Surface Analysis of Failed Oral Titanium Implants

Marco Esposito,1,2 Jukka Lausmaa,3 Jan–M. Hirsch,2,4 Peter Thomsen1

1
Institute of Anatomy and Cell Biology, Göteborg University, Göteborg, Sweden

2
Department of Oral and Maxillofacial Surgery, Uppsala University Hospital, Uppsala, Sweden

3
Department of Chemistry and Materials Technology, Swedish National Testing and Research Institute SP, Borås, Sweden

4
Department of Oral and Maxillofacial Surgery, Göteborg University, Göteborg, Sweden

Received 18 February 1998; revised 25 August 1998; accepted 5 October 1998

Abstract: The aim of the present study was to investigate the surface topography, compo-
sition, and oxide thickness of consecutively failed, oral Brånemark implants in order to
determine possible causes for failure. The failure criterion was lack of osseointegration
manifested as implant mobility. Ten implants were retrieved before loading (early failures)
and 12 during a period of function up to 8 years (late failures). At retrieval, early losses did
not display any clinical sign of infection. All late failures were radiographically characterized
by peri-implant radiolucency and did not show infectious signs with one exception. No implant
seemed to be lost due to peri-implantitis (plaque-induced progressive marginal bone loss).
Twelve implants were analyzed by scanning electron microscopy (SEM), Auger electron
spectroscopy (AES), and depth profiling using a blind protocol. Two pristine fixtures, which
underwent the same preparation as the failed implants, were used as controls. In the SEM,
control samples were essentially free from macroscopic contamination, whereas failed im-
plants contained varying amounts of tissue residues. AES showed that all surfaces consisted of
Ti oxide and varying amounts of additional elements, with C dominating in most cases.
Nitrogen and sometimes Na, Ca, P, Cl, S, and Si were detected. The Si contamination was most
likely due to ion leaching from the glass vials used for storage. Depth profiles showed a typical
oxide thickness of 5– 8 nm for all samples. In conclusion, no significant changes in the oxide
layer composition or thickness as a result of implantation were observed. The results do not
indicate any material-related cause for the failures of these implants. Possible reasons for these
failures were impaired healing, asymptomatic infection, and overload. © 1999 John Wiley & Sons,
Inc. J Biomed Mater Res (Appl Biomater) 48: 559 –568, 1999

Keywords: dental implants; surface spectroscopy; medical device failure; osseointegration;

titanium

INTRODUCTION covered by a carbon-dominated contamination layer and trace

Based on the predictable results obtained by commercially pure
titanium (CP Ti) implants ad modum Brånemark,1–5 the reha-
bilitation of edentulous patients with implants according to the
osseointegration concept is an accepted and expanding treatment
modality. The integration of Ti implants in bone has been partly
ascribed to the biocompatibility of the surface oxide layer.6 The
surface of Ti consists of a thin (2–6 nm) oxide (mainly TiO2)
Correspondence to: M. Esposito, Institute of Anatomy and Cell Biology, Göteborg
University, Box 420, SE-405 30 Göteborg, Sweden
Contract grant sponsor: Swedish Medical Research Council; contract grant number:
9495
Contract grant sponsors: National Research Council of Italy (CNR); Einar and
Greta Asker Foundation; C. M. Lerici Foundation; Hjalmar Svensson Foundation;
Faculty of Medicine and the Faculty of Odontology, Göteborg University, Sweden
© 1999 John Wiley & Sons, Inc. CCC 0021-9304/99/040559-10
amounts of N, Ca, P, Cl, S, Na and Si.7,8
Several authors have underscored the importance of
implant surface cleanliness. For instance, it has been re-
ported that a small amount of fluorine contamination can
dramatically alter the surface oxide of Ti implants during
autoclaving.9 It has been hypothesized that surface con-
taminants may be released from the implant surface, en-
hancing and perpetuating the inflammatory response, thus
altering the healing process and possibly provoking the
dissolution of Ti.10 –21 It has also been suggested that
single monolayers of contaminants from the environment
or bulk material can invalidate the utility of implants.22 It
is, therefore, possible that some early implant failures
(when osseointegration fails to occur) may be caused by
the presence of contaminants on the implant surface,
whereas some late losses (when the achieved osseointe-
559
560 ESPOSITO ET AL.

TABLE I. Clinical Data of Patients with Failed Implants

Implant Positiond
Patient Implant Length and Primary Resorption Anatomy Implantation Possible Reason for Failure
Sample Datac and Diameter Stability and Bone Qualitye Time Patient’s Notes and Symptoms

