Air and O 2 -Assisted Catalytic VACNT Growth Optimization for Uniformity and Throughput

: The development of an optimized air or O 2 -assisted multi-wall vertically aligned carbon nanotubes (VACNT) process that adjusts the vertical height proﬁle of a standard H 2 O vapor-assisted VACNT process is reported. The effect of the air or O 2 chemical vapor deposition (CVD) precursor ﬂow rate, the catalytic Fe layer thickness, the process growth temperature, and the H 2 /C 2 H 4 ratio on VACNT length was ﬁrst investigated to ﬁnd the optimum growth conditions. Spatial distribution height mapping of VACNT structures on six patterned 4 (cid:48)(cid:48) catalyst Si wafers prepared with a 70–90 min long O 2 -assisted growth step shows an average growth height of 1.8–2.2 mm, with a standard deviation of less than 10%. Characterization techniques included Raman spectroscopy, scanning electron microscopy (SEM), and spatial height mapping analysis for a range of Fluid channel Array Brick (FAB) components with a length of 30 mm, a width range of 2.5–15 mm, a ﬂuid channel diameter range of d = 5–100 mm, and a ﬂuid channel closest gap range of g = 5–50 mm. A signiﬁcant ﬁnding is that the O 2 -assisted VACNT growth process optimization efforts enable 2 mm parts processing with square edges, ﬂat top surfaces, uniform height tolerances, and maximum catalyst wafer utilization for application in engineering devices.


Introduction
Since its introduction in 2004 by Hata et al. [1] H 2 O vapor-assisted vertically aligned carbon nanotubes (VACNT) growth, commonly known as the super-growth method for carbon nanotube (CNT) synthesis, is often used to increase the growth rate, catalyst lifetime, and growth height of VACNTs.There has been substantial research into the kinetics and mechanism of VACNT growth [2].It has been determined that the catalyst-substrate interaction in the tip growth mechanism of VACNT occurs when hydrocarbon decomposes on the metal's top surface, allowing carbon to diffuse through the metal catalyst, causing CNT to precipitate out across the metal bottom, dislodging the entire catalyst particle from the substrate.As long as additional hydrocarbon breakdown can occur at the metal's top, CNT will continue to grow vertically.The metal's catalytic activity declines and CNT growth stops after it is entirely coated with excessive carbon [2][3][4].The super-growth method has been used, for example, to grow thus far the tallest (up to 21 mm tall) VACNT structures with a specialized catalyst structure on cm-size catalyst wafers [5].Although Sugime et al. reported a carbon nanotube forest with a length of 14 cm in a 26 h growth step, the carbon nanotubes are not vertically aligned [6].Odunmaku et al. showed a chlorine-assisted chemical vapor deposition (CVD) method for preparing up to 2 mm multi-wall VACNT, filled with catalyst particles when vapor flow rates were lower than 15 mL/min, as seen in FIG 2B [7].There is a significant lag between scientific discovery and technological implementation concerning CNTs [8][9][10].Recently, we used a modified super-growth method to improve the growth height and growth rate for >4 mm VACNT authors found that high-temperature growth (∼775 • C) yields the highest-quality SWCNTs, whereas controllable growth of double-and few-walled nanotubes can also be achieved at lower temperatures (550-600 • C).The use of CO as a feed gas had been introduced two decades ago by Tang et al., who used a hot tungsten filament in a CVD for the activation of CO molecules in the synthesis of carbon nanotube structures [23].Li et al. reported the use of the ethanol-assisted two-zone CVD for the rapid growth of VACNT arrays by setting one zone at 850 • C and the other at 760 • C for VACNT deposition.However, the CNT structure in FIG 2a shows non uniform heights while such structures suffer aggregation, as shown in FIG 8 (a, c, e) [24].The effect of other CVD parameters on the VACNT aspect ratio had been discussed in detail [25][26][27][28][29].In this paper, we present our investigation into our novel air or O 2 -assisted VACNT growth process and CVD system standard process operation for highly uniform 2 mm tall VACNT structures grown simultaneously on up to six 4 Si catalyst wafers for enhanced throughput inside a relatively simple, atmospheric pressure operated VACNT CVD growth systems.We report here on our process optimization efforts of (1) the air or O 2 gas flow rate, (2) the H 2 /C 2 H 4 ratio, and (3) the CVD process temperature optimization for up to 1.5 h long VACNT growth process step for growing ≥2 mm tall VACNT forest structures.

