聚(3-己基噻吩)薄膜電晶體的摻雜效應與特性強化之研究

Abstract

本論文中，我們採用規則性聚(3-己基噻吩) (P3HT)作為薄膜電晶體之主動層的主要材料，探討不同氣體分子對聚(3-己基噻吩)主動層引起的摻雜效應對電特性與可靠度的影響，此外，為提升電晶體特性，我們針對改善載子遷移率與次臨界特性分別提出了不同的方法，成功的將載子遷移率大幅提升，也深入探討了導致次臨界特性變差的體電流效應(bulk current effect)之機制，並明顯改善次臨界特性包括次臨界漏電流和次臨界擺幅(subthreshold swing)與開關電流比(on/off current ratio)。 藉由對聚(3-己基噻吩)主動層進行各種不同的後處理，或在量測過程加入氮氣流，並觀察引起的電性變化，發現水分子與氧分子對聚(3-己基噻吩)主動層有兩種摻雜效用，我們分別稱為穩態摻雜(stable state dopant)與暫態摻雜(transient state dopant)；穩態摻雜增加將於主動層中引致更多電洞，導致主動層導電性升高。然而，次臨界特性將變差，遲滯(hysteresis)電壓寬度變窄，而暫態摻雜增加將使遲滯現象更明顯，且可藉由將元件至於鈍氣環境中而被輕易移除；不論後處理時的溫度高低，穩態摻雜可藉由大量提供水或氧氣分子而增加。然而，也可藉著在鈍氣中退火而減少，由氮氣退火後遲滯增加得知，暫態摻雜將反向增加；意即穩態摻雜與暫態摻雜是有關聯並可互換的。這個觀點可於定電壓應力測試(constant voltage stress)的結果中再次確認。在不同極性的閘極電壓的定電壓應力測試後，可觀察到相反趨勢的特性；意即負閘極電壓應力測試後，有較大的遲滯與稍低的開啟電流(on-current)，而正閘極電壓應力測試後，則回復為近似定電壓應力測試前的特性。另外，若於定電壓應力測試期間兼具通入氮氣流，仍可觀察到近似的趨勢。不同點在於，負閘極電壓應力測試後的遲滯相對較小於無通入氮氣的測試的結果。而正閘極電壓應力測試後，除了遲滯相對較小外，關閉電流(off-current)與次臨界擺幅亦較小；這些結果反映出，穩態摻雜與暫態摻雜是有關聯並可互換的，另外，暫態的摻雜可輕易的移除。總而言之，我們有系統的探討了，廣泛存在於環境中的水分子與氧分子對聚(3-己基噻吩)主動層的摻雜效應與其機制。 此外，由於較高的載子遷移率意味著能提供較大的驅動電流，因而有利於許多的應用。不同於水與氧分子的摻雜，我們也採用奈米碳管(carbon nanotubes)對聚(3-己基噻吩)主動層進行摻雜。除主動層摻雜外，我們同時也將源汲極電極材料改為使用多壁奈米碳管(multi-walled carbon nanotubes)。目的同是為了改善載子遷移率；將聚(3-己基噻吩)作為主動層之電晶體的電極改用多壁奈米碳管後，可有效降低元件的接觸阻抗(contact resistance)。這個現象和我們稍早前以並五苯(pentacene)作為主動層的電晶體搭配多壁奈米碳管電極之後的結果相符；另外，相較於一般未摻雜的電晶體特性，藉由將官能基化的單壁奈米碳管(functionalized single-walled carbon nanotubes)摻雜入主動層中，載子遷移率可被提升約三倍。另外，藉由我們提出的一個新技巧，可將多壁奈米碳管有效分離並摻雜入聚(3-己基噻吩)中，同樣的，載子遷移率可被提升約三倍。因為載子在奈米碳管摻雜的主動層中，可經由遷移率較高的碳管移動，進而提高整體的載子遷移率。值得一提的是，較低重量比的多壁奈米碳管摻雜就可達成提高載子遷移率的效果。另外，在採用相同重量比的狀況下，主動層以較長的多壁奈米碳管摻雜之元件展現出較佳的載子遷移率；這可歸因於載子在較長的碳管的傳遞將遭遇到較少的載子散射(scattering)。摻雜奈米碳管的主動層有較低的接觸阻抗，這可歸因於主動層中的金屬性之奈米碳管直接接觸金屬電極所致；當將官能基化的單壁奈米碳管與多壁奈米碳管摻雜之主動層元件同時輔以多壁奈米碳管之電極，載子遷移率可進一步分別被提升至0.149與0.088(cm2/Vs)。同樣的，在採用相同摻雜重量比的狀況下，主動層以較長的多壁奈米碳管摻雜之元件展現出較佳的載子遷移率。 然而，無論從水分子或氧分子對主動層的摻雜效應，或在主動層中利用奈米碳管摻雜的元件特性，皆可觀察到次臨界特性變差，這現象明顯不利於薄膜電晶體的主要應用。為此，我們探討了近期文獻中提到的體電流效應，此效應被認為和較差的次臨界特性有關。藉由將鈦金屬薄膜覆蓋在聚(3-己基噻吩)電晶體的電極上，並改變主動層厚度，我們發現，體電流可由電極上表面(top-face)注入與側壁(side-wall)注入；當將主動層薄膜變薄後，因其片電阻提高，可使體電流下降。因此，可獲得較佳的次臨界擺幅與較低的次臨界漏電流，這部份結果與文獻提出的結果相符。而當固定主動層厚度的情況下，電晶體的電極改為鈦金屬薄膜覆蓋的結構後，可觀察到次臨界特性被進一步的改善。這可歸因於鈦金屬薄膜與聚(3-己基噻吩)間極高的載子注入能障抑制了上表面注入的體電流，卻也因此導致開啟電流有所降低；這現象與接觸阻抗增加相符。當採用較薄的主動層與鈦金屬薄膜覆蓋之電極，且閘極偏壓在零伏特時，全由體電流組成的汲極(漏)電流下降可達二十倍，而次臨界擺幅則下降兩倍。雖開啟電流稍微降低，但是開關電流比仍增加達五倍；接著，我們也將這新發現的觀念，套用到多壁奈米碳管電極上，藉由先將鈦金屬薄膜覆蓋在奈米碳管的催化金屬層上，再進行多壁奈米碳管的成長，成功製作出側向成長多壁奈米碳管電極。比較採用此電極與傳統(垂直成長)多壁奈米碳管電極的電晶體特性，仍可觀察到明顯的次臨界特性改善，也再次驗證了我們提出的觀點。 最後，根據鈍氣中退火可移除主動層中的分子摻雜的現象，我們發展一種新型主動層結構，製作方法是在已於鈍氣中退火之第一層主動層上塗覆第二層聚(3-己基噻吩)主動層，稱為雙層主動層(double-coated channel layers)。相較於單層主動層結構的電晶體，雙層主動層之電晶體展現出改善超過一百倍的開關電流比與顯著降低的次臨界擺幅。為瞭解這種特別的改善的機制，我們分別改變第一層與第二層的厚度。當固定第二層厚度時，若已鈍氣退火的第一層厚度增加，特性改善程度將降低；而當調變第二層厚度但固定第一層厚度與第二層厚度相同時，若厚度增加，開關電流比改善幅度降低。由交叉比對這兩種調變的結果，發現開關電流比的改善程度與第二層之厚度亦有關連，且開啟電流大小和第二層呈正相關。另外，我們將不同厚度之主動層的電晶體進行鈍氣退火，並萃取各種元件特性與接觸阻抗，發現退火後主動層中的摻雜濃度降低，因而接觸阻抗升高兩倍。又藉由觀察將第四章所提出的使用鈦金屬覆蓋之電極的電晶體進行鈍氣退火後的結果，應證出鈍氣退火後的高接觸阻抗可歸因於較高的載子注入能障；這些現象有助於抑制體電流效益。另外，利用簡化的等效電路與能帶結構圖，我們進一步說明了使用雙層主動層之電晶體特性改善，尤其是開關電流比與次臨界擺幅的改善，並前述一些現象的原因。藉由鈍氣退火的第一層主動層與電極間的高載子注入能障，可使體電流效應降低；意即較佳的次臨界特性與較低的關閉電流。當元件操作在次臨界區時，第一層主動層的片電阻隨著閘極電壓升高而降低，因而有利於載子從源極經第一層進入第二層中傳遞，在經由第一層進入汲極形成電流。當第一層厚度較薄與第二層厚度較厚時，皆有益於此電流形成，故可觀察到正偏移的臨界電壓。此電流估計可持續增加直到第一層中的累積通道層形成為止，又汲極電流先經由第二層或同時與累積通道層一起導通，使次臨界擺幅有所改善，如此也解釋了電流特性與第一層與第二層後度相關；這樣的論點可在後續的實驗結果中再次被肯定，根據第三章中的研究，將第二層主動層置換成奈米碳管摻雜的主動層後，可觀察到電流因此提升，驗證了雙層主動層電晶體特性與第二層有關。而相較於採用奈米碳管摻雜的單層主動層電晶體，此雙層主動層電晶體的開關電流比與次臨界特性皆有所改善，再次應證了我們提出的運作機制。 總結，我們深入探討了以規則性聚(3-己基噻吩)作為主動層的主要材料之薄膜電晶體的特性、可靠性與體電流效應，也提出不同的特性改善的方法，包含富含於空氣中的水分子與氧分子導致的摻雜效應、摻雜效應造成的特性變動、大幅提高載子遷移的方法、體電流效應的機制與改善方法和一種改善特性的主動層結構。