Aa 46 m 7 mm L2 maxilla C T(6) 8 years Bruxist Overload

3.75 mm good 3 pain at percussion

B 48 m 13 mm L3 maxilla E T(7) 8 monthsf Bone graft Impaired healing

3.75 mm bad 4 Le Fort I pain at percussion

C — 10 mm — — 0 (control) — —
3.75 mm

D1 60 m 18 mm R1 maxilla E T(2) 18 months Overdenture Overload

3.75 mm good 3 pain at percussion

D2 18 mm L1 maxilla E 18 months Overdenture Overload

3.75 mm good 4 pain at percussion

Eb 77 m 15 mm L3 maxilla B T(6) 8 monthsf — Impaired healing

3.75 mm good 4 —

F1 56 m 10 mm L4 maxilla E T(7) 16 months Bone graft Overload/impaired healing

4 mm optimal 4 sinus lift pain at percussion

F2a 15 mm R3 maxilla E 16 months Bone graft Overload/impaired healing

3.75 mm optimal 4 sinus lift pain at percussion

F3a 13 mm R2 maxilla E 16 months Bone graft Overload/impaired healing

3.75 mm optimal 4 sinus lift pain at percussion

F4a 10 mm R1 maxilla E 16 months Bone graft Overload/impaired healing

3.75 mm optimal 4 nasal lift pain at percussion

F5 10 mm L1 maxilla E 18 months Bone graft Overload/impaired healing

3.75 mm optimal 4 nasal lift pain at percussion

F6 15 mm L2 maxilla E 18 months Bone graft Overload/impaired healing

3.75 mm optimal 4 nasal lift pain at percussion

F7 15 mm L3 maxilla E 18 months Bone graft Overload/impaired healing

3.75 mm optimal 4 sinus lift pain at percussion

G 54 f 10 mm R2 maxilla C T(6) 6 monthsg Thin bone Impaired healing

3.75 mm good 2 use of barrier pain at percussion

H 76 m 18 mm R2 mandible B T(5) 3 monthsf — Impaired healing/iatrogenic

3.75 mm optimal 3 —

I — 10 mm — — 0 (control) — —
3.75 mm

J1 68 m 13 mm R3 maxilla E T(6) 10 monthsf Bone graft Impaired healing/iatrogenic

3.75 mm bad 3 buccal onlay —

J2 10 mm L1 maxilla E 10 monthsf Bone graft Impaired healing/iatrogenic

3.75 mm bad 3 nasal lift —

(continued)

gration is lost after many years of function), could be due
to progressive dissolution of Ti over time.
It has been reported that, after implantation, a continu-
ous oxide growth occurs over the years (from 50 Å before
implantation up to 2000 Å, after 6 years of function).23
These preliminary findings were confirmed by others.24,25
According to Eliades,26 who referred to other studies, the
highest rate of oxide growth was found in areas of the
implant located in the bone marrow, whereas the lowest
rate was observed in cortical bone. It has been hypothe-
sized that the oxide growth could be induced by oxidizing
radicals liberated by cells undergoing metabolic processes
at the interface.24,26 –28 These oxidative processes could be
responsible for the release of Ti ions (indeed negligible)
observed in vivo29,30 and were considered to be indicative
of biological activity of Ti.23,26 –28
Based on in vitro observations, it has also been hypothesized
that hydrogen peroxide (H2O2) released during an inflammatory
process may react with the Ti surface, thus forming a Ti-
gel.28,31,32 The formation of this gel was suggested to be related
to the extensive oxide growth observed in vivo24 and was con-
sidered to be an important factor for the biocompatibility of Ti.
 SURFACE ANALYSIS OF FAILED ORAL TITANIUM IMPALNTS 561

TABLE I. (continued)
Implant Positiond
Patient Implant Length and Primary Resorption Anatomy Implantation Possible Reason for Failure
Sample Datac and Diameter Stability and Bone Qualitye Time Patient’s Notes and Symptoms

K 70 f 13 mm R1 maxilla E T(8) 12 monthsf — Impaired healing/iatrogenic

3.75 mm bad 3 —

La 52 m 10 mm R1 maxilla C P(3) 3 years — Overload

3.75 mm good 3 pain at percussion

M 37 m 15 mm L1 mandible B P(5) 8 months — Overload

3.75 mm optimal 2 pain at percussion

N 54 f 15 mm L3 maxilla B T(6) 6 monthsf — Impaired healing

3.75 mm good 3 —

O1 70 f 15 mm R2 maxilla B T(6) 6 monthsf — Impaired healing

3.75 mm good 4 —

O2a 15 mm R1 maxilla B T(6) 6 monthsf — Impaired healing

3.75 mm good 4 —
a
Implant accidentally dropped on the surgical sheet.
b
Implant handled with surgical gloves.
c
Age at implant installation; m ! male; f ! female.
d
R ! right; L ! left; 1, 2, 3, respectively ! mesial, medial and distal implant.
e
Jaw shape and quality39: A ! most of the alveolar ridge is present; B ! moderate residual jaw resorption; C ! only basal bone remains; D ! some resorption of the basal
bone has occurred; E ! extreme resorption of the basal bone; 1 ! almost the entire jaw is composed of homogeneous compact bone; 2 ! a thick layer of compact bone surrounds
a core of dense trabecular bone; 3 ! a thin layer of cortical bone surrounds a core of dense trabecular bone of favorable strength; 4 ! a thin layer of cortical bone surrounds a core
of low density trabecular bone of poor strength; T ! total and P ! partial edentulous, in parenthesis the total number of inserted implants.
f
Removed at abutment connection.
g
Removed 2 weeks after abutment connection.
Patient B lost another implant at the abutment connection which was accidentally lost.
Patient D (diabetes, high blood pressure and by-pass) refused bone grafting, therefore only 2 implants could be placed in the canine areas to support an overdenture. Two weeks
after the operation, dehiscences developed around both implants, which healed 2 weeks later. This complication was related to diabetes.
Patient F (2 heart attacks, high blood pressure and heavy smoker) was bone grafted and lost all implants (11) already twice.
Patient K (ankylosing spondylitis and high blood pressure) was bone grafted in the sinus and nasal cavity; however, the failed implant was not in graft.
Patient L was bone grafted in the sinus, but the failed implant was not in the graft. The provisional reconstruction fractured in the middle and the implant holding a cantilever
unit failed.

