非揮發記憶體用之高分子/核殼奈米顆粒奈米複合材料之合成與元件製備(III)

行政院國家科學委員會專題研究計畫 成果報告

非揮發記憶體用之高分子/核殼奈米顆粒奈米複合材料之合

成與元件製備(3/3)

研究成果報告(完整版)

中 華 民 國 99 年 11 月 17 日

行政院國家科學委員會補助專題研究計畫

■ 成 果 報 告

□期中進度報告

非揮發記憶體用之高分子/核殼奈米顆粒奈米複合材料之合

成與元件製備

計畫類別：□ 個別型計畫 ■ 整合型計畫

計畫編號：NSC 98-2218-E-009-003

執行期間：96 年 8 月 1 日至 99 年 7 月 31 日

執行單位：國立交通大學 材料科學與工程學系

計畫主持人：韋光華

共同主持人：許鉦宗、徐雍鎣

計畫參與人員：陳振嘉、邱茂源、陳家閔、陳韋達、楊婷婷

成果報告類型(依經費核定清單規定繳交)：□精簡報告 ■完整報告

本成果報告包括以下應繳交之附件：

□赴國外出差或研習心得報告一份

□赴大陸地區出差或研習心得報告一份

□出席國際學術會議心得報告及發表之論文各一份

□國際合作研究計畫國外研究報告書一份

處理方式：本計畫可公開查詢

中 華 民 國 年 月 日

行政院國家科學委員會專題研究計畫成果報告

利用 core/Shell 結構之 CdSe@ZnSe 量子點與 P3HT 混合當

作主動層，做成有機薄膜電晶體，成功地完成光學寫入與電

子抹除之記憶體元件

An optical programming/electrical erasing memory drvice:

Organic thin filme transistors incorporating core/shell

CdSe@ZnSe quantum dots and ploy(3-hexylthiophene)

計畫編號：NSC 98-2218-E-009-003

執行期限：96 年 8 月 1 日至 99 年 7 月 31 日

計畫主持人：韋光華 交通大學材料系

共同主持人：許鉦宗、徐雍鎣 交通大學材料系

Abstract

An optical programming/electrical erasing memory device was fabricated by adopting organic thin film transistors incorporating core/shell CdSe@ZnSe quantum dots (QDs) and poly(3-hexylthiophene) (P3HT) as active layers. After illumination, the presence of quantum well-structured core/shell CdSe@ZnSe QDs within the P3HT film enhanced the maximum ON/OFF ratio substantially to 2700; this value was maintained for 8000 s without noticeable decay. The ON state current could be erased effectively when using a single pulse of the gate voltage (-10 V). This fabrication approach opens up the possibility of improving the memory performance of polymeric materials prepared at low cost using simple processes.

採用包覆核殼CdSe@ZnSe 量子點(QDs)和 3-己烷噻吩(P3HT)為主動層之有機薄膜電晶體去

製作光學寫入/電子抹除記憶體元件。照光後，在 P3HT 膜內存在有量子牆結構核殼 CdSe@ZnSe 量子點(QDs)強化 ON/OFF 比例達 2700;此值可維持達 8000 秒而沒有明顯衰 退。當使用單一閘極電壓脈衝(-10 V)，這 ON 狀態電流可以有效地被抹除。此製作方式開 啟改善高分子材料製備在簡單低價值製程過中之記憶體效果的可能性。

關鍵字：optical programming (光學寫入), electrical erasing(電子抹除), core/shell quantum dots (核殼量子點)

1. Introduction

The development of conjugated polymers for use in organic optoelectronic devices is an area of intense investigation.Several research groups have recently reported the photoresponse and memory functions of organic thin film transistors (OTFTs) [1-9]. One such early device took advantage of the illumination of poly(3-hexylthiophene)(P3HT) with light at a wavelength of 632.8 nm to create electric charges that were later trapped at the polymer–dielectric interface [1]; this system

featured an ON/OFF ratio of ca. 30 for the memory window at a gate voltage (VGS) of 60

V under a light intensity of 70 mW/cm2. This device exhibited a loss of 70% in its drain current and a short retention time after turning the light off. An alternative approach involves the use of conjugated polymers or quantum dots (QDs) as photosensitive materials along with carbon nanotube (CNT)-based field effect transistors for the fabrication of optoelectronic memory devices that function through optical programming and electrical erasing.One such

memory device featuring a CNT transistor coated with poly(3-octylthiophene) (P3OT) exhibited an ON/OFF ratio of ca. 103 after lasing at a wavelength of 457 nm at a value of VGS of 4 V under a laser power of ca.190

mW/cm2. For this system, a several percentage loss of the ON state current occurred 40 s after the laser light was turn off [2]. The same research group reported that for a similar device structure that exposed to a much higher laser power [3], it showed an absence of current decay after light turn-off or there exist two regimes: one decaying and one non-volatile [4].

Another optoelectronic memory device,

comprising a conjugated polymer coating a CNT transistor irradiated with UV light at 365 nm, provided an ON/OFF ratio of ca. 4 at a value of VGS of 4 V, but with a retention time of

over 16 h [5].

It appears that commercially viable TFT memory devices exhibiting high ON/OFF current ratios and long retention times (particularly when the gate voltage is turned off) are difficult to prepare. Approaches toward improving device performances while simplifying their fabrication processes are, therefore, necessary for the development and application of future commercial memory devices.

Bulk heterojunctions [10], in which n-type (e.g., fullerene) and p-type (e.g., conjugated polymer) materials are intimately mixed on the nanometer scale to form interpenetrated networks, have been adopted recently to achieve efficient photoinduced charge generation and separation[11,12]. On the other hand, semiconductor nanocrystal quantum dots (QDs) have also been used in such organic optoelectronic devices as solar cells [13] and light-emitting diodes (LEDs) [14]. Recently, the first bulk heterojunction photoresponsive OTFT memory device incorporating P3HT and CdSe QDs by our group was reported to have an ON/OFF ratio of ca. 102; because the CdSe QDs served as trap centers, the memory effect of the device was maintained for 1 h-even without a gating voltage [15].

Type-I core/shell structured QDs, in which the conduction and valence bands of the shell material are higher and lower, respectively,

than the corresponding values of the core material, feature a quantum well structure that can confine both holes and electrons in the core

[16]_{. Hence, the quantum well structure of such}

QDs enhances their electroluminescence (EL) in LED applications [17].

In this paper, we report bulk heterojunction polymer TFT memory devices exhibiting long retention times and high ON/OFF ratios that we fabricated using quantum well-structured QDs comprising CdSe cores and ZnSe shells (CdSe@ZnSe). To the best of our knowledge, this system is the first to employ a quantum well structure to enhance the memory effect of polymer TFTs.

2. Experimental Section

2.1. Materials

Regioregular poly(3-hexylthiophene) was obtained from Rieke Metals and used as received. Cadmium acetate dehydrate [Cd(OAc)2 · 2H2O] was obtained from Fisher Chemicals. Selenium (Se, 99.999%) and hexadecylamine (HDA, tech. 90%) were obtained from Aldrich. Trioctylphosphine oxide (TOPO, 98%), n-octylphosphonic acid (OPA, 98%), and trioctylphosphine (TOP, tech. 90%) were purchased from Alfa Aesar. Zinc stearate was obtained from J.T. Baker. The solvents heptane, toluene, methyl alcohol, and chloroform (HPLC-grade) were obtained from commercial sources.TOPO-capped CdSe and CdSe@ZnSe QDs were synthesized using a modification of a procedure reported previously

[18,19]_{: A mixture of Cd(OAc)2 · 2H2O (105 mg),}

HDA (1.39 g), OPA (225 mg), and TOPO (1.95 g) was heated in a 25-ml three-neck flask at 270 ºC under an argon flow to obtain a colorless, clear solution. At this temperature, the Se solution (100 mg in 2.4 ml TOP) was injected rapidly. The growth temperature was maintained at 270 ºC for 100 s and then the reaction mixture was cooled to room temperature. The CdSe QDs were collected as powders after their precipitation with MeOH. A colloidal solution of the CdSe QDs (ca. 20 mg) in heptane (4 ml) was heated in a 25-ml three-neck flask under an argon flow. After

addition of TOPO (2.5 mg) and HDA (1.5 mg), the mixture was heated at ca. 190 ºC to completely remove the heptane. Zinc stearate (316 mg) was dissolved in toluene (2.5 ml) at ca. 60 ºC. After cooling to room temperature, the resulting 0.2 M solution was mixed with TOP (2.5 ml) and Se (39.48 mg). This mixture was injected via syringe pump (0.085 ml/min) into the reaction flask containing the CdSe QDs at ca. 190 ºC. After the addition was complete, the crystals were annealed at 190 ºC for an additional 1 h. The CdSe@ZnSe QDs were collected as powders after their precipitation with MeOH.