Materials and Methods
For all our experiments discussed here, we used either a small size FirstNano ® EasyTube ® 2000 with a 3 outer diameter process tube or a medium size FirstNano ® EasyTube ® 3000 system with an outer diameter process tube of 5 , SI Figure S1.Both systems had a three-zone motorized clamshell furnace, atmospheric process capability, an automated sample loading capability, and a precursor delivery line connected to a 3-way valve for either Ar or N 2 inert gas delivery that connected to a single mass flow controller (MFC) (SI: Automated MFC Range Control).For each system, the precursor delivery line additionally included delivery options with MFCs for H 2 (regular tank grade or UHP grade), C 2 H 4 ultra high purity grade (UHP grade), C 2 H 4 (research grade), and air or O 2 (UHP grade).For most VACNT process runs, research-grade C 2 H 4, and UHP grade H 2 process gas was used, which further allowed to improve the growth height and its reproducibility compared to UHP grade C 2 H 4 and lower grade H 2 process gases which showed a daytime-dependent O 2 /H 2 O content from sun-heated, outside located gas tanks.In addition, each system was equipped with an air pump and a 500 mL H 2 O bubbler.
The modified standard operational procedures developed included a Clean Run (SI: Clean Run Process) after each carbon process run and before a new process run after an extended shutdown.Alternating a carbon process run with a Clean Run helped significantly to reduce growth height variation from one process run to the next.The ET 2000 system was additionally modified to include airflow from a filtered air pump to provide sufficient pressure to drive the required airflow through the respective Air MFC.The ET3000 system was modified to include an additional MFC connected to a pressure UHP O 2 gas tank to provide the required O 2 flow when needed.Additionally, the combined precursor delivery line was always heated to ≥150 • C to prevent moisture accumulation.This further helped to stabilize the modified super-growth VACNT growth method when the H 2 O vapor precursor flow option was used for some of the tests and to eliminate any leftover moisture from prior super-growth process tests.The relocation of the 1000 Torr pressure sensor, see SI, and the standard operating procedure change of adding a constant gas flow of » 0.5 standard liter per minute (SLPM) of inert gas through the heated precursor delivery line, even when not in process mode, was done to minimize oxygen contamination of the process tube from the pressure sensor line.To minimize air back diffusion through the atmospheric exhaust line, a constant flow of 8 SLPM of N 2 was used during the VACNT process run until the process temperature was cooled to below 200 • C.
As further discussed in our previous work [11] all VACNT catalyst wafers were singleside polished 4 Si wafers, with or without a patterned photoresist layer, onto which a three-layer catalyst film stack (comprised of a 20 nm SiO 2 film, a 10 nm Al 2 O 3 film, and 0.5-2 nm Fe film) was deposited in a single batch run with an ebeam system.After loading the to-be-processed samples and closing the endcap of the respective ET 2000 or 3000 system, the system was purged with inert gas so that in 8 min, a total flow of 3× the process volume of the respective process tube was flown to purge most of the air and H 2 O moisture from the process tube and gas delivery lines.The process tube was then heated in furnace mode under the respective inert gas flow of either Ar or N 2 until the center zone process thermocouple reached 10 • C below the targeted VACNT process temperature.Thereafter, the system was switched to cascade temperature control and the set point for the temperature controller for the load zone (LZ), center zone (CZ), and end zone (EZ) got set to the targeted VACNT process temperature.After that, the system was further annealed for 2 min before the process gas was changed to a mixture of inert gas and H 2 to deoxidize the Fe catalyst layer and everything was held constant for 3 min.Thereafter, the VACNT growth period started with the flow of the respective precursor gas mixture.When Fluid channel Array brick (FAB) precursors on Si wafers were manufactured with this system, a carbon infiltration process set followed the VACNT growth step [30].After all the respective carbon processing was completed, the process tube was cooled down by turning off the furnace heating power and by opening the clamshell oven in multiple stages (<750 • C, <500 • C).A mixture of inert and H 2 was flown above 2 SLPM into the process tube until the process temperature reached 400 • C to prevent any oxygen back diffusion from the exhaust line.Thereafter only inert gas at the same flow rate was flown until the process temperature was below 200 • C, after which 0.5 SLPM of inert gas was flown through the process tube.The operator then was allowed to offload the test sample by opening the endcap on the manual command.For the ET 2000/ET 3000 system with a 3 /5 diameter process tube, the optimized VACNT process flow gas was as follows unless otherwise specified below: 1 SLPM Ar/3 SLPM N 2 , 0.9/3 SLPM H 2 , 4.1 sccm air/3.0sccm O 2 , 1 SLPM, 0.7/2.5 SLPM C 2 H 4, and a constant flow at all times of less than 0.5 SLPM Ar to bypass the H 2 O to prevent any moisture from getting in there and/or accumulating over time.

