Sei sulla pagina 1di 10

An Investigation of the Eects of Varying Growth Conditions on Induim-Arsenide Quantum Dots

Adam Podell1,
1

Department of Physics, Rochester Institute of Technology, Rochester, NY 14623-5603 (Dated: February 21, 2011)

Stranski-Krastanov (SK) grown Indium Arsenide Quantum Dots (InAs QDs) have been studied and their optical and physical properties compared. Four independent experiments were conducted, varying a series of growth parameters, c , InAs coverage, Tg , growth temperature, GRI, Growth rate interrupt and V/III ratio. Test structures were analyzed using atomic force microscopy (AFM) evaluated using Scanning Probe Image Processor (SPIP) software and statistical results were compared to optical results using photoluminescence (PL) spectroscopy. It was seen that an increased InAs coverage tended to increase average QD height with subsequent decrease in areal number density with average heights ranging from 1.60.5nm to 3.90.5nm and areal number densities ranging from (1.280.2)x1010 cm 2 to (3.000.2)x1010 cm 2 . Statistical results from the V/III ratio series showed that increased ratio produced higher number densities with lower average heights, ranging from 1.70.3nm to 4.70.3nm. PL data showed that a V/III ratio of 12 led to the best optical quality dots. Growth temperature studies showed that higher growth temperature produced lower number densities of dots with higher average heights at each point measured, with heights ranging from 2.50.3nm to 5.50.3nm and number densities ranging from (0.590.01)x1010 cm 2 to (2.570.1)x1010 cm 2 . GRI studies determined that a 30 second interrupt led to both higher densities of dots and higher average heights than a 10 second interrupt with heights ranging from 2.30.3nm to 2.50.3nm and number densities ranging from (4.30.1)x1010 cm 2 to (7.00.1)x1010 cm 2 .

I.

INTRODUCTION

Nanoparticles such as Quantum Wells (QWs), Quantum Wires (QWRs) and Quantum Dots (QDs) are structures with at least one dimension on the order of the de Broglie wavelength of an electron in the material into which they are incorporated. They are quantum conning structures; that is to say that they conne electron (or hole) movement in three dimensions (QDs), two dimensions (QWRs) or one dimension (QWs) and with this connement a change is seen in the density of states from a continuum of energy states in the case of a bulk material (i.e. no nanostructures) to a set of discrete or quantized energy states in the case of the quantum box or QD as shown in Fig. 1 [8]. Electrons in QD states can be approximated by the Schrdinger equation for a particle in a three dimensional o innite potential box. ~2 2 r (r) + V (r) (r) = E (r) 2m (1)

FIG. 1. Density of states (DOS) versus energy for bulk material and nanostructures. It is seen that the DOS function changes to from a continuum of states in the case of the bulk material to a set of quantized or discrete states in the quantum dot.

where ~ is the reduced Planks constant, m is the effective electron mass in material, is the electron wavefunction, V(r) is the potential inside the box and E is the total energy. This equation can be simplied to the case where the potential inside the box is zero and can

be solved for the energies present in the material as " 2 2 # 2 ~2 2 nx ny nz Ex,y,z = + + 2m Lx Ly Lz Here nx , ny and nz are the quantum numbers for the energy states and Lx , Ly and Lz are the dimensions of the box which determine the scale of the energy states. Note that this is poor estimate of the exact energy levels in the QDs as a more accurate model requires a treatment of overlap between electron and hole wavefunctions and the conning potential is not innite, however, it does

