Tightening characteristics for screwed joints in

osseointegrated dental implants

Richard L. Burguete, MEng,a Richard B. Johns, PhD, LDSRCS,b
Toby King, BA,e and Eann A. Patterson, BEng, PhDd
University of Sheffield and Institute for Dental Implants, Sheffield, and University of Cambridge,
Cambridge, England

The significance of tightening abutment screws and gold cylinders to osseointe-

grated fixtures with the correct torque is demonstrated, and a simple relationship
between applied torque and screw preload is derived by use of mechanical
engineering principles. The principles of a number of tightening methods are
outlined and assessments made of their accuracy. The dierence between optimum
and design torque is highlighted. The necessity and means of achieving optimum
torque to ensure a reliable joint in clinical practice is discussed. (J PBOSTHET DENT
1994;71:692-9.)

W. ith an increased understanding of the mechan- of the preload in the screwed joint. The screw can be
ical principles related to screw tightening and an appreci- thought of as a spring, stretched by the preload, for which
ation of the fallibility of the intuitive “feel” when applying the stretch is maintained by the friction forces in the
torque by hand, various mechanical and electronic devices threads. Any transverse or axial external force that causes
have been developed for use with implants. However, little a small amount of slippage between the threads, no matter
information has been published about the merits of various how small, releases some of the stretch and some of the
tightening techniques. In addition, the more fundamental preload is lost. At this stage the greater the joint preload
question of exactly what level of tightness is desirable (up to a maximum equal to the ultimate strength), the
needs to be answered. This article addresses both issues. greater will be the resistance to loosening, because the fric-
The examples used in this article refer to the Branemark tion forces between the threads will be greater and a large
osseointegrated dental implant system (Nobelpharma, external force is required to cause slippage. In the second
Gothenborg, Sweden), but the principles are applicable to stage of loosening, the preload is below a critical value so
other screw-retained implant systems. that external forces and vibrations cause the mating
threads to turn or “back off.” Once this stage has been
REVIEW OF LITERATURE reached, the screwed joint ceasesto perform the function
The purpose of tightening any screwed joint is self-evi- for which it was intended and has failed.
dent; if a screw is not tight it cannot achieve the function The other major consideration in selecting an appropri-
of clamping together component parts. However, the ate level of tightness is the fatigue life of the screw. If the
appropriate level of tightness required in a particular sit- screw is tightened until it is “snug tight,” meaning all mat-
uation is much less obvious, although a specific torque is ing parts of the joint are in contact, then the screw or bolt
recommended for each screw in the implant system. The will experience all of the external load applied to the
long-held belief has been that to avoid loosening of the clamped parts to separate them. However, as the tighten-
screw, the torque applied when tightening the screw should ing torque is increased above the snug tight level, the pre-
be as high as possible. Bickfordi provides a detailed load increases and the screw or bolt will gradually receive
description of the process of loosening, which occurs in two increased protection against the external load. This pro-
stages.Initially external forces applied to the screwed joint, tection is increasingly beneficial to the fatigue performance
for instance, during chewing, lead to the effective erosion of the screw until the total load experienced by the screw
as a result of the preload and the external load is approx-
imately equal to the yield of the screw. When this load level
is exceeded, the fatigue performance of the screw decreases
aResearch Associate, Department of Mechanical and Process En-
gineering, University of Sheffield. drastically. The relationship between preload and fatigue
bProfessor, Institute for Dental Implants. life is illustrated for a gold alloy restorative screw in Fig. 1.
=Department of Engineering, University of Cambridge. These concepts have been described in the context of den-
dReader, Department of Mechanical and Process Engineering, tal implants by Patterson and Johns2
University of Sheffield.
Copyright @ 1994 by The Editorial Council of THE JOURNALOF
Therefore the aim in tightening a screwed joint is to
PROSTHETICDENTISTRY. achieve the optimum preload that will maximize the
0022-3913/94/$3.00+0. 10/1/53442 fatigue life while offering a reasonable degree of protection