Results
Initially, a single quartz tray with a single catalyst wafer was used for the first feasibility trials to demonstrate that free-standing VACNT FABs can be manufactured and can survive liquid exposure and subsequent drying without structural changes.

H 2 O-Assisted VACNT Growth
An unmodified ET 3000 has been used for these initial trials.SI Figure S2 shows the outcome of such an initial test with a CVD VACNT precursor gas flow rate of 0.5 SLPM N 2 , 2 SLPM H 2 , 1.5 SLPM (UHP) C 2 H 4 and 0.75 SLPM of N 2 flow into the H 2 O bubbler heated to 57 • C. The growth temperature was 750 • C, and the growth time was 45 min.The isolated FAB precursors had a fall-off near their edges, with the sharp corners having the most significant drop.This can also be seen more dramatically in the image when looking at the two small-size witness samples from a cut from two non-patterned catalyst wafers made with different ebeam runs, demonstrating that this effect was due to the CVD procedure and not the sample type or ebeam batch.Based on our previous work [11], we expected an upwards curvature for the overall top surface profile.The obtained downwards curvature made it more difficult to seal these convex-shaped parts with a top gasket for fluid flow testing.In addition, the height of these parts often varied from one run to the next (0.5-1.5 mm).This was due to an accidental air leak caused by a degrading O-ring that was used to seal the quartz paddle shaft to the endcap.
We repeated the H 2 O-assisted super growth test after fixing the detected air leak.During the 70 min multi-wall VACNT growth a 750 • C process temperature was used with these precursor flows: »6 SLPM Ar, »2 SLPM H 2 , »1.5 SLPM (UHP) C 2 H 4 , »0.25 SLPM Ar into H 2 O bubbler heated to »57 • C. SI Figure S3 shows the resulting FAB height variations measured at the center and 5 mm from the edge of the 30 mm long FABs grow with the specified process conditions.The height dip in each FAB center (opposite to the ones observed in SI Figure S2) was again as expected.This reassured us that ET 3000 system was behaving normally again.In SI Figure S4, three catalyst wafers were processed with 0.5 SLPM of Ar flow into the H 2 O bubbler.This time a much more significant VACNT height variation, both internal to each catalyst wafer, and from wafer to wafer, was observed, suggesting that the super-growth process was significantly limiting the possible obtainable yields and that possible other process changes might be needed to reduce the edge distortion effects observed and further increase production capacity.

Air and O 2 -Assisted VACNT Growth
Before fixing the accidental air leak caused by a degrading O-ring that was used to seal the quartz paddle shaft to the endcap in the ET 3000 system, we tested VACNT growth with no flow in and out of the H 2 O bubbler with the following precursor flow of » 4 SLPM N 2 , » 3 SLPM H 2 , » 2.7 SLPM C 2 H 4 (UHP), and no flow into the H 2 O bubbler circuit.With 50 min growth time, a similar growth height of 1-1.2 mm was achieved, with a considerable variation from run to run.The result of such a test run is shown in SI Figure S5, which shows much taller and uniform VACNT compared with the H 2 O-assisted super-growth process.This indicates that O 2 might be able to counteract the "valley" shaped top surface of the H 2 O-assisted super-growth process in the absence of water vapor.
To replace pure H 2 O vapor with air in the ET 2000 system, we changed the standard operating procedures to minimize any possibility of trapped air or moisture getting into the process tube during the VACNT growth process.To eliminate process variables effects, we used small Si wafer chips from the same catalyst wafer.After some preliminary process dial-in, we settled on a range of process values to be fine-tuned, with all parameters kept fixed, except one, for each test series.There was an observed variation of the VACNT growth height for two cm size samples located in the middle of the 8 sample holder in dependence on the airflow rate.The presence of air was beneficial for the growth height, with an optimum value near the 4.1 sccm range for the remainder of the process conditions.Also, note that the maximum VACNT growth height obtained with the ET 2000 system was 2.6 mm, Figure 1.
Processes 2022, 10, x FOR PEER REVIEW 5 of 13 Ar into H2O bubbler heated to »57 °C.SI Figure S3 shows the resulting FAB height variations measured at the center and 5 mm from the edge of the 30 mm long FABs grow with the specified process conditions.The height dip in each FAB center (opposite to the ones observed in SI Figure S2) was again as expected.This reassured us that ET 3000 system was behaving normally again.In SI Figure S4, three catalyst wafers were processed with 0.5 SLPM of Ar flow into the H2O bubbler.This time a much more significant VACNT height variation, both internal to each catalyst wafer, and from wafer to wafer, was observed, suggesting that the super-growth process was significantly limiting the possible obtainable yields and that possible other process changes might be needed to reduce the edge distortion effects observed and further increase production capacity.

