張穎驍,張梓楊,宋龍飛,2,李曉剛
Ti80合金及其熱模擬組織在含氟模擬海水中的力學(xué)電化學(xué)行為研究
張穎驍1a,張梓楊1a,宋龍飛1a,2,李曉剛1
(1.北京科技大學(xué) a.新材料技術(shù)研究院 b.“腐蝕與防護(hù)”教育部國防科技重點(diǎn)實(shí)驗(yàn)室,北京 100083;2.廣州大學(xué) 化學(xué)化工學(xué)院,廣州 510006)
研究應(yīng)變、環(huán)境、組織對Ti80合金在含氟海水中電化學(xué)行為的影響,為海洋工程裝備的安全服役提供數(shù)據(jù)支持。使用熱處理的方式模擬Ti80合金焊接熱影響區(qū)組織,并通過拉伸機(jī)加載至不同應(yīng)變狀態(tài),進(jìn)行開路電位和極化曲線的測試,最后通過機(jī)器學(xué)習(xí),挖掘應(yīng)變、環(huán)境、組織與電化學(xué)行為的關(guān)系。在添加和不添加0.001 mol/L F–的模擬海水(pH=2)中,開路電位隨應(yīng)變的增加而負(fù)移。在添加0.01 mol/L F–的模擬海水中,應(yīng)變對開路電位沒有明顯影響,應(yīng)變增加整體上提高維鈍電流密度。受應(yīng)變影響最大的是添加0.01 mol/L F–模擬海水中的1 500 ℃熱模擬組織,其最大應(yīng)變狀態(tài)下維鈍電流密度是無應(yīng)變狀態(tài)下的3倍左右。陰極塔菲爾斜率最大值出現(xiàn)在屈服點(diǎn)附近。F–濃度增加顯著提高維鈍電流密度。決策樹和梯度提升樹算法預(yù)測極化曲線電流值較為準(zhǔn)確,隨機(jī)森林算法的準(zhǔn)確度較差。塑性變形顯著提高Ti80在模擬海水中的電化學(xué)活性,而彈性變形的影響并不明顯。F–濃度增加顯著提高電化學(xué)活性。決策樹和梯度提升樹算法預(yù)測準(zhǔn)確度高于隨機(jī)森林算法。在相對重要性對比中,F(xiàn)–濃度對電化學(xué)行為的影響最大,應(yīng)變狀態(tài)次之,組織的影響最小。
Ti80合金;力學(xué)電化學(xué);機(jī)器學(xué)習(xí)
近年來,隨著海洋資源的開發(fā),對海洋工程裝備的性能和安全提出了更高的要求[1]。鈦及其合金質(zhì)輕、高強(qiáng)、耐蝕,是海工裝備的理想材料[2-4],其中Ti80合金由于更高的比強(qiáng)度和良好的焊接性能在工程結(jié)構(gòu)材料中得到了廣泛的應(yīng)用[5-6]。鈦合金表面致密的氧化膜使其具有優(yōu)異的耐蝕性[7],然而環(huán)境中的氟離子濃度、應(yīng)力應(yīng)變狀態(tài)、焊接熱輸入導(dǎo)致的組織劣化都可能使這層氧化膜失效,進(jìn)而導(dǎo)致鈦合金面臨嚴(yán)重的腐蝕風(fēng)險[8-13]。
應(yīng)力應(yīng)變可改變晶格中原子間距,改變鈍化膜半導(dǎo)體特性,也可形成位錯、層錯等缺陷,為陰陽極反應(yīng)提供活性位點(diǎn),從而影響材料的電化學(xué)行為[18-20]。Cui等[10]研究了塑性變形對X70管線鋼在近中性pH環(huán)境中電化學(xué)的影響,塑性變形增加了電極表面粗糙度,使得電化學(xué)活性增加,尤其是陰極反應(yīng)受到明顯促進(jìn)作用。Jandaghi等[21]的研究表明,Al-Mn-Si合金的強(qiáng)烈變形導(dǎo)致晶粒細(xì)化,加速了其在NaCl溶液中的腐蝕。Krawiec等[11]的研究表明,陰極反應(yīng)優(yōu)先發(fā)生在表面缺陷處,塑性變形產(chǎn)生的滑移帶導(dǎo)致陰極電流增加。Li等[22]的研究結(jié)果表明,塑性變形可增加TC2在模擬海水中的電化學(xué)活性,降低其耐蝕性。
金屬材料在使用中經(jīng)常需要焊接,而焊接熱影響區(qū)的組織、力學(xué)性能、電化學(xué)特性與母材有很大差異[23-24]。Or?owska等[25]的研究結(jié)果表明,鋁合金攪拌摩擦焊樣品中,熱影響區(qū)組織比母材更耐蝕。Ma等[26]研究了E690鋼焊接接頭在含SO2海洋大氣環(huán)境中的電化學(xué)行為,結(jié)果表明,臨界熱影響區(qū)的腐蝕電流遠(yuǎn)高于其他區(qū)域。