2.2. Device fabrication and measurement

A solution of P3HT in CHCl3 was blended with a solution of the QDs in CHCl3; the

P3HT/QD composite weight ratio was 1:0.1. The P3HT and P3HT/QD TFT devices were fabricated in a bottom-gate configuration. An n+ silicon wafer (<0.005 Ω cm) was used as the substrate and gate; 900-Å thermal SiO2

(capacitance: 38.4 nF/cm2) was the gate insulator; a photolithographically patterned Au/Cr layer (thickness: 600/50 Å) functioned as the source and drain electrodes (W= 1000 μm; L = 10 μm). Octadecyltrichlorosilane (OTS) was deposited by immersing the substrate in 1 mM heptane solution for 10 min; the substrates were rinsed with heptane and isopropanol, followed by drying with N2. The P3HT and

P3HT/QD films (thickness: ca.60 nm) were deposited through spin-coating and then they were annealed at 150 ºC for 5 min inside a glove box under N2 atmosphere. The samples

were then transferred to a cryogenic probe station (VFTTP4, Lakeshore). The performance of each device was measured under vacuum (<1 × 10-5 torr) in the dark using a Hewlett– Packard 4156C semiconductor parameter analyzer. The devices were illuminated under vacuum using a tungsten halogen lamp (2.75 mW/cm2).

2.3. Characterization

TEM images were obtained using an FEI Tecnai Spirit TWIN apparatus operated at 120

keV. For TEM analysis, the devices were placed into 1% HF solution; after the active layers had floated to the solution surface, they were transferred to the TEM grid. A Hitachi U-4100 spectrophotometer was used to obtain optical absorption spectroscopy in the UV–vis range; a Hitachi F-4500 FL spectrophotometer was employed to obtain photoluminescence spectra.

3. Results and Discussion

To discern the effect that the ligands on the QDs had on the performance of the memory devices, we investigated trioctylphosphine oxide (TOPO) as the ligand because TOPO-capped CdSe QDs feature large barriers that prevent electron tunneling into P3HT and better dispersion than pyridine-capped CdSe QDs in P3HT (see Fig. S1, Supplementary material).

Fig. 1a presents the UV–vis and photoluminescence (PL) spectra of CdSe and CdSe@ZnSe QDs that we prepared from the same concentrations in toluene. The first excitonic absorption peak of the CdSe QDs appeared at ca. 540 nm, suggesting an average particle size of ca.2.85 nm [20], which is consistent with the dimensions obtained from transmission electron microscopy (TEM) image analyses. The presence of the ZnSe shell caused a red shift in the absorption spectrum and enhanced the PL intensity.

Fig. 1b displays a schematic energy level diagram of the CdSe and ZnSe bulk materials, P3HT, and Au electrodes [21,22]. Because the lowest unoccupied molecular orbital (LUMO) and the highest occupied molecular orbital (HOMO) of P3HT lie above the conduction band (CB) and valance band (VB) edges of the CdSe and ZnSe materials, respectively, the P3HT–ZnSe interface forms an offset band heterojunction; in contrast, the CdSe core and ZnSe shell form a type-I heterojunction. When illuminated, excitons were generated in the QDs and P3HT, charge separation occurred at the P3HT–QDs interface, and

then electrons and holes were transferred into the QDs and P3HT, respectively [23]. The work function of Au (5.1 eV) matched the HOMO of P3HT (4.9 eV); therefore, an Ohmic contact formed for hole injection, resulting in hole-only transport in the P3HT/QD TFTs.

Fig. 2a displays the transfer curves (drain-to-source voltage, VDS = _20 V) of the P3HT-only and P3HT/QD (including CdSe and CdSe@ZnSe) bulk heterojunction TFTs in the dark and under white light (2.75 mW/cm2). All of the devices exhibited the characteristic behavior of p-channel field-effect transistors. The hole mobility lh was obtained using the following equation

[24]_:

IDS =W/2LμhC(VGS - Vth)2 (1)

where Vth is the threshold voltage, W is the channel width, L is the channel length, C is the gate oxide capacitance perunit area, and VGS is the gate voltage. In the saturation

regime, the hole mobilities of the P3HT-only, P3HT/CdSe QD, and P3HT/CdSe@ZnSe QD devices were 4 × 10-4, 8 × 10-4, and 3 × 10-3

cm2 V-1 s-1, respectively. Since the HOMO of the P3HT is much higher than the VB of CdSe QDs, this energy difference constitutes a larger energy barrier that prevents holes transferring from P3HT to CdSe QDs

[25]_{.Thus, incorporation of the CdSe and}

CdSe@ZnSe QDs into the P3HT enhanced the hole mobility of the devices slightly, owing to the fact that it is possible that the QDs reduced the density of traps in the polymer [the concentration of the QDs in the P3HT film was 10% (w/w)]. The hole mobility of each of these three composites was lower than that reported recently for pure P3HT (ca. 10-2 to 10-3 cm2 V-1 s-1) [26]. Note, however, that several factors, such as the molecular weight of the P3HT, its methods of preparation and purification, the channel dimensions, and the substrate treatment conditions, can influence the characteristics of a TFT device. The values

of Vth of the P3HTonly,P3HT/CdSe QD, and P3HT/CdSe@ZnSe QD devices were 0.8, 3.4, and 3.8 V, respectively; i.e., those of the blended devices shifted to more-positive values, indicating the existence of a permanent electric field at the interface.The increase in the drain current of the polymer/QD blends under illumination resulted from accumulation of the majority carriers (holes) inside the active layer; these holes tended to drift toward the drain electrode, whereas the electrons will stay in the QDs or the insulator layer. It is well-known that the properties of the interface between the insulator and the semiconductor can critically influence the device performance. When the OTFT is illuminated, the electrons are attracted to the interface with the gate dielectric by a positive gate voltage; they are then trapped either in the dielectric layer or at the interface.Acceptor-like traps, when the traps are filled by electrons, leads to a positive threshold voltage. After turning the light off, electron detrapping under a negative gate bias indicates a returning to the initial state.

Fig. 2b present plots of the ratio Ilight/Idark

versus VGS. We obtained the ratio Ilight/Idark

from the transfer curves of the drain current for samples either in the dark or irradiated under white light (2.75 mW/cm2). The

Ilight/Idark ratio depends on the gate bias for a

given drain bias; it decreases as the gate bias is applied above or below the switch-on voltage. In the depletion regime, the maximum Ilight/Idark ratios for the P3HT-only,

P3HT/CdSe, and P3HT/CdSe@ZnSe devices were 1.6 × 102, 7.1 × 102, and 2.7 × 103,

respectively; i.e., the photosensitivity of the P3HT/CdSe@ZnSe device was higher than that of the P3HT-only and P3HT/CdSe devices. Because the switch-on voltages, which were defined by the maximum Ilight/Idark ratios, for the P3HT/CdSe and P3HT/CdSe@ZnSe devices were 5.6 and 8 V, respectively (Fig. 2b), we chose to operate these two devices at a value of VGS of 5 V

for our subsequent time-response studies. Fig. 3a displays the evolution of the normalized drain current for the P3HT/CdSe

and P3HT/CdSe@ZnSe devices at values of VDS and VGS of -20 and 5 V, respectively,

after they were subject to a light of 2.75 mW/cm2 with a duration of 100 s. At the onset of illumination, the drain current of both the P3HT/CdSe and P3HT/CdSe@ZnSedevices increased dramatically, and then dropped off to form

plateaus when light is turned off. After the light had been turned off for 400 s, the drain current of the P3HT/CdSe@ZnSe device had decreased by 14%, compared with a corresponding loss of 60% for the P3HT/CdSe device. We suspect that the mechanism underlying this behaviors involved two relaxation processes: (i) rapid decay corresponding to the recombination of closely spaced charge carriers and (ii) slow decay resulting from the recombination of well-separated carriers [27] (Fig. 3b). The slow decay might be manifested by the fact that, in a heterojunction device, the spatially separated holes and electrons will move differently-the holes drifting toward the channel and then reaching the drain electrode, the electrons mostly confined in the QDs and at the P3HT-SiO2 interface.