amp5726@rit.edu

2 serve to illustrate trends seen as the physical dimensions change. Nanostructures such as QDs have application in many dierent solid state devices such as lasers, LEDs, photodetectors and solar cells as their inclusion makes for production of high quality devices. In lasers QDs aid in ground state population inversion which is imperative for stimulated emission. QD LEDs can be produced which output 50-100 times brighter than CRT and LCD displays with longer lifetimes. [9] Research in photovoltaics (PV) has been of great interest in the past thirty years and increased demand for e ciency has caused technology to change from the previous standard, the single junction crystalline solar cell to the triple junction solar cell. The maximum possible e ciency for a lattice-matched triple junction cell, however is 33% in the Shockley-Queisser limit [12] which refers to the theoretical maximum e ciency of a solar cell using a p-n junction model and detailed balance. It has been shown that by incorporating QDs to alter the bandgap of the middle junction, the e ciency can be boosted past this limit [4]. QD synthesis can be performed both chemically, resulting in colloidal QDs which have application in organic and thin lm PV or by organometallic vapour phase epitaxy (OMVPE) and molecular beam epitaxy (MBE) [5], producing very high quality, coherent dots used in crystalline PV. There are three main heteroepitaxial growth methods, each resulting from dierent interfacial energies and substrate mismatches. [1] Frank-Van der Merwe growth (2D) occurs in systems where the energies of the epilayer and interface sum to a lower energy than the substrate, i.e. the deposited material and substrate material are lattice-matched. In such a system atoms wet the surface and form smooth fully formed layers. Volmer-Weber growth (3-D) occurs in systems whose epilayer and interface energies sum to a greater energy than the substrate. In other words the lattice constants of the substrate and the deposited material are mismatched resulting in 3-D island formation causing dislocations and defects. The third growth mode is an intermediary type of growth known as Stranski-Krastanov (SK) growth [10] which is characterized as a 2-D to 3-D type of growth and is the growth method of QDs in this paper. SK growth occurs when the substrate and deposited material are slightly (<10%) lattice-mismatched. In these systems, monolayers (half of the lattice constant, 3 for InAs) A of the deposited material are grown directly on the substrate where the crystals compress so that their lattice constant matches the substrate below and thus compressively straining the molecules. Adding more layers of material becomes energetically unfavorable after some critical thickness, crit , and as more material is deposited, the the crystals relax to their natural lattice constant. After this point material bonds to the top layer as it is more energetically favorable, resulting in the formation of coherent islands. These islands are called quantum dots. The amount of InAs owed during growth can have a signicant impact on overall size of QDs as well as densities. It has been seen that almost no QDs form on the substrate when InAs of coverage below a certain thickness, crit is applied. After this point is reached, QD growth tends to occur with number densities increasing as more InAs is added. It has been seen that as InAs is added in excess of crit , coherent QDs can coalesce in a process called Ostwald ripening. [11] In this process, smaller QDs which are highly mobile on the surface of the substrate during growth will combine to form larger clumped islands which may not be optically active as a coalescence leads to dislocations which reduce radiative recombination. GaAs grown by OMVPE is governed by the pyrolysis equation AsH3 + (CH3 )3 Ga ! GaAs + 3CH4 (2)

where trimethylgallium and arsene are the chemical precursors. Heat is the driving energy for this reaction, breaking the hydrogen and methyl bonds and the methane is exhausted. In a very similar reaction, InAs can be grown on the GaAs substrates by AsH3 + (CH3 )3 In ! InAs + 3CH4 (3)

The lattice constant of InAs is 6.05 while that of GaAs A is 5.65 leading to a fractional mismatch of 7%, thus A InAs can form on the substrate in the SK growth mode as previously described, resulting in coherently strained three dimensional islands. These islands are in fact the QDs and in solar cells, very often multiple (5-60) layers of QDs are capped and stacked as shown in Fig. 6. for more desired optical and electrical properties. Quantum dot growth depends on many parameters and changing any parameter during growth can change the way the QDs form on the surface. The growth process is entirely thermodynamically driven in that the process itself would not even take place unless there is enough thermal energy present and thus, QD growth inside an OMVPE reactor, described by eq. 3 usually takes place around 500 . Temperatures too much in excess of 500 can lead to desorption of Indium as it is highly volatile on the surface of the GaAs. Thus, QD growth is highly temperature sensitive so growth temperature is a good parameter to study in this investigation. Aside from the growth temperature, parameters such as V/III ratio can be of importance to growth when bonding becomes important [13]. V/III ratio describes the ratio of Arsenic to Indium. In OMVPE, Indium and Arsenic are owed into the chamber, however Arsenic does not bond as readily with the indium so it is often necessary to have more Arsenic present for formation of InAs. Another technique practiced in growth of QDs is the growth rate interrupt (GRI) where a growth pause is implemented during growth [6]. InAs is highly mobile on the surface of the GaAs and by shutting o ow of gas

