Optimization of CdS Nanowire Growth
Most people have a general understanding of what a solar is and what it does. However, what many people may not realize is that solar cells, while clearly beneficial, are not all that efficient. While there are many variables as to why this is the case, one important aspect of solar cell efficiency is physical size.
Schematic diagram of a photovoltaic cell (solar cell).
As the saying goes, size matters. However, the interesting thing about technology is that this motto reigns supreme, but in a somewhat backwards way: in the world of electronic gizmos, smaller is better. Don’t believe me? Let’s start with the earliest functional computer. It literally filled an entire room! Fast-forward to the first “personal” computer, which, while much smaller than its earliest ancestor, was still a behemoth when compared to the computers of today: sleek desktop computers and monitors, ultra-slim laptops and notebooks, and even PDAs and cell phones [which are essentially pocket-sized computers] serve as proof. And, on a somewhat more comical note, think back to the days of the original floppy disk [if you even can!]. A relatively enormous piece of data storage that held relatively nothing when compared to today’s USB flash drives. In the world of technology, things keep getting radically smaller while their efficiency grows exponentially greater.
Figure 2: Original
floppy disk [left] and a modern USB flash drive [right]. The floppy disks of the past
held < 1 megabyte of data, while the flash drives of today are capable of holding multiple gigabytes.
The ultimate goal of my research project is rather simple: to effectively scale-down a solar cell in order to increase its energy conversion efficiency.
The previous discussion hopefully got you thinking about technology, and what can happen to it when we make it smaller. One way to scale-down any electronic device is to build it using smaller wires. And this is the primary focus of my part in the bigger research project of which I was a part over this past summer: I essentially made nanowires! Not only did I make them, however, I optimized them.
A crucial aspect of quality CdS nanowire growth [at least as performed in this project] comes from a look at the binary phase diagram of gold-cadmium [Au-Cd] below1:
Binary phase diagram of gold/cadmium [Au-Cd].
The most important aspects of the diagram above are the eutectic points for Au-Cd, which have been encircled and labeled. The cadmium in the metalorganic precursor and the gold colloid on the substrate theoretically bond together at this point as the temperature of the mixture cools. This bond serves as the primary nucleation site for CdS nanowires. After inspection of the diagram, it becomes clear why consistent results are not always possible.
Synthesis of nanowires in this research was accomplished by means of MOCVD [metalorganic chemical vapor deposition]. Chemical vapor deposition [CVD] is a process in which a chosen chemical precursor [often in powder form] is heated within a gas flow chamber along with a chosen substrate [often a solid metallic chip]. The chemical precursor is vaporized, transported through the chamber, and then deposited onto the substrate where, theoretically [and, more importantly, hopefully ], a particular substance [in this case, CdS nanowires] form and grow2. The process is illustrated in the schematic diagram which follows:
Figure 5: Schematic diagram of nanowire growth gas flow chamber setup.
In this research project, the following conditions and substances were employed:
• The metalorganic precursor used was cadmium diethyldithiocarbamate, with the following molecular formula: 2(C5H10NS2) Cd
• The metallic substrate which served as the nucleation site for the nanowires was a gold/palladium sputter-coated thin wafer of silicon metal
• The precursor and substrate chips were placed at opposite ends of a glass furnace tube [gas flow chamber] and placed in an industrial furnace which was then heated to 790 °C; argon gas [chosen for its inertness] flowed through the tube to transport the precursor to the substrate
• The glass ring which held the precursor was pushed into the furnace tube incrementally during the nanowire growth process to allow the precursor to evaporate and thus be transported to the substrate in small, even amounts
• The nanowires that grew on the substrate chips were approximately 100 nm thick and several μm long and composed of CdS [cadmium sulfide]
Additionally, several gold/palladium sputter coated silicon substrate chips were placed side-by-side in the furnace tube during each nanowire growth process. This served two main purposes:
1] to maximize the amount of nanowire growth during a single run; and
2] to isolate and alter the variable of sputter coating time [the different chips were sputter coated for different lengths of time] while holding the other variables constant.
The precursor was uniformly spread into a thin line parallel to the flow of the gas through the chamber. In order to achieve the best results, the precursor was pushed further and further into the chamber towards the substrate during synthesis of the nanowires. In order to achieve this mobility, the precursor was laid in a circular glass ring which fit within and easily slid through the glass furnace tube. It was pushed along during the process by the aid of a metal bolt which was manipulated by means of a magnet on the outside of the tube. The substrate chips were laid directly in the tube at the other end of the furnace, directly above the heating coils for maximum heat intake which served to melt the thin layer of gold palladium sputter coating [this was important, as the nanowires needed a nucleation site]. The glass furnace tube was evacuated as much as possible prior to the start of the nanowire growth process in order to reduce unwanted gas-phase reactions and also to improve the uniformity of the nanowire growth across the substrate. Finally, the essential step which allowed the vaporized precursor to reach and ultimately deposit upon the substrate was the meticulously controlled flow of argon gas through the pressurized chamber [which was maintained at a constant pressure of 275 Torr].