592 VOLUME71 NUMBESI

BURGUETRETAL. THEJOURNALOFPROSTHETICDENTISTRY

0 200 400 600 800 1000

Number of cydes to failure (millions)

Fig. 1. Relationship between preload and fatigue life for gold restorative screw in Brfine-
mark osseointegrated implant. Fatigue tialysis is taken from Patterson and Johns2 for
screw with highest tensile load in prosthetic bridge supported by five implants. Joint stiff-
ness has been taken as one quarter of screw stiffness, eccentric loading of implant has been
assumed, and no confidence limits have been applied.

against loosening. When torque is applied to a “snug-tight” only indirectly proportional to one another because of the
joint, it is opposed by the forces required to stretch the gold influence of the friction forces under the head of the screw,
alloy screw and compress the clamped parts, such as the as explained in the preceding section. Fig. 2 shows the re-
prosthesis and upper regions of the abutment. It is also op- lationship between the torque reacted in the threads, R,
posed by frictional forces under the head of the screw and and the preload, P. The gradient of the graph is a function
in the mating threads, which induce reactive torques, H of the coe5cient of friction, geometry, and material prop-
and R, respectively. The relative size of these reactive erties; however, the variation in the geometry and material
torques is dependent on the frictional and lubricant prop- properties is usually small compared with that of the coef-
erties at their source. It is the reactive torque in the threads ficient of friction. The coefficient of friction is dependent
that produces the preload in the screw and is given by: on the hardness of the threads, the surface finishes, the
quantity and properties of the lubricant, and the speed of
R=T-H (1)
tightening. The designer has good control of the hardness
where T is the applied torque. Because the HIR ratio var- of the materials and the quality of the surface finishes pro-
ies with a large number of factors, the applied torque duced during manufacture. The coefficient of friction, ~1,
required to achieve the optimum preload is difficult to increases (Fig. 2) with the hardness of the materials and the
specify because it will vary from one screw to another, even surface roughness. The implant system designer has little
though the screws and their joints might be nominally the control over the lubricant or the speed of tightening,
same. This difficulty leads to the concept of an optimum although the use of electronic or powered torque control-
torque and a design torque; the former is that which actu- lers helps to maintain the speed of tightening within lim-
ally achieves the optimum preload, whereas the latter is the its. The quantity and condition of the lubricant in the
torque specified by the designer as necessary to achieve the abutment and restorative screws is almost unpredictable
optimum preload on the basis of nominal properties. It is unless there is no lubricant present, namely, the implant
common practice, particularly in proprietary literature, to and abutment threads are completely dry when assembled.
use the term “optimum torque”; however, in most in- The coefficient of friction will increase as the quantity of
rlinces this is misleading because it is the design torque lubricant decreases.
bout devices achieve, and the two terms are rarely synon- At a local level at the flanks of the threads, the friction
grprous. force F is related to the contact force, N, as follows:
F=pN (2)
The importance of tightening is the application of the The friction force, F, is equivalent to a torque, t, acting
@mum prelaad, but the operation of tightening involves about the axis of the screw where:
4;beapplication of torque. Applied torque and preload are t=F.r (3)

#UNE 1994 593

 BURGIJETE El’ N.

ILI =O.l , well lubricated metals

unlubricated Ti-Au

p=2.5, Au-Au in

i
Thread Torque, R (Ncm)
Fig. 2. Relationship between preload and thread torque for gold restorative screw based
on equation (6). Values of coefficient of friction taken from Bowden and TaborlO and from
Abkowitz et a1.l’

r is the distance from the axis of the screw to the center of

action of the forces (Fig. 3), which is nominally the pit&
radius of the screw. The sum or integral of t for all poinb
along the helix of the thread is equal to the thread torque,
R. The contact force, N, when resolved along the axis of the
screw, as shown in Fig. 3, and then integrated along the ho-
lix of the thread is equal to the preload, P, namely:
P = EN cos i3 (4)
and from equations 2 and 3
R = z (r.pN) (6)
hence a simplified relationship between thread torque and
preload can be obtained:

P=
[.I
$ XR
(6)
where B is the thread half angle and the term in parenthe-
ses is the gradient of the graph in Fig. 2. Equation 6 illue-
trates the influence of the geometry on the relationship,
namely, that the gradient of the graph in Fig. 2 will change
with the radius of the screw and the shape of the thresd.
Hence every screw design will have a different preload-
torque relationship.
The material properties are not explicitly included in
equation 6; however, the coefficient of friction, p, is depea-
dent on them. During loading the threads will deform and
change the geometric factors r and /3.These changes will ba
B functions of the modulus of elasticity, E, the Poissons ra-
tio (=lateral strain/longitudinal strain), and the yield
Fig. 3. Forces induced during tightening of a screw (A) stress, because both gross elastic deformation and local
thread interface and (B) head. plastic deformation in the roots of the thread will occur.

594 VOLUME 71 NUMBER 0

 BURGUETE ET AL. THE JOURNAL OF PROSTHETIC DENTISTRY

Yii Torque
------------m--e

Torque/Rotation Gradient

Rotation, 8
Fig. 4. Idealized torque-rotation curves for screw joint during tightening process.

A similar analysis can be performed for the underhead METHODS OF TIGHTENING

torque, H, with Fig. 3, so that the friction forces under the
head are equivalent to the torque, namely: There are three classesof tightening methods: (1) torque
control, (2) angle control, and (3) torque/angle control. To
H = G.d12 (7) our knowledge only the first class of methods is used in
assuming the contact forces, H, are uniformly distributed prosthetic dentistry. This class of methods aims to apply a
under the head. In addition: design torque calculated from equation 11 or its more rig-
= FH.M orous forms. The torque can be applied by hand or by a
G (8)
powered device. In the former instance the torque can be
and equating forces in the axial direction:
limited by a mechanical device such as a ratchet, whereas
M=Wi-P (9) in powered devices the torque is controlled electronically,
I where W is the force applied along the screw during the usually by limiting the power supplied to the device’s mo-
tightening process (“to push the screw in”). Hence com- tor.
I bining equations +I.,8, and 9: Angle control relies on the relationship between the
forces in the screwed joint and its deformation in terms of
H = ,‘ff( W + P)d/2 (10) the angular rotation of the screw. The relationship between
It is therefore clear that the underhead torque is also a the preload and the angular rotation, 8, is given by4:
function of friction, geometry, and material properties’lt
should be noted that the coefficient of friction, pi, may be
(12)
different from that found in the threads.
Finally, the applied torque is equal to the sum of the un- where a is the pitch of the thread, and KS and Kc are the
derhead torque, H, and the thread torque, R, thus combin- stiffnesses of the screw and the clamped components,
ing equations 1, 6, and 10: respectively. Estimates of these stiffnesses have been made
for the fixture screw and its clamped parts by Patterson
T=
[co@
E
1 p+“H(W+P)d
2 (11)
and Johns2; however, accurate determination of these
stiffnesses can usually only be made by experiment. Fur-
The derivation of this equation has been performed with ther difficulty arises because the angle 0 needs to be mea-
a number of assumptions concerning the uniformity of sured from the “snug-tight” condition, and this is difficult
I loads and simplifications of the geometry, so it should be to identify precisely. In general engineering practice5 the
used with caution to calculate torque and preload. How- method is found to give good accuracy only when the screw
ever, it is useful for examining trends and influences. A is tightened into the plastic region, but this is extremely
more rigorous derivation and resulting equation is pro- detrimental to the fatigue performance of the screw.
vided for a general case by Juvinall and Marshak.3 The previous two methods of tightening are aimed at

JUNE 1994 595

THE JOURNAL OF PROSTHETIC DENTISTRY BURGUETE ET AI.