3 temporarily, the InAs is allowed to move around the surface of the substrate briey and settle in a more energetically favorable nucleation site.
II. METHODS OF DATA ANALYSIS

QDs are nanoparticles and their scale makes it di cult to image optically. In 1986, Binnig, Schockley and Quate invented the atomic force microscope based on the technology used in the scanning tunneling microscope (STM) [2]. The AFM has a wide range of use since the samples being imaged need not be semiconducting or conducting whereas STM requires a conductive of the sample. An AFM uses a small cantilever with an ne tip which relies on Van der Waals and atomic forces between the surface of the sample and the tip. There are two primary modes of operation for an AFM: contact mode in which the AFM tip is dragged across a sample and its height is tracked by a laser which reects from the tip into a photodetector calibrated to read zero voltage. As the tip encounters a surface feature the laser is deected and the photodetector reads a nonzero voltage. A small piezoelectric cylinder underneath the sample stage then either contracts or expands to keep the voltage at zero and based on this change, an image can be formed showing the topography of the sample. The second mode is tapping mode, shown in Fig. 2, in which the cantilever arm oscillates at a set voltage and a laser reecting o the tip creates a strip read at a position detector. As the tip encounters a feature on the surface the oscillations are dampened and a piezoelectric cylinder under the surface expands or contracts to maintain a constant amplitude of oscillation. From this an image can be formed in the same manner as with the contact mode AFM. [7] Samples imaged using AFM in this paper were done so using the tapping mode as tapping mode tips have a higher radius which can lead to sharper resolution in images obtained. Optical data can be collected through photoluminescence spectroscopy (PL) in which a focused laser beam is used to inject a high density of photons whose energy is greater than the bandgap of the sample. The incident photons create electron hole pairs (EHP) and as the excited carriers relax and recombine, they emit electrons at the transition wavelength shown in Fig. 3. The measured peak of emitted photon energies correspond to the band gaps present in the material. [3] Electrons and holes can be quantum conned by the potential well that they experience at the location of a quantum structure. Here, excited electrons in the conduction band and holes in the valence band can fall into the wells in high densities and recombine to emit light with an energy of the bandgap of the conning structure as opposed to that of the bulk material. The quantum connement helps more carriers recombine at desired wavelengths.

FIG. 2. Schematic of a tapping-mode AFM. A laser incident on an oscillating cantilever arm causes a laser strip incident on two photodetectors, A and B. From this, a voltage dierence is measured and set as the constant voltage for the tool to maintain.

FIG. 3. Band diagram of a QD showing photoluminescence. Incident light excites an electron into the conduction band and the electron then recombines in the valence band. The wavelength of the emitted photon is the transition wavelength which corresponds directly to the energy gap of the QD state in which the electron was conned.

III.

EXPERIMENTAL SETUP

All samples were grown at the NASA Glenn Research Facility using low-pressure OMVPE. The OMVPE reactor was a 3 2" Veeco D125LDM. Standard precursors of trimethylgallium (TMGa), trimethylindium (TMIn), trimethylaluminum (TMAl) and arsine (AsH3) were used for alkyl and hydride sources, respectively. Growth temperature was monitored in-situ using emissivity corrected optical pyrometery[? ]. All samples were imaged using a Veeco Dimension-3000 Atomic Force Microscope in the tapping mode. Images were taken at three dierent points on the wafer in the radial direction from the center of the reactor as shown

4 in Fig 4.

