder, by filling the thread
depth (approximately 0.2 mm) with material, slightly
stiffened the implant assembly and counteracted the
aforementioned effect.
e39 Volume 31, Number 3, 2016
© 2016 BY QUINTESSENCE PUBLISHING CO, INC. PRINTING OF THIS DOCUMENT IS RESTRICTED TO PERSONAL USE ONLY.
NO PART MAY BE REPRODUCED OR TRANSMITTED IN ANY FORM WITHOUT WRITTEN PERMISSION FROM THE PUBLISHER.
Fig 11  Gap width dis-
F
tribution on the implant-
abutment interface after
loading with 200 N for FEA
models A, B, and C; the
location of graphs is des-
ignated in the view of the
whole model on the right.
Another factor influencing the FEA results is the choice
of contact conditions for the interfaces between differ-
ent parts of the implant assembly. While friction has an
important impact on stresses in the implant components
and on microgap formation between the implant and
abutment,31–33 the exact coefficient of friction for the
implant materials titanium and titanium alloy is unknown
and depends on several conditions, including surface
topography and the presence of lubricant agents. Friction
coefficients given in the literature differ significantly,
ranging from 0.3 to 0.5.34–36 In the present analysis, a
friction coefficient of 0.4 was assumed to simulate dry
conditions between contacting surfaces. If lubricants
were present in the respective interfaces, the coefficient
of friction would decrease. This would result in higher
screw preloads, as evaluated in a previous study,33 and
therefore lead to somewhat smaller microgap widths and
slightly larger regions of possible plastic deformation
than determined here.
Different material models were chosen to represent
the behavior of the various model constituents. Linear
elastic properties were assigned to the load cap (NiCr
alloy) and embedding resin (PUR). Because of the rela-
tively low stresses in the load cap, this seems to be most
appropriate for NiCr. However, von Mises stresses in the
embedding resin of model A, for example, turned out
to be in a range of approximately 30 to 130 MPa after
loading with 200 N and thus sometimes exceeded the
compressive strengths of reinforced polyurethanes, which
lie between 80 and 170 MPa.37 This suggests that plastic
deformation might have occurred in the resin, which is
consistent with the more flattened course of experimental
load-displacement curves beyond a force of approxi-
mately 80 N, as compared with the curves determined
by FEA (Fig 5). Nevertheless, this effect should have had
a greater influence on the load-displacement behavior
of the implant-abutment complex as a whole than on
the stresses and gap formation within the complex.
In contrast to most other studies, in which linear
elastic material behavior was assumed for all materials
of the implant-abutment complex,18–20,38,39 the present

study introduced bilinear stress-strain curves for titanium
and titanium alloy to simulate material behavior more
realistically, thus taking plastic deformation into account.
This approximation, though incomplete because of the
lack of availability of true stress-strain curves for the
materials in question, proved to be necessary, as local
stresses indeed exceeded the material’s yield strength
(see Figs 8 and 10).
As expected, the most complex finite element model
(C) with the most realistic contact definitions delivered
the best approximation to the nonlinear experimental
load-displacement behavior of the embedded implant-
abutment complex (Fig 5). The differences between
the corresponding loads on the theoretical load-dis-
placement curve and the nearest experimental curve
are negligible for small displacements, but gradually
increase to a maximum of approximately 15% for higher
displacements. This is understandable in light of the
aforementioned limitations of the FEA, and the overall
agreement of experimental and numerical results is
regarded as reasonable evidence for the validity of
model variant C. If, for example, possible plastic behavior
of PUR could have been incorporated in the model, the
degree of approximation would have been even better,
as relatively low stiffness of the embedment would have
allowed higher implant displacements. In contrast, the
neglect of the possibility of detachment in the implant/
embedding resin and implant-abutment screw interface
regions in model B led to the load-displacement curve
being approximately 15% steeper than that calculated
with model C. This underlines the importance of care-
fully defining contact conditions in complex multipart
assemblies in this kind of FEA. Moreover, disregarding
the pretension of the screw (model A) allowed the abut-
ment more play in its socket. This increased force-driven
displacement and, finally, resulted in almost as good of
an approximation to experimental curves as with model
C, but in completely different results for stresses and gap
widths in the assembly.