Air and O2-Assisted VACNT Growth
Before fixing the accidental air leak caused by a degrading O-ring that was used to seal the quartz paddle shaft to the endcap in the ET 3000 system, we tested VACNT growth with no flow in and out of the H2O bubbler with the following precursor flow of » 4 SLPM N2, » 3 SLPM H2, » 2.7 SLPM C2H4 (UHP), and no flow into the H2O bubbler circuit.With 50 min growth time, a similar growth height of 1-1.2 mm was achieved, with a considerable variation from run to run.The result of such a test run is shown in SI Figure S5, which shows much taller and uniform VACNT compared with the H2O-assisted super-growth process.This indicates that O2 might be able to counteract the "valley" shaped top surface of the H2O-assisted super-growth process in the absence of water vapor.
To replace pure H2O vapor with air in the ET 2000 system, we changed the standard operating procedures to minimize any possibility of trapped air or moisture getting into the process tube during the VACNT growth process.To eliminate process variables effects, we used small Si wafer chips from the same catalyst wafer.After some preliminary process dial-in, we settled on a range of process values to be fine-tuned, with all parameters kept fixed, except one, for each test series.There was an observed variation of the VACNT growth height for two cm size samples located in the middle of the 8" sample holder in dependence on the airflow rate.The presence of air was beneficial for the growth height, with an optimum value near the 4.1 sccm range for the remainder of the process conditions.Also, note that the maximum VACNT growth height obtained with the ET 2000 system was 2.6 mm, Figure 1.   Figure 1 also shows the respective VACNT growth height dependence for the ET 3000 system on O 2 flow for samples located at the center of a 4 quartz tray (center zone) for 5 mm size catalyst square wafer samples for different ebeam runs as well as for reference for the ET 2000 system that had about 1/3.5× lower flow for each process gas, which corresponds to the ratio of the cross-sectional area change between the two systems.Scaling the 4.1 sccm air flow rate from the ET 2000 to the ET 3000 system results in a » 3 sccm O 2 flow rate.As Figure 1 shows, with an O 2 -assisted VACNT process, the VACNT growth height first increases then plateaus off and finally decreases with increasing O 2 -flow rate.Also notable is that some variations from one ebeam batch to another suggest that further optimization of the ebeam deposition process is needed.Even if the same ebeam catalyst batch (eb1244) was used, the air-assist small diameter ET 2000 system produced better growth than the medium-size O 2 -assisted ET 3000 system under the tested fixed process parameters.The test results obtained by varying the process temperature at 770 • C, in Figure 2, show the optimum growth height for the tested conditions.Three different positions (beginning (LZ), middle (CZ), and end (EZ)) along the 8 quart tray were tested simultaneously with cm size catalyst samples from the same ebeam batch run.There was very little growth height variation observed along the 8 process tube length, with relative height variations increasing with process temperature.The maximum observed growth height for small samples with a 45 min growth time was ~2.7 mm for the ET 2000 system, which is more uniform than a test run with an H 2 O-based modified super growth process where we observed up to 300 µm height change from one end to the other of the 8 paddle for some not fully optimized growth process conditions.
for 5 mm size catalyst square wafer samples for different ebeam runs as well as for reference for the ET 2000 system that had about 1/3.5× lower flow for each process gas, which corresponds to the ratio of the cross-sectional area change between the two systems.Scaling the 4.1 sccm air flow rate from the ET 2000 to the ET 3000 system results in a » 3 sccm O2 flow rate.As Figure 1 shows, with an O2-assisted VACNT process, the VACNT growth height first increases then plateaus off and finally decreases with increasing O2-flow rate.Also notable is that some variations from one ebeam batch to another suggest that further optimization of the ebeam deposition process is needed.Even if the same ebeam catalyst batch (eb1244) was used, the air-assist small