FIG. 5. Schematic of PL setup. A 514nm Ar-ion is directed by mirrors onto a sample and emitted photons are sent through a series of optics for detection by a monochrometer. An optical chopper coupled with a lock-in amplier is used to eliminate background noise. FIG. 4. Schematic of OMVPE reactor showing three two inch wafers. Distances labeled represent points imaged using AFM, moving radially from the center of the reactor.

All images taken using AFM were processed using a metrology software program Scanning Probe Image Processor (SPIP) designed to detect features on images. The software allows the user to dene parameters in order to detect particles of certain sizes to within a tolerance and to manually select/void pixels which may or may not be features in order to count surface features. The software collects data such as maximum height, minimum height, height range, diameter and surface area and records maximum, minimum, mean and standard deviation for each parameter for all features detected. Height range data was used as the feature height since growth steps on the surface of the wafers do not provide a completely level surface at the nanoscale. Height data was exported to Microsoft EXCEL to bin results in histogram form in order to determine statistical properties such as average QD height per sample. Areal number density was determined using the number of QDs detected using SPIP and a conversion factor from QDs per square micron to QDs per square centimeter to get a statistical number density. PL measurements were performed using a 514nm Argon-ion laser and JY Horiba monochrometer. An optical chopper coupled to a phase locked lock-in amplier was implemented in order to eliminate background noise aecting the laser signal as shown in Fig. 5. In order to study the optical properties using PL, ve layers of QDs were grown and capped (see Fig. 6) to enhance the PL signal.
IV. RESULTS AND DISCUSSION

FIG. 6. Diagram of a test structure used for both PL and AFM. QDs stacked below surface are for PL measurements and uncapped QDs on the surface are for AFM measurements.

A series of linear experiments was conducted to determine the eect of an individual parameter on the growth of the QDs. The parameters investigated were InAs coverage, , growth temperature (Tg ), growth rate interrupt (GRI) and V/III ratio. The series consisted of ve samples varying in InAs coverage: 1.8ML, 2.1ML, 2.17ML, 2.24ML, 2,31ML. The Tg series had two wafers whose

reactor temperatures were 557/540 and 555/476 (corresponding to the inner/outer heaters respectively in the reactor). The GRI series had two wafers whose growth pauses were 10 seconds and 30 seconds respectively. The V/III ratio series had three wafers whose V/III ratios were 12, 38.8 and 48 respectively. The histogram results from AFM of the series are shown for the center of the wafer, see Fig. 4. This point showed the most uniform results across all samples. Statistical results were obtained by plotting data collected

5 using SPIP as a histogram. For low InAs (1.8ML) coverage, a narrow full-width half-max (FWHM) was seen. Since the data represents a statistical binning of QDs of a specic size range, the FWHM of the peak indicates a relative uniformity of QD heights. PL was taken at the center of the wafer for the 1.8ML and the 2.1ML sample in the series for comparison. PL data was intended to be compared to mechanical data taken using AFM to make conclusion about the average QD sizes and number densities across samples. Since PL involves exciting carriers into QD states for radiative recombination, the intensity of a peak is related to the number of optically active QDs and its location corresponds roughly to QD size as is evident by the solution to Schrdingers equation previously stated. o Shown below (see Fig. 9) are the PL results for both the 1.8ML and the 2.1ML samples for comparison. These samples were chosen to illustrate a dierence in PL corresponding to an increase in . The peak for the 2.1ML sample is shifted towards higher wavelengths indicating a larger average QD size in the region tested, as longer wavelengths correspond to lower energies and therefore a larger quantum box. The dierence in peak intensity can be accounted for in the material quality since the 1.8ML sample, corresponding to a higher density of emitters produced a lower PL signal than did the sample with fewer emitters. Thus the 1.8ML sample had a lower overall material quality.

