Сборник материалов VIІІ международной научной конференции студентов и молодых ученых «Наука и образование 2013»
жүктеу/скачать 15,22 Mb. Pdf көрінісі
|
- Бұл бет үшін навигация:
- PASSIVATION OF SMALL CdS CLUSTERS
- Results and Discussion
Рисунок 3. Спектр короткоживущего (1, 2) и долгоживущего компонентов ИКЛ (3) при 300К (1) и при 15 К (2,3) кристалла LiF-OH облученные при 200 К.
величины характеристического времени затухания ИКЛ в кристалле LiF-OH
Итак, в кристаллах содержащих кислород присутствуют полосы в области 3.1 эВ в спектрах ИКЛ. Это полоса обусловлен атомарным ионом кислорода в виде О 2- , расположенный в регулярных узлах решетки, как это доказано для кристалла LiF-O в [12]. Этот центр обладает сцинтилляционными свойствами: ее световыход есть величина постоянная в интервале доз выше (10 2 -10 4 ) Гр и в области температур 15-250К. Активаторные кислород-водородные центры, оптически активные в области 0.45 эВ, присутствуют только в кристаллах LiF-ОН (кристаллы LiF и LiF-O прозрачны в этой области спектра.) Этот тип дефектности, в отличие от описанной выше, под действием радиации претерпевает изменения, приводящие к следующим результатам: 1) к уменьшению интенсивности полос в области 0.45 эВ; 2) к возникновению новой широкой полосы в области 0.27 эВ; 3) к образованию молекулярных ионов кислорода О 2 –
которых свидетельствует появление ЭКС. Как и в других ЩГК [например,13], процесс радиационного создания ионов О 2 –
Выводы. С использованием техники измерения с высоким временным и спектральным разрешением при исследовании люминесценции кислородсодержащих кристаллов LiF, LiF-О и LiF-OH нами установлено присутствие во всех исследованных кристаллах активаторного кислородного центра, представляющего собой ион О 2- в узле решетки с излучательным переходом в области 3.1 эВ. 187 Другой тип активаторных центров в виде водород - кислородных соединений ОН – ,
2 О 2 с поглощательными переходами в области 0.45 эВ присутствуют только в кристаллах LiF-OH. Этот тип дефектности под действием радиации приводит к образованию молекулярных ионов кислорода О 2 – . Установлено, термоактивированный процесс создания молекулярных ионов О 2 –
кристалле LiF-OH.
1
performance // Phys. Stat. Sol.(b). – 2008. – V. 245, № 9. – P. 1701-1722. 2
Слободин Б.В., Сурат Л.Л., Самигулина Р.Ф., Ищенко А.В., Шульгин Б.В., Черепанов А.Н. Термохимия и люминесценция двойных пированадатов M+2M+2V2O7 // Неорганические материалы. – 2010. – Т. 46, № 5. – С. 590-597. 3
Антонов-Романовский В.В. Кинетика фотолюминесценции кристаллофосфоров. – М.: Наука, 1966. – 324 с. 4
283 с. 5
Лисицын В.М., Корепанов В.И. Спектральные измерения с временным разрешением: учебное пособие. – Томск: Изд-во ТПУ, 2007. – 94 с. 1.
– 1970. – V. 31, № 6. – Р. 1291-1294. 2.
7.Wedding В., Klein M. V., Infrared Absorption of the Hydroxyl Ion in Alkali Halide Crystals // Phys. Rev. – 1969. – V.177, №3. – P.1274-1288 3.
соединений. М.: Мир. 1991. – 381 с.. 4.
9.Архангельская В.А., Гусева Е.В., Зингер Г.М., Королев Н.Е., Рейтеров В.М. //Оптика и спектр. 1986. T.61. №3. -C. 542-549. 5.
Т.105, - С. 531-537. 6.
11. Лисицына Л.А., Корепанов В.И., Елисеев А.Е., Даулетбекова А.К., Абдрахметова А.А. Импульсная катодолюминесценция F 2 -центров в чистых и активированных кристаллах LiF // Неорганические материалы. – 2011. – T. 47, № 5. – С. 600-604. 7.
12. Егранов А.В., Раджабов Е.А. Спектроскопия кислородных и водородных примесных центров в щелочно-галоидных кристаллах. Новосибирск, Наука. 1992. - 160 с. 8.
– doped NaF // J.Phys.Chem.Solids. Pergamon Press. – Printed in Great Britain, 1968. – V. 29. – P. 1119-1125. Meistrich M.L. // J. Phys. Chem. Solids. 1968. V. 29. P. 1119.