diameter ET 2000 system produced better growth than the medium-size O2-assisted ET 3000 system under the tested fixed process parameters.The test results obtained by varying the process temperature at 770 °C, in Figure 2, show the optimum growth height for the tested conditions.Three different positions (beginning (LZ), middle (CZ), and end (EZ)) along the 8" quart tray were tested simultaneously with cm size catalyst samples from the same ebeam batch run.There was very little growth height variation observed along the 8" process tube length, with relative height variations increasing with process temperature.The maximum observed growth height for small samples with a 45 min growth time was ~ 2.7 mm for the ET 2000 system, which is more uniform than a test run with an H2O-based modified super growth process where we observed up to 300 µm height change from one end to the other of the 8" paddle for some not fully optimized growth process conditions.Figure 2 shows the growth height dependence on process temperature for both the ET 2000 system at 2.7 mm (at 770 °C) with air or at 1.7 mm (at 760 °C) with O2 and ET 3000 system at 2.0 mm (at 770 °C) with O2.While the ET 2000 system showed a higher growth height with air compared to O2 for the equivalent process parameters, both systems showed relatively similar peak values with O2 at similar parameter values.
Next, we tested the VACNT growth height dependence on the H2/CH4 ratio.Figure 3 shows the observed slightly H2/CH4 ratio-dependent result with a maximum near H2/CH4 = 1.33.Note that a different catalyst wafer batch was used for this test series but with the same targeted catalyst film layer thicknesses of 1.0 nm.The ET 2000 system with air shows a higher growth VACNT height of 2.6 mm compared to a 2.0 mm for the ET 3000 system with O2 for the equivalent process parameters, where both systems show similar trends.Figure 2 shows the growth height dependence on process temperature for both the ET 2000 system at 2.7 mm (at 770 • C) with air or at 1.7 mm (at 760 • C) with O 2 and ET 3000 system at 2.0 mm (at 770 • C) with O 2 .While the ET 2000 system showed a higher growth height with air compared to O 2 for the equivalent process parameters, both systems showed relatively similar peak values with O 2 at similar parameter values.
Next, we tested the VACNT growth height dependence on the H 2 /CH 4 ratio.Figure 3 shows the observed slightly H 2 /CH 4 ratio-dependent result with a maximum near H 2 /CH 4 = 1.33.Note that a different catalyst wafer batch was used for this test series but with the same targeted catalyst film layer thicknesses of 1.0 nm.The ET 2000 system with air shows a higher growth VACNT height of 2.6 mm compared to a 2.0 mm for the ET 3000 system with O 2 for the equivalent process parameters, where both systems show similar trends.
Then we tested the VACNT growth height dependence on the Fe catalyst layer thickness.Figure 4 shows the observed test results suggesting a strong dependence on Fe catalyst layer thickness for both systems.
While a process temperature of 770 • C results in a higher peak growth of ~2.5 mm for the ET 2000 system with air and ~2.1 mm for the ET 3000 system with O 2 at 0.8 nm of Fe catalyst layer thickness, the de-rated process temperature of 750 • C results in a much lower process parameters sensitivity, thereby indicating that a de-rated peak process temperature might be a more suitable process parameter for growing uniform growth height over much larger spatial process area, i.e., along a more extended length zone of the respective process tube.Then we tested the VACNT growth height dependence on the Fe catalyst layer thickness.Figure 4 shows the observed test results suggesting a strong dependence on Fe catalyst layer thickness for both systems.While a process temperature of 770 °C results in a higher peak growth of ~ 2.5 mm for the ET 2000 system with air and ~ 2.1 mm for the ET 3000 system with O2 at 0.8 nm of Fe catalyst layer thickness, the de-rated process temperature of 750 °C results in a much lower process parameters sensitivity, thereby indicating that a de-rated peak process temperature might be a more suitable process parameter for growing uniform growth height over much larger spatial process area, i.e., along a more extended length zone of the respective process tube.