In this dissertation, regioregular poly (3-hexylthiophene) (P3HT) is employed as the channel layer of thin film transistors (TFTs). Influences of water and oxygen molecules induced doping effect in the P3HT channel layer on the device characteristics and reliability are investigated. Moreover, we propose different approaches to improve device performances, including mobility and subthreshold performance. Significant enhancement in the mobility has been accomplished successfully. Also, the detail of the bulk current effect, resulting in poor subthreshold performance, is deeply discussed, and, accordingly, dramatic improvements in the subthreshold performance, including subthreshold leakage current and subthreshold swing, and on/off current ratio are obtained. By either implementing different post treatments on the P3HT active layer or incorporating N2 blow during electrical characterization, the performance differences indicate that the doping effect caused by water and oxygen molecules can be discriminated between two kinds of dopant, so-called stable state dopant and transient state dopant. Stable state dopants could induce holes in the channel layer contributing to the better conductivity. However, poorer subthreshold performance and smaller hysteresis width can be observed. On the other hand, transient state dopants can trigger the formation of polarization agents producing the hysteresis, and can be removed easily; for instance, by placing devices in an inert environment. Stable state dopants will increase as providing sources of water and/or oxygen molecules regardless the temperature of these treatments, and can be reduced by thermal annealing in the inert atmosphere. In contrast, after annealing in N2, there is the increased hysteresis indicating the raised amount of transient state dopants. In other words, there shall be relationship between stable state dopants and transient state dopants that they are interchangeable. This can be confirmed by the results of constant voltage stress. Characteristics in the results of constant positive and negative gate voltage stresses exhibit obviously opposite tendencies. After stressing by negative gate voltage, there are enlarged hysteresis width and little lowered on-current, and they will be recovered after the following positive gate voltage stress. Otherwise, if the stress is carried out with N2 blow, similar tendency can be observed. Nevertheless, the hysteresis width is less enlarged for the result of negative gate voltage stress, and, moreover, both off-current and subthreshold swing are lowered after the positive gate voltage stress besides smaller hysteresis. All these phenomena reply the relationship between stable state dopants and transient state dopants and their interchangeability. Also, the transient state dopants can be obviated easily. To sum up, we systematically investigated the doping effect, induced by water and oxygen molecules generally existing in air atmosphere, in P3HT channel layer and its mechanisms. Next, higher mobility will contribute to higher driving current that benefit different applications. To attain the goal of improving mobility, distinct to doping effect caused by water and oxygen molecules, carbon nanotubes (CNTs) are blended into P3HT active layer as the dopants, and, besides to dope the channel layer, multi-walled carbon nanotubes (MWCNTs) are employed to replace conventional source and drain electrodes (S/D). Contact resistance is effectively reduced by using MWCNTs as the electrodes of P3HT based TFTs (PTFTs). Similar effect was observed in our recent works, in which pentacene based TFTs are employing MWCNT S/D. In addition, compared with the PTFTs, the mobility can be raised for three times by doping functionalized