Table I. Typical errors for different torque-controlled

tightening methods
Tightening method Torque setting Percentage error

Hand torque wrench Torque from estimates 23% -28%

of friction
Power nut runner 23%-28%
Torque wrench Torque estimatedfrom 17%-23%
strain-gaugedsample
Power nut runner 26%-43%

Data derived from Junker and Wallam

can be seen in Fig. 4. The power to the torque wrench is

10 20 32 45
cut off when the gradient drops to half its maximum value.
Torque Setting (Nan) The relaxation of the torsional stress, when the torque
A is removed, ensures that the screw is not left above its
yield point. Burguete7 has designed a prototype instru-
ment based on these principles for use in prosthetic
dentistry. His results suggest that optimum preload could
be reliably achieved in the gold alloy screws with this
method.
The method described by Hagiwara and Ohashi4 consists
of measuring tightening and loosening torques and then
estimating the clamping force from them. An initial esti-
mate of the tightening torque, Tt, is made, for instance,
with equation 11 and is applied to the screw. The torque Tt
required to loosen the screw by an arbitrary angle, y, is then
measured and, according to Hagiwara and Ohashi,4 the
preload actually achieved by Tt is given by:
P = -(r/u) (Tt - IT& (13)
assuming that (~/cos /3)2is 10.05, which generally holds for
-c metal International Standards Organization threads (typ-
10 20 32 45 ically in the gold alloy screw: (&OS /3)2= (0.2lcos
Torque Setting (Nan) 30)2 = 0.05). If the preload achieved by Tt is within accept-
6 able limits of the optimum value, then the screw is rotated
Fig. 6. Set torque against actual torque achieved by No- through -y to retighten it. If the preload achieved by Tt is
belpharma Torque Controller device during (A) low-speed outside the limits, the screw is rotated through an angle
and (B) high-speed tests. greater than or less than y, which depends on whether the
achieved preload was less than or greater than the optimum
value. The screw is then loosened by turning through y, and
achieving the design torque, because they rely on achieve- a new estimate is made of preload by use of equation 13 and
ment of a specified torque or angular rotation. The third the retightening and reloosening torques. This method has
class of methods is aimed at achieving the optimum torque, the advantage that any desired preload can be achieved
and hence does not use a preset torque or rotation. There without knowledge of the condition of the screwed joint.
are two main methods within this class: one is known as
joint control and is described by Boys and Wallace5 and ACCURACY OF TIGHTENING METHODS
Junker and Wallace,Gand the other has been most recently Torque control methods are the most widely used class
described by Hagiwara and Ohashi.4 Joint control works on of tightening methods in both general engineering and
the principle that the optimum preload corresponds to the dentistry. Junker and Wallace6 have analyzed most torque
yield point of the screw and sensesthe onset of yield in the control techniques to establish the possible magnitudea of
screw. The sensing is done by monitoring the applied errors, and the results are reproduced in Table I. Boys and
torque and rotation. A typical torque-rotation curve is Wallace5 have compared the torque/angle control method
illustrated in Fig. 4. At the onset of yield there is a dramatic with torque control and angle control. Torque control
change in the gradient of the torque-rotation curve, as methods yielded much higher scatter (= f 20 % ) compared

596 VOLUME 71 NUMBLI l

BURGUETE ET AL. THE JOURNAL OF PROSTHETIC DENTISTRY

(with masses hw@ng from it, in the

second series of tests it was rigidly
clamped by a lever arrangement)

I yj I
\ ’ /
TORQUE METER‘I 1 P _ OutputtoStrainGauge
I I Amplifier and Displays

I I 1
I 1 I

SQVAREMUM

_I
II COUPLING
F
(in the second series of tests tbii was
fitted with an im&nt system)

t
DRIVE FROM
SCREW-DRIVER

Fig. 6. Diagram of test rig used for measuring torque supplied by torque controller de-
vice.