FIG. 7. Histogram plot of count versus average QD height for 1.8ML InAs coverage sample with Gaussian t. Narrow FWHM indicates uniform QD heights across image.

As more InAs was deposited during growth, a change in size distribution was seen. In samples of higher (2.2ML) coverages, a second peak was seen in the histogram plot of QD height. The existence of multiple peaks indicates that the heights of QDs became poly-dispersed and in the case of the 2.24ML sample (Fig. 8), a bimodal size distribution was realized, thus shifting the average QD height to larger values as shown in table I.

FIG. 9. Comparison of PL responses for 1.8ML and 2.1ML samples in the series. A shift in wavelength of peak indicates a change in average size of QDs. Peaks at 870nm and close to 950nm correspond to the GaAs band edge and wetting layer respectively.

FIG. 8. Histogram plot of count versus average QD height for 2.24ML InAs coverage sample. Large peak near 2nm indicates one mode and a second peak can be seen around 12nm indicating a shift to a bimodal size distribution.

Shown below (see Fig. 10, Fig. 11) are the statistical results for the wafers in the series. Results are shown for the center of the wafer in each case and show how average QD height and number density varied by changing the growth parameter . From these results, it is seen that as coverage increased, the average height of QDs tended to increase. For the same increase in coverage, it was seen that the number density tended to decrease which may be evidence of the coalescence of smaller QDs forming larger islands. All results from this data are tabulated (Table I) for reference.

6
TABLE I. Quantum Dot Statistical Properties at Wafer Centers for Series Sample Average Height Number Density 1.8ML 1.6nm 3.00x1010 cm 2 2.1ML 1.7nm 2.38x1010 cm 2 2.17ML 2.6nm 2.15x1010 cm 2 2.24ML 2.9nm 1.28x1010 cm 2 2.31ML 3.9nm 1.44x1010 cm 2

FIG. 10. Average QD height vs. InAs coverage. Note that as coverage increases, average height tends to increase; a result consistent with photoluminescence results.

series and compared to determine the eect of growth temperature on the optical quality of quantum dots. PL results shown in Fig. 15 were taken at approximately the same spot on both wafers for comparison. PL data conrmed AFM results as shown, with a higher growth temperature relating to a lower PL peak. As PL intensity is related to number of conning structures, the lower peak agrees with a decrease in number density.

FIG. 12. AFM images of wafers in the Tg series. Left: lower growth temperature sample, Right: higher growth temperature sample. Both images taken from center point on wafer. Note the decrease in number density corresponding to a higher growth temperature.

FIG. 11. Areal number density vs. InAs coverage. Note that as coverage increases, number density tends to decrease. This may be evidence of a coalescence of QDs on the surface.

Statistical data and PL for the Tg , GRI and V/III ratio series was taken and plotted versus location. All locations refer to radial distance from the center in the OMVPE reactor. In the Tg series, it was seen that a higher overall temperature led to lower areal number densities at each point imaged, while the average QD height increased at each point. The AFM images of the location PL was taken for each wafer is shown in Fig. 12. The loss in number density (Fig. 13) may have been due to desorption of the indium at higher temperatures or QD coalescence due to its high mobility on the surface of the substrate. Coalescence may also describe the increase in average height (Fig. 14) of the QDs. PL was taken for the wafers in the

A similar trend was seen in the V/III ratio series, whose AFM scans are shown in Fig. 16. Analysis showed an increase in number density as V/III ratio increased and a subsequent decrease in average QD height as shown in Fig. 17 and 18 respectively. As previously mentioned, the inability for arsenic to bond with the indium led to a need to increase the ratio of As to In for the statistical likelihood of InAs formation. Thus, an increase in In may have led to the formation of more InAs, leading to more QD formation. Arsenic, however has a relatively large atomic radius compared to indium and so more As on the surface of the wafer may have led to a di culty in nucleation of In on the surface, decreasing the probability of coalescence and potentially leading to the decrease in average QD height. The PL results for the V/III ratio series correspond dierently to the AFM results than did the PL results of the Tg series. Shown in Fig. 19, a V/III ratio of 12 led to the highest PL peak intensity while having the smallest areal number density of QDs. This indicates that despite having fewer emitters, the V/III ratio of 12 led to the highest material quality for

FIG. 13. Comparison of number densities of QDs at three points imaged for Tg series. Locations refer to radial distance from the center in the OMVPE reactor.

