Molecular ion sources for semiconductor ion

implantationA. Hershcovitch1, V. I. Gushenets2, D. N. Seleznev3, A. S. Bugaev2, S. Dugin4, E. M. Oks2, T. V. Kulevoy3, O. Alexeyenko4, A. Kozlov3, G.N. Kropachev3, R. P. Kuibeda3, S. Minaev3, A. Vizir2, and G. Yu. Yushkov2

1Brookhaven National Laboratory, Upton, New York 11973, USA 2High Current Electronics Institute, Siberian Branch of Russian Academy of Sciences, Tomsk 634055, Russia3Institute for Theoretical and Experimental Physics, Moscow 117218, Russia 4State Scientific Center of the Russian Federation State Research Institute for Chemistry and Technology of Organoelement Compounds, Moscow, Russia

•The work was supported in part by Plasma Sources LTD, Tomsk, Russia, under a grant from the Skolkovo Foundation, and in part under Contract No. DE-AC02-98CH1-886 with the US Department of Energy.

Molecular Ion Sources For Semiconductor Ion Implantation

As semiconductors become smaller shallow implantation is desired and ion energy needed for implantation decreases, resulting space charge (intra-ion repulsion) effects, which reduced beam currents and production rates. To increase production rates, molecular ions are used. Boron and phosphorous (or arsenic) ion implantation are needed for P-type and N-type semiconductors respectively. Carborane, which is the most stable molecular boron ion leaves unacceptable carbon residue on extraction grids. A self-cleaning carborane compound was synthesized: the m isomer of m-carborane-1,7-dicarboxylic acid (C4H12B10O4), when utilized in the ITEP Bernas ion source resulted in large carborane ion output, without carbon residue. Ion source acid operation at high temperatures still had carbon residue, which was remedied by a special O2 elliptical cross section dissociator that injected miniscule amounts of O unto the grid prevented carbon deposition without loading up power supplies. Pure gaseous processes are desired for enabling rapid switch among ion species. Molecular phosphorous was generated by introducing phosphine in dissociators via 4РН3 = Р4 + 6Н2 generated molecular phosphorous in a pure gaseous process (same applies for arsenic AsH3). Molecular phosphorous was then injected into the HCEI Calutron-Bernas ion source, from which Р4

+ ion beams were extracted. For avoiding the use of ovens in deep implantation, high charge state phosphorous and arsenic can be generated by driving the HCEI Calutron-Bernas ion source harder and absorbing the hydrogen molecules in Pd after the dissociator. Results from devices and some additional concepts are presented.

Historically: Ion Sources for High Energy;

Now Focus is on Low Energy Ion Sources Various types of ions, but mostly B, P, Sb, and As, are implanted, over a wide range of energies into some of the materials used in the construction of semiconductors. These energies range from as low as approximately 100 eV for shallow surface implantations, to as high as multi-MeV for deep implantation into the substrate. State of the art ion sources meet industry needs for the energy range of about 10 keV to about 300 keV. But at the two extremes (100’s of eV and at multi-MeV) of the energy range, there is a lot of room for improvement due to space charge limitations at the low energy range and due to inefficiency in acceleration at the higher energy range. Originally, the collaboration started to develop, E-MEVVA, pulsed metal vapor ion sources, to enhanced charge states of Lead and Bismuth, which previously achieved doubly charged ions. These elements were ionized to ion charge states of Pb7+ & Bi8+ with ion currents exceeding 200 mA. Next, step was to adapt these charge enhancement characteristics to ion sources that generate steady state multi-charged B, P, As, and Sb ions. Technically these enhancements can be adapted to DC ion implanters in order to improve upon present day high-energy ion implanters that use rf accelerators. This endeavor has resulted in record steady state output currents of higher charge state Antimony and Phosphorous ions: P2+ (8.6 pmA), P3+ (1.9 pmA), and P4+ (0.12 pmA) and 16.2, 7.6, 3.3, and 2.2 pmA of Sb3+ Sb4+, Sb5+, and Sb6+ respectively. However, the semiconductor industry has greater needs in the area of low energy (100’s of eV) ion implantation, where space charge problems associated with lower energy ion beams limit implanter ion currents, thus leading to low production rates. To tackle the space charge problem, two approaches were followed: using molecular ions and ion beam deceleration with gasless/plasmaless space charge compensation. 3 emA of positive decaborane ions were extracted at 4 keV and 0.2 mA of negative decaborane ions were also extracted. Now our focus is on clean carborane (C2B10H12) and molecular phosphorous (P4

+) ion sources, on pure boron plasma immersion, gas-fed universal P ion source, and on beam deceleration with gasless plasmaless space charge compensation.

Molecular Phosphorous• A modified Bernas-Calutron ion source was converted to a truly

universal ion source by generating molecular phosphorous P4+.

The extract current density was somewhat above 1 mA/cm2, which matches peak previous results. It’s the same ion source that generated record high charge phosphorous and 41 mA were extracted, of which singly charged boron made up over 70% of the total ion beam, as well as high charge state P.