with the angle and torque/angle gradient control (+ 8%). friction in the strain-gauged torque meter could be ac-
Golleen et a1.shave recently investigated the accuracy with counted for in the tests.
which persons can achieve a specified torque in a trial In the Cambridge tests the electronic torque controller
where clinicians were asked to apply 10, 20, and 32 N/cm was initially tested by the application of 10% of the set
torque on the appropriate implant component with hand- torque followed by a steady increase over approximately
held screwdrivers. The implant was attached to the sensor 0.5 second until the controller cut off the power. In 0.5 sec-
of an electronic torque meter. Golleen et aLa found a wide ond at the low-speed setting the screwdriver rotated
variation in the ability of the clinicians to perceive torque. through approximately 60 degrees. At each combination of
Each clinician was asked to apply each torque six times, torque and speed setting, three tests were performed and
and the values ranged from 0.71 to 13.1 N/cm, 1.4 to 33.7 the mean values are shown in Fig. 5. Both this series of tests
N/cm, and 8.2 to 36.2 N/cm for 10,20, and 32 N/cm torque, conducted in Cambridge and those performed in Sheffield
respectively. Tests have also been conducted on an elec- indicate that the torque controller typically achieves a
tronic torque controller, that is, a steady torque was torque equivalent to between 65% and 80% of the set
applied to the output of the screwdriver until the control- torque at the low-speed setting. The high-speed setting is
ler shut off the power at the preset limit. This test was con- only recommended for unfastening of screws. Additional
ducted at both speed settings and all three torque settings, tests were performed with a modified version of the appa-
and the results are s.hown in Fig. 5. These tests are not rep- ratus in Fig. 6 to investigate the effect of retightening a
resentative of the tightening process but indicate the screw. The winding spool was rigidly clamped with a lever
steady state torques that the device will apply. After these arrangement, and an implant system was fixed to the cou-
initial tests a second series of tests were performed at the pling so that the screwdriver could be used to tighten gold
University of Cambridge. The apparatus used in these tests alloy screws (10 N/cm) with both 3 and 4 mm abutment
is shown in Fig. 6, and the applied torque is displayed on cylinders, and also abutment screws (20 N/cm) and single-
both an oscilloscope and an analogue meter. The system tooth implant abutment screws (32 N/cm). The test was
was extensively calibrated with known torques, so that the performed at the low-speed setting for all torque settings.

JUNE 1994 597

THE JOURNAL OF PROSTHETIC DENTISTRY BURGUETE ET AL

Torque Setting
1 ONcm (3mm Cylinder)
----+- 1 ONcm (4mm Cylinder)
.. . . ..~...m..
20Ncm
40
--+-- 32Ncm
r

0 2 4 6 8 10 12

NO. of Pedal hm?S

Fig. 7. Torque achieved by Torque Controller device during repeated activation of con-
trol pedal.

700

600
Resulting range
of prabad
4 500 aoheii
v
.
; 400
T
o. 300

100

Fig. 8. Relationship between preload and applied torque for gold alloy indicates possible
degree of scatter in preload resulting from 25% scatter in applied torque.

An initially loose screw was tightened until the torque con- The results are shown in Fig. 7, and each point representi
troller cut out; the control pedal was then pressed a second the mean of three tests.
time and the applied torque was measured. The pedal was In the preceding tests the accuracy with which the design
repeatedly pressed until there was no change in the torque. torque could be achieved was investigated. The tests did