On the other hand, experimental and clinical studies have
yielded contradictory results both in terms of surface contam-
ination and oxide growth. In fact, it was reported that, in
some of the early animal experiments, when the implant
surface was not chemically cleaned, there was a black or
grayish slight discoloration of the tissue adjacent the abut-
ment.33 However, no relationship with implant losses was
reported. Ameen et al.34 analyzed CP Ti plates as received
from the manufacturer. After “handling contamination” with
stainless steel tweezers, no obvious change in the surface
chemistry was detected. Thus, the authors questioned the
validity of rigorous protocols in handling Ti implants, and
concluded that the biological relevance of surface contami-
nants for osseointegration (if any) could only be speculated
upon. Another experimental investigation21 did not demon-
strate any statistical difference in bony contact between bio-
logically contaminated implants and standard controls. How-
ever, it was still recommended to avoid soft tissue contami-
nation of implants in the clinical situation.
The chemical composition of 11 failed Brånemark fix-
tures, after cleaning and autoclaving, has previously been
investigated.35 No substantial oxide thickness increase was
found (both test and control samples had an oxide thickness
in the range of 10 nm). In addition, reduced oxide thickness
of Ti flame-sprayed IMZ implants was reported 1 year after
installation.36
The purpose of the present study was to analyze the
surface properties of consecutively retrieved Brånemark im-
plants in order to determine the possible role of surface
contamination and oxide growth for the failure of Ti im-
plants.
MATERIALS AND METHODS
CP Ti implants (Brånemark System, Nobel Biocare AB,
Göteborg, Sweden) were inserted at the Department of Oral
and Maxillofacial Surgery, Uppsala University Hospital,
Sweden (with the exception of sample G), to support fixed
prostheses according to the general principles given by Adell
et al.37 in partially and totally edentulous patients. Bone
grafting procedures were implemented when the available
bone volume was insufficient. After connection of the supra-
structure, patients were checked at 6 months, 1 year, and
thereafter annually. Periodical radiographs were taken after
fitting the suprastructure, at the 1, 3, and 5 year controls using
a standardized method with the parallel, long-cone tech-
nique.38
562 ESPOSITO ET AL.
Patient and Implant Characteristics
The criterion for implant removal was lack of osseointegra-
tion recorded as the slightest mobility tested by rotating
and/or moving every implant back and forward using instru-
ments. All fixtures were checked for mobility at abutment
connection and at each visit before bridge insertion. How-
ever, once the bridge was fitted, it was not removed unless
radiographs showed peri-implant radiolucency or excessive
marginal bone loss. The implants were in such cases tested
for mobility. No implant had to be removed for progressive
marginal bone loss during the study period (October 1996 –
April 1997). All the retrieved implants were mobile and were
surrounded by a radiolucent line on radiographs. Twenty-two
failed implants were consecutively removed from 13 patients
(Table I). The mobile implants were retrieved under local
anesthesia by gently unscrewing them with stainless steel
forceps, which were carefully positioned on the cover screw
or on the abutment in order to avoid any possible contami-
nation of the implant surface. The implants were rinsed with
sterile physiological saline (NaCl) solution and were imme-
diately immersed in such a solution contained in plastic vials
used for transporting tissue samples for histopathologic ex-
amination. Special care was taken to avoid any possible
source of contamination. However, in 6 cases the implant
dropped accidentally on the surgical sheet and, in one case,
the implant was touched with gloves (Table I). The implant
sites were carefully curetted from remaining soft tissue and
flaps raised to achieve primary closure.
The failed implants were retrieved after an insertion time
ranging from 3 months up to 8 years. The implants had
different lengths (from 7–18 mm). Twelve fixtures had been
placed in bone of poor density (type 4)39 and 13 in extremely
resorbed maxillas (type E).39 The majority of the losses (20)
occurred in maxillas of totally edentulous (9) or bone grafted
patients (10). Pain at percussion was experienced at 14 sites.
With the exception of patient L, no signs of infection such as
marginal mucosal swelling, redness, or suppuration were
present at retrieval. In patient L, abundant plaque accumula-
tion, redness, and marked bleeding tendency were observed at
the failed implant site.
Sample Preparation
The failed implants were immediately inspected for macro-
scopic soft tissue remnants, which were carefully removed
using CP Ti tweezers. All retrieved implants and 2 nonim-
planted controls were ultrasonically cleaned, separately, for
10 min in clean glass containers filled with sterile saline. No
macroscopic contamination or discoloration was visible to the
naked eye after this procedure. The failed and the 2 control
implants were placed in clean and sterile plastic or glass vials
(for details see Table II), filled with saline solution. There-
after, all vials were labeled with a code number and stored in
a refrigerator prior to mail postage to the surface analyst, who
did not have access to any clinical information.
Methods of Surface Analysis
Two control and 12 retrieved samples were analyzed with a
scanning Auger microprobe (SAM, Physical Electronics, PHI
660, USA), with facilities for scanning electron microscopy
(SEM) and Auger electron spectroscopy (AES). SEM was
used to investigate the implant surfaces with respect to sur-
face topography and the possible presence of macroscopic
films or particulates and also for selecting areas for AES
analysis. Auger survey spectra (30 –1030 eV) were acquired
from at least 2 different locations on each sample. The
analysis areas were selected either at the tip, flank, or bottom
of threads, the neck or the cylindrical portion of the implant,
or the apical cone (see Fig. 1). All survey spectra were
measured using a beam energy of 5.0 keV, beam currents
around 200 nA, and a CMA resolution of 0.6%. The electron
beam was defocused to a diameter of 200 !m, which was the
survey analysis spot size. The survey spectra were evaluated
semi-quantitatively, by noting the relative peak heights of the
detected elements and dividing these into 3 different levels
(see legend of Table II).
Overlayer thicknesses (organic films and surface oxide)
were measured by depth profiling, using 1.5 keV argon ions
for sputtering. Alternating sputtering and analysis cycles
(0.25 or 0.5 min of sputtering per cycle) were used. To ensure
a uniform and reproducible sputter rate and to minimize
artifacts due to sample geometry and surface roughness, the
analyzing electron beam was focused to 0.5 !m and posi-
tioned at an appropriate area on the sample, over which the
ion beam was rastered (1.5 " 1.5 mm2). Prior to each series
of analyses, the sputtering rate was calibrated against anod-
ized Ti samples with known oxide (TiO2) thicknesses. Typ-
ical sputter rates for TiO2 were around 5 nm/min. The oxide
thickness was defined as the depth at which the oxygen signal
had decreased to 50% of that at the oxide surface. In those
cases where a thick carbon overlayer was present, the thick-
ness of the latter was subtracted.
RESULTS
Surface Composition
In the SEM, the control samples were essentially free from
macroscopic contamination. Different degrees of organic res-
idues, appearing mainly as dark stains, were detected on most
of the retrieved samples (Figs. 2 and 3). The surface topog-
raphy was dominated by grooves and ridges along the ma-
chining direction and appeared essentially unchanged on the
retrieved samples, as compared to the controls.
The AES analyses showed that all surfaces consisted of Ti
oxide, with varying amounts of additional elements. Table II
gives the relative amounts, in 3 different levels, of the ele-
ments (except Ti and O) that were detected. All control
implants showed clear Ti signals, while for the retrieved
samples only weak or no Ti signals were detected at the
outermost surface. Carbon was the dominant signal in the
AES spectra for most cases. O, N, and sometimes Ca, P, Cl
 SURFACE ANALYSIS OF FAILED ORAL TITANIUM IMPALNTS 563

