Tritium retention inside the vacuum vessel has emerged as a potentially ser
ious constraint in the operation of the International Thermonuclear Experim
ental Reactor (ITER), In this paper we review recent tokamak and laboratory
data on hydrogen, deuterium and tritium retention for materials and condit
ions which are of direct relevance to the design of ITER. These data, toget
her with significant advances in understanding the underlying physics, prov
ide the basis for modelling predictions of the tritium inventory in ITER. W
e present the derivation, and discuss the results, of current predictions b
oth in terms of implantation and codeposition rates, and critically discuss
their uncertainties and sensitivity to important design and operation para
meters such as the plasma edge conditions, the surface temperature, the pre
sence of mixed-materials, etc. These analyses are consistent with recent to
kamak findings and show that codeposition of tritium occurs on the divertor
surfaces primarily with carbon eroded from a limited area of the divertor
near the strike zones. This issue remains an area of serious concern for IT
ER. The calculated codeposition rates for ITER are relatively high and the
in-vessel tritium inventory limit could be reached, under worst assumptions
, in approximately a week of continuous operation. We discuss the implicati
ons of these estimates on the design, operation and safety of ITER and pres
ent a strategy for resolving the issues. We conclude that as long as carbon
is used in ITER and more generically in any other next-step experimental f
usion facility fuelled with tritium - the efficient control and removal of
the codeposited tritium is essential. There is a critical need to develop a
nd test in situ cleaning techniques and procedures that are beyond the curr
ent experience of present-day tokamaks. We review some of the principal met
hods that are being investigated and tested, in conjunction with the R&D wo
rk still required to extrapolate their applicability to ITER. Finally, unre
solved issues are identified and recommendations are made on potential R&D
avenues for their resolution. (C) 1999 Elsevier Science B.V, All rights res
erved.