After the light was turned off, the devices existed in a non-equilibrium state; some of the photogenerated holes presumably recombined with some residual electrons that were not confined in QDs, causing a reduction in the drain current,eventually reaching a metastable state. Because the coverage of the CdSe QDs surfaces by TOPO was only ca.55% [21], the ZnSe shell layer between the CdSe core and the P3HT polymer in the P3HT/CdSe@ZnSe devices resulted in an additional tunneling barrier that prevented the electrons from tunneling back to P3HT, leading to a smaller decrease in the drain current and a larger retention time relative to those of the CdSe QDs devices, as indicated by the slope of the drain currents at 600 s.

Because low power consumption is an important feature for non-volatile memory applications, it is preferable to operate optoelectronic memory devices in the absence of a gate voltage. Fig. 3c displays

the evolution of the normalized drain currents at values of VDS and VGS of -20 and

0 V, respectively, for the devices subjected to a light of 2.75 mW/cm2 with a duration of 100 s. After turning off the light, the P3HT/CdSe@ZnSe and P3HT/CdSe devices exhibited losses of 8 and 35%, respectively, of their ON state currents, with ON/OFF ratios of 36 and 1.5, respectively. Therefore, we conclude that Type-I heterojunction core/shell QDs are more suitable than homogenous QDs for memory applications both with and without applied gate voltages.

Fig. 4 presents TEM images of the CdSe and CdSe@ZnSe QDs dispersed in the P3HT matrix at a P3HT-to-QD weight ratio of 1:0.1. The bright appearance of the P3HT regions relative to dark QD regions in the contrast image was probably due to the large difference in their respective electron densities. The CdSe QDs were distributed rather homogenously in the P3HT matrix; we suspect that the lower homogeneity of the P3HT/CdSe@ZnSe film was due to a loss of TOPO coverage during the growth of the shell. The TEM images in the insets to Fig. 4 reveal that the CdSe and CdSe@ZnSe QDs had average sizes of ca. 2.9 and 4.3 nm, respectively. The ZnSe shell thickness was ca. 0.7 nm, i.e., slightly larger than the critical penetration length of electrons (ca. 0.5 nm) [21].

To determine the optimal operating conditions, we fabricated 10 devices from

three independently prepared P3HT/CdSe@ZnSe films; the optimal ON/OFF ratio was greater than 1000 at a value of VGS of 10 V. Fig. 5 reveals that the

P3HT/CdSe@ZnSe devices exhibited a high ON/OFF ratio of 2700 at a value of VGS of 10 V-without any noticeable decay after the light had been turned off for 8000 s. This result indicates that incorporating core/shell QDs into a conjugated polymer significantly extends the lifetime of the memory states of the resulting polymer TFTs. Moreover, the inset to Fig. 5 also displays the dynamic responses of the optical programming and electrical erasing of the P3HT/CdSe@ZnSe device. The ON state current could be erased

effectively when using a singlepulse of the gate voltage (-10 V) for a short duration (1 s).When this negative pulse gate bias was applied, the Fermi level of CdSe and ZnSe modulated up towards the conduction band, thereby reducing the barrier height. As a result of the decrease in the barrier hight, electron jump back into the P3HT and recombine with hole. Thus, we suspect that the trapped electrons were induced by the electric field to move out of the QDs and recombine with the holes to reform the OFF state. Based on this protocol of operation, it was possible for us to program the P3HT/CdSe@ZnSe device optically and then erase it electrically.

4. Conclusions

In summary, we have examined the optoelectronic properties and memory effects of polymer TFTs incorporating P3HT/CdSe and P3HT/CdSe@ZnSe QDs as an active layer. After illumination, the presence of the quantum well-structured core/shell CdSe@ZnSe quantum dots in the P3HT film substantially enhanced the ON/OFF ratio to 2700, maintaining this value for 8000 s without noticeable decay. This fabrication approach opens up the possibility of improving the memory performance of polymeric materials prepared at low cost using simple processes.

Acknowledgment.

We thank the National Science Council of Taiwan for funding (NSC 97-2120-M-009-0 06) and (NSC97-2218-E009-004).

References

[1] S. Dutta, K.S. Narayan, Adv. Mater. 16 (2004) 2151.

[2] J. Borghetti, V. Derycke, S. Lenfant, P. Chenevier, A. Filoramo, M.Goffman, D. Vuillaume, J.P. Bourgoin, Adv. Mater. 18 (2006) 2536.

[3] J.P. Bourgoin, J. Borghetti, P. Chenevier, V. Derycke, A. Filoramo, L.Goux, M.F. Goffman, S.

Lyonnais, K. Nguyen, G. Robert, S. Streiff, J.M. Bethoux, H. Happy, G. Dambrine, S. Lenfant, D. Vuillaume, Proc. Int. Electron. Dev. Meeting (2006) 435.

[4] C. Anghel, V. Derycke, A. Filoramo, S. Lenfant, B. Giffard, D.Vuillaume, J.-P. Bourgoin, Nano Lett. 8 (2008) 3619.

[5] A. Star, Y. Lu, K. Bradley, G. Grüner, Nano Lett.4 (2004) 1587.

[6] L. Hu, Y.L. Zhao, K. Ryu, C. Zhou, J.F. Stoddart, G. Grüner, Adv. Mater. 20 (2008) 939.

[7] S. Dutta, K.S. Narayan, Phys. Rev. B: Condens. Matter Mater. Phys. 68(2003) 125208.

[8] K.S. Narayan, M. Rao, R. Zhang, P. Maniar, Appl.Phys. Lett. 88 (2006) 243507.

[9] Y. Shi, H. Tantang, C.W. Lee, C.H. Weng, X. Dong, L.J. Li, P. Chen, Appl. Phys. Lett. 92 (2008) 103310.

[10] (a) N.S. Sariciftci, L. Smilowitz, A.J. Heeger, F. Wudl, Science 258(1992) 1474;(b) G. Yu, J. Gao, J.C. Hummelen, F. Wudl, A.J. Heeger, Science 270 (1995) 1789.

[11] N. Marjanovic, T.B. Singh, G. Dennler, S. Gunes, H. Neugebauer, N.S. Sariciftci, R. Schwodiauer, S. Bauer, Org. Electron. 7 (2006) 188.

[12] S.M. Mok, F. Yan, H.L.W. Chan, Appl. Phys. Lett. 93 (2008) 023310.

[13] W.U. Huynh, J.J. Dittmer, A.P. Alivisatos, Science 295 (2002) 2425.

[14] M. Gao, B. Richter, S. Kirstein, Adv. Mater. 9 (1997) 802.

[15] C.C. Chen, M.Y. Chiu, J.T. Sheu, K.H. Wei, Appl. Phys. Lett. 92 (2008) 143105.

[16] X. Peng, M.C. Schlamp, A.V. Kadavanich, A.P. Alivisatos, J. Am. Chem. Soc. 119 (1997) 7019. [17] (a) J. Lim, S. Jun, E. Jang, H. Baik, H. Kim, J.

Cho, Adv. Mater. 19 (2007) 1927;(b) Z. Tan, F. Zhu, J. Xu, A.Y. Wang, J.D. Dixon, L. Li, Q. Zhang, S.E. Moheney, J. Ruzyllo, Nano Lett. 7 (2007) 3803;(c) M.T. Harrison, S.V. Kershaw, A.L. Rogach, A. Kornowski, A. Eychmuller, H. Weller, Adv. Mater. 12 (2000) 123.

[18] H. Skaff, K. Sill, T. Emrick, J. Am. Chem. Soc. 126 (2004) 11322.

[19] P. Reiss, J. Bleuse, A. Pron, Nano Lett. 2 (2002) 781.

[20] W.W. Yu, L. Qu, W. Guo, X. Peng, Chem. Mater. 15 (2003) 2854.

[21] J. Bleuse, S. Carayon, P. Reiss, Physica E 21 (2004) 331.

[22] S. Cho, J. Yuen, J.Y. Kim, K. Lee, A.J. Heeger, Appl. Phys. Lett. 90 (2007) 063511.

[23] N.C. Greenham, X. Peng, A.P. Alivisatos, Phys. Rev. B 54 (1996) 17628.