• Breakthrough: now this ion source is purely gas fed that generates, without ovens, molecular phosphorous P4

+.

• It can also generate gas-fed P++++

and P+++.

Gas Fed Molecular Phosphorous• Phosphine gas PH3 flows through

dissociator (comprising of a metal coil heated up to approximately 500º C) for the following reaction to occur 4 PH3 (gas) P4 (gas) + 6H2 (gas); i.e. molecular phosphorous (P4) exits the dissociator in gaseous state and injected into an ion source for ionization and extraction.

HCEI Calutron-Bernas ion source for molecular phosphorous generation: working gas phosphine into special dissociator. Parameters: discharge voltage – 500 V, discharge current 140-160 mА, ion beam current 3 mA; extraction voltage 15 kV, gas pressure in vacuum vessel near the entrance to bending magnet and it was as low as 6×10-5 Tor. Parameters for optimal operating mode to maximize Р4

+ fraction of the ion beam current; emission slit is 1× 40 mm2.

Schematic of the ion source (1–7) and dissociator (8–11). 1 – discharge chamber, 2 – directly heated cathode, 3 – shield, 4 – anicathode, 5 – supressor, 6 – accelerating electrode, 7 – vapor line, 8 – case, 9 – thermal shield, 10 – quartz tube, 11 – spiral heater.

Ion beam composition with phosphine flow under the different temperature of the heater. Red line – heater current is less then 1,6 A, blue – about 1,7 A, black – more than 1,8 A.

Green line is the bending magnet current. The dissociator works rather fine and with higher temperature any PH items disappeared.

-1 0 1 2 3 4 5 6 7 8 9

0,00

0,02

0,04

0,06

0,08

0,10 P+

2

P+

3

I be

ams

(mA

)

time (s)

P+

4

P+

PH+

PH+

3

small window

0

6

12

18

24

30

36

I m

agni

t (A

)

Recently Current was doubled to over 1 mA P4

+ where P4+ fraction of 40%

In the HCEI embodiment, about half of the extracted ion beam is P4+. Very specific

embodiment and operating parameters are required to achieve these parameters:

1. Proper transfer line between the dissociator and the ion source that has specific combination of temperature, pressure & path length & made of Russian made stainless steel that facilitates recombination of P2+ P2= P4, to generate optimal P4 flux. Tube dimensions are: ID 4 mm, OD 5 mm, length L of at least 20 cm. Not possible to use Ta or W tubes because these materials are catalysts for P4 decomposition. Average pressure <p> in tube can be in the range of 10-100 mTorr. Hence, <p>xL = 200 to 2000 mTorr-cm. Tube temperature of range of 300 C to 800 C. These rather restrictive parameters needed to facilitates optimal recombination of P2+ P2= P4 Physics: 3-body collisions are required for recombination of P2+ P2= P4 due to momentum conservation (collisions can be also involve the wall). Reason for a minimal <p>xL is that too small <p>xL does not generate enough P4, other reaction in dissociator there other reactions 2PH3->P2+3H2 and 2PH3->2P+3H2. Temp range cold many P4 stick to wall, too hot a wall tube causes break up of molecules.

2. Cathode structure must be relatively small and be located near extractor (asymmetric to the source axis that’s perpendicular to beam direction) Thin plasma/e-beam might be an even better option.

3. For this particular ion source specific ion source plasma parameters in previous slide for optimal generation P4

+ are following: Discharge current density - about 1 A/cm2

Many previous attempts had poor results: R. Chow and Y. Chai, J. Vac. Sci. Technol. A 1, 49 (1983), where P4 is only a small fraction of the total flux (usually much less than 1%).

High Charge State P

Source can also generate gas-fed P++++ and P+++.

Truly Universal Ion Source

Carborane

Due to its stability carborane (C2B10H12) is a very attractive molecule for molecular boron implantation. However, in the process of generating carborane ions, carbon residue remains in the source and it’s especially problematic when such deposition occur on the grids, since it can lead to unacceptable “shadow” on wafers during implantation

Methods for mitigating graphite deposition were developed: self-cleaning carborane compounds.

Attempted Self-Cleaning Compounds

Three possible processes of chamber cleaning during the operation, based on reaction with carbon, were tested. First was fluorine, which is used by the ion implantation industry (usually between ion source operations). Fluorinated carborane (C2B10H11-F), in which one hydrogen atom was replaced by a fluorine. Second cleaning mechanism was based on so called Boudouard reaction where source surface is cleaned from carbon by CO2 produced in the discharge chamber as result of C4H12B10O4 or C3H12B10O2 (о-carborane 1-carboxylic acid) molecular fragmentation by the reaction СО2+С2СО, which goes from left to right at temperatures higher than 4000С. Pure oxygen, from molecular fragmentation, can clean surfaces by forming СО2 and СО that are pumped out. The difference between C3H12B10O2 and C4H12B10O4 is that the first contains one C-O-O-H acid molecule, while the latter is bi-acid.