VACNT-FAB Growth Process Optimization
Initially, a modest target height of 1.2 mm tall VACNT structure inside was set, and two photolithography masks were designed for growing 9 individual 30 mm × 15 mm Then we tested the VACNT growth height dependence on the Fe catalyst layer thickness.Figure 4 shows the observed test results suggesting a strong dependence on Fe catalyst layer thickness for both systems.While a process temperature of 770 °C results in a higher peak growth of ~ 2.5 mm for the ET 2000 system with air and ~ 2.1 mm for the ET 3000 system with O2 at 0.8 nm of Fe catalyst layer thickness, the de-rated process temperature of 750 °C results in a much lower process parameters sensitivity, thereby indicating that a de-rated peak process temperature might be a more suitable process parameter for growing uniform growth height over much larger spatial process area, i.e., along a more extended length zone of the respective process tube.

VACNT-FAB Growth Process Optimization
Initially, a modest target height of 1.2 mm tall VACNT structure inside was set, and two photolithography masks were designed for growing 9 individual 30 mm × 15 mm

VACNT-FAB Growth Process Optimization
Initially, a modest target height of 1.2 mm tall VACNT structure inside was set, and two photolithography masks were designed for growing 9 individual 30 mm × 15 mm wide cross-sectional growth patterns with a 2 mm edge exclusion zone and 50 mm or 100 mm fluid channel diameters with a similar gap spacing.The separation between these 9 growth patterns was designed to be about 2 mm to ease further the subsequent removal of the FAB precursors from the growth substrate.
SI Figure S6 shows the results obtained when the optimized (de-tuned) O 2 -assisted VACNT growth process with an 0.8 nm Fe catalyst layer thickness was used to process the same 12 "long quartz tray holding three 4" catalysts wafers in a row.Note that, compared to the results shown in SI Figure S5, all 9 FAB precursors showed similar and uniform growth height across each part and for all FAB precursors located on the same catalyst wafer.They also showed similar height uniformity from the first to the third catalyst wafer.Also, all the edges of the FAB precursors and the small witness samples were sharp and square, indicating further capacity scale-up potential.
Figure 5 shows a Raman scan of VACNTs grown with the H 2 O-assisted and the O 2 -assisted growth processes described above.The Raman scan shows a smaller D/G ratio for the O 2 -assisted growth process compared to the H 2 O-assisted growth process indicating that O 2 -assisted growth is causing a smaller defect level and therefore produces better quality VACNT growth samples.
VACNT growth process with an 0.8 nm Fe catalyst layer thickness was used to process the same 12 " long quartz tray holding three 4" catalysts wafers in a row.Note that, compared to the results shown in SI Figure S5, all 9 FAB precursors showed similar and uniform growth height across each part and for all FAB precursors located on the same catalyst wafer.They also showed similar height uniformity from the first to the third catalyst wafer.Also, all the edges of the FAB precursors and the small witness samples were sharp and square, indicating further capacity scale-up potential.
Figure 5 shows a Raman scan of VACNTs grown with the H2O-assisted and the O2assisted growth processes described above.The Raman scan shows a smaller D/G ratio for the O2-assisted growth process compared to the H2O-assisted growth process indicating that O2-assisted growth is causing a smaller defect level and therefore produces better quality VACNT growth samples.SI Figure S7 shows the mask design innovation where we designed a mask that (i) minimized the spacing between adjacent FAB precursor areas to about <100 µm compared  VACNT growth process with an 0.8 nm Fe catalyst layer thickness was used to process the same 12 " long quartz tray holding three 4" catalysts wafers in a row.Note that, compared to the results shown in SI Figure S5, all 9 FAB precursors showed similar and uniform growth height across each part and for all FAB precursors located on the same catalyst wafer.They also showed similar height uniformity from the first to the third catalyst wafer.Also, all the edges of the FAB precursors and the small witness samples were sharp and square, indicating further capacity scale-up potential.