TABLE II. Results from Surface Analyses of Retrieved Implants

Samplea Packaging Material Analysis Locationb Contaminantsc dox (nm)d SEM Observations and Other Notes
A Glass B1 C Na Cl Ca P (Si) Organic film on cylinder
6 weeks B2 C Na Cl (P) (Ca)
T2 C (Ca) (P) (S)
F2 C (Ca) (P) (S) (Na) (Cl)
Cyl C (Ca) (P) (S)
Cyl C Ca (P) (S) Analysis of bright stain
Cyl C N O (S) Analysis of dark stain
T2 6
B Glass B1 C Cl (P) (Si) Films on threads
3 weeks B3 C Cl (P) (Si)
T1 C Si (P)
T2 C Si (P) (Ca)
F1 C (Cl) (P) (Si) (S)
Cyl C Si P (Cl) Analysis of dark area
Cyl C Na Cl P Si Analysis of bright area
T3 C NO Analysis of dark area
T3 6
Cyl 6
C Glass T1 C Si (Cl) (Ca) No macroscopic contamination
Control T3 C Si
3 weeks B1 C Si Na Cl Ca (P)
B3 C Si Na Cl Ca (P)
T4 C Si Na Cl Ca
T3 6–7
T8 6–7
D1 Glass F2 C Si Cl S Thick films on threads
3 weeks T6 C Si Cl S
B2 C N O Na Cl S (Si) No Ti detected
T3 #15 Carbon overlayer
T8 #10 Carbon overlayer
E Glass B1 C Si Mg Na Cl Ca
36 weeks B2 C Si Mg Na Cl Ca
T3 C Si Mg Na Cl Ca
Cyl 7–8
T6 8–11 Carbon overlayer
F1 Glass B1 C Si Ca Na Cl Patchy surface
32 weeks B3 C Si Ca Cl (P)
T5 C Si Ca
T12 C Si S Na (Mg)
T5 5
F6 Plastic B1 C Cl Na P (Ca) (Si) No films, stains or particulates
36 weeks B4 C Cl Na P
T5 C Cl P (Si)
T7 C Cl P
T2 6
T7 6
F7 Plastic T3 C Cl Na P Bright stains, dark films on cylinder
36 weeks T6 C Cl Na P
B5 C Cl Na P (Si)
Cyl C N Na Cl Analysis of dark film
T2 6
T8 6
(continued)
564 ESPOSITO ET AL.

TABLE II. (continued)

Samplea Packaging Material Analysis Locationb Contaminantsc dox (nm)d SEM Observations and Other Notes

I Glass B1 C Na Cl Si Ca (P) (F) Drying stains, no films

Control B3 C Na Cl Si Ca P F
31 weeks T4 C Si Na Cl Ca P F
T5 C Na Cl Si Ca P F Analysis of drying stain
Cyl 5
T5 7
J1 Plastic B1 C (Na) (Cl) (Si) (Ca) (P) (F) Dark films
29 weeks B4 C (Na) (Cl) (S) (P)
T4 C (Si) (Cl) (P)
T3 C N O Cl S Analysis of dark film
Cyl 5
T5 5
K Plastic B2 C Na Cl (S) Dark films
29 weeks B4 C Na Cl P
T5 C Na Cl (Ca)
T5 C N Cl S Analysis of dark film
Cyl 7
T5 7
L Plastic B2 C S P (Ca) (Na) (Cl) Patchy surface, dark films
26 weeks B4 C Ca P S (Na) (Cl)
T4 C Ca P S (Na) (Cl)
T6 C Ca P (S)
Cone 5
T4 7
M Plastic B3 C Ca P Na Cl (S) No macroscopic contamination
20 weeks B5 C Ca P (S) (Na) (Cl)
T6 C Ca P (Si) (Na) (Cl)
Cyl 7–8
T3 7
T6 7
O1 Plastic B2 C Ca S P (Cl) No macroscopic contamination
18 weeks B5 C Ca P (S)
T4 C Ca P (S)
T2 7
T7 5–6
a
Time elapsed between retrieval and analysis (storage time).
b
Denotes the location on the implant where analysis was performed (B1 ! bottom of 1st thread; T2 ! tip of 2nd thread; etc.; F ! flank; Cyl ! cylindrical part of fixture; etc.;
see also Fig. 1).
c
Elements (except Ti and O) that were detected. Dominant elements ($10%) are written in bold type, intermediate levels (# 1–10%) in normal type, and trace amounts (%1%)
in parentheses.
d
Surface oxide thicknesses, based on depth at which oxygen signal has decreased to 50% of intensity at oxide surface and assuming sputtering rate calibrated against TiO2.

