The Oak Ridge National Laboratory (ORNL) SITEX (Surface Ionization with Transverse Extraction) negative ion source utilizes a 100-V/20-A reflex arc discharge in a 1300-gauss magnetic field to generate Cs+ ions and H+ or D+ ions, depending on the beam required. A shaped molybdenum plate is placed directly behind the arc column. Cesium coverage on this plate is used to minimize the surface work function, which requires two-thirds of a monolayer coverage. Cesium coverage is adjusted both by cesium flow control into the arc discharge chamber and by temperature control of the converter using gaseous-helium cooling channels in the converter plate. Normal converter operational temperatures are 300° to 500°C. H/D beams are generated at the biased converter surface (− 150 V with respect to the anode) by Cs+ sputtering of absorbed hydrogen or deuterium and by the reflection-conversion mechanism of H+/D+ ions which strike the converter surface at 150 eV. The negative ions are accelerated through the 150-V plasma sheath at the converter surface and are focused by the converter geometry and magnetic field so as to pass through the exit aperture with minimum angular divergence. The ion optics of the SITEX accelerator has been calculated using the ORNL 3-D optics code and results in a divergence perpendicular to the slot of θ⊥rms = 0.35° and parallel to the slot of θ‖rms = 0.18°. This beam divergence should be adequate for injection into a radio frequency quadrupole (RFQ) for further acceleration. H beams of 650 mA, 18 kV, and 9 s, along with similar D beams have been accelerated at a 10% duty cycle. Development is under way to permit steady-state operation. The extracted electron to extracted H/D current ratio of 15%/5% has been achieved. All extracted electrons are recovered on the source with an electron-recovery system at an energy that is 10% of the first acceleration gap potential energy difference. The arc efficiency is 5 kW of ion source power per 1 A of accelerated H/D beam. We propose a method to produce intense Li beams for further acceleration and neutralization to be used in a charge-exchange alpha particle diagnostic scheme. Li ions would be generated by Cs+ sputtering of lithium adsorbed on the converter, with Li accelerated by a system whose performance is similar to that described above. Beams of fewer than 10 µA have been produced in accelerator sources by mechanisms similar to those we propose. If the Li production efficiency is as good as expected, we will be able to produce 100-mA Li beams with an emittance suitable for final acceleration by an RFQ accelerator. Arc efficiency, electron control, and optics are projected to be as good as those for H/D beams already produced. A proof-of-principle experiment will soon be conducted at modest cost by modifying existing SITEX equipment to measure the production efficiency of Li ions.