Figure 5 shows a Raman scan of VACNTs grown with the H2O-assisted and the O2assisted growth processes described above.The Raman scan shows a smaller D/G ratio for the O2-assisted growth process compared to the H2O-assisted growth process indicating that O2-assisted growth is causing a smaller defect level and therefore produces better quality VACNT growth samples.SI Figure S7 shows the mask design innovation where we designed a mask that (i) minimized the spacing between adjacent FAB precursor areas to about <100 µm compared SI Figure S7 shows the mask design innovation where we designed a mask that (i) minimized the spacing between adjacent FAB precursor areas to about <100 µm compared to the original 2 mm gap, (ii) allowed the placement of at least one of the FAB corners up to 3 mm from the catalyst wafer edge (compare to >5 mm before), and (iii) utilized a FAB area gap of <300 µm to grow a sacrificial VACNT in the adjacent unused wafer area.SI Figure S7 shows such a throughput improved mask layout after an O 2 -assisted growth process that allowed the placement of 11 FAB parts in one version of mask design and later allowed 12 parts, in an optimized version, to be placed on the same 4 catalyst wafer.As SI Figure S8 shows, the scale up to 12/9 = 133% higher packing did not cause yield losses.Also, these FAB precursors were still subsequently separated without yield loss, even when very tight gap separation was used.This allowed maximization of the usable real estate of a given catalyst wafer size with an optimized mask layout design.
Figure 7 shows the VACNT growth height for full-size wafers with the de-tuned 750 • C process temperature and the optimized Fe catalyst layer thickness of 0.8 nm for 4 size catalyst wafers.The optimum of 3 sccm for the O 2 precursor flow can be seen under the remaining chosen process conditions.Therefore, we used an O 2 = 3 sccm flow rate for all subsequent O 2 -assisted VACNT growth experiments and the Fe catalyst layer of 0.8 nm.(Note that after we rebuilt the ebeam system, we discovered that the quartz crystal monitor position was out of place.After returning it to normal conditions and optimizing the effective Fe layer thickness, we got a new effective optimum Fe layer thickness of 1.2 nm.This shows that for each ebeam-CVD system combination, the Fe catalyst layer thickness must be optimized to the available system.
As Figure SI S8 shows, the scale up to 12/9 =133% higher packing did not cause yield losses.Also, these FAB precursors were still subsequently separated without yield loss, even when very tight gap separation was used.This allowed maximization of the usable real estate of a given catalyst wafer size with an optimized mask layout design.
Figure 7 shows the VACNT growth height for full-size wafers with the de-tuned 750 °C process temperature and the optimized Fe catalyst layer thickness of 0.8 nm for 4" size catalyst wafers.The optimum of 3 sccm for the O2 precursor flow can be seen under the remaining chosen process conditions.Therefore, we used an O2 = 3 sccm flow rate for all subsequent O2-assisted VACNT growth experiments and the Fe catalyst layer of 0.8 nm.(Note that after we rebuilt the ebeam system, we discovered that the quartz crystal monitor position was out of place.After returning it to normal conditions and optimizing the effective Fe layer thickness, we got a new effective optimum Fe layer thickness of 1.2 nm.This shows that for each ebeam-CVD system combination, the Fe catalyst layer thickness must be optimized to the available system.The histogram shown in Figure 8 shows the narrow height distribution achieved with a ± 6% (1s) relative height variation across all six wafers, with each FAB component measured in three length locations along the middle axis of each FAB.This confirmed that the developed O2-assisted VACNT process can achieve tight height yields, despite a 600% capacity increase.SI Figure S9   The histogram shown in Figure 8 shows the narrow height distribution achieved with a ± 6% (1 s) relative height variation across all six wafers, with each FAB component measured in three length locations along the middle axis of each FAB.This confirmed that the developed O 2 -assisted VACNT process can achieve tight height yields, despite a 600% capacity increase.SI Figure S9 shows the six-wafer's custom height map shown in SI Figure S8.After an initial dial-in with a 70 min growth, we determined that the average growth height was on average in the 1.8 mm range, slightly shorter than the targeted >2mm height range.Therefore, we increased the growth time to 90 min and ran processed photolithographic catalyst wafers with the highest parts density of 12 FAB parts per a 4" wafer (as shown in SI Figure S8a).This resulted in an average growth height of »2.2 mm with a standard deviation of less than 10% and with a >50% total FAB part yield.The outliers After an initial dial-in with a 70 min growth, we determined that the average growth height was on average in the 1.8 mm range, slightly shorter than the targeted >2 mm

Figure 1 .
Figure 1.VACNT growth height dependence on the precursor air or O2 flow rate for both ET 2000 and ET 3000 systems, with 45 min growth time at 770 °C, H2/C2H4 = 1.33, and Fe = 1.0 nm.