FIG. 15. Comparison of PL responses for wafers in the Tg series. The peak at 870nm corresponds to the GaAs excitation while the peaks around 1100nm correspond to QD response. Note that the QD peak for the 555/476 wafer is a lower intensity than the GaAs response. This indicates the formation of fewer optically active QDs or a lower overall material quality.

FIG. 14. Comparison of average QD height at three points imaged for Tg series. Locations refer to radial distance from the center in the OMVPE reactor.

the QDs. This can be seen when looking at the ratio of PL peak intensity to number of QDs in the sample shown in table II.
TABLE II. Ratio of PL Peak intensity to Number of QDs for V/III Ratio Series FIG. 16. AFM images of wafers in the V/III ratio series. Left: V/III ratio = 12, Right: V/III ratio = 38.8, Center: V/III ratio = 48. All images taken from center point on wafer.

V/III Ratio Peak Intensity Number Density Intensity per QD 12 9.91mV 277m 2 0.035 well as average QD height (see Fig. 22) at each position 38.8 1.38mV 288m 2 0.005 measured corresponding to a longer interrupt which is 48 3.82mV 395m 2 0.009 consistent with the idea that InAs is allowed a longer

Data from the GRI series, AFM images shown in Fig. 20, showed an increase in number density (see Fig. 21) as

time to settle into an energetically favorable nucleation site, allowing for more dots to form. PL data for the GRI series, shown in Fig. 23, showed agreement with AFM data. A higher PL peak was seen for a 30 second GRI

FIG. 17. Comparison of number densities at three points imaged for V/III ratio series. Locations refer to radial distance from the center in the OMVPE reactor.

FIG. 19. Comparison of PL responses for V/III ratio of 12, 38.8 and 48. The height of the PL peaks indicate the optical quality of the material. Note that a V/III ratio of 12 yielded an order of magnitude stronger PL signal at the peak wavelength than that of the 38.8 ratio indicating higher quality QDs.

FIG. 18. Comparison of average QD heights at three points imaged for V/III ratio series. Locations refer to radial distance from the center in the OMVPE reactor.

FIG. 20. AFM images of wafers in the GRI series. Left: 10 second GRI sample, Right: 30 second GRI sample. Both images taken from center point on wafer. Note the increase in number density and average QD size corresponding to a longer growth interrupt.

than for a 10 second GRI occurring at a slightly higher wavelength, indicating a larger average QD size.
V. CONCLUSIONS

An investigation of the eect of InAs coverage, growth temperature, growth rate interrupt and V/III ratio on statistical and optical properties of InAs QDs on GaAs substrates was conducted. Data was obtained using two methods of analysis: atomic force microscopy and photoluminescence spectroscopy. It was seen that as coverage increased, average QD height tended to increase. An inverse trend was seen in the areal number density of QDs, attributed to QD coalescence which appeared as

a decrease in overall feature count when using SPIP for analysis. Furthermore a change in size distribution from a mono-dispersion to a poly-dispersion or bimodal distribution of sizes was seen corresponding to an increase in InAs coverage. Photoluminescence conrmed AFM data in that a shift in wavelength of the QD PL peak was seen indicating a change in QD size. It was seen that a higher growth temperature led to lower areal number densities at each point imaged with a subsequent increase in average QD height at each point. PL data for the Tg series conrmed AFM results. It was determined that as V/III ratio increased, the density of QDs tended to increase while the average height tended to decrease at each point imaged. PL data for the V/III series showed that a V/III ratio of 12 produced the highest PL signal in the QD regime indicating a higher optical QD qual-

