Development of o-carborane boronization for MST

Transcription

1 Development of o-carborane boronization for MST J. Ko, D. J. Den Hartog, J. A. Goetz, P. J. Weix, and S. T. Limbach University of Wisconsin Madison, WI, USA 52nd APS DPP Meeting Chicago, Illinois, USA Nov 2010

2 Abstract A boronization technique using o-carborane sublimation, which is much safer and simpler than those using diborane or trimethylboron, is being developed for MST. Due to the pulsed nature of the wall-conditioning discharges used in MST, efforts have been made to increase the repetition rate of the discharges in order to increase the boronization duty cycle. The gaseous o-carborane, created by heating a powder, can be injected in a pulsed or continuous manner. In the former, throughput is enhanced by the flow of helium, the working gas of the discharge, but is limited by the short opening time of the puffing valve. Density control and plasma parameters such as D-alpha and impurity radiation will be compared before and after the boronization. Two aluminum coupons flush with the inner wall of the MST vessel, one located near the o-carborane injection port and the other opposite to it, will be used to analyze the film and to examine the uniformity of the coating. This work is supported by the U.S. Department of Energy. Special thanks to F. L. Tabares (TJ-II) and S. Anderson (HSX) for valuable advice and suggestions. 2

3 Coating the wall with low Z materials is a common practice in magnetic fusion experiments The vacuum wall can be covered by a layer of absorbed gas from the surrounding atmosphere (both impurities and plasma species). Impurities desorbed from the wall dominate the energy balance by radiating most of the power put into the plasma. Plasma species desorbed from the wall can lead to lack of density control. These absorption/desorption processes can be minimized by covering the wall with low Z materials such as carbon, boron, and lithium in addition to baking, various discharge cleanings, and gettering. Low-Z wall coating is relatively quick and reproducible compared to other wall conditioning techniques. Successful wall conditioning can lead to a wider range of plasma operating conditions, especially, higher densities without excessive radiation. 3

4 Various boronization schemes 4

5 Concerns in MST Toxic and explosive substances (general characteristics in most of boron compounds) are not allowed as a boron-carrier: MST machine resides in a college building with hundreds of students coming and going daily. The boundary of the machine area is not gas-sealed. Neither glow discharge nor heating the vessel is recommended as a method to decompose the boron-carrier molecules: Damage to Viton seals at the gaps in VCV Damage to the array of magnetic sense coils MST uses the Pulsed Discharge Cleaning (PDC) as a routine wall-conditioning technique: 5 A variation of Taylor discharge cleaning 2 ~ 3 msec of pulse with 5 second interval Working gas (helium) is pulse-puffed as well several tens of msec before the discharge (Equilibrium pressure ~ 0.1 mtorr) Ip ~ 100 ka, Te ~ 10 ev, ne ~ 10^19 /m3

6 O-carborane has been selected and tested for MST MST Ethanol

7 Pulsed nature of decomposing discharge is not favorable in terms of money and time This is fundamentally different from others (glow discharge of several Amperes). If we use the current PDC duty cycle, it will cost several hundred thousand dollars per boronization (~ 7 kg of o- carborane) and ~ 200 hours of PDC to make a 50 nm boron film. So the focus has been on how to increase the PDC duty cycle (the PDC repetition rate). The cost will be the reduction in plasma current (and maybe density due to the shorter puffing) which seems to be okay based on the parameters from other devices. sec Time to take for 50 nm B/C film Q_oCB: 0.07 torr l/sec Q_He: 12 torr l/sec (in nominal cases) Q_oCB / Q_He 0, model_d.ps, Wed Mar 3 13:27:

8 Small modification of current PDC power supply reduces the cycle by a factor of 10 Standard Modified Bt capacitance Bp capacitance Voltage (Bt & Bp) Peak power Helium puffing 5 x 240 uf 8 x 240 uf 4.7 kv 10 ~ 20 KJ/sec 38 msec 3 x 240 uf 6 x 240 uf 4.7 kv < 10 KJ/sec 3 msec Capacitor charging time 3.88 sec 0.52 sec ip (ka) x 10 n_fir (1e19/m3) x tlp1_te (ev) sec press (mtorr) sec 0, pdc_diag_time.ps, Fri May 28 11:11:

9 To maintain higher Ip with shorter PDC, fewer caps and shorter puffing help 25 colors: # of caps, symbols: puffing He puff 38 msec 30 msec 20 msec 10 msec 5 msec 3 msec 2 msec Bt/Bp caps 5/8 4/7 3/6 ip,ka 10 Shorter puffing 9 5 Longer puffing PDC PSO (msec) 4, bzn_ _ip.ps, Fri May 28 02:35:

10 To maintain higher density with shorter PDC, fewer caps and longer puffing help 1.2 colors: # of caps, symbols: puffing n_fir,1e19/m He puff 38 msec 30 msec 20 msec 10 msec 5 msec 3 msec 2 msec Bt/Bp caps 5/8 4/7 3/ Longer puffing PDC PSO (msec) Shorter puffing 4, bzn_ _n_fir.ps, Fri May 28 02:34:

11 To maintain higher Te with shorter PDC, fewer caps and shorter puffing help 16 colors: # of caps, symbols: puffing tlp1_te,ev He puff 38 msec 30 msec 20 msec 10 msec 5 msec 3 msec 2 msec Shorter puffing Bt/Bp caps 5/8 4/7 3/ Longer puffing PDC PSO (msec) 4, bzn_ _tlp1_te.ps, Fri May 28 02:34:

12 Two injection schemes under consideration Pulsed injection to save the amount of o-carborane pumped out during the non-plasma period (99.8 %) in a PDC cycle In this case, the puffing can be combined with the pressurized (~ 30 psi) puffing of the working gas (helium) to enhance the o-carborane throughput. Continuous puffing without any throughput enhancement In theory, 99.8 % of o-carborane injected will be wasted without participating in forming the film. Note this scheme may save money, but not time. He oven He oven : Piezoelectric pulsed puffing valve (PEV) 12 : Diaphragm (continuous) valve

13 Two injection schemes under consideration Rough pumping line Gate valve to MST 13 He supply O-carborane oven PEV (wrapped by heating tape) Oven temperature controller Diaphragm valve

14 Pulsed puffing scheme tests in vacuum b_he35c50b N He ocb b_he20c58 N He ocb O2 pressure (torr) time (sec) 0, rga_time_read_ b_he35c50b_a.ps, Fri May 28 11:49: pressure (torr) time (sec) 0, rga_time_read_ b_he20c58_a.ps, Fri Jun 4 10:01: Oxygen levels were also monitored from the RGA on It is suspected that the finite nitrogen and oxygen signals are due to the not-low-enough pressure in the oven, not a leak. 14

15 Finite He injection does help o-carborane throughput 2.5 ocb 2.5 ocb relative increase (55C) 35 psi 25 psi 15 psi puff valve on (msec) 2, rga_time_pv_mst_ c.ps, Fri Jun 4 15:11: relative increase (58C) 35 psi 30 psi 20 psi 35 psi (20C) puff valve on (msec) 2, rga_time_pv_mst_ a.ps, Fri Jun 4 15:11: Pulsed o-carborane injection with helium into vacuum (no plasma, no boronization) Both results qualitatively demonstrate 15 Finite He injection helps o-carborane injection. But pressure dependence is weak. No pulsed o-carborane injection without the pressurized oven

16 Residence times are longer than expected but reasonable and consistent. tau (sec) He (55C) 35 psi 25 psi 15 psi tau (sec) ocb cryo turbo C2 C1 RGA 6.0 tau (sec) He puff valve on (msec) (58C) 35 psi 30 psi 20 psi 35 psi (20C) puff valve on (msec) 6, rga_time_pv_mst_ c.ps, Fri Jun 4 15:11: , rga_time_pv_mst_ a.ps, Fri Jun 4 15:11: tau (sec) ocb puff valve on (msec) puff valve on (msec) 7, rga_time_pv_mst_ c.ps, Fri Jun 4 15:11: , rga_time_pv_mst_ a.ps, Fri Jun 4 15:11: Pumps 3 turbo s Cryo 1 Cryo 2* ocb oven Pumping speed (l/s) 1000 N *Not used during PDC 1000 H2O H, He Therefore, the residence time for He = V/S = 8 m3 / 3000 l/s ~ 2.7 sec 16