Na, S, and Si were detected in the organic overlayers. The
main difference in surface composition between the retrieved
implants and the controls was that the former had signifi-
cantly higher amounts of C at the surface. The only exception
was sample E, which displayed an unusual Mg contamina-
tion. As can be observed in Table II, the high Si levels are
strongly correlated to the material (glass) used for storing the
implants prior to analysis.
Surface Oxide Thickness Measurements
The depth profiles showed typical oxide thicknesses of 5– 8
nm for all samples (Table II), with the exception of samples
D1 and E for which overlayer thicknesses up to 15 nm were
measured. However, both these samples were characterized
by significant organic overlayers, which led to depth profile
broadening and over-estimates of the oxide thickness.
DISCUSSION
Surface analysis investigations of failed implants have the
advantage that they do not provoke any additional discomfort
for the patients in contrast to histological studies, where, in
order to obtain useful information, an adequate amount of
 SURFACE ANALYSIS OF FAILED ORAL TITANIUM IMPALNTS
Figure 1. Schematic illustration showing different analysis loca-
tions on the implants and the numbering of threads.
tissue has to be retrieved and properly oriented. In addition,
these failed implants, being surrounded by a soft tissue cap-
sule, could be efficiently cleaned to allow for reliable analy-
ses, in contrast to failed implants that have previously been
embedded in plastic (unpublished observations) or successful
implants integrated in bone, for which analysis is hampered
by plastic and/or tissue residues.23,24,40 AES was here pre-
Figure 2. SEM micrograph of implant A showing no macroscopic
contamination and normal topography. The round dark areas were
caused by the electron beam during AES analysis. Scale bar ! 500
!m.
565
Figure 3. SEM micrograph of implant D1 showing an area with
pronounced organic residues (dark areas). The bright areas represent
parts of the surface with less organic material. Scale bar ! 500 !m.
ferred to other similar techniques such as XPS (X-ray photo-
emission spectroscopy), also called ESCA (electron spectros-
copy for chemical analysis), because it offers elemental anal-
ysis with high lateral resolutions (&0.1–1.0 !m) and precise
selection of the analysis area (by SEM).
The majority of failed implants were inserted in resorbed
maxillas, which in many cases had to be bone grafted prior to
implant placement. Pain at percussion was the only symptom
in 14 cases. No macroscopic signs of infection or inflamma-
tion were observed, with exception of patient L. This patient
had not showed up at the scheduled controls, but was seen in
the emergency room due to pain and bridge mobility. The
failed implant site was characterized by abundant plaque
accumulation, redness, and heavy bleeding tendency on prob-
ing. Radiographically the implant was entirely surrounded by
a radiolucent line. At clinical inspection, the 4-unit provi-
sional maxillary bridge was found to be fractured in the
middle and the mesial implant, supporting 1 premolar and a
mesial cantilever, was clearly mobile. The patient referred
that the routine hygienic measures in the area had stopped not
to further compromise the situation. Such a clinical picture
may indicate a superficial infection superimposed to a failure
etiology due to overload. Therefore, the late failed Brånemark
implants analyzed in this study had clinical characteristics
similar to the late failures investigated in a recent immuno-
histochemical investigation.41 In addition, the failure charac-
teristics and patterns of these implants are in agreement with
those observed in a recent systematic overview on the same
implant system,42 and is in accordance with the etiologies
(overload in relation to nonoptimal bone characteristics) that
are believed to play a major role for the losses of this type of
implants.43
All the implant surfaces analyzed consisted of Ti oxide,
with varying amounts of contaminants or organic matter
(mainly C and N, but also traces of Ca, P, Na, Cl, S, and Si).
The main difference observed between retrieved and control
implants was that the former had significantly higher C lev-
els, which is consistent with the thin organic films frequently
566 ESPOSITO ET AL.

TABLE III. Oxide Thicknesses of Nonimplanted and Implanted Brånemark Implants as Reported in the Literature

Implantation Time
Authors Nonimplanted (nm) Implanted (nm) (Range) Methodologyf
McQueen et al., 1982a,23 5 Af up to 200Bf (6 months–8 years) A and B
Sundgren et al., 1986c,24 3.5 up to 63b (6 months–8 years) C
Lausmaa et al., 1988d,25 2–6 up to 16b Not specified D and E
Klauber et al., 199014 2 — — D
Binon et al., 199248 17 Af; 10.8 Df — — A and D
Aparicio et al., 1992e,35 10 10–15b (3–20 months) D
Olefjord et al., 199317 4.2–4.6 — — E
Esposito et al., 1999 5–7 5–8 or 7–15b (3 months–8 years) D
a
No calibration of sputtering rate.
b
Organic overlayer on the implant probably leading to depth profile broadening.
c
Same material as in McQueen et al.23
d
Oxide thickness for nonimplanted samples depending on the sterilization procedure. Presented depth profile indicate similar oxide thicknesses for nonimplanted and retrieved
samples, if C overlayer is accounted for.
e
Failed implants cleaned and sterilized twice: before and after insertion by the authors.
f
A: Oxide thickness determined from the depth at which O signal has decreased to a concentration of 10% to 20%.
B: Oxide thickness determined from the depth at which S, P, and Ca profiles terminate.
C: Oxide thickness determined from O/Ti signal intensity ratio, taking into account escape depth and sputtering induced effects.
D: Oxide thickness determined from the depth at which the O signal has decreased to half its value in the bulk of the oxide.
E: Oxide thickness estimated from relative intensities of metallic and oxide contributions in XPS Ti2p spectra.