As a means to determine the specific criteria that effects the outcome of each nanowire growth cycle, variables were identified and then methodically altered while the results were observed and qualitatively analyzed. Two crucial variables, temperature and pressure, were varied slightly, but no noticeable changes were observed. It seemed as if, as long as a minimal temperature [in order to vaporize the precursor and to melt the sputter coating on the substrate] and a maximal pressure was achieved, the resulting nanowire growth came out in much the same way. The most noticeable change in quality and quantity of nanowire growth occurred when the length of time during which the sputter coating process took place was altered. Theoretically, the sputter coating process lays down a relatively uniform layer of the desired coating [in this case, gold/palladium] on the substrate, with the thickness of this layer being directly proportional to the amount of time the sputter coating took place. The results of differing the time intervals of the sputter coating process were recorded using both optical and SEM [scanning electron microscope] images. The different time intervals of sputter coating employed were in 5 s intervals, starting with 5 s and going all the way up to 30 s. Brief discussions on the results are presented below the image galleries which follow:
Click on the image above to view a web album of optical images of my CdS nanowires.
Click on the image above to view a web album of SEM images of my CdS nanowires.
5 s Sputter Coating: Although it may be difficult to determine from the optical image at left, this particular sample is covered in a dense layer of relatively long CdS nanowire growth. The SEM image shows this better, and the depth [which appears out of focus] is proof of this.
10 s Sputter Coating: Again, just as in the 5 s sputter coating sample from above, this 10 s sample shows a large amount of thick, long CdS nanowire growth. Yet again, the optical image appears somewhat blurry, while the SEM image reveals why this is [dense growth].
15 s Sputter Coating: As evidenced through these images, it would appear that there is little difference between 5, 10, and 15 s sputter coating time intervals as it pertains to the amount and quality of CdS nanowire growth. All these samples came out with excellent growth.
20 s Sputter Coating: Starting with the 20 s sputter coating, the nanowire growth seemed to begin to decline. The optical image, taken at the edge of the substrate, shows less density and length than the previous samples. The SEM image reveals their scarcity and reduced quality.
25 s Sputter Coating: Seemingly as a steady decline, the 25 s sample revels even less quantity and overall quality of CdS nanowire growth. The SEM image, focused on a relatively barren patch of substrate, reveals abnormalities on the substrate itself.
30 s Sputter Coating: The longest sputter coating, 30 s, follows in the pattern of the previous two samples. A relatively low quantity, along with reduced quality, patch of nanowire growth resulted from longer sputtering times. This SEM image reveals the most irregularities yet.
As stated previously in the Data/Results section, multiple variables were isolated and systematically altered in order to determine which were most important in the overall optimization of the CdS nanowire growth. Also as previously stated, although temperature and pressure were essential in even beginning the nanowire growth process, small variations in either variable seemed to have little effect on the overall outcome. Indeed, critical temperatures [i.e. the eutectic points in the binary phase diagram] and pressures had to be obtained, but slight alterations produced no noticeable effects that could be definitively associated with these particular changes, as other slight changes would inevitably occur which seemed to have a much more significant effect.
In all, the following variables, although important to the entire process, did not seem to significantly effect either the quality or quantity of nanowire growth when altered slightly:
• Temperature – at first, a temperature of 710 °C was employed. Later in the project, the temperature was increased to 790 °C; however, both temperatures seemed to produce very similar results [as both were well above the eutectic point temperatures]. It should be noted, however, that the images presented in the previous Data/Results section of the best nanowire growth were obtained using a temperature of 790 °C.
• Pressure – similar to temperature, as long as the pressure was kept at a relatively low level [below 300 Torr], slight variations in this variable did not seem to produce noticeable differences in the overall quality or quantity of nanowires. The system was set to maintain a constant pressure of 280 Torr; yet, due to technical issues, the pressure varied slightly [under its own volition] during most growth processes. Yet, the variations were slight enough that, even when this was the only variable which changed, no noticeable effects were observed as a result.
The following variables, by contrast, did seem to significantly effect either the quality or quantity of nanowire growth when altered slightly:
• Sputter Coating Time – this variable, which seemed to be the most important, was addressed and thoroughly discussed in the previous Data/Results section.
• Amount of Precursor – this variable was found to be rather important near the end of the project. At the beginning of the project, a much smaller amount of precursor was employed, but as the project went on, more and more precursor was used to test whether or not this was an important variable. Indeed, it was: both the quantity and quality of nanowire growth seemed to improve when more precursor was employed.
• Initial Location of Precursor – following an understandable trend, it was found towards the end of the project that the initial location of the precursor was crucial in obtaining the best results. When the precursor was placed too close to the furnace’s extreme heat in the initial phase of the process, it was found that most, if not all, of the precursor evaporated before the “tapping” process even began. This resulted in little, if any, nanowire growth. Instead, a dusting of cadmium powder often covered the substrates.
• Initial Uniformity of Precursor Line – the initial thin line of precursor seemed to have a profound influence on the overall outcome of nanowire growth. If the initial line was too clumpy, the resulting nanowires grew too thick, resulting in structures known appropriately as “belts” [as illustrated in the images below]. The best nanowires resulted from runs in which the precursor was laid out in a rather thin, smooth, uniform line.
1 Binary alloy phase diagrams. Thaddeus B. Massalski; editors, Joanne L. Murray, Lawrence H. Bennett, Hugh Baker. Metals Park, Ohio: American Society for Metals, 1986.
2 Organometallic Vapor-Phase Epitaxy: Theoy and Practice (2nd ed.). Gerald B. Stringfellow. Academic Press, 1999.