17 The ratio of ocb to helium increases with increased temperature exponentially 17 peak_ocb / peak_he ocb/he (55C) 35 psi 25 psi 15 psi (58C) 35 psi 30 psi 20 psi 35 psi (20C) puff valve on (msec) Very rough exponential fits imply the ratio will be ~ (yellow arrow) when the temperature is around 80 degc. More measurements at 70 degc (yellow box) falls on to the lower exponential. 4, rga_time_pv_mst_ a.ps, Fri Jun 4 15:11: , rga_time_pv_mst_ c.ps, Fri Jun 4 15:11: ocb/helium ocb/helium exp(c0*t) + c1, 20 msec puffing with 35 psi helium c0, c , , temperature (degc) exp(c0*t) + c1, 20 msec puffing with 35 psi helium c0, c , , , (30 psi) (20 psi) temperature (degc) 0, rga_time_pv_mst_temp_est_a.ps, Sat Jun 5 00:40: , rga_time_pv_mst_temp_est_b.ps, Fri Jun 11 16:01:

18 2, rga_temp_pv_mst_b.ps, Fri Jun 11 16:16: , rga_temp_pv_mst_b.ps, Fri Jun 11 16:16: Temperature dependence is consistently exponential in other puffing durations and helium pressures 10 msec puffing 20 msec puffing 30 msec puffing ocb/he ocb/he ocb/he , rga_temp_pv_mst_b.ps, Fri Jun 11 16:16: psi 35 psi 0, rga_temp_pv_mst_b.ps, Fri Jun 11 16:16: temperature (degc) temperature (degc) temperature (degc) 60 msec puffing 50 msec puffing 40 msec puffing ocb/he ocb/he ocb/he , rga_temp_pv_mst_b.ps, Fri Jun 11 16:16: , rga_temp_pv_mst_b.ps, Fri Jun 11 16:16: temperature (degc) temperature (degc) temperature (degc) 18

19 Continuous puffing scheme tests in vacuum (0.2 g) (1.8 g) C2 ON b_during_ocb_time 10-6 He (4) C1 & C2 ON C1 ON (same as PDC) N (28) b_during_ocb_time 10-6 N (28) pressure (torr) He (4) 0, rga_time_read_ b_during_ocb_time_all.ps, Tue May 4 02:31: Ocb (70) pressure (torr) Oven closed , rga_time_read_ b_during_ocb_time_all.ps, Sat Apr 17 08:34: Ocb (70) time (sec) time (sec) Oven open Pressures normalized by its maximum (background-to-signal ratio) Oven closed Oven closed time 2 % 8 % 15 % 35 % 0 t close + 30 min Oven open Oven open 19

20 Decay times are in the order of several minutes (too long) (0.2 g) (1.8 g) b_during_ocb_time /- 2.3 sec (2.83); 82.9 l/sec /- 42. sec (2.63); 23.0 l/sec b_during_ocb_time torr estimated pumping speed 2 of the fit /- 5.1 sec (2.28); 47.0 l/sec / sec (1.63); l/sec sec 0, rga_time_ b_during_ocb_time_b1.ps, Tue Apr 20 16:18: torr sec 0, rga_time_ b_during_ocb_time_a1.ps, Tue May 4 02:05: Fit function: P(t) = Po exp(-(t to) / ) + Pc, where Pc is fixed as the background level before opening the oven. is the decay time. Two decay times can be characterized. It is speculated that The shorter time constant (black in the plots: ~ 3 min): o-carborane residence time The longer time constant (red in the plots: 7 ~ 8 min): the characteristic time for the slow outgasing from the condensed (and sticky) o-carborane in the rough Al vessel wall 20

21 Systematic study implies there are two decay times rga_time_scan_ b_during_ocb_time_i2395d20s10.sav The decay saturates so that the decay time reaches ~ 350 min. pressure (torr) rga_time_scan_ b_during_ocb_time_i2395d20s10.sav , rga_time_scan_ b_during_ocb_time_e15.ps, Tue May 4 01:54: sec 400 decay time (sec) pumping speed (l/sec) , rga_time_scan_ b_during_ocb_time_e15.ps, Tue May 4 01:54: sec