[24] G. Horowitz, Adv. Mater. 10 (1998) 365. [25] Z.X. Xu, V.A.L. Roy, P. Stallinga, M. Muccini,

S. Toffanin, H.F. Xiang,C.M. Che, Appl. Phys. Lett. 90 (2007) 223509.

[26] R.J. Kline, M.D. McGehee, E.N. Kadnikova, J. Liu, M.J. Frechet, Adv. Mater. 15 (2003) 1519. [27] K.S. Narayan, N. Kumar, Appl. Phys. Lett. 79

Figure. 1. (a) Absorption and photoluminescence spectra of the CdSe and CdSe@ZnSe QDs. (b)

Energy level diagram for the CdSe and ZnSe bulk materials, P3HT, and the electrode materials.

Figure 2. (a) Transfer characteristics (VDS = -20 V) of TFTs incorporating P3HT-only,

P3HT/CdSe, and P3HT/CdSe@ZnSe blend films and operated in the dark and under white light (2.75 mW/cm2).(b) Relative Ilight/Idark ratios of the transfer characteristics of the drain current in

Figure 3. Time responses of the normalized drain currents of the P3HT/CdSe and

P3HT/CdSe@ZnSe devices (VDS = -20 V) illuminated by a light of 2.75 mW/cm2 with a duration

of 100 s at values of VGS of (a) 5 and (c) 0 V. (b) Schematic representation of the relaxation

processes within the bulk heterojunction P3HT/CdSe@ZnSe active layers in TFT devices. Red: slow decay process; blue: fast decay process. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Figure 4. TEM images of (a) the CdSe QDs and (b) the CdSe@ZnSe QDs dispersed in the P3HT

matrix Insets: TEM images of the CdSe and CdSe@ZnSe QDs (scale bar: 20 nm).

Figure 5. Time response of the value of IDS of the P3HT/CdSe@ZnSe device at VGS = 10 V and

VDS = -20 V illuminated by a light of 2.75 mW/cm2 with a duration of 100 s. Inset: Time response

of the P3HT/CdSe@ZnSe device after a negative gate voltage pulse (VGS = -10 V) was applied at

行政院國家科學委員會專題研究計畫成果報告

P3HT/CdSe 量子點(Quantum dot)薄膜電晶體之光學反應與

記憶體之效應

Photoresponses and memory effects in organic thin film

transistors incorporating poly„3-hexylthiophene…/CdSe

quantum dots

計畫編號：NSC 98-2218-E-009-003

執行期限：96 年 8 月 1 日至 99 年 7 月 31 日

計畫主持人：韋光華 交通大學材料系

共同主持人：許鉦宗、徐雍鎣 交通大學材料系

Abstract

This paper describes the optical responses and memory effects of poly(3-hexylthiophene) P3HT/ CdSe quantum dot (QD) thin-film transistors (TFTs). TFTs incorporating P3HT/CdSe QD blends as the active layer exhibited higher photocurrents than did the corresponding P3HT-only devices because the heterojunction between P3HT and the CdSe QDs enhanced the separation of excitons. Moreover, the CdSe QDs served as trap centers so that the memory effect was maintained for several hours, even when the device was operated without a gating voltage. Here, we demonstrate the potential applicability of such P3HT/CdSe QD TFTs through repeated optical programming and electrical erasing.

此篇論文描素這P3HT/CdSe 量子點薄膜電晶體之光學反應與記憶體效果。薄膜電晶體包覆

有P3HT/CdSe 量子點為主動層比僅有 P3HT 元件存在有較高的光電流，因為 P3HT 與 CdSe

量子點之間的異質界面強化這激子的分離。而且，即使當元件被操作在沒有閘極電壓下，

這CdSe 量子點可提供一個陷阱中心以便於記憶體效果可維持幾個小時。這裡我們證明這樣

P3HT/CdSe 量子點薄膜電晶體經由光學寫入/電子抹除潛在應用性。

關鍵字：optical responses (光學反應), memory effects (記憶體效應), quantum dots (量子點)

1. Introduction

During the past few years, a tremendous amount of effort has been devoted to studies of polymer-based optoelectronic devices. The use of polymer memory has several advantages, including ease of fabrication and low cost. Many research teams are actively pursuing polymer phototransistors,but few are focusing on memory effects in the polymer devices, especially for polymer devices operated using optical programming and electrical erasing. A simple optically writeable memory device incorporating poly(alkylthiophene) as the active layer has been proposed, with the optically induced charge being trapped at the

polymer-dielectric interface.1 Carbon nanotube networks have also been coated with polymers to form optoelectronic memory devices that are written optically and read and erased electrically, but these blended polymer/carbon nanotube devices lost their memory capability because the carbon nanotubes separated from the substrate and because the nanotube bundles

contained metallic nanotubes.2 In

polymer-based memory devices, the dynamic switch phenomenon depends strongly on the gate effect.3 The “on” state can be returned to the “off” state by removing the gate voltage.4 Unfortunately, the on states of these devices were not maintained for very long in the

absence of an applied gate voltage, which limits their potential use in commercial applications. In this paper, we describe poly(3-hexylthiophene) (P3HT)/CdSe quantum dot (QD) thin-film transistors (TFTs) that exhibit long retention times for their on states even in the absence of a gate voltage.This behavior differs significantly from that reported previously.

Composite films of CdSe QDs and P3HT have been widely used in photovoltaic devices5 in which exciton dissociation and charge separation occur at the interface between the CdSe QDs and P3HT. Based on this principle, the electrons become trapped in the CdSe QDs when the light is turned off. This behavior enhances the retention time of polymer-based phototransistors. In this paper, we demonstrate that hole-only transport occurs in P3HT/CdSe QD TFTs. The drain-source currents (IDS) of

both the P3HT and P3HT/CdSe TFTs increased by up to several orders of magnitude upon irradiation with light when operated in the depletion mode. After turning off the light, the current decayed to a metastable state, where it remained for several hours, when the devices were subjected to a positive or zero gate voltage.

Trioctylphosphine oxide (TOPO)-capped CdSe QDs were synthesized using a modification of a procedure reported previously.6 A solution of P3HT (Rieke Metals, used as received) in chloroform (5 mg/mL) was blended with a solution of CdSe QDs (diameter of 3.5 ± 0.5 nm).The P3HT and P3HT/CdSe TFT devices were fabricated in a bottom-gate configuration (Fig.1). An n+silicon wafer (< 0.005 Ω cm) was used as the substrate and gate. 900 Å thermal SiO2 (capacitance of 38.4

nF/cm2) was the gate insulator.It was hydrophobically modified using hexamethyldisilazane vapor. The source and

drain fingerlike electrodes (W=3000 μm and

L=10 μm) were defined using standard

photolithography. A bilayer of Au/Ti (thickness of 1000/100 Å) was thermally evaporated and then lifted off. The P3HT and P3HT/CdSe films (thickness of 100 nm) were deposited through spin coating. The density of CdSe QDs in the thin P3HT film is about 9.93×1017 cm−3.

The films were subsequently annealed at 150 °C under N2 for 5 min. The performance of

each device was measured under vacuum (< 1×10−5 torr) in the dark using a Hewlett-Packard 4156C semiconductor parameter analyzer and a cryogenic probe station (VFTTP4, Lakeshore). The devices were illuminated under vacuum using a tungsten halogen lamp.

Figure 2 displays the transfer curves (drain-to-source voltage VDS=−20 V) of the

P3HT-only and P3HT/CdSe QD blend TFTs in the dark and under a white light of 0.26 mW/cm2. Both the P3HT and P3HT/CdSe TFTs exhibit the characteristic behavior of

p-channel field-effect transistors.

Ambipolar-type TFTs based on blended donor- and acceptor-type materials have been reported previously.7 The work function of Au (5.1 eV) matched the highest occupied molecular orbital of P3HT (4.9 eV),8 forming an Ohmic contact for hole injection. The drain currents measured by sweeping gate voltage at VDS=20 V. We

observed a weak electron current in the P3HT/CdSe and pure P3HT TFTs. The electron current remains weak even if higher positive gate voltages. Whereas Au strongly suppressed electron injection into CdSe [lowest unoccupied molecular orbital (LUMO):4.3 eV]9 and P3HT (LUMO: 3.0 eV)8 because of a large mismatch between its work function and the LUMO band of CdSe and P3HT. In addition, the current in the PH3T/CdSe TFTs is about one order of magnitude larger than for the pure PH3T TFTs in the whole voltage range. Incorporation of CdSe QDs into the P3HT lightly enhances the hole mobility of the devices, it is possible that CdSe reduced the density of traps in the polymer.10

The inset of Fig. 2 displays phase images—obtained by AFM (Multimode, DI) in the tapping mode—of both the P3HT and P3HT/CdSe thin films. The P3HT/CdSe composite thin film had a rougher surface morphology (rms roughness of 3.6 nm) in comparison with that of the P3HT-only film (2.0 nm). The surface morphology of the film incorporating the CdSe QDs was rough because of CdSe aggregation.The contrast in the phase image of the film blend indicates phase

separation. The transmission electron microscopy image of the blend film (not shown here) revealed that the CdSe QDs had unexpectedly aggregated into clusters. To avoid aggregation of the CdSe QDs, a suitable ligand must be used to passivate the QDs to enhance the performance of the memory devices.