598 VOLUME ‘71 NUMBFiR 0

BURGUETE ET AL. THE JOURNAL OF PROSTHETIC DENTISTRY

not indicate the accuracy with which the optimum preload CONCLUSION
was achieved. The preload in the gold alloy or the abutment The importance of tightening the screwed joints in den-
screw would have to be measured directly, for instance, by tal implants to the optimum torque as opposed to the de-
a strain gauge located on the shank of the screw. This is not sign torque has been discussed. It is noted that most pro-
practical in an implant system. However, an estimate of the prietary literature loosely uses the term “optimum torque”
variation in preload under typical conditions can be ob- when “design torque” would be the more appropriate term.
tained with the aid of Fig. 8, which shows the relationship The methods currently available for tightening screw joints
between preload, P, and applied torque, T, based on equa- have been described, and those used in dentistry have been
Con 11 for both a poorly and a well-lubricated joint, in
investigated and found to be of limited value. The impor-
both instances with no axial force applied through the tance of considering each joint separately has been high-
torque device; thus N = 0. A scatter band of 25% on the
lighted, and the influence and detection of misfits between
applied torque can be seen to translate with additional components have been considered. It is suggested that the
variation of frictional properties to a potential scatter of
development of an appropriate torque device based on
approximately 500 N. The figure is based on the properties torque/angle control methods is needed.
of the gold alloy screw for which the suggested preload is
300 N.g
EFFECT OF MISFITS
It has thus far been assumed that a “snug-tight” condi- REFERENCES
tion is achieved at the onset of tightening and that in this 1. Bickford JH. An introduction to the design and behavior of bolted
condition all the Imating parts fit together perfectly. When joints. New York: Marcel Dekker, Inc, 1981.
these conditions (do not exist the preload-torque relation- 2. Patterson EA, Johns RB. Theoretical analysis of the fatigue life of fix-
ture screws in osseointegrated dental implants. Int J Oral Maxillofac
ship will be different. For a small degree of misfit, such as Implants 1992;7:26-33.
occurs when the screw hole in the prosthesis does not align 3. Juvinall RC, Marshak KM, Fundamentals of machine component de-
with that in the abutment, the geometric factors in equa- sign. 2nd ed. New York: John Wiley & Sons, 1991.
4. Hagiwara M, Ohashi N. A new tightening technique for threaded fas-
tion 11 will be changed as a result of the small-scale defor- teners, Proc ASME Offshore Mech Arctic Eng 1992;111-B:371-6.
mation of the fixture screw. When a prosthesis does not 5. Boys JT, Wallace PW. Design and performance of an automated con-
make complete contact with an abutment, there will be a trol system for fastener tightening. Proc In&n. Mech. Engrs., 1977;
191:371-80.
gap between all or part of the mating surfaces. If there is 6. Junker GH, Wallace PW. The bolted joint: economy of design through
no contact the preload is used to bring the mating surfaces improved analysis and assembly methods. Proc Instn. Mech Engrs
closer together and perhaps into contact. In this situation 1984;198Bz255-66.
I. Burguete RL. The design and construction of a prototype miniature
virtually no fatigue protection is obtained, because any ex- automatic yield sensing torque wrench [MEng thesis]. Sheffield (UK):
ternal load applied to separate the prosthesis from the The University of Sheffield, 1990.
abutment causes further tension in the screw and is not 8. Golleen KL, Vermilyea SG, Vossoughi J, Agar JR. Torque generated
using hand-held screwdrivers and torque drivers for osseointegrated
dissipated by relieving compressive stresses in the clamped implants. Proceedings of the Eighth Annual Meeting of the Academy of
parts. This is because the gap gives the clamped parts a low Osseointegration. San Diego, 1993:103-4.
degree of stiffness. Although torque control methods will 9. Rangert B, Gunne J, Sullivan DY. Mechanical aspects of a Branemark
implant connected to a natural tooth: an in vitro study. Int J Oral Max-
not detect this situation, both angle control and torque/ illofac Implants 1991;6:177-66.
angle control methods can be used to identify this type of 10. Bowden FP, Tabor D. The friction and lubrication of solids. Oxford:
misfit, because the angle of rotation from the point at which Oxford University Press, 1950.
11. Ahkowitz S, Burke JJ, Hiltz RH. Titanium in industry. New York: Van
significant torque is applied will be much greater than that Nostrand Co Inc, 1955.
in the usual situ.ation. This is because a greater axial
deflection of the clamped parts is necessary to bring them Reprint requests to:
DR. EANN A. PATTERSON
into contact. Therefore there are significant potential ad- DEPARTMENT OF MECHANICAL AND PROCESS ENGINEERING
vantages in the ‘use of torque/angle control to tighten UNIVEWITY OF SHEFFIELD
screws in dental implants, both in terms of improved qual- MAPPIN ST.
SHEFFIELD
ity in the tightening operation and sensing misfits in the Sl 3JD
system. ENGLAND

JUNE 1994 599