9 a greater number of QDs with a larger average height which was conrmed by PL data for the GRI series.
ACKNOWLEDGMENTS

I would like to make special acknowledgement to Dr. Seth Hubbard for his tremendous help and support throughout the planning of this project. I would also like to acknowledge Dr.s John Andersen, Vern Lindberg, Ron Jodoin, and Dawn Hollenbeck for their help answering any questions and their guidance. I would like to thank fellow physics student Wyatt Strong for valid

FIG. 21. Comparison of number densities at three points imaged for GRI series. Locations refer to radial distance from the center in the OMVPE reactor.

FIG. 23. Comparison of PL responses for wafers in the Tg series. The peak at 870nm corresponds to the GaAs excitation while the peaks around 1100nm correspond to QD response. Note that the QD peak for the 555/476 wafer is a lower intensity than the GaAs response. This indicates the formation of fewer optically active QDs or a lower overall material quality.

FIG. 22. Comparison of average QD heights at three points imaged for GRI series. Locations refer to radial distance from the center in the OMVPE reactor.

ity than that of the 38.8 and 48 ratio samples. It was also determined that a longer growth interrupt produced

A input in proofreading and for his knowledge of L TEX. Also, a special thank you to the graduate students in the RIT Nanopower Research Laboratory, Chris Bailey, Steve Polly and Zachary Bittner. And thank you to anyone who helped in the planning of the project and/or the proofreading of this paper.

[1] Dieter Bimberg, Marius Grundmann, and Nicolai N. Ledenstov. Quantum Dot Heterostructures. John Wiley and Sons, 1999. [2] G. Binnig, C. F. Quate, and Ch. Gerber. Atomic force microscope. Phys Rev Lett., 56:930933, March 1986. [3] T. H. Gfroerer. Photoluminescence in analysis of surfaces and interfaces. In Encyclopedia of Analytical Chemistry. John Wiley and Sons, 2000. [4] S.M. Hubbard, C. Plourde, Z. Bittner, C.G. Bailey, M. Harris, T. Bald, M. Bennett, D.V. Forbes, and R. Raf-

faelle. Inas quantum dot enhancement of gaas solar cells. In Photovoltaic Specialists Conference (PVSC), 2010 35th IEEE, pages 001217001222, June 2010. [5] Omar Manasreh. Semiconductor Heterojunctions and Nanostructures. McGraw-Hill, 2005. [6] Yoshitaka Morishita, Koh Osada, and Tomoaki Hasegawa. Eects of growth interruption during growth of inas wetting layer on formation of inas quantum dots. Journal of Applied Physics, 44(5A):29252928, January 2005.

10
[7] University of Michigan Electron Microbeam Analysis Laboratory. Digital instruments nanoscope iiia-phase atomic force microscope. [8] The University of Warwick. Electronic band structure. [9] Joanne Okvath. The eects of substrate miscut orientation on inas/gaas quantum dot radiative and defective properties examined by photoreectance (pr) and deep level transient spectroscopy (dlts). Masters thesis, Rochester Institute of Technology, 2010. [10] Alberto Pimpinelli and Jacques Villain. Physics of Crystal Growth. Cambridge University Press, 1998. [11] Peter W. Ratke, Lorentz; Voorhees. Growth and Coarsening: Ostwald Ripening in Material Processing. Springer, 2002. [12] William Schockley and Hans J. Queisser. Detailed balance limit of e ciency of p-n junction solar cells. Journal of Applied Physics, 32:510519, 1961. [13] K. Sears, H. H. Tan, J. Wong-Leung, and C. Jagadish. The role of arsine in the self-assembled growth of inas/gaas quantum dots by metal organic chemical vapor deposition. J. Appl. Phys., 99:0449081, 2006.

Potrebbero piacerti anche