We observed photoconductivity and the photovoltaic effect in the active layer of the transistors upon illumination.11 We attribute the significant increase in the drain current in the off state, when the device was being illuminated, to the enhancement of the drain current caused by the excitons in the polymer and nanocrystal. When illuminated, the excitons were generated in the CdSe QDs and P3HT. The electrons and holes eventually separated as a result of the electrical field. The drain currents of the P3HT/CdSe devices were larger than those of the P3HT-only devices because of the built-in field present at the P3HT-CdSe interface. The photoexcitation hole density within the thin film also contributed to the drain current and increased the threshold voltage to a large positive value.

Threshold voltages (Vth) were determined

from the intercepts of the √IDS-VGS plot. In

general, the values of Vth of the blend

P3HT/CdSe devices shifted to more-positive values,indicating the existence of a permanent electric field at the interface. The IDS-VGS curve

shifted toward a positive voltage under illumination. As shown in Fig. 2, the value of

Vth of the P3HT/CdSe device shifted to 10.0 V

(illuminated) from 4.2 V (darkness), i.e., Δ

Vth=5.8 V. In contrast, the value of ΔVth of the

P3HT-only device was ~1.9 V. The Δ Vth

extracted from the backward sweep curve is more pronounced than that extracted from the forward sweep. For the forward sweep, when an initial negative gate voltage is applied, the trapped electrons in the P3HT/SiO2 interface

and CdSe QDs would easily detrap and recombine with the hole in P3HT, leading to a reduction in the carrier density. Whereas, the recombination process of electrons and holes in backward sweep took place in the region that is not concerned with the memory functionality.

The increase in the value of the carrier density N* in the active layer could be

estimated using the equation N*=CiΔVth/e,

where Ci is the capacitance per unit area of the

dielectric layer, Δ Vth is the shift of the

threshold voltage, and e is the elementary charge. The values of N* of the P3HT/CdSe device and the P3HT-only device were ~1.39 ×1012 /cm2 and ~4.53×1011 /cm2, respectively, indicating that the P3HT/CdSe devices were more efficient at separating excitons, presumably because of the heterojunction at the P3HT-CdSe interface in the active layer.

Upon sweeping the voltage from positive to negative and then back to positive, we observed obvious hysteresis under illumination, but indistinct hysteresis in the dark (Fig. 2).This behavior was a consequence of the trapped charges present at the polymer-dielectric interface or in the dielectric material. The hysteresis for a P3HT device was typically less than 1.1 V. For the P3HT/CdSe devices, it was ~0.8 V. These similar values indicate a very small difference in their interface trap densities. The manifest hysteresis under illumination has also been observed in P3HT/PCBM phototransistors and organic capacitors.10,12 Apparent hysteresis occurred upon illumination as a result of an increased number of carriers in the active layer becoming easily trapped, either at the P3HT-SiO2 interface or in SiO2 itself. In

the P3HT/CdSe devices, some of the carriers were trapped in the CdSe QDs, resulting in the reduced degree of hysteresis.

Hysteresis in the illuminated devices resulted from trapped electrons. When applied a positive gate voltage of 7.5 V (where the hysteresis is most pronounced in the Fig.2), Fig. 3(a) indicates that the current in the P3HT-only devices reached a metastable state after turning off the pulse light source (30 s). This behavior is consistent with previous observations.1–3 Trapping of electrons in SiO2 or at the

P3HT-SiO2 interface screened the back-gate

potential, resulting in the metastable state. We attribute the slow current decay to bulk recombination, which is an indication of the slow nonexponential relaxation process inherent to polymer-based devices. After turning off the light source, the drain current decayed rapidly when the gate voltage was equal to zero. The current moved back to the

initial state in the P3HT devices after turning off the light (100 s). In the absence of a gate voltage, there was no external electrical field to induce the electrons; thus, the trapped electrons escaped from the trap centers (i.e., from the SiO2 or the P3HT-SiO2 interface) into the

active layer, resulting in decay of the drain current in the P3HT devices.

Relative to the P3HT-only devices, the P3HT/CdSe devices displayed entirely different behavior, as shown in Fig.3(b). Under illumination, the electrons were trapped not only in SiO2 and at the P3HT-SiO2 interface but

also in the CdSe QDs. After photoexcitation, the trapped electrons escaped from the trapping centers (SiO2 or P3HT-SiO2) after a few

minutes. In contrast, the highly localized electrons inside the CdSe QDs had difficulty jumping back into the polymer. Also, the presence of TOPO on the surface of the CdSe QDs enhanced the trapping of electrons. This metastable state was maintained for several hours, even in the absence of a gate bias. The P3HT/CdSe devices exhibited higher ION/ IOFF

ratios (>100) than did the P3HT-only devices (10). A large on/off ratio can also be achieved for the P3HT/CdSe and P3HT-only devices providing an appropriate gate voltage is applied. Our observations show that the on/off ratio of P3HT/CdSe TFTs is at least one order of magnitude larger than that the pure P3HT TFTs near the threshold voltage. Moreover, there is a much long retention time for the P3HT/CdSe devices than the P3HT-only devices in absence of a gate voltage. Although the P3HT/CdSe devices exhibited a memory effect in the absence of a gate voltage, their ION/ IOFF ratios

were too low to meet the required memory window.

Upon illumination with white light (2.75 mW/cm2), the drain current of the P3HT/CdSe device rose from 1.5 to 415 nA (Fig. 4). After turning off the white light, the drain current dropped slowly and eventually settled at a metastable state of 260 nA. Moreover, this metastable state could be erased efficiently using a single pulse of a gate voltage for a short duration (−15 V, 100 ms). When this negative pulse gate bias was applied, trapped electrons quickly recombined with the majority carriers

from the trap centers. After applying the negative electrical field, the Fermi level of CdSe also modulated up toward the conduction band, reducing the built-in field and, hence, enhancing recombination. Based on this mode of operation, we could repeatedly program the P3HT/CdSe device optically and electrically erase it.

In summary, we have investigated the electrical and optical properties of polymer memory TFTs incorporating P3HT and P3HT/CdSe as active layers. Upon illumination, the P3HT/CdSe TFTs exhibited stronger carrier induction in the channel and greater electron trapping ability than did the P3HT-only devices. This phenomenon resulted in a relatively high

ION/ IOFF ratio. After introducing the CdSe QDs

as electron trap centers, the retention time of the metastable memory state of the P3HT/CdSe TFT improved in the absence of a gate voltage. We are currently optimizing the working point of Vth at values of VGS near 0 V to maximize the

photoresponse so that the memory window can be widened further. We are also investigating replacing the TOPO molecules with long alkyl chains to enhance the retention time.

We are grateful to the National Science Council (NSC-96-2218-E-009-011) and the MOE-ATU Program in Taiwan for financial support.

1

S. Dutta and K. S. Narayan, Adv. Mater. (Weinheim, Ger.) 16, 2151 (2004).

2

A. Star, Y. Lu, K. Bradley, and G. Gruner, Nano Lett. 4, 1587 (2004).

3

J. Borghetti, V. Derycke, S. Lenfant, P. Chenevier, A. Filoramo, M. Goffman, D. Vuillaume, and J. P. Bourgoin, Adv. Mater. (Weinheim, Ger.) 18, 2535 (2006).

4

V. Podzorov, V. M. Pudalov, and M. E. Gershenson, Appl. Phys. Lett. 85, 6039 (2004).

5

W. U. Huynh, J. J. Dittmer, and A. P. Alivisatos, Science 295, 2425 (2002).

6

H. Skaff, K. Sill, and T. Emrick, J. Am. Chem. Soc.

7

E. J. Meijer, D. M. De Leeuw, S. Setayesh, E. Van Veenendaal, B. H.Huisman, P. W. M. Blom, J. C. Hummelen, U. Scherf, and T. M. Klapwijk, Nat. Mater.