L.Gumilyov Eurasian National University, Astana Colloidal quantum particles have a number of outstanding properties that make them useful for technical applications in wide range of areas such as bio imaging, photocatalyst, sensor and solar cell technology (1-5). However, the high ratio of surface area to volume makes such particles prone to imperfectly passivated surfaces, which introduce trap states. Such trap states may have several unwanted consequences, such as a promoting recombination of charge carriers that are detrimental to solar cell device performance. It is therefore essential to obtain understanding of the
188 structure-property relations that are associated with the occurrence of trap states and how QD surfaces can be passivated, either by the composition of the quantum dots themselves, or by the use of special passivating ligands. The study of optical properties offers a possibility to shed the light on trap states and the role of imperfection of QDs. The availability of optical data for small model clusters for QDs which at the same time are reachable by first principles electronic structure calculations, makes such QDs suitable for studies of trap states. Herein we study the optical properties of CdS nanocsrytals with time dependent density functional theory to accomplish that purpose. We first test the parametrization of the computational model in terms of density functional and basis sets for the optical absorption for the pure clusters with well determined band gaps, and use the results of that for the further study of imperfection and trap states. Particularly we have considered the effect of oxygen surface impurities on the optical properties of CdS clusters. Cu atoms constitute another kind of impurity which according to known experimental data is responsible for deep trap states. In this paper we study the spectra of wurtzite structure CdS cluster containing 17 Cd atoms, surface passivation by S or SH groups, the role of surface impurities such as oxygen and copper. Methods Structure CdS crystals have three stable crystal structures: wurtzite (hexagonal), zinc blende (cubic) and rock salt. The third one exists only at extreme conditions. Bulk CdS crystals have wurtzite structure, while nano-sized CdS crystals may exhibit both zinc blende and wurtzite structures depending on synthesis conditions (6, 7). Quantum mechanical calculations performed by Frenzel and Joswig et al. (8, 9) have shown that wurtzite and zinc blende structures of the same size produce almost the same spectra. The only difference is that in the case of wurtzite structure the absorption peaks are doublets which relates to the symmetry difference of the two structures. The shape of the cluster, however, was found to be of minor importance for the optical properties (8, 9). We have considered various hexagonal wurtzite clusters, all based on the Cd 17 S 10 cluster (Fig. 1) (10, 11). All clusters considered in our study were constructed according to this principle. It is interesting to note that bulk and nano-sized CdS crystals have been found to have very similar cell parameters from experiments (see, for example Ref. (10, 11)). Optimization of the clusters was not performed since in order to maintain the hexagonal structure, either the application of periodicity condition is required, or unfeasibly large (for quantum chemistry) clusters need to be used.
All calculated absorption spectra were obtained at TD DFT theoretical level (12-14). Excitation energies in this approximation are determined as solutions of non-Hermitian eigenvalue problem
189 (A-B)(A+B)|X+Y> = w 2 |X+Y> where A matrix is configuration interaction restricted to single excitations, B matrix involves excitation and de-excitation elements, X and Y are linear responses of density matrix.
17 S 10 cluster.
We have used different functionals to determine the one which provides optimal results for the bare cluster in terms of energy band gap: the widely used B3LYP (15, 16), the Becke three- parameter hybrid functional with the non-local correlation provided by Perdew/Wang 91 B3PW91 (17, 18) as has been suggested for band gap calculation of bulk semiconductors like CdS (19) as well as some long range correlated functionals LC-wPBE (20-23), wB97XD (24). The basis sets used in all calculations were 6-31+G(d) for S, O, Cu, 3-21G for H and DZ with LanL2 effective core potential for Cd (25). It should be noted that in all our calculations the charge of the system was determined by the formula 2*(g-j) – k for a Cd g S j (SH)
k cluster, i.e. in case of Cd 17 S
the charge is +14. Gaussian 09 Revision C.01 program package was used for the calculations (26). The dipole moment is reported relative to the center of charge for the nuclei.
One of the unique property of nanosized crystals is the profound effect of surface atoms on the optical properties since the ratio of surface atoms to the number of inner atoms is high compared to bulk crystals. Therefore it is important to pay attention to the correct surface passivation when modeling quantum dot clusters. Several authors have provided data and strategies for surface passivation for theoretical models (27, 28, 8, 9). All these data clearly show that dangling bonds of the surface atoms provide trap states in the band gap something that lowers the available LUMO level.