observed on retrieved samples44 and from samples exposed to
serum solutions.45 Some contamination may also have orig-
inated from several sources, e.g., the fabrication process, as a
result of the cleaning and sterilization procedures, from the
environment during handling and storage,7,9,46,47 at implant
insertion, or during retrieval and analysis preparation proce-
dures. In particular, the Si contamination was observed only
on the samples stored in glass vials (both tests and controls)
and not for those placed in plastic vials. In order to confirm
this hypothesis, 2 additional pristine implants, as received
from the manufacturer, were analyzed. No traces of Si were
found. Therefore, Si being one of the major constituents of
glass, the observed Si contamination was mainly attributed to
ion dissolution from the glass vials.
The only major exception in the surface composition of
failed implants was sample E, where Mg was found to be a
dominant contaminant. The sample was incautiously handled
with gloves. One possible source for this contamination
could, therefore, have been talc powder on surgical gloves.
In view of the present and previous findings,34 it is prob-
able that the slight surface contamination possibly induced by
accidentally dropping six of the implants on the surgical sheet
was not relevant for the analytical procedures used, because
no differences could be detected comparing the “accidentally
contaminated” sample to the remaining samples. This may
indicate, as previously suggested,34 that the clinical validity
of some rigorous protocols for handling Ti implants could be
carefully revisited.
With regard to surface contamination, the present findings,
for both test and control implants, are in good overall agree-
ment with other studies on unused7,14,25,48,49 and retrieved CP
Ti implants.23–25,35,36,44 However, this is clearly not the case
for the measured oxide thicknesses, for which contradictory
results, based on retrieved functioning implants, have been
presented. In fact, some authors reported a constant and
marked increase of the oxide layer over time for functioning
CP Ti implants.23–25 No significant oxide growth was ob-
served in the group of failed implants analyzed in the present
investigation, not even for the implant that had been in
function up to 8 years. It could be argued that the previous
studies were not dealing with failed implants and, conse-
quently, that the lack of oxide layer growth could have
conditioned the failure process. On the other hand, it should
also be recognized that reliable depth profiling of screw-
shaped machined (rough) samples requires careful position-
ing of the ion and electron beams to areas on the surface that
minimize artifacts due to sample geometry and roughness. In
addition, in order to obtain reliable oxide thickness measure-
ments, the area that is subjected to analysis must be free from
tissue or plastic remnants. It is noteworthy that in those cases
where increased oxide thicknesses were recorded,23,25 and for
sample D1 and E in the present investigation, the implant
surface was covered by a significant organic overlayer. Fail-
ure to respect these analysis conditions leads to severe profile
broadening and misleading results (i.e., over-estimated oxide
thicknesses). This may also explain the different oxide thick-
nesses reported by McQueen et al.23 and Sundgren et al.24
analyzing the same material. In fact, in the latter report, the
sputtering was calibrated and areas with less organic residuals
were selected for analysis; see Table III for details. In addi-
tion, also the length of the alternating sputtering and analysis
cycle was found to significantly affect the measurements of
oxide thickness.7
Instead, our results corroborate the finding of Aparicio and
Olive,35 who reported an essentially unchanged oxide layer
thickness. These authors, when discussing their findings in
relation to those of McQueen et al.,23 postulated that the
differences could be due to the shorter implantation periods
of their own implants (between 3–20 months). However, it is
also true that the latter specimens were carefully cleaned and
 SURFACE ANALYSIS OF FAILED ORAL TITANIUM IMPALNTS
not embedded, rendering the presence of macroscopic con-
tamination minimal, thus facilitating the analysis procedure.
Interestingly, other studies on retrieved Ti flame-sprayed
implants (IMZ), after accurate cleaning, displayed a signifi-
cant decrease of the oxide layer thickness compared to un-
used controls, which was interpreted as a loss of part of the
oxide layer in the surrounding tissues or during sample prep-
aration.36
The difference in oxide thicknesses reported by several
investigators (Table III) may also be partly attributed to
different methodologies adopted to measure the Ti oxide
layer, which might have led to different results. In fact, there
is no absolute standardized means of measuring Ti oxide
thickness on retrieved implants.
The present findings might have important consequences
for the hypothesis of oxide growth and the suggested Ti-gel
formation at the implant-tissue interface in vivo. Recently,
Eliades26 elaborated an intriguing hypothesis to explain the
contribution of biological substances to the oxide growth,
based on the postulate that the site of implantation affected
the thickness of the oxide layer (i.e., Ti oxide thickness in
bone cortex was believed to be 3– 4 times thinner than in bone
marrow). However, no evidence of altered Ti oxide growth in