2, 678 (2003).

8

Y. Kim, S. A. Choulis, J. Nelson, D. D. C. Bradley, S. Cook, and J. R. Durrant, Appl. Phys. Lett. 86, 063502 (2005).

9

R. A. M. Hikmet, D. V. Talapin, and H. Weller, J. Appl. Phys. 93, 3509 (2003).

10

Z. X. Xu, V. A. L. Roy, P. Stallinga, M. Muccini, S. Toffanin, H. F. Xiang, and C. M. Che, Appl. Phys. Lett.

90, 223509 (2007).

11

N. Marjanovic, T. B. Singh, G. Dennler, S. Gunes, H. Neugebauer, N. S. Sariciftci, R. Schwodiauer, and S. Bauer, Org. Electron. 7, 188 (2006).

12

D. M. Taylor, J. A. Drysdale, I. Torres, and O. Fernandez, Appl. Phys. Lett. 89, 183512 (2006).

Figure 1. (Color online) Schemes of devices structure and organic active layers. (a) Schematic

representation of the bottom-gate organic TFT configuration with an active polymer layer and interdigitated source and drain (S for source, D for drain; and G for gate). (b) Schematic
representations of the P3HT-only and P3HT/CdSe blend films.

Figure 2. (Color online) Transfer characteristics obtained for the (a) P3HT-only and (b)

P3HT/CdSe blend films (i) initially (in the dark) and (ii) under illumination. Under illumination, all of the devices displayed hysteresis. Transfer characteristics of the devices in the dark and under illumination (0.27 mW/cm2) were measured at VDS=−20 V. Inset: AFM phase images of

Figure 3. (Color online) Time responses of the drain current at VGS=7.5 V and VGS=0 V of the (a)

P3HTonly

and (b) P3HT/CdSe devices to a light pulse (2.75 mW/cm2, 30 s). Inset: Illustrations of the electron trap mechanisms for the (a) P3HT-only and (b) P3HT/CdSe devices.

Figure 4. (Color online) Dynamic responses of the optical programming and electrical erasing of

a typical P3HT/CdSe device. Light was turned on at t=80 s and turned off at t=90 s. A short (100 ms) negative gate voltage pulse was applied at t=260 s to erase the memory.

行政院國家科學委員會專題研究計畫成果報告

Au-CdS 核殼奈米晶體其介面載子傳輸動力學研究

Interfacial Charge Carrier Dynamics in Core-Shell Au-CdS

Nanocrystals

計畫編號：NSC 98-2218-E-009-003

執行期限：96 年 8 月 1 日至 99 年 7 月 31 日

計畫主持人：韋光華 交通大學材料系

共同主持人：許鉦宗、徐雍鎣 交通大學材料系

Abstract

The interfacial charge carrier dynamics for core-shell Au-CdS nanocrystals with various shell thicknesses were investigated and presented. Due to the difference in band structures between Au and CdS, a pronounced photoinduced charge separation took place at the interface of Au and CdS, resulting in the electron-charged Au core and the hole-enriched CdS shell. The electron-charging of Au core in Au-CdS nanocrystals was revealed with the corresponding XPS analysis and photocurrent measurement. Time-resolved PL spectra were measured to quantitatively analyze the electron transfer event between CdS shell and Au core for Au-CdS nanocrystals. An increase in the electron-transfer rate constant was observed for Au-CdS nanocrystals with increasing shell thickness, probably due to the less pronounced electron-hole interaction of thicker CdS, which enabled a fuller extent of participation of photoexcited electrons in the charge separation process. On the other hand, the hole-enriched CdS shell of Au-CdS nanocrystals upon light illumination was characterized with a photocatalytic process. The photocatalytic activity of Au-CdS nanocrystals was found to increase with increasing shell thickness, attributable to the greater capability of light absorption achieved by the extensive growth of CdS shell. The correlation of photocatalytic activity with the shell thickness of Au-CdS nanocrystals corresponded well with that of electron-transfer rate constant. As compared to the relevant commercial products like N-doped P-25 TiO2 and CdS powders, the as-synthesized Au-CdS nanocrystals exhibited superior

photocatalytic performance under visible light illumination, demonstrating their potential as an effective visible-light-driven photocatalyst. Furthermore, the result of performance evaluation under natural sunlight shows that the present Au-CdS nanocrystals can be used as highly efficient photocatalysts which may practically harvest energy from sunlight.

此研究主題為探討不同殼層厚度的金-硫化鎘(Au-CdS)核殼奈米晶體其界面間載子傳輸的 動力學。由於硫化鎘與金其能帶結構上的差異，可有效地將光激發所產生的電子侷限於金 奈米粒子中，而電洞則會相對的累積於硫化鎘殼層材料。電子儲存在金核層的現象係利用 X 光光電子能譜儀(XPS)與光電流(Photocurrent)量測設備來解析。另一方面，我們係利用時 間解析螢光光譜儀(Time-resolved PL spactra) 來定義分析出硫化鎘與金界面間電子傳輸現 象，其電子傳輸速率常數(electron-transfer rate constant)隨著硫化鎘殼層厚度的增加而增加， 推測可能的原因為於較厚的硫化鎘殼層情況下，其界面間電子電洞對交互作用較不明顯。 另一方面，光激發程序所產生與累積在硫化鎘殼層材料中的大量電洞也進一步利用於光催 化系統中。於實驗中可得到其光催化活性隨著硫化鎘殼層厚度的增加而增加，其原因可歸 屬於其成長較厚的硫化鎘殼層具有較佳的光吸收性能力，其光催化活性與其界面間電子傳 輸速率常數也可得到相呼應的現象。相較於氮-參雜的 P-25 TiO2粒子與其硫化鎘商用品的 比較下，Au-CdS 核殼奈米晶體於可見光的光催化系統中同樣展現出過人優秀的催化活性，

此結果也展現其材料具有潛力可發展於可見光催化的系統中。此外，於利用自然太陽光的

結果，Au-CdS 核殼奈米晶體則也具有潛力應用作為太陽能轉化作為高效率的光催化材料。

關鍵字：charge carrier dynamics (載子動力學), core-shell nanocrystals (核殼奈米晶體), charge separation (載子分離), photocatalysis (光催化)

1. Introduction

Modulation of charge carrier dynamics for semiconductors is important to the development of light-energy conversion systems.1 _{In general, the fast recombination of}

charge carriers in semiconductors would diminish the resulting photoelectric conversion efficiency. To effectively gain energy from light, the photoexcited electrons and holes of semiconductors must be separated to suppress the direct recombination of them. Previous studies have shown that charge separation of semiconductors can be essentially promoted through the introduction of suitable electron acceptors, such as metals,2 _{carbon derivatives,}3

and other semiconductors with appropriate band structures.4 _{By adopting these composite}

systems, a significant enhancement in the photoconversion efficiency can be attained. For example, CdSe quantum dots can show 2-3 orders of magnitude improvement in photocurrent generation once they were capped with a molecular shell of C60.3a This is due to

the sufficiently positive reduction potential of C60, which ensured a quick electron transfer

from excited CdSe to C60 and thus the

successful collection of electrons. Besides, with the attachment of CdS nanoparticles, ZnO nanowires exhibited enhanced photocatalytic activities.4b _{This enhancement resulted from the}

band offsets between CdS and ZnO, which may retard charge recombination to capture more charge carriers for participation in photocatalysis.

Photocatalysis is a valuable approach to practically utilize the solar power. In the last two decades, much research effort has been expanded to develop semiconductor photocatalysts because of their capability of converting light energy into chemical energy. Among the various composite systems, metal/semiconductor combination is of particular interest to photocatalytic applications.2,5 _{For metal/semiconductor}

composites, the presence of

metal-semiconductor interface may induce effective charge separation to favour the subsequent photocatalysis. The most commonly used semiconductor photocatalysts have been metal oxides like TiO2, which exhibits

ultraviolet absorption ability only due to its large bandgap energy. Many efforts have thus been made to modify TiO2 such that it can

absorb visible light to carry out photocatalytic reactions. For example, through the doping of suitableelements, an additional electronic level can be created and located in the energy gap of TiO2, extending its light absorption range from

ultraviolet to visible regions.6 _{In addition to the}

doped TiO2 photocatalysts, many other

semiconductors that possess suitable bandgap energies are found to show fascinating photocatalytic activities upon sunlight illumination.7 CdS is one of the most popular visible-light-driven photocatalysts since it has a bandgap energy of 2.5 eV that corresponds well with the visible light. Furthermore, its conduction band at relatively negative potential

(-1.0 versus NHE)8 _{offers CdS good}

photocatalytic activities. Till now, many structural forms of CdS including nanoparticles,9 _{hollow nanospheres,}10 _porous

nanocrystals,11 nanowires,12 nanotubes,13 and nanocomposites14 _{have been proven effective in}

relevant photocatalytic processes.