Figure 1
Figure 1 also shows the respective VACNT growth height dependence for the ET 3000 system on O2 flow for samples located at the center of a 4" quartz tray (center zone)

Figure 1 .
Figure 1.VACNT growth height dependence on the precursor air or O 2 flow rate for both ET 2000 and ET 3000 systems, with 45 min growth time at 770 • C, H 2 /C 2 H 4 = 1.33, and Fe = 1.0 nm.

Figure 4 .
Figure 4. VACNT growth height dependence on Fe catalyst layer thickness for ET 2000 and ET 3000 systems with H2/C2H4 = 1.33.Both 750 °C and 770 °C process temperatures were tested for the ET 3000 system.

Figure 4 .
Figure 4. VACNT growth height dependence on Fe catalyst layer thickness for ET 2000 and ET 3000 systems with H2/C2H4 = 1.33.Both 750 °C and 770 °C process temperatures were tested for the ET 3000 system.

Figure 4 .
Figure 4. VACNT growth height dependence on Fe catalyst layer thickness for ET 2000 and ET 3000 systems with H 2 /C 2 H 4 = 1.33.Both 750 • C and 770 • C process temperatures were tested for the ET 3000 system.

Figure 5 .
Figure 5. Raman scans of (a) H2O-assisted and (b) O2-assisted VACNT grown on a catalyst wafer showing less D/G ratio for the O2-assisted growth process, i.e., less CNT damage.

Figure 6
Figure 6 shows SEM images of VACNT growth with the carbon infiltration process and Fluid channel Array Brick (FAB) components with a fluid channel diameter range of d = 5-100 mm and a fluid channel closest gap range of g = 5-50 mm.

Figure 6 .
Figure 6.Left: SEM images of VACNT (inset: growth with carbon infiltration process), Right: Fluid channel Array Brick (FAB) components with a fluid channel diameter range of d = 5-100 mm, and a fluid channel closest gap range of g = 5-50 mm.

Figure 5 .
Figure 5. Raman scans of (a) H 2 O-assisted and (b) O 2 -assisted VACNT grown on a catalyst wafer showing less D/G ratio for the O 2 -assisted growth process, i.e., less CNT damage.

Figure 6
Figure 6 shows SEM images of VACNT growth with the carbon infiltration process and Fluid channel Array Brick (FAB) components with a fluid channel diameter range of d = 5-100 mm and a fluid channel closest gap range of g = 5-50 mm.

Figure 5 .
Figure 5. Raman scans of (a) H2O-assisted and (b) O2-assisted VACNT grown on a catalyst wafer showing less D/G ratio for the O2-assisted growth process, i.e., less CNT damage.

Figure 6
Figure 6 shows SEM images of VACNT growth with the carbon infiltration process and Fluid channel Array Brick (FAB) components with a fluid channel diameter range of d = 5-100 mm and a fluid channel closest gap range of g = 5-50 mm.

Figure 6 .
Figure 6.Left: SEM images of VACNT (inset: growth with carbon infiltration process), Right: Fluid channel Array Brick (FAB) components with a fluid channel diameter range of d = 5-100 mm, and a fluid channel closest gap range of g = 5-50 mm.

Figure 6 .
Figure 6.Left: SEM images of VACNT (inset: growth with carbon infiltration process), Right: Fluid channel Array Brick (FAB) components with a fluid channel diameter range of d = 5-100 mm, and a fluid channel closest gap range of g = 5-50 mm.
shows the six-wafer's custom height map shown in SI Figure S8a,b.

Processes 2022 , 13 Figure 8 .
Figure 8. Binned histogram of height mapping in 100 µm increments on three locations of each 30 mm long FAB precursor (at the middle and 3 mm from each long edge) for 6 wafers processed as shown in SI Figure S10a,b.

Figure 8 .
Figure 8. Binned histogram of height mapping in 100 µm increments on three locations of each 30 mm long FAB precursor (at the middle and 3 mm from each long edge) for 6 wafers processed as shown in SI Figure S8.