single-walled carbon nanotubes into the active layer. Moreover, by a new approach that profit to separate MWCNTs effectively and to uniformly blend into P3HT, the mobility has threefold increase too. These can be referred to that carriers are rapidly transferred through the CNTs, of which the conductivity is very prominent, inside the channel layer so that increased mobility can be realized. Noteworthily, similar enhancement effect on mobility can be accomplished by using the MWCNTs/P3HT composite channel, of which the weight percentage is lower than composite of f-SWCNTs/P3HT. Otherwise, if the weight percentage is controlled to be the same for the MWCNTs/P3HT and f-SWCNTs/P3HT composite channels, there is higher mobility due to the reduced scattering in the longer MWCNTs. Moreover, after incorporating CNTs into channel layer, contact resistance is reduced because of direct contact between metallic CNTs and electrodes. Furthermore, as f-SWCNTs/P3HT and MWCNTs/P3HT composite channel based TFTs are assisted by MWCNT S/D, the mobility of 0.149 and 0.088 cm2/Vs, respectively, has been attained. Similarly, the channel layer embedding longer MWCNTs shows better mobility as the situation employing the same weight percentage. However, inferior subthreshold performance can be observed in the results of not only the doping effect induced by water and oxygen molecules but also the CNTs embedded P3HT channel layer. To alleviate this issue, we investigate the bulk current effect, which has been proposed in the recent literatures and correlated to poor subthreshold performance. By employing Ti capped electrodes for PTFTs and varying channel layer thicknesses, the bulk current can be injected from the top-face and side-wall of electrodes. As the active layer is thinned, the bulk current will be lowered due to increase of channel sheet resistance so that better subthreshold swing and lower subthreshold leakage current can be achieved; this result is consistent with the results in the previous literatures. On the other hand, as the channel layer is fixed, after the electrodes of PTFTs are capped by a Ti layer, the subthreshold performance can be further improved. This can be referred to the restriction of top-face injected bulk current by the high carrier injection barrier between Ti layer and P3HT. Nevertheless, this also results in a decrease of on-current, a result compatible to increased contact resistance. As the channel layer and electrodes of PTFTs are thin and Ti capped, twentyfold decrease of drain current fully composed of the bulk current as the gate voltage is 0 V and twofold improvement in subthreshold swing are observed. Although on-current is little decreased, on/off current ratio is enlarged for five times. Furthermore, we apply the new concept to MWCNT electrodes. By capping a Ti layer upon MWCNT catalytic layer and executing following MWCNT growth, lateral grown MWCNT electrodes are fabricated successfully. Comparing to PTFTs with conventionally vertical grown MWCNT S/D, PTFTs with lateral grown MWCNT electrodes show obvious improvement in subthreshold performance. This second confirm our inference. At last, according to the de-doping effect induced by N2 annealing, we develop a new channel structure, formed by coating a second P3HT layer on top of the N2-annealed first P3HT layer and so-called double-coated channel layers. As compared to the conventional, single channel layer, PTFTs, devices with double-coated channel layers (DPTFTs) show more than two orders of magnitude of improvement in the on/off current ratio and a significantly lowered subthreshold swing. To explain such