We previously showed that core-shell Au-CdS nanocrystals exhibited pronounced charge carrier separation upon light illumination.15 For Au-CdS nanocrystals, Au core can serve as an effective electron scavenger for CdS shell due to its lower Fermi energetic level (+0.5V versus NHE) than the conduction band of CdS (-1.0V versus NHE).8

Consequently, the photoexcited electrons in CdS shell would preferentially transfer to Au core, leaving positively-charged holes in CdS domain to achieve charge carrier separation. In this work, we explored the interfacial charge carrier dynamics for Au-CdS nanocrystals with various shell thicknesses. By probing the

emission lifetime of CdS, the electron transfer event between CdS shell and Au core for Au-CdS nanocrystals was quantitatively analyzed. A higher electron-transfer rate constant was observed for Au-CdS nanocrystals with thicker CdS shell, probably due to the less pronounced electron-hole interaction of thicker CdS that enabled a fuller extent of participation of photoexcited electrons in the charge separation process. On the other hand, the hole-enriched CdS shell of Au-CdS nanocrystals upon light illumination was characterized with a photocatalytic process. The photocatalytic activity of Au-CdS nanocrystals was found to increase with increasing shell thickness, attributable to the greater capability of light absorption achieved by the extensive growth of CdS shell. As compared to the relevant commercial products like N-doped P-25 TiO2 and CdS powders, the

as-synthesized Au-CdS nanocrystals exhibited superior photocatalytic performance under visible light illumination, demonstrating their potential as an efficient visible-light-driven photocatalyst. Furthermore, the photocatalytic performance under natural sunlight was also examined, and the result shows that the present Au-CdS nanocrystals can be used as highly efficient photocatalysts which may practically harvest energy from sunlight.

2. Experimental Section

Chemicals. All chemicals were analytic grade reagents and used without further purification. Special attention should be paid when dealing with the hazardous cadmium source and the highly poisonous potassium cyanide (KCN). Preparation of Au-CdS Nanocrystals. The detailed synthetic approach and relevant characterizations of Au-CdS nanocrystals used here can be found in our previous work.15 Briefly, Au colloids and L-cysteine-Cd2+ complexes (Cys/Cd) with a suitable molar ratio were mixed and allowed for hydrothermal reaction at 130oC for 6 h. The product (Au-CdS nanocrystals) was then centrifuged and washed with distilled water and ethanol to remove remaining ions. By increasing the volumes of Cys/Cd mixed with Au colloids, Au-CdS nanocrystals with increasing shell thickness can

be obtained. In this work, Au-CdS nanocrystals with three different shell thicknesses (9.0, 14.0 and 18.6 nm) were prepared and compared. The shell thickness of Au-CdS nanocrystals was determined by examining dozens of nanocrystals from the low-magnification TEM image. A number-averaged value was then calculated and represented.

Preparation of CdS Counterpart Nanocrystals. CdS counterpart nanocrystals were prepared by treating Au-CdS nanocrystals with 0.1 M KCN solution, resulting in the removal of Au core and the preservation of CdS shell (hollow structures).

Preparation of N-doped P-25 TiO2. N-doped

P-25 TiO2 was prepared by annealing Degussa

P-25 TiO2 powder (1130 mg) in the mixed

atmosphere of Ar (200 sccm) and NH3 (10

sccm) at 500oC for 2 h.16 The x value of the product (TiO2-xNx) was about 0.28 as estimated

from the XPS measurement.

Photocurrent Measurement. Photocurrent measurement for Au-CdS nanocrystals was conducted in a photoelectrochemical system under white light irradiation (xenon lamp, 250 W, with a light intensity of 100 mW/cm2). Spin-coated film of Au-CdS nanocrystals on fluorine-doped tin oxide (FTO) substrate was used as the photoanode in the three-electrode cell which consisted of Pt counter electrode, Ag/AgCl reference electrode, and 0.1 M Na2S

redox couple.

Photoluminescence Lifetime Measurement. Time-resolved photoluminescence (PL) spectra were measured using a home-built single photon counting system. GaN diode laser (408 nm) with the pulse duration of 50 ps was used as the excitation source. The signals collected at the excitonic emission of CdS (λ = 495 nm) were dispersed with a grating spectrometer, detected by a high-speed photomultiplier tube, and then correlated using a single photon counting card. The emission decay data were analyzed with the biexponential kinetics in which two decay components were derived. The lifetimes (τ1 and τ2), preexponential factors

(A1 and A2), and intensity-average lifetime (<τ>)

for Au-CdS nanocrystals and the corresponding CdS counterparts were determined and summarized in Table 1.

Photocatalytic Activity Measurement. The photocatalytic activity of Au-CdS nanocrystals was evaluated by the photodegradation of rhodamine B (RhB, C28H31N2O3Cl) under

visible light illumination. A quartz tube with a capacity of 30 mL was used as the photoreactor vessel. The optical system used for photocatalytic reaction consisted of a xenon lamp (500 W, with a light intensity of 175 mW/cm2) and a bandpass filter (with the bandwidth of 400-700 nm) that ensured the irradiation in visible range. All the photocatalysis experiments were conducted at room temperature in air. Four kinds of photocatalysts including N-doped P-25 TiO2,

commercial CdS powders (Aldrich-Sigma, with the particle size of 10-20 nm), CdS counterpart and Au-CdS nanocrystals were used and compared in the photodegradation of RhB. Typically, 5.4 mg of photocatalyst was added into 18 mL of RhB aqueous solution (1.0 × 10-5 M) in the photoreactor vessel. Prior to irradiation, the suspension was stirred in the dark for 30 min to reach the adsorption equilibrium of RhB with photocatalyst. At certain time intervals of irradiation, 1.5 mL of the reaction solution was withdrawn and centrifuged to remove photocatalyst particles. The filtrates were analyzed with a UV-Visible spectrophotometer to measure the concentration variation of RhB through recording the corresponding absorbance of the characteristic peak at 553 nm. Furthermore, photodegradation of RhB (1.0 × 10-5 M) under natural sunlight by using Au-CdS nanocrystals (5.4 mg) as photocatalyst was also examined. Characterizations. The morphology and dimensions of Au-CdS nanocrystals were examined with a high-resolution TEM (HRTEM, JEOL JEM-3000) operated at 300 kV. X-ray photoelectron spectroscopy (XPS) data were recorded with a VG Scientific Microlab 350 electron spectrometer using Mg Kα (hυ = 1253.6 eV) as X-ray source under a base pressure of 1.0 × 10-9 Torr. The spectrum resolution of XPS was 0.1 eV, and the pass energy for survey and fine scans was 40 eV. All the binding energies were calibrated by C 1s at 284.6 eV. Photocurrent signals were recorded with a Keithley 2400 semiconductor analyzer.

UV-Visible spectra were collected using a Hitachi 3900H at room temperature under ambient atmosphere.