relation to the anatomical location can be found in the liter-
ature or in the present material. No evidence supporting
directly or indirectly the hypothesis that Ti-gel formation may
occur in vivo31,32 can be found either. The argument that the
lack of Ti oxide growth was observed only for failed and not
for functioning implants may tentatively be used to explain
the failure etiology. However, oxide thickness measurements
on failed implants are more accurately performed. In addi-
tion, the same cell-derived oxidizing products that have been
used in the literature24,26 –28 to explain the oxide growth
around successfully bone-integrated implants are, in fact,
likely to be found in a higher concentration at a failed implant
interface due to the presence of inflammatory cells.41 There-
fore, the hypotheses used to explain the Ti oxide growth in
vivo24,26 –28 need to be carefully revisited, because there
might be no significant oxide growth occurring in vivo. The
presence of analysis artifacts may have induced some authors
to reach erroneous conclusions. In order to clarify this issue,
new surface analytical studies on retrieved successful im-
plants are required.
Since the clinical scenario and the results suggest that the
role of surface contamination and peri-implantitis41,43,50 may
be excluded, the present findings support the assumption that
the failures hereby analyzed may be attributed to impaired
healing, asymptomatic infection, or disruption of a weak
bone-to-implant interface51 (early losses) and to overload in
relation to the host characteristics (late failures).41
Nobel Biocare Norden AB is acknowledged for donating the control
implants.
REFERENCES
1. Adell R, Lekholm U, Rockler B, Brånemark PI. A 15-year study
of osseointegrated implants in the treatment of the edentulous
jaw. Int J Oral Surg 1981;10:387– 416.
567
2. Lekholm U, van Steenberghe D, Herrmann I, Bolender C,
Folmer T, Gunne J, Henry P, Higuchi K, Laney WR, Lindén U.
Osseointegrated implants in the treatment of partially edentu-
lous jaws: A prospective 5-year multicenter study. Int J Oral
Maxillofac Implants 1994;9:627– 635.
3. Jemt T, Chai J, Harnett J, Heath MR, Hutton JE, Johns RB,
McKenna S, McNamara DC, van Steenberghe D, Taylor R,
Watson RM, Herrmann I. A 5-year prospective multicenter
follow-up report on overdentures supported by osseointegrated
implants, Int. J Oral Maxillofac Implants 1996;11:291–298.
4. Henry PJ, Laney WR, Jemt T, Harris D, Krogh PHJ, Polizzi G,
Zarb GA, and Herrmann I. Osseointegrated implants for single-
tooth replacement: A prospective 5-year multicenter study. Int
J Oral Maxillofac Implants 1996;11:450 – 455.
5. Friberg B. Sterile operating conditions for the placement of
intraoral implants. J Oral Maxillofac Surg 1996;54:1334 –1336.
6. Kasemo B, Lausmaa J. Surface science aspects on inorganic
biomaterials. CRC Crit Rev Biocomp 1986;2:335–380.
7. Lausmaa J, Kasemo B, Mattsson H. Surface spectroscopic char-
acterization of clinical titanium implant materials. Appl Surf Sci
1990;44:133–146.
8. Lausmaa J. Surface spectroscopic characterization of titanium
implant materials. J Electr Spectr Rel Phen 1996;81:343–361.
9. Lausmaa J, Kasemo B, Hansson S. Accelerated oxide growth on
titanium implants during autoclaving caused by fluorine con-
tamination. Biomat 1985;6:23–27.
10. Baier RE, Meyer AE, Natiella JR, Natiella RR, Carter JM.
Surface properties determine bioadhesive outcomes: methods
and results. J Biomed Mater Res 1984;18:337–355.
11. Kasemo B, Lausmaa J. Biomaterial and implant surfaces: A
surface science approach. Int J Oral Maxillofac Implants 1988;
3:247–259.
12. Kasemo B, Lausmaa J. Biomaterial and implant surfaces: On
the role of cleanliness, contamination, and preparation proce-
dures. J Biomed Mater Res 1988;22:145–158.
13. Baier RE, Meyer AE. Implant surface preparation. Int J Oral
Maxillofac Implants 1988;3:9 –20.
14. Klauber C, Lenz LJ, Henry PJ. Oxide thickness and surface
contamination of six endosseous dental implants determined by
electron spectroscopy for chemical analysis: A preliminary re-
port. Int J Oral Maxillofac Implants 1990;5:264 –271.
15. Smith DC. Surface characterization of implant materials: bio-
logical implications. In: Davies JE, editor. The bone-biomaterial
interface. Toronto: Univ Toronto; 1991. p 3–18.
16. Smith DC. Dental implants: Materials and design consider-
ations. Int J Prosthodont 1993;6:106 –117.
17. Olefjord I, Hansson S. Surface analysis of four dental implant
systems. Int J Oral Maxillofac Implants 1993;8:32– 40.
18. Brånemark PI, Lausmaa J, Ericson L, Thomsen P, Brånemark
R, Skalak R. Anatomy of osseointegration and the transfer of
load. In: Fonseca RJ, Davis WH, editors. Reconstructive pre-
prosthetic oral and maxillofacial surgery. 2nd Ed. Philadelphia:
WB Saunders; 1995. p 165–224.
19. Zöller GO, Zentner A. Initial attachment of human gingival
fibroblast-like cells in vitro to titanium surfaces pretreated with
saliva and serum. Clin Oral Implants Res 1996;7:311–315.
20. Ericsson I, Persson LG, Berglundh T, Edlund T, Lindhe J. The
effect of antimicrobial therapy on periimplantitis lesions. An
experimental study in the dog. Clin Oral Implants Res 1996;7:
320 –328.
21. Ivanoff CJ, Sennerby L, Lekholm U. Influence of soft tissue
contamination on the integration of titanium implants. An ex-
perimental study in rabbits. Clin Oral Implants Res 1996;7:
128 –132.
22. Bjursten LM, Gould TRL, Skalak R, Baier RE, Beertsen W,
Ducheyne P, Kasemo B, Pilliar RE. Basic science committee.