190 Furthermore recent studies (8, 9) have demonstrated that only single bonded Cd atoms are responsible for deep surface trap sates and that the use of S 2- ions and SH - groups can passivate single bonded Cd atoms. We have followed this way of passivation in the present study. Most of theoretical studies of CdS quantum dots do not provide direct comparison of predicted and experimental values of band gap. This is mostly due to dependence of UV-vis spectrum of nanocrystalls on the cluster size. Therefore it is necessary to establish well defined parameter for linking calculation and experiment. The authors of (8, 9) have demonstrated good agreement between measured spectra and the calculated ones for clusters possessing the same number of cadmium atoms. This is probably due to the fact that cations Cd 2+ define the size of space available for electron localization in a CdS cluster, i.e. particle size correlates with number of cations. Therefore it is reasonable to consider number of Cd atoms as the main parameter affecting the value of band gap of the CdS clusters. Hence we have used this criteria for comparing theoretical and experimental data. To demonstrate the role of passivation initially we have calculated the UV-vis spectrum of the unit cell [Cd 17 S
] 14+
which contains 13 single bonded Cd atoms (Fig. 1). The selection of this cluster for our theoretical consideration is explained by two practical reasons. First, this cluster can be considered as a cluster of relatively small size for modern available computational resources which makes it possible to consider this system at the DFT level. Second, there are available experimental results for CdS nano-crystals containing 17 Cd atoms therefore it is possible to make direct comparison of our model system with the real one. Furthermore based on systematic theoretical study provided by (8, 9) it is reasonable to suggest that basic tendencies obtained for this small cluster will also hold true for the clusters of bigger size. In Table 1 the predicted wavelengths of the lowest optical transition (λ 1 ) and the intense low energy transition (λ strong
) are presented as well as experimental results for Cd 17 S 4 (SCH
2 CH 2 OH) 26
cluster (29).
Table 1. Calculated wavelengths of the lowest energy transition (λ 1 energy transition (λ strong
) for [Cd 17 S 10 ] 14+ Wavelengt h (nm) B3LYP
LC-wPBE B3PW91 wB97XD Experiment (29) λ 1 450 508
511 532
290 λ strong 408 409
422 392
290
As one can see from Table 1 all theoretical data are far away from experimental result. The presence of λ 1 , which differs from λ strong , demonstrates the existence of trapped surface states in the band gap, which is represented by λ strong
. As a next step we have passivated some of the single bonded Cd atoms by adding S 2- or SH
- group. Structures with partial passivation are presented in Fig. 2. Addition of 6 SH - groups to 6 single bonded Cd atoms of [Cd 17 S 10 ] 14+
cluster leads to formation of [Cd 17 S
(SH) 6 ] 8+ cluster (Fig. 2) . We have also considered a cluster without single 191 bonded Cd atoms where all single bonded Cd atoms were passivated by adding 3 S 2- ions to the [Cd 17 S 10 (SH)
6 ] 8+ cluster, i.e. [Cd 17 S 13 (SH)
6 ] 2+ (Fig. 2). The UV-vis spectra were calculated for the clusters presented in Fig. 2 and the results are presented in Table 2. Here N singCd
represents number of single-bonded Cd atoms in the cluster, in particular N singCd =7 refers to [Cd 17 S 10 (SH) 6 ] 8+ and N
singCd =0 refers to [Cd 17 S
(SH) 6 ] 2+ . Comparison of theoretical results with the experiment (Table 2) demonstrates that long range corrected functional LC-wPBE gives more accurate results than the other three functionals do. Therefore we have used the LC-wPBE functional in the following calculations. The failure of B3LYP functional is due to the well known problem of standard exchange-correlation DFT functionals (30, 39). The small deviation of theoretical (cluster Cd 17 S
(SH) 6 ) and experimental (cluster Cd 17 S 4 (SCH
2 CH 2 OH) 26 ) data is probably due to difference of the nature of passivating agents and contribution of size distribution of nano-particles to the measured spectrum. It should be noted that the addition of sulfur atoms, which leads to size increase of the cluster, does not provide red shift of band gap in theoretical model as may be expected from quantum confinement effect. This is in agreement with calculations in the literature that it is that number of Cd atoms that defines the band gap value in all considered clusters as mentioned in the introduction.
Table 2. Calculated wavelengths of the lowest transition (λ 1 ) and first intensive transition (λstrong) for passivated clusters. N singCd
means number of single-bonded Cd atoms. Wavelengt h (nm) B3LYP
LC-wPBE B3PW91
wB97XD Experime
nt (29)