22 Rough expectation from both methods hour Time to take for 50 nm B/C film 0.5 sec rep Q_He = 15 torr l/sec tau_he = 5.8 sec tau_ocb = 10 sec Q_oCB / Q_He 0, model_e.ps, Wed Jun 9 12:31: pressure (torr) continuous puffing (1.8 g) He (4) N (28) Ocb (70) time (sec) b_during_ocb_time 0, rga_time_read_ b_during_ocb_time_all.ps, Tue May 4 02:31: Pulsed puffing (3 msec PDC / 0.5 sec) Continuous puffing (3 msec PDC / 0.5 sec) Saves o-carborane, does not save time. However, enhancement of O-carborane throughput by helium is not large. 8-hour PDC will make about 1 nm B/C film, consuming 0.08 g of o-carborane. Neither o-carborane nor time is saved. 5-gram o-carborane loading will form 0.3 nm B/C film. Maybe a little hope from longer decay time experimental observed. 22

23 Three boronization attempts so far BZN Run BZN Run BZN Run Pulsed o-carborane puffing with He flow for 28 msec before each pulse. Pulsed discharge for 2 msec every 1 sec: Ip = 10 ka, ne = 5e17 / m3, Te = 5 ~ 10 ev. The total boronization time: 5 hours. Took 20 reference RFP plasmas shots right before and after the boronization session. Same as run conditions. The total boronization time: 11 hours. Took 20 reference RFP plasmas shots right before and after the boronization session. Continuous o-carborane puffing without He flow. Pulsed discharge condition same as run. The total boronization time: 2 hours Took 20 reference RFP plasma shots right before and after the boronization session. 23

24 Continuous puffing with fast-pulsed discharge looks promising Before/After ref BZN Before/After ref BZN Before/After ref BZN BIV Before After tot(ima_biv) Open: F = -0.3 Close: F = -0.2 The total amount of BIV and OIV within shots are plotted as a function of shotaveraged density tot(ima_oiv) OIV 0, bzn_read_ref_2010.ps, Wed Oct 13 02:58: No boron increase, no oxygen decrease observed in two pulsed-puffing tests. An order of magnitude increase in boron and a slight decrease in oxygen throughout this density range are observed. No noticeable decrease in other high z impurities in all cases. 20 No noticeable difference in density signals in all cases avg(n_co2) 1e19/m avg(n_co2) 1e19/m avg(n_co2) 1e19/m3 24

25 Continuous puffing with fast-pulsed discharge looks promising 10 5 Before/After ref BZN BIV Before/After ref BZN Before/After ref BZN Before After tot(ima_biv) /n_co2 (1e19/m3) 10 4 Open: F = -0.3 Close: F = -0.2 The total amount of BIV and OIV within shots are plotted as a function of shotaveraged density. tot(ima_oiv) /n_co2 (1e19/m3) OIV 0, bzn_read_ref_norm_2010.ps, Wed Oct 13 05:30: No boron increase, no oxygen decrease observed in two pulsed-puffing tests. An order of magnitude increase in boron and a slight decrease in oxygen throughout this density range are observed. No noticeable decrease in other high z impurities in all cases. No noticeable difference in density signals in all cases shot index shot index shot index

26 Al coupons analyzed by XPS show 5 ~ 7% boron contents in the continuous puffing test Sample A Sample B O-carborane injection intensity Sample A (far from the injector) Al 2p B 1s C 1s N 1s O 1s BE (ev) 0, bzn_read_xps_ a.ps, Wed Oct 13 12:56: intensity Sample B (near the injector) Al 2p B 1s C 1s N 1s O 1s BE (ev) 0, bzn_read_xps_ b.ps, Wed Oct 13 12:56:

27 Al coupons analyzed by XPS with Ar+ ion beam etching show > 3 nm B/C coating in the continuous puffing test 60 Sample A from (Far from the injector) 60 Sample B from (Near the injector) For SiO2 films, an 1- minute etching erodes 3 nm (+ 100% / - 50%). Atomic concentration (%) B O C Al Atomic concentration (%) O 30C B Al O C B Al Based on this, the thickness is about > 3 nm. This is 40 times thicker than expected with the pulsed discharge duty cycles in this test. The coating is slightly reversely non-uniform: may be due to pulsed nature of the discharge. 0 0 Background level (< 2%) Minute Minute 27