Of the different boundary conditions in the model
variants, the introduction of screw preload in models B
and C had the highest influence on stresses in both the
implant and abutment. Tightening the screw understand-
ably resulted in higher stresses at the screw seat (Fig
7), pressed the abutment into the fixture, and thereby
reduced the ability of the abutment to tilt under the
influence of oblique forces. As a consequence, the high
stresses, which were caused by leverage effects at the
lower hexagon edge of the abutment at the side of force
application in model A, were alleviated by up to 20%,
and the risk of exceeding the yield strength (483 MPa)
was considerably reduced (Fig 8). At the same time, the
better fit of the abutment and socket after screw tighten-
ing distributed stresses more evenly and explained the
absence of local stress maxima in curves for models B
© 2016 BY QUINTESSENCE PUBLISHING CO, INC. PRINTING OF THIS DOCUMENT IS RESTRICTED TO PERSONAL USE ONLY.
NO PART MAY BE REPRODUCED OR TRANSMITTED IN ANY FORM WITHOUT WRITTEN PERMISSION FROM THE PUBLISHER.
Jörn et al
and C, as compared with that for model A (Fig 10). The
region with stresses above the yield strength, however,
stretched for about one-quarter of the path along the
implant collar in all three models, thus indicating possible
plastic deformation in this area due to horizontal forces.
It must be pointed out here that all analyses presented
were carried out with implants embedded only up to a
level of 3 mm below its collar, thus simulating the clinically
unfavorable case of substantial bone loss as prescribed
by ISO 14801.24 Therefore, the numerical results are to be
treated with caution and cannot readily be transferred
to the clinical situation. With an implant fully inserted
into alveolar bone and thus better supported against
lateral forces, maximum stresses as well as gap widths
and displacements are most likely to be smaller than
calculated in the analysis of the most realistic model
C, and the risk of plastic deformation would therefore
considerably diminish.
In comparison with the introduction of screw pre-
load, allowing detachment at the interfaces between
the implant and embedding resin as well as between
the implant and abutment screw, though essential for
the quality of approximation to experimental load-
displacement curves, resulted in only a small effect on
stresses (compare results for models B and C in Figs 6
to 10). The biggest difference can be observed at the
screw seat opposite to the side of force application
(Fig 7), where stresses rose by approximately 20%, as
a greater share of the load must be supported at this
side when detachment is allowed, and fixed contacts at
these interfaces were omitted that would have assured
support against bending.
FEA revealed a detailed picture of gap width for-
mation between the implant and abutment under
an oblique load. For the most realistic of the three
models investigated, namely, model C, taking into
account the abutment screw preload, the gap width
was largest (10 µm) at the side of load application near
the implant collar, but vanished within a distance of
approximately 1.5 mm from that point (Fig 11). Thus,
continuous access for bacteria to the dead space in the
implant underneath the abutment was not created.
This would have allowed steady presence, growth,
and colony forming of anaerobic pathogens there
and would have presented by far a higher risk for
the peri-implant health than the possible temporary
entering of oral bacteria into the said small semicir-
cular gap. A free access to the space underneath the
abutment, however, was indicated in the analysis of
model A, in which screw preload was neglected and,
as a consequence of missing counteraction against
the external tilting force, the maximum gap width was
twice as high (approximately 20 µm) as for the more
realistic model C. Since Pessoa et al,18 in an FEA for a
similar implant under almost identical external load
The International Journal of Oral & Maxillofacial Implants e40
Jörn et al
(200 N under 40 degrees) and with similar contact
definitions as in the present study, did not account
for the screw preload, their calculated gap widths
(18.4 µm and 16.1 µm, according to the simulated
healing state of the implant) corresponded roughly
with those presented here for model A, but might also
have been overestimated. The enormous influence on
calculated gap widths clearly shows the necessity of
incorporating screw-tightening torque into FEAs of this
kind when realistic results are to be obtained.
