Robofurnace automated CVD system. (a) Top-view schematic showing the major subsystems. Blue lines, from left to right, represent motion of the transfer arm, sample magazine, and tube furnace. (b) Photograph of prototype Robofurnace on a standard laboratory bench, including LabVIEW software interface. The system is ∼9 feet wide and 2.5 feet deep.
Magazine and transfer arm sample load sequence. (a) Magazine in the extended position for loading of samples. (b) Single etched Si sample holder (“boat”) resting on the quartz transfer arm. (c) Motion sequence where the transfer arm places a boat (holding substrate samples) in the reactor tube. The transfer arm enters the tube, then lowers, and retracts, leaving the boat in the tube.
Vision system. (a) Photograph of the vision system looking down the central axis of the reactor tube. (b) Image of a CNT forest during growth as taken by the vision system.
Queued state machine architecture implemented using LabVIEW. User inputs, alarms, and sensors can generate events that are inserted into the queue. The machine controller loop then reads from the top of the queue and executes that action/event. Alarms are inserted to the top of the queue for immediate execution.
LabVIEW software interface to Robofurnace, with recipe entry interface, real-time process data, and system status.
Architecture for thermal control of the system. (a) System model relating the position and motion of the furnace to the sample temperature. (b) Algorithm used to control the position of the furnace relative to the sample.
Measurement of accuracy and repeatability of the pick-and-place operation using the quartz transfer arm and silicon boat. A boat was placed and removed from the reactor tube 10 times. (a) The position of the sample along the reactor tube was measured relative to a reference mark. (b) The angular deviation is shown for the same sample being placed in the tube 10 times, as measured by the angle between the line formed by the boat's contact points with the tube and the horizontal reference axis.
Repeatability and accuracy of sample height measurement using the vision system to image a CNT forest on a silicon substrate. (a) Side view image with height measurement as seen through the vision system. The red bars indicate the top of the forest and the bottom of the boat. Yellow shows the profile of the forest. (b) Exemplary SEM image of CNT forest. (c) Ten independent measurements taken of this sample, with unloading and re-loading of the sample between measurements. The average height during a 10 s measurement cycle is shown with standard deviations. The blue band represents one standard deviation (5 μm) above and below the 10 trial average height. (d) Comparison of height measurements taken using edge-finding with the vision system and using digital calipers in the SEM.
Characterization of the thermal control system for sample cooling. (a) shows the sample temperature cooling at 15 °C/min, 30 °C/min, and 45 °C/min compared to predicted values. (b) Temperature versus time when the furnace is on the sample and hot, retracted off the sample, and retracted off the sample with a fan turned on, representing the limiting rates of cooling when the control algorithm is not utilized.
Real-time process data recorded by the system during synthesis of a CNT forest. (a) Furnace temperature (built-in control thermocouple in heating coil) and sample temperature (thermocouple in contact with furnace tube) versus time. (b) Input flow rate of each gas. (c) Moisture level of input gas flow, measured using hygrometer upstream of reactor at position shown in Figure 1 .
Use of the vision system for real-time analysis of CNT growth kinetics. (a) A view of the substrate before and after the experiment. The CNT forest can be seen as a dark layer on top of the Si substrate. (b) A profile of the CNT film is plotted every 30 s (for a different experiment). (c) During the synthesis step in which a carbon source is introduced the vision system measures the thickness of a CNT film in μm. (d) The rate of growth of the film in μm/s, which first increases rapidly to a maximum, then gradually decays, and then abruptly stops after 10 min when the growth step concludes.
A comparison CNT forest growth statistics obtained using a standard manual tube furnace and an automated tube furnace under the same conditions, via histograms of (a) height and (b) volumetric density. The manual growth study and the analysis of variation are described by Oliver et al. 12 Reprinted with permission from C. Ryan Oliver, E. S. Polsen, E. R. Meshot, S. Tawfick, S. J. Park, M. Bedewy, and A. J. Hart, ACS Nano7, 3565 (Year: 2013). Copyright 2013 American Chemical Society.
(a) Schematic of the dynamic growth recipe with time. (b) Recorded processing parameters for dynamic recipe showing temperature of sample, furnace, and position of the furnace against time.
(a) SEM images of a forest from a manual furnace and using Robofurnace. (b) A comparison of the volumetric density of CNT forests synthesized using a manual furnace static recipe and after optimization using Robofurnace dynamic recipe at 750 °C.
Motion system specifications.
Motion systems calculated torque and resolution values.
Parameter values for depth of field analysis.
Typical carbon nanotube experimental recipe.
Article metrics loading...
Full text loading...