extraordinary improvements, different thicknesses of the first and second active layers were used to study the mechanism. If the thickness of second layer is fixed, the performance improvement decrease with increased thickness of N2 annealed P3HT first layer. In contrast, if the thicknesses of the second layer of DPTFTs were controlled to be the same as that of the first layer, there is less improvement in on/off current ratio. As compared to the foresaid result, the effect of improvement in on/off current ratio also correlates with the thickness of P3HT second layer, and positive proportion of the on-current to the thickness of second layer is observed. In addition, N2 annealing is carried out on the PTFTs, of which the channel layers are using different thicknesses, and characteristics and contact resistance are extracted. Accordingly, after N2 annealing, the dopant concentration in P3HT active layer is reduced so that the contact resistance has a twofold increase. By the results of N2 annealed PTFTs with Ti capped electrodes, proposed in Chapter 4, the reason of increased contact resistance after N2 annealing is identified with high carrier injection barrier. These are contributive to suppress the bulk current effect. Furthermore, by using a simple equivalent circuit and a speculated band structure, we interpret the improvements in performances of double-coated channel based TFTs, especially improvements in on/off current ratio and subthreshold swing, and the cause of abovementioned phenomena. As the P3HT first layer is N2 annealed, a high carrier injection barrier will result in the restraint of bulk current effect so that superior subthreshold performance and lower off-current are obtainable. As devices are operated in the subthreshold region, the channel sheet resistance of the first P3HT layer will lower with increasing gate bias so that a current can be formed by carriers transferring from source electrode through the first layer into second layer and then again through first layer into drain electrode. This current is facilitated if the first and second P3HT layers are thin and thick, respectively. Consequently, a positive shift threshold voltage is observed. Supposedly, this current can increase until the formation of accumulated channel layer. Moreover, besides the accumulated channel layer, the drain current can be conducted by this current so that the subthreshold swing can be improved. These consist with the correlation between the drain current and thicknesses of first and second P3HT layer. The inference is secondly confirmed by the results of following experiments. According to the investigation in Chapter 3, as replacing second layer by CNTs/P3HT composite channel, there is enlarged on-current that substantiate the relation between characteristics of double-coated channel based TFTs and the second layer. Compared to the conventional TFTs with CNTs/P3HT composite channel, the devices with double-coated channel layers reveal improvements in on/off current ratio and subthreshold performance that is consistent to the mechanism we proposed. In summary, we deeply discussed the characteristics, reliability, and bulk current effect of TFTs, of which the channel layer is the regioregular P3HT, and have proposed many approaches for the performance improvement, including the doping effect and characteristics fluctuation induced by water and oxygen molecules generally existing in air atmosphere, novel techniques to significantly enhance the mobility, study of the bulk current effect and proposed method to suppress it, and a new channel structure to improve performances.