實(shí)現(xiàn)商業(yè)上可行的太陽(yáng)能電力對(duì)于確保長(zhǎng)期經(jīng)濟(jì)增長(zhǎng)和減輕氣候變化的影響至關(guān)重要。金屬鹵化物鈣鈦礦(MHP),尤其MAPbI3 (MA=CH3NH3),是目前研究最多的太陽(yáng)能電池材料,其功率轉(zhuǎn)換效率(PCE)約為25.2%,超過(guò)了目前商業(yè)化的太陽(yáng)能電池,如多晶硅(c-Si,21.3%)、碲化鎘(CdTe,22.1%)和銅銦鎵硒(CIGS,22.3%)。但是,與傳統(tǒng)的太陽(yáng)能電池材料相比,MHP的主要優(yōu)點(diǎn)是它們易于大規(guī)模合成且成本相對(duì)較低。此外,其在可見光區(qū)域的吸收系數(shù)》3.0×104cm?1 、激子結(jié)合能低所導(dǎo)致的自由電子和空穴量子產(chǎn)率高、長(zhǎng)的電子-空穴擴(kuò)散長(zhǎng)度、電子良性點(diǎn)和晶界缺陷。MHP是串聯(lián)太陽(yáng)能電池,能將寬帶隙的“頂部電池”與窄帶隙材料(如硅)耦合為“底部電池”。鑒于晶體硅具有1.1 eV的帶隙,需要具有1.75 eV帶隙的材料才能使兩個(gè)結(jié)的電流匹配。當(dāng)前的研究重點(diǎn)是探尋成本低廉、穩(wěn)定且無(wú)鉛的MHPs單個(gè)吸收器或串聯(lián)太陽(yáng)能電池最佳材料,但材料設(shè)計(jì)的化學(xué)空間仍然過(guò)于寬泛,需更有效的搜索方法來(lái)尋找不同帶隙范圍的鈣鈦礦結(jié)構(gòu)。
美國(guó)匹茲堡大學(xué)機(jī)械工程與材料科學(xué)系的Wissam A. Saidi教授等,系統(tǒng)地開發(fā)了一個(gè)包含862個(gè)MHP結(jié)構(gòu)和帶隙特性的數(shù)據(jù)集,目的是開發(fā)一個(gè)預(yù)測(cè)性ML模型,以捕獲該化學(xué)空間的復(fù)雜趨勢(shì)和相關(guān)性。他們主要通過(guò)關(guān)注有機(jī)分子的變化來(lái)構(gòu)建材料設(shè)計(jì)空間,與無(wú)機(jī)體系相比,有機(jī)分子的結(jié)構(gòu)具有極大數(shù)量的可能性。作者為MHP開發(fā)了一種ML方法,該方法將:
使用易于計(jì)算的描述符準(zhǔn)確估算帶隙;
克服目標(biāo)值分布不平衡的小型數(shù)據(jù)集問題,即目標(biāo)值的某些范圍部分可能樣本太多,而其他部分可能樣本太少或沒有;
使用簡(jiǎn)單的ML方法,這些方法在計(jì)算上并不苛求,并且相對(duì)容易理解和控制。
以此ML方法研究表明,由相對(duì)較小的神經(jīng)網(wǎng)絡(luò)元素組成的分層神經(jīng)網(wǎng)絡(luò)體系結(jié)構(gòu),可以解決所有這些約束。在這種情況下,ML元素的排列方式應(yīng)使每個(gè)部分在預(yù)測(cè)過(guò)程中發(fā)揮特定的作用。每個(gè)元素都是使用卷積神經(jīng)網(wǎng)絡(luò)(CNN)框架構(gòu)建的,并且除了其他元素之外,還針對(duì)其預(yù)設(shè)作用進(jìn)行了獨(dú)立訓(xùn)練。這樣簡(jiǎn)化了神經(jīng)網(wǎng)絡(luò)的學(xué)習(xí)過(guò)程,也避免了對(duì)具有諸多隱藏層的更復(fù)雜的網(wǎng)絡(luò)體系結(jié)構(gòu)的需求。
Machine-learning structural and electronic properties of metal halide perovskites using a hierarchical convolutional neural network
Wissam A. Saidi, Waseem Shadid & Ivano E. Castelli
The development of statistical tools based on machine learning (ML) and deep networks is actively sought for materials design problems. While structure-property relationships can be accurately determined using quantum mechanical methods, these first-principles calculations are computationally demanding, limiting their use in screening a large set of candidate structures. Herein, we use convolutional neural networks to develop a predictive model for the electronic properties of metal halide perovskites (MHPs) that have a billions-range materials design space. We show that a well-designed hierarchical ML approach has a higher fidelity in predicting properties of the MHPs compared to straight-forward methods. In this architecture, each neural network element has a designated role in the estimation process from predicting complex features of the perovskites such as lattice constant and octahedral till angle to narrowing down possible ranges for the values of interest. Using the hierarchical ML scheme, the obtained root-mean-square errors for the lattice constants, octahedral angle and bandgap for the MHPs are 0.01??, 5°, and 0.02?eV, respectively. Our study underscores the importance of a careful network design and a hierarchical approach to alleviate issues associated with imbalanced dataset distributions, which is invariably common in materials datasets.
責(zé)任編輯:pj
-
太陽(yáng)能電池
+關(guān)注
關(guān)注
22文章
1186瀏覽量
69466 -
多晶硅
+關(guān)注
關(guān)注
3文章
241瀏覽量
29331 -
神經(jīng)網(wǎng)絡(luò)
+關(guān)注
關(guān)注
42文章
4777瀏覽量
100959
發(fā)布評(píng)論請(qǐng)先 登錄
相關(guān)推薦
評(píng)論