N singCd
=7 N singCd =0 N singCd
=7 N singCd =0 N singCd =7 N singCd =0 N singCd =7 N singCd = 0 λ 1 430
373 396
310 427
496 384
387 290
λ strong
390 356
328 298
389 460
330 366
290
192
Fig. 2. Structures of clusters [Cd 17 S 10 (SH) 6 ] 8+ and [Cd
17 S 13 (SH) 6 ] 2+ .
Calculated with LC-wPBE functional UV-vis spectra of [Cd 17 S 10 ] 14+ , [Cd 17 S 10 (SH)
6 ] 8+ and [Cd
17 S 13 (SH) 6 ] 2+ clusters are shown in Figure 3. The band halve-width of 0.2 eV was used to reproduce theoretical spectra of all clusters. Table 1, Table 2 and Fig. 3 show for LC-wPBE functional that gradual reduction of number of single bonded Cd atoms by adding S atoms and SH groups, while keeping the total amount of Cd atoms constant leads to an increase of the band gap. As one can see from Fig. 3 the predicted onset of absorption intensities for the three clusters varies significantly. Adding of 6 SH - groups to 6 single bonded Cd atoms, i.e. transformation of [Cd 17 S 10 ] 14+ into [Cd 17 S 10 (SH)
6 ] 8+ , causes substantial increase of intensity of long-wavelength peak. Analysis of transitions of the [Cd 17 S
(SH) 6 ] 8+ cluster shows that the λ strong intensity is of the same order of magnitude as for [Cd 17 S 10 ] 14+ . The difference in the intensities shown in Fig. 3 for the clusters [Cd 17 S
] 14+
and [Cd 17 S 10 (SH)
6 ] 8+ is resulted from presence of second intensive transition 193 located close to λ strong of [Cd
17 S 10 (SH) 6 ] 8+ cluster. This transition has much higher probability than λ strong
of [Cd 17 S 10 (SH)
6 ] 8+ and relates to orbitals located on SH - groups. Further passivation of single bonded Cd atoms of [Cd 17 S 10 (SH)
6 ] 8+ provided by addition of 3 S atoms, namely transformation of [Cd
17 S 10 (SH) 6 ] 8+ into [Cd 17 S
(SH) 6 ] 2+ , results in substantial decrease of band gap absorption intensity, although [Cd 17 S 13 (SH)
6 ] 2+ contains SH groups. This indicates that only presence of SH -
groups does not provide very intensive transition near λstrong but presence of single bonded Cd atoms and SH groups does.
17 S 10 ] 14+
, [Cd 17 S 10 (SH)
6 ] 8+ and [Cd 17 S 13 (SH)
6 ] 2+
According to population analysis of unoccupied orbitals responsible for transitions located between λ 1 and λ strong have significant contribution from single bonded Cd atoms for clusters [Cd 17
10 ] 14+ and [Cd 17 S 10 (SH)
6 ] 8+ and this is in accordance with results (8, 9). For [Cd 17 S 10 (SH)
6 ] 8+ SH groups make marked contribution to some of these orbitals. Highest occupied orbitals relate to two coordinated S atoms (not SH groups). Prediction of λ 1 which differs from λ strong for
[Cd 17 S 13 (SH)
6 ] 2+ cluster (Table 2) demonstrates presence of shallow trap state but not a deep one since these two transitions are close to each other in energy. Employment of the effective mass approximation model (32-34) for estimating the value of the band gap in our case is not appropriate, as this approximation overestimates band gap value significantly for small clusters (diameter less than 30 Å) (35), i.e. estimated wavelength of band gap transition is blue shifted comparing to the real wavelength. We note that in our case the [Cd 17
13 (SH)
6 ] +2 cluster without single bonded Cd atoms has a diameter of about 10 Å. Despite of the large abundance of papers on CdS nanoparticles published in recent almost 30 years there are some discrepancies in interpretation of experimental data. For instance, some authors have assumed that SH groups at surface of CdS nanoparticles form trap states which quench band gap emission and provide weak longwave emission (36-39), while other authors based on
194 experimental (40-43) and theoretical (8, 9, 44) data suggested that non-passivated Cd atoms at the surface, i.e. S 2- vacancies, are responsible for deep trap states and corresponding red emission. Our results support the suggestion that single bonded Cd atoms provide deep trap states in the band gap region. In our case, on the one hand non-passivated [Cd 17 S
] 14+
and partially passivated [Cd
17 S 10 (SH) 6 ] 8+ complexes with single bonded Cd atoms have deep trap states, while on the other hand the presence of SH - groups on the surface of [Cd 17 S 13 (SH) 6 ] 2+ complex do not exhibit deep trap states. Calculations show that number of trap states m (number of transitions located between λ 1 and
λ strong
) correlates with dipole moment of nanocrystals. For cluster [Cd 17 S 10 ] 14+ dipole moment is 13.58 D and m=3. Addition of six SH - groups to [Cd 17 S 10 ] 14+
forming [Cd 17 S 10 (SH)
6 ] 8+ increased dipole moment up to 16.53 D and increased m up to 5 although band gap was shifted to higher energy. And finally three S 2- ions added to [Cd 17 S 10 (SH) 6 ] 8+ forming [Cd 17 S
(SH) 6 ] 2+ cluster provided decrease of dipole moment down to 7.76 D and m=1. Thus increase of dipole moment leads to increase of m and vice-versa. жүктеу/скачать 15,22 Mb. Достарыңызбен бөлісу:
|