28 The fuel efficiency is better than expected MST area = 30.8 m2 Results: ~ 3 nm coating which corresponds 0.27 g / 3 g (2 hours) = This corresponds to 2 msec / 22 msec PDC cycle) The current pulsed discharge scheme (2 msec of plasmas every 1 sec) would give 10 % fuel efficiency. B/C film (nm) May be related to the longer decay time of o- carborane in continuous puffing Expectation: 3 g (2 hours) x = g would make 0.07 nm coating from 2 msec/1sec PDC cycle o-carborane (g) 0, plot_film_thick_mst_ ps, Wed Oct 13 13:10:

29 Summary Boronization by sublimating o-carborane, which is much safer and simpler to use than other conventional boronization materials, is under development. Due to the pulsed nature of the decomposing discharge in MST, it is important to carefully design the pulsed discharge, main focus being reducing the pulse cycle. Two schemes pulsed puffing assisted by the flow of discharge working gas and independent continuous puffing were initially investigated. The pulse puffing with helium does not improve the o-carborane throughput. The o-carborane decay is much slower than its residence time in the continuous puffing Being consistent with these two observations, the boronization using the continuous puff scheme yields more than 40 times the initially estimated fuel efficiency. 29

30 Future plan PDC power supply modification Target: AC glow-like discharge. Scan of the PDC duty cycle using the existing PDC power supply to scale its dependence on the fuel efficiency. For example, a 4-hour PDC with 80 degc sublimation temperature with two injectors will give Expected coating (nm) PDC cycle (sec) Installation of cold trap for possible recycling of o-carborane. H & D concentration measurements on the coupons using ERDA (Elastic Recoil Detection Analysis) or NRA (Nuclear Reaction Analysis). 30

31 XPS: X-ray photoelectron spectroscopy 1/3 31

32 XPS: X-ray photoelectron spectroscopy 2/3 32 Also known as ESCA (Electron Spectroscopy for Chemical Analysis). Measurement capability: Elemental composition of the surface Empirical formula of pure materials Chemical or electronic state of each element Uniformity of elemental composition across the top surface (line profiling or mapping) Uniformity of elemental composition along the depth by ion beam etching (depth profiling) or by ARXPS (Angle Resolved XPS) Measures the kinetic energy and number of electrons that escape (< 10 nm) X axis: binding energy of the electrons; Y axis: the number of electrons The characteristic peaks: the electron configuration of the electrons within the atoms (1s, 2s, 2p, ) The number of detected electrons in each characteristic peak: the amount of element within the irradiated area

33 XPS: X-ray photoelectron spectroscopy 3/3 Accuracy: Under optimum conditions, 90 95% of the atom % values of each major peak (1 atom % = ppm) Analysis time: Typically, 1 10 min; 1 4 hours for a depth profile for 4 5 elements with usual final depth of 1000 nm. Detection limits: Typically, atom %; 0.01 atom % with 8 16 hours Analysis area limits: Typically, 10 to 200 um Sample size limits: 1 x 1 to 3 x 3 cm for old devices; 30 x 30 cm or 30 cm wafers for recent devices 33

34 ERDA: Elastic recoil detection analysis Ion beam analysis for quantitative analysis of light elements in solids From the measured energy spectrum of recoiled atoms, a concentration depth profile is calculated. Very suitable method for Hydrogen profile in thin films, which is impossible to measure either XPS or AES. Accuracy ~ 10% Time requirements ~ the order of an hour Max. depth range: ~ 1000 nm Best dept resolution: ~ 10 nm Detection limit: ~ Atom % Sample size range: 2 x 5 mm^2 ~ 15 x 15 mm^2 34

35 NRA: Nuclear reaction analysis Obtains concentration vs depth distributions for certain target chemical elements in a solid thin film. Known initial kinetic energy of the projectile nucleus + stopping power => determines the depth at which the concentration of the target nucleus is measured. Measures the ionizing radiation from the reaction Another good way to measure H (or light particles) concentration profile along with ERDA 35