In contrast to FEA, radiologic methods are limited
in resolution and are not able to reliably identify gaps
if they are not at least as wide as two pixel or voxel
edge lengths. For this reason, the micro-CT experi-
ments in this investigation (with 6.5-µm voxel edge
length) could not directly corroborate the maximum
10-µm gap width calculated in the FEA, but only allow
the indirectly supportive statement that gaps, if they
existed, must be considerably smaller than 13 µm.
The discernability of gaps might have been improved
by using longer scanning times reducing background
noise and, eventually, by more sophisticated filtering
minimizing beam-hardening artifacts. Unfortunately,
both options were not feasible due to time limitations
for experiments and because of software restrictions.
More advanced hardware would have offered better
spatial resolution, theoretically down to a voxel size of
approximately 1 µm,40 given the specimen dimension
used, but were not at the authors’ disposal. Abdel-
hamed et al12 and Zipprich et al,16 using micro-CT
experiments and microfocus x-ray videos, respectively,
could also not detect any gap wider than the resolu-
tion limit of their systems (9.7-µm voxel edge length
and 4-µm pixel edge length, respectively) for the same
Astra Tech implant as used in this study. While the study
of Zipprich et al16 was undertaken under equal loading
conditions to those presented here (though the load
of 200 N was dynamically applied during the record-
ing of the video), Abdelhamed et al12 did not apply
external load to the implants in their micro-CT studies,
but, in supplemental experiments, also with unloaded
implants, could prove the existence of gaps with dye
solutions as an indicator. Jansen et al5 also verified the
presence of gaps in unloaded Astra Tech implants by
using bacteria suspensions. This suggests that gaps
had probably existed but, due to limited resolution,
could not be reliably detected by the x-ray methods
applied in these studies. In these cases, the detection
of a gap might have been made even more difficult by
beam-hardening effects in regions of greater material
thickness.41 FEA does not show these disadvantages
and promises to be well suited for investigating in
detail gap formation between the dental implant and
abutment. The essential prerequisite, however, is a
well-designed and validated finite element model.
e41 Volume 31, Number 3, 2016
© 2016 BY QUINTESSENCE PUBLISHING CO, INC. PRINTING OF THIS DOCUMENT IS RESTRICTED TO PERSONAL USE ONLY.
NO PART MAY BE REPRODUCED OR TRANSMITTED IN ANY FORM WITHOUT WRITTEN PERMISSION FROM THE PUBLISHER.
CONCLUSIONS
By using 3D nonlinear FEA, a maximal gap width of
10 µm between the implant and abutment could be
predicted for a specific system under an oblique load
of 200 N. The areal extension of the gap, however, was
determined to be too small as to allow oral pathogens
to enter the dead space underneath the abutment,
so the risk for peri-implantitis developing due to the
intermediate existence of this gap during loading may be
deemed minor. Because of resolution-based limitations,
the FEA results could not be directly corroborated by
micro-CT experiments. Thus, FEA proved to be superior
to the applied radiologic method for detailed studies
of gap formation in dental implants. Moreover, FEA is
capable of forecasting displacements and stresses in
loaded implants, thereby identifying, for example, local
stress concentrations with risk for plastic deformation
or even fracture. The quality of the predictions is greatly
enhanced by careful consideration of the material prop-
erties, including nonlinear constitutive laws, by proper
design of interface contact conditions and, especially,
by simulation of abutment screw tightening.
ACKNOWLEDGMENTS
This study was supported within the framework of “Son-der-
for-schungs-be-reich 599, Teilprojekt D12” by a grant from the
German Research Foundation. The authors thank Mr Richard
Werth for development and construction of the special loading
jig used for the micro-CT experiment. Special thanks are due
to Dentsply Implants (formerly Astra Tech), for kindly providing
geometry data of the implant system and additional support.
Finally, the assistance of Viscom with the micro-CT experiments
is gratefully acknowledged. The authors reported no conflicts of
interest related to this study.
REFERENCES
  1. Brånemark PI, Adell R, Breine U, et al. Intra-osseous anchorage of
dental prostheses. I. Experimental studies. Scand J Plast Reconstr
Surg 1969;3:81–100.
  2. Weber HP, Sukotjo C. Does the type of implant prosthesis affect
outcomes in the partially endentulous patient? Int J Oral Maxillofac
Implants 2008;23:140–180.