New Compound: m-Carborane-1.7-dicarboxylic acid. Copper ion source with new compound operation (left photo) versus pure carborane operation in cooled ion source (right photo)

New Compound: m-Carborane-1.7-dicarboxylic acid. Molybdenum ion source chamber after uncooled ion

source operation on (right photo). Molybdenum chamber extraction slit after operation & extraction of

carborane beam (left photo).

Identical results were obtained with the “O” isomer

• To mitigate the graphite residue deposition problem for the o isomer of carborane the compound o-carborane- 1,2-dicarboxylic acid was sythesized. The results were identical to those of the m isomer: residue free ion source after over 4 hours of continous operation.

• No experiments were performed with p-isomers, since they are very hard and expensive to synthesize. Hence the p-isomers are of no practical use, and most likely of no commercial value.

Commercial Viability

Although initial self-cleaning carborane compound operations resulted in residue free ion sources and their extractors, the carborane output was much lower than that of pure carborane operation to a point where larger time-integrated implanted dose is obtained when using pure carborane and cleaning an ion source residue, especially when switching between sources (one operates, while the other is cleaned).

New operating modes were developed, which rectified the problem.

Initial results: low carborane output

Present results: carborane output comparable to pure

carborane operation

Parameters for Obtaining Good Results Experiments performed ITEP Bernas ion source with extraction slit of 1x20 mm2. Set operating parameters (to enable comparison between compounds) were extraction Voltage of 10 kV and arc Voltage of 250 Volt. In most operations, the pressure (outside the source chamber) was 6x10-5 Torr. Ion source operation was 4 hours steady state.

Oven & transfer line (to discharge chamber) temperatures were very critical: they ranged from 140-400ºC in the transfer line, and 200-500ºC in the oven depending on the compound. Temperature control required care, since transfer line heating affected the oven temperature and vice versa; uncooled chamber operation had only slight affect. The transfer line was heated first.Surprisingly, carborane current extracted from the ITEP Bernas ion source during o-carborane-1,7-dicarboxylic acid operation was higher than carborane current extracted during pure o-carborane operation. Total current extracted during acid operation was 0.3 mA, while during pure carborane operation it was 0.1 mA. The reason for a higher carborane output during o-carborane-1,7-dicarboxylic acid operation than pure carborane, is most likely due to higher discharge current 20 mA versus 16 mA and/or higher plasma density in the ion source chamber; most likely the two parameters are interrelated..

Pure Boron Discharge• A magnetron with a 2” diameter pure boron cathode was brought into

self sputtering mode by superimposing high current pulses over a DC heating discharge used to heat the cathode, or with hot conductive boron cathode. Ion beam was extracted from the plasma discharge. Beam analysis was performed: all the boron ions are single charged. The maximum boron ion fraction in the beam is greater than 99%, and the mean boron ion fraction, time-integrated over the whole pulse length, is about 95%. Up to 500 mA (in 100 μs pulses) were extracted at 15 kV. And later, boron ion implantation into silicon wafer samples was done.

• This kind of a boron ion source could be competitive and useful for semiconductor industry as an alternative to BF3.

• Utilization as a primary source for a plasma immersion implantation. • Heating by lamps or

Start with gas discharge

Until cathode exceeds

500OC

Pure Boron Discharge for Plasma Immersion Implantation

•Lower quality semiconductors are implanted by plasma immersion, where boron implantation is performed by sputtering boron with a pulsed argon discharge. However, extracted argon loads up power supplies, which does not occur in pure boron discharges. But, at room temperature boron is an insulator; cannot serve as cathode.

•Pure boron magnetron discharge with thermally isolated boron target in self-sputtering mode was achieved at HCEI utilizing two methods to heat a 2” boron cathode to at least 500C: heating the cathode with halogen lamps and a reflector; warming the boron in an argon discharge until the cathode can sustain the discharge in pure boron.

•Boron was self- extracted from the plasma discharge. Analysis was performed: all the boron ions are single charged with boron ion fraction greater than 99%. And, boron ion implantation into silicon wafer samples was performed. Original surface resistance of wafers was in the range of megaohms. After implantation, the resistance predictably reduced to several kiloohms, which indirectly confirms that there is no boron surface deposition.

Molecular Beam Transport and Deceleration

A novel LEBT is being developed to provide better molecular beam transmission through a bending magnet, and to facilitate beam slowing down. An electrostatic focusing system with long focus length was developed and tested for improved beam transmission through the separating magnet. And, after the bending magnet, an electrostatic undulator is being developed as a transport channel for total carborane beam transmission to the target.

Gasless Plasmaless Deceleration Though extremely difficult conceptually a gasless plasma-less space-charge

mitigation system work, it would be the ideal way to enhance low energy implantation rate. Originally it was claimed to be high risk, extremely high payoff. Now there is hope that undulators with a long focus system can be used to slow down ions with gasless plasmaless space charge mitigation.

ITEP Long focus system (Photo below)

Undulator above is being fabricated.

Simulation of the combined system showed that a 0.75 mA carborane (C2B10H12+)

ion beam was successfully extracted at 10 KV and decelerated down to 2 KV without space charge “blow-up” (beam expansion).