3. Results and Discussion

First, core-shell nanocrystals of Au-CdS were prepared in a hydrothermal process by using the pre-synthesized Au colloids and Cys/Cd complexes as the starting materials. The formation of Au-CdS nanocrystals involved the binding of Cys/Cd complexes toward Au nanoparticles, followed by the decomposition of Cys/Cd in the hydrothermal reaction and the subsequent growth of CdS onto the surfaces of Au. By suitably modulating the experimental parameters such as the volumes of Cys/Cd added, a controllable shell thickness of Au-CdS nanocrystals can be achieved. In this work, Au-CdS nanocrystals with three different shell thicknesses (9.0, 14.0 and 18.6 nm) were prepared and compared. The present Au-CdS nanocrystals provide an ideal platform to study the interfacial charge carrier dynamics for metal-semiconductor core-shell heterostructures. Due to the difference in band structures between Au and CdS, a pronounced photoinduced charge separation took place at the interface of Au and CdS, resulting in the electron-charged Au core and the hole-enriched CdS shell. The electron-charging of Au core in Au-CdS nanocrystals can be revealed with the corresponding XPS analysis and photocurrent measurement. As shown by the XPS spectra of Figure 1(a), a binding energy of 84.1 eV of Au 4f7/2 peak was found for pure Au colloid sample,

which is in good agreement with the value of bulk metallic Au.17 However, a negative binding energy shift of around 0.5 eV of Au

4f7/2 peak was observed for Au-CdS

nanocrystals, indicating a significant difference in electronic structures between Au and CdS and a strong electronic interaction therein.18 Similar phenomenon was ever reported in Au-SnO2 core-shell nanocrystal system, in

which the binding energy shift of Au 4f was attributed to the effective electron transfer from SnO2 to Au.19 Here we ascribed the negative

binding energy shift of Au 4f observed in Au-CdS nanocrystals to the electron-charging of Au core that resulted from the occurrence of

charge separation. To further elucidate the effect of Au on the charge separation of CdS for the present core-shell nanocrystals, we compared the photocurrent response of Au-CdS nanocrystal and CdS counterpart electrodes by inserting them in a photoelectrochemical cell. Note that CdS counterpart was composed of CdS hollow structures, which were prepared by dissolving the Au core of Au-CdS nanocrystals.15 Figure 1(b) depicts the photocurrent generation for Au-CdS nanocrystal and CdS counterpart electrodes subjected to the white light irradiation. Both electrodes showed prompt response to the on/off cycles of light illumination, demonstrating the effective charge transfer and successful electron collection for the samples within the photoelectrochemical cell. More importantly, Au-CdS nanocrystals exhibited lower photocurrents than CdS counterpart upon light irradiation. We believed that the significant electron transfer from CdS shell to Au core accounted for such an evident photocurrent depression found in Au-CdS nanocrystals.

If the observed photocurrent depression as well as the XPS binding energy shift for Au-CdS nanocrystals indeed involved the electron transfer from CdS to Au, we should be able to reveal this event in the excitonic emission decay profile of CdS. Figure 2 represents the time-resolved PL spectra for two Au-CdS samples with different shell thicknesses. The emission decay data were analyzed with biexponential kinetics in which two decay components were derived. For Au-CdS nanocrystals with the shell thickness of 14.0 nm, emission lifetimes of both components were shorter than those of the corresponding CdS counterpart (τ1=0.40 ns,

τ2=2.95 ns for Au-CdS versus τ1=0.53 ns,

τ2=2.98 ns for CdS counterpart). The

intensity-average lifetime was then calculated to make an overall comparison of the emission decay behaviour.20 The difference in the average emission lifetime between Au-CdS (<τ> = 0.57 ns) and CdS counterpart (<τ> = 1.06 ns) indicates the emergence of a nonradiative pathway from the interaction between CdS and Au. This proposition can be

confirmed by the emission quenching of CdS observed for Au-CdS sample.15 Such difference became more noticeable as the shell thickness of Au-CdS nanocrystals further increased to 18.6 nm (<τ> = 0.48 ns for Au-CdS versus <τ> = 1.51 ns for CdS counterpart), inferring a much more significant electronic interaction between CdS and Au. If electron transfer from CdS to Au was the predominant process that dictated the emission quenching of CdS, we can then estimate the electron-transfer rate constant (ket) of Au-CdS nanocrystals from the

emission lifetime data by the following equation: ) CdS ( 1 ) CdS Au ( 1 > < − − > < = τ τ et k

Using the lifetime values listed in Table 1, we obtained the electron-transfer rate constants as 0.36×109, 0.51×109, and 1.42×109 s-1 for Au-CdS nanocrystals with the shell thickness of 9.0, 14.0 and 18.6 nm, respectively.

It should be noted that the electron-transfer rate constant of Au-CdS nanocrystals increased with increasing shell thickness. The less pronounced interaction between electrons and holes in the thicker CdS shell may contribute to such an increase in electron-transfer rate constant for Au-CdS nanocrystals with increasing shell thickness. We noticed that the emission lifetimes of the three CdS counterpart samples were substantially different, with CdS of larger characteristic size (thickness) showing longer emission lifetime, as can be seen in Table 1. This size-dependent correlation of exciton lifetime has been widely reported for CdS nanocrystals.21 It is generally believed that the significant interaction between electrons and holes, which is due to the confinement of electrons and holes in a particle of reduced size, may induce additional pathways for nonradiative recombination.22 A shortened exciton lifetime would consequently be observed for CdS nanocrystals with reduced size. The electron-hole interaction in small particles is related to the trapping of excitons by the abundant surface states that may further act as alternative sites for nonradiative charge recombination.23 For the present Au-CdS

nanocrystals, it is reasonable to presume a less pronounced electron-hole interaction for Au-CdS with thicker CdS shell since they possessed a larger characteristic size of CdS and thus a less amount of surface states. Such less pronounced electron-hole interaction in fewer surface states prohibited charge carriers from being consumed in nonradiative recombination, which further enabled a fuller extent of participation of photoexcited electrons in the charge separation process. Accordingly, an increase in the electron-transfer rate constant was observed for Au-CdS nanocrystals with increasing shell thickness.

Since Au-CdS nanocrystals exhibited pronounced charge separation upon light illumination, it is worth studying the potential application that this property may bring. Owing to the effective electron transfer from CdS shell to Au core, photogenerated holes with an abundant amount were existent in CdS shell and would transfer to the surfaces of Au-CdS nanocrystals. These highly reactive holes could oxidize water to produce hydroxyl radicals that can further decompose organic pollutants through an oxidation process. A spectacular capability of photocatalytic oxidation is therefore expected at the surfaces of Au-CdS nanocrystals. A series of photocatalysis experiments were performed in this work to investigate the photocatalytic properties of the as-synthesized Au-CdS nanocrystals. RhB, a typical dye that can be decomposed by hydroxyl radicals,24 was used as the test pollutant to monitor the photocatalytic oxidation progress for Au-CdS nanocrystals. The time-dependent UV-visible spectra of RhB solutions under visible light illumination in the presence of Au-CdS nanocrystals with a shell thickness of 14.0 nm were first shown in Figure 3(a). It can be seen that the intensity of the characteristic absorption peak at 553 nm decreased dramatically with the irradiation time. Besides, a concomitant blue shift in the absorption maximum was observed after the solution was irradiated for 20 min. It is well known that the photodegradation of RhB undergoes two competitive processes.25 One is the destruction of dye chromogen, which is characteristic of the loss of absorbance at 553

nm. The other is the N-demethylation reaction that produces a series of N-demethylated intermediates, accompanied by a blue shift in the absorption maximum from 553 to 498 nm. In the current case, RhB concentration was determined by referring to the absorbance of the characteristic peak at 553 nm. To quantitatively understand the reaction kinetics of RhB photodegradation for our samples, we analyzed the normalized concentration of RhB (C/Co) as a function of irradiation time. As

shown in the inset of Figure 3(a), an exponential decay of RhB concentration with the irradiation time was evident for Au-CdS

nanocrystal photocatalyst. The photodegradation process was then fit to

pseudo-first-order reaction, in which the value of the apparent rate constant (kRhB) is equal to

the slope of the fitting line according to the following expression:26

ln(C/Co) = - kRhB t, where Co and C are the

concentrations of RhB at initial and at a certain irradiation time t, respectively.

For Au-CdS nanocrystals with a shell thickness of 14.0 nm, kRhB is found to be 0.026 min-1. The

mechanism for RhB photodegradation by using Au-CdS nanocrystal photocatalyst can be described by the following four pathways:

Au–CdS + hν Æ Au(e–_)–CdS(h+_{) (1)}

Au(e–_)–CdS(h+_{) + H}

2O Æ Au(e–)–CdS + H+ + ·OH (2)

RhB + ·OH Æ oxidation products (3) Au(e–_{)–CdS + O}

2 Æ Au-CdS + ·O2– (4)

Under visible light illumination, charge separation occurred within Au-CdS nanocrystals, resulting in an electron-charged Au core and a hole-enriched CdS shell (1). Subsequently, the photogenerated holes transferred to the surfaces of nanocrystals and reacted with water to produce hydroxyl radicals (2). RhB molecules were then decomposed by hydroxyl radicals through an oxidation process (3). Once the photogenerated holes were depleted in photocatalysis, Au-CdS nanocrystals underwent a Fermi level