  3. Goodacre CJ, Kan JY, Rungcharassaeng K. Clinical complications of
osseointegrated implants. J Prosthet Dent 1999;81:537–552.
  4. Broggini N, McManus LM, Hermann JS, et al. Peri-implant inflam-
mation defined by the implant-abutment interface. J Dent Res
2006;85:473–478.
  5. Jansen VK, Conrads G, Richter EJ. Microbial leakage and marginal
fit of the implant-abutment interface. Int J Oral Maxillofac Implants
1997;12:527–540.
  6. Steinebrunner L, Wolfart S, Bössmann K, Kern M. In vitro evalu-
ation of bacterial leakage along the implant-abutment inter-
face of different implant systems. Int J Oral Maxillofac Implants
2005;20:875–881.
  7. Gross M, Abramovich I, Weiss EI. Microleakage at the abutment-im-
plant interface of osseointegrated implants: A comparative study.
Int J Oral Maxillofac Implants 1999;14:94–100.

  8. Brägger U, Aeschlimann S, Bürgin W, Hämmerle CH, Lang NP. Bio-
logical and technical complications and failures with fixed partial
dentures (FPD) on implants and teeth after four to five years of
function. Clin Oral Implants Res 2001;12:26–34.
  9. Malathi GK, Chandra RP. Mechanical complications with implants
and implant prostheses. IJDA 2011;3:555–558.
10. Piattelli A, Scarano A, Paolantonio M, et al. Fluids and micro-
bial penetration in the internal part of cement-retained versus
screw-retained implant-abutment connections. J Periodontol
2001;72:1146–1150.
11. Lorenzoni FC, Coelho PG, Bonfante G, et al. Sealing capability and
SEM observation of the implant-abutment interface. Int J Dent
2011;2011:864183. doi: 10.1155/2011/864183
12. Abdelhamed MI, Galley JD, Bailey MT, et al. A comparison of zirconia
and titanium abutments for microleakage. Clin Implant Dent Relat
Res 2015;17(suppl 2):e643–e651.
13. Quirynen M, Bollen CM, Eyssen H, van Steenberghe D. Microbial
penetration along the implant components of the Brånemark
system. An in vitro study. Clin Oral Implants Res 1994;5:239–244.
14. Rack A, Rack T, Stiller M, et al. In vitro synchrotron-based radiogra-
phy of micro-gap formation at the implant-abutment interface of
two-piece dental implants. J Synchrotron Radiat 2010;17:289–294.
15. Meleo D, Baggi L, Di Girolamo M, et al. Fixture-abutment connec-
tion surface and micro-gap measurements by 3D micro-tomo-
graphic technique analysis. Ann Ist Super Sanita 2012;48:53–58.
16. Zipprich H, Weigl P, Lange B, Laue HC. Erfassung, Ursachen und
Folgen von Mikrobewegungen am Implantat-Abutment-Interface.
Implantologie 2007;15:31–46.
17. Geng JP, Tan KB, Liu GR. Application of finite element analysis
in implant dentistry: A review of the literature. J Prosthet Dent
2001;85:585–598.
18. Pessoa RS, Muraru L, Júnior EM, et al. Influence of implant con-
nection type on the biomechanical environment of immediately
placed implants - CT-based nonlinear, three-dimensional finite
element analysis. Clin Implant Dent Relat Res 2010;12:219–234.
19. Saidin S, Abdul Kadir MR, Sulaiman E, Abu Kasim NH. Effects of
different implant-abutment connections on micromotion and
stress distribution: Prediction of microgap formation. J Dent
2012;40:467–474.
20. Çaglar A, Bal BT, Karakoca S, et al. Three-dimensional finite element
analysis of titanium and yttrium-stabilized zirconium dioxide abut-
ments and implants. Int J Oral Maxillofac Implants 2011;26:961–969.
21. Chun HJ, Shin HS, Han CH, Lee SH. Influence of implant abutment
type on stress distribution in bone under various loading condi-
tions using finite element analysis. Int J Oral Maxillofac Implants
2006;21:195–202.