In: van Steenberghe D, Albrektsson T, Brånemark PI, Henry PJ,
Holt R, Lidén G, editors. Tissue integration in oral and maxillo-
568 ESPOSITO ET AL.
facial reconstruction. Amsterdam: Excerpta Medica; 1986. p
511–513.
23. McQueen D, Sundgren JE, Ivarsson B, Lundström I, af Eken-
stam B, Svensson A, Brånemark PI, Albrektsson T. Auger
electron spectroscopic studies of titanium implants. In: Lee
AJC, Albrektsson T, Brånemark PI. editors. Clinical applica-
tions of biomaterials. Chichester: Wiley; 1982. p 179 –185.
24. Sundgren JE, Bodö P, Lundström I. Auger electron spectro-
scopic studies of the interface between human tissue and im-
plants of titanium and stainless steel. J Colloid Interface Sci
1986;10:9 –20.
25. Lausmaa J, Kasemo B, Rolander U, Bjursten LM, Ericson LE,
Rosander L, Thomsen P. Preparation, surface spectroscopic and
electron microscopic characterization of titanium implant ma-
terials. In: Ratner BD, editor. Surface characterization of bio-
materials. Amsterdam: Elsevier; 1988. p 161–174.
26. Eliades T. Passive film growth on titanium alloys: physico-
chemical and biologic considerations. Int J Oral Maxillofac
Implants 1997;12:621– 627.
27. Kasemo B, Lausmaa J. Metal selection and surface character-
istics. In: Brånemark PI, Zarb GA, Albrektsson T, editors.
Tissue-integrated prostheses. Chicago: Quintessence; 1985. p
99 –116.
28. Tengvall P, Lundström I. Physico-chemical considerations of
titanium as a biomaterial. Clin Mater 1992;9:115–134.
29. Ducheyne P, Willems G, Martens M, Helsen J. In vivo metal-
ion release from porous titanium-fiber material. J Biomed Mater
Res 1984;18:293–308.
30. Osborn JF, Willich P, Meenen N. The release of titanium into
human bone from a titanium implant coated with plasma-
sprayed titanium. In: Heimke G, Soltész U, Lee AJC, editors.
Clinical implant materials. Advances in biomaterials. Amster-
dam: Elsevier; 1990. p 75– 80.
31. Tengvall P, Elwing H, Sjöqvist L, Lundström I, Bjursten LM.
Interaction between hydrogen peroxide and titanium: a possible
role in the biocompatibility of titanium. Biomat 1989;10:118 –
120.
32. Tengvall P, Lundström I, Sjöqvist L, Elwing H, Bjursten LM.
Titanium– hydrogen peroxide interaction: model studies of the
influence of the inflammatory response on titanium implants.
Biomat 1989;10:166 –175.
33. Brånemark PI, Breine U, Adell R, Hansson BO, Lindström J,
Ohlsson A. Intra-osseous anchorage of dental prostheses. I.
Experimental studies. Scand J Plast Reconstr Surg 1969;3:81–
100.
34. Ameen AP, Short RD, Johns R, Schwach G. The surface anal-
ysis of implant materials. 1. The surface composition of a
titanium dental implant material. Clin Oral Implants Res 1993;
4:144 –150.
35. Aparicio C, Olivé J. Comparative surface microanalysis of
failed Brånemark implants. Int J Oral Maxillofac Implants
1992;7:94 –103.
36. Kuliralo M, Pireaux JJ, Caudano R, Dourov N. Surface studies
on titanium IMZ implants. J Biol Buccale 1991;19:247–253.
37. Adell R, Lekholm U, Brånemark PI. Surgical procedures. In:
Brånemark PI, Zarb GA, Albrektsson T, editors. Tissue-inte-
grated prostheses. Chicago: Quintessence; 1985. p 211–232.
38. Strid KG. Radiographic procedures. . In: Brånemark PI, Zarb
GA, Albrektsson T, editors. Tissue-integrated prostheses. Chi-
cago: Quintessence; 1985. p 317–327.
39. Lekholm U, Zarb GA. Patient selection and preparation. In:
Brånemark PI, Zarb GA, Albrektsson T, editors. Tissue-inte-
grated prostheses. Chicago: Quintessence; 1985. p 199 –209.
40. Lausmaa J, Linder L. Surface spectroscopic characterization of
titanium implants after separation from plastic-embedded tissue.
Biomater 1988;9:277–280.
41. Esposito M, Thomsen P, Mölne J, Gretzer C, Ericson LE,
Lekholm U. Immunohistochemistry of soft tissues surrounding
late failures of Brånemark implants. Clin Oral Implants Res
1997;8:352–366.
42. Esposito M, Hirsch JM, Lekholm U, Thomsen P. Biological
factors contributing to failures of osseointegrated oral implants.
(I) Success criteria and epidemiology. Eur J Oral Sci 1998;106:
527–551.
43. Esposito M, Hirsch JM, Lekholm U, Thomsen P. Biological
factors contributing to failures of osseointegrated oral implants.
(II) Etiopathogenesis. Eur J Oral Sci 1998;106:721–764.
44. Mouhyi J, Sennerby L, Pireaux JJ, Dourov N, Nammour S, Van
Reck J. An XPS and SEM evaluation of six chemical and
physical techniques for cleaning of contaminated titanium im-
plants. Clin Oral Implants Res 1998;9:185–194.
45. Healy KE, Ducheyne P. Hydration and preferential molecular
adsorption on titanium in vitro. Biomat 1992;13:553–561.
46. Doundoulakis JH. Surface analysis of titanium after steriliza-
tion: role in implant-tissue interface and bioadhesion. J Prosthet
Dent 1987;58:471– 478.
47. Keller JC, Draughn RA, Wightman JP, Dougherty WJ, Meletiou
SD. Characterization of sterilized CP titanium implant surfaces.
Int J Oral Maxillofac Implants 1990;5:360 –367.
48. Binon PP, Weir DJ, Marshall SJ. Surface analysis of an original
Brånemark implant and three related clones. Int J Oral Maxil-
lofac Implants 1992;7:168 –175.
49. Edwards BN, Gold BR. Analysis of surface cleanliness of three
commercial dental implants. Biomat 1992;13:775–780.
50. Isidor F. Histological evaluation of peri-implant bone at im-
plants subjected to occlusal overload or plaque accumulation.
Clin Oral Implants Res 1997;8:1–9.
51. Esposito M, Thomsen P, Ericson LE, Lekholm U. Histopatho-
logical observations at early failed oral implants. Int J Oral
Maxillofac Implants, to appear.