22. Wu T, Liao W, Dai N, Tang C. Design of a custom angled abutment
for dental implants using computer-aided design and nonlinear
finite element analysis. J Biomech 2010;43:1941–1946.
23. Canullo L, Pace F, Coelho P, Sciubba E, Vozza I. The influence of
platform switching on the biomechanical aspects of the implant-
abutment system. A three dimensional finite element study. Med
Oral Patol Oral Cir Bucal 2011;16:e852–e856.
© 2016 BY QUINTESSENCE PUBLISHING CO, INC. PRINTING OF THIS DOCUMENT IS RESTRICTED TO PERSONAL USE ONLY.
NO PART MAY BE REPRODUCED OR TRANSMITTED IN ANY FORM WITHOUT WRITTEN PERMISSION FROM THE PUBLISHER.
Jörn et al
24. International Organization for Standardization (ed). Standard ISO
14801:2007 Dentistry - Implants - Dynamic fatigue test for endosse-
ous dental implants. 2007; International Organization for Standard-
ization, Geneva, Switzerland.
25. Deutsches Institut für Normung e.V. (ed). Standard DIN 17 862
Stangen aus Titan und Titanlegierungen, Technische Lieferbedin-
gungen. 1993; Beuth Verlag, Berlin, Germany.
26. ASM International (ed). Metals and alloys: Materials and coatings for
medical devices: Cardiovascular. 2009; 1st ed. ASM International,
Materials Park, Ohio, USA.
27. Dittmer MP. Ermittlung der Spannungsverteilung in viergliedri-
gen Zirkoniumdioxidbrücken mit Hilfe der Methode der Finiten
Elemente. Dissertation (Dr. med. dent.), Medizinische Hochschule
Hannover, Hannover, Germany, 2007.
28. Jörn D, Waddell J, Swain M. The influence of opaque application
methods on the bond strength and final shade of PFM restorations.
J Dent 2010;38(suppl 2):e143–e149.
29. Dittmer S, Dittmer MP, Kohorst P, et al. Effect of implant-abutment
connection design on load bearing capacity and failure mode of
implants. J Prosthodont 2011;20:510–516.
30. Bickford JH. An Introduction to the Design and Behavior of Bolted
Joints. New York: Marcel Dekker, 1995.
31. Sakaguchi RL, Borgersen SE. Nonlinear contact analysis of
preload in dental implant screws. Int J Oral Maxillofac Implants
1995;10:295–302.
32. Merz B, Hunenbart S, Belser U. Mechanics of the implant-abutment
connection: An 8-degree taper compared to a butt joint connec-
tion. Int J Oral Maxillofac Implants 2000;15:519–526.
33. Jörn D, Kohorst P, Besdo S, et al. Influence of lubricant on screw
preload and stresses in a finite element model for a dental implant.
J Prosthet Dent 2014;112:340–348.
34. Keith O, Kusy RP, Whitley JQ. Zirconia brackets: An evaluation of
morphology and coefficients of friction. Am J Orthod Dentofacial
Orthop 1994;106:605–614.
35. Seitzman LE, Bolster RN, Singer IL. IBAD MoS lubrication of titanium
alloys. Surf Coat Tech 1996;78:10–13.
36. Abkowitz S, Burke JJ, Hiltz RH. Titanium in Industry. New York: Van
Nostand, 1995 .
37. Gegauff AG, Garcia JL, Koelling KW, Seghi RR. Thermoplastic
composites for veneering posterior teeth—a feasibility study. Dent
Mater 2002;18:479–485.
38. Alkan I, Sertgöz A, Ekici B. Influence of occlusal forces on stress
distribution in preloaded dental implant screws. J Prosthet Dent
2004;91:319–325.
39. Lang LA, Kang B, Wang RF, Lang BR. Finite element analysis to
determine implant preload. J Prosthet Dent 2003;90:539–546.
40. Burghardt AJ, Link TM, Majumdar S. High-resolution computed
tomography for clinical imaging of bone microarchitecture. Clin
Orthop Relat Res 2011;469:2179–2193.
41. Barrett JF, Keat N. Artifacts in CT: Recognition and avoidance. Radio-
graphics 2004;24:1679–1691.
The International Journal of Oral & Maxillofacial Implants e42