氟離子濃度、應(yīng)力應(yīng)變狀態(tài)、焊接過程中的熱輸入都是鈦合金電化學(xué)行為的重要影響因素。然而Ti80合金作為一種新型鈦合金,近幾年才得到關(guān)注,研究內(nèi)容多集中在組織和力學(xué)性能調(diào)控,關(guān)注其電化學(xué)行為的研究較少。隨著Ti80在海工裝備中的廣泛應(yīng)用,其在苛刻服役環(huán)境下面臨的腐蝕風(fēng)險理應(yīng)受到重視。
電化學(xué)技術(shù)已被廣泛應(yīng)用于材料的腐蝕行為研究,但電化學(xué)試驗(yàn)費(fèi)時費(fèi)力,同時由于試驗(yàn)條件的局限性,一旦研究條件有所變化,就需要重新進(jìn)行試 驗(yàn)[27-30]。通過機(jī)器學(xué)習(xí),建立電化學(xué)回歸模型,預(yù)測多種條件下的電化學(xué)數(shù)據(jù),可降低研究成本,提高材料開發(fā)、設(shè)計(jì)效率[31-32]。在這方面已有許多研究。Gong等[33]在Python的scikit-learn模塊使用多種算法構(gòu)建了極化曲線和阻抗譜,結(jié)果表明,隨機(jī)森林的預(yù)測效果最好,輸入權(quán)重分析結(jié)果和傳統(tǒng)電化學(xué)結(jié)果一致。Pei等[34]比較了隨機(jī)森林、人工神經(jīng)網(wǎng)絡(luò)和支持向量回歸模型對于預(yù)測瞬時大氣腐蝕的準(zhǔn)確性,結(jié)果表明,隨機(jī)森林模型的精度更高。Yang等[35]通過腐蝕大數(shù)據(jù)技術(shù)闡明了Cr元素對耐候鋼耐蝕性能的動態(tài)影響,該過程同時受環(huán)境因素和銹層反應(yīng)的影響。這些研究充分證明了使用大數(shù)據(jù)技術(shù)分析腐蝕和電化學(xué)數(shù)據(jù)的先進(jìn)性和必要性,然而通過機(jī)器學(xué)習(xí)研究Ti80電化學(xué)行為的結(jié)果尚未見報道。
本文通過熱處理的方式制備了Ti80合金的熱模擬組織,在不同F(xiàn)–濃度的模擬海水中,對不同應(yīng)變狀態(tài)下的Ti80合金及其熱模擬組織進(jìn)行電化學(xué)測試,并通過機(jī)器學(xué)習(xí)方法,挖掘應(yīng)變、環(huán)境、組織與電化學(xué)行為的關(guān)系,為保障海洋工程裝備安全服役提供數(shù)據(jù)支持。
所用材料為Ti80合金,其化學(xué)成分見表1。使用熱處理的方式模擬焊接熱影響區(qū)組織,根據(jù)Su等[36]的研究結(jié)果,Ti80合金在831 ℃開始β轉(zhuǎn)變,在1 011 ℃轉(zhuǎn)變結(jié)束。同時,根據(jù)Ti-Al相圖[37],Ti80的熔點(diǎn)在1 700 ℃附近,將熱處理溫度選在900、1 500 ℃。將Ti80合金分別置于900 ℃和1 500 ℃的爐內(nèi)保溫 5 min,隨后取出空冷至室溫,所得組織分別稱為900 ℃熱模擬組織和1 500 ℃熱模擬組織。900、1 500 ℃的選擇是從受熱溫度區(qū)間出發(fā)考慮的,并非有針對性地模擬某一特定區(qū)域的組織。熱影響區(qū)由原始的母材組織受到短時高溫?zé)釠_擊后形成,而根據(jù)焊接接頭形態(tài),熱量由焊縫一側(cè)單向輸入,因此在熱影響區(qū)中距離焊縫中心距離越遠(yuǎn)的位置,受到的熱沖擊溫度越低,整個熱影響區(qū)受到的熱沖擊溫度區(qū)間將覆蓋831~1 700 ℃。根據(jù)這個規(guī)律,選取了接近該溫度區(qū)間兩端的數(shù)值作為熱處理溫度,用以模擬對應(yīng)位置處的組織。
表1 Ti80合金的化學(xué)成分
Tab.1 Element compositions of Ti80 alloy wt.%
所用溶液為ASTM D1141-98(2013)模擬海水。由于應(yīng)力腐蝕裂紋尖端[38]、裝配產(chǎn)生的縫隙內(nèi)部[39]、海生物和微生物膜的附著[40]導(dǎo)致局部環(huán)境與整體的差異,其中以陽極溶解產(chǎn)生金屬陽離子水解而導(dǎo)致環(huán)境酸化為主。因此,本文中的模擬海水使用鹽酸將pH值調(diào)至2,加上近海工業(yè)污染的影響[9,14-15],另外向其中加入不同質(zhì)量的NaF,使其F–濃度分別增加0.001、0.01 mol/L。
3種組織經(jīng)電火花線切割至圖1中的試樣尺寸,用碳化硅砂紙逐級打磨至2000目,之后用丙酮、酒精超聲清洗,并吹干待用。試樣端部焊接銅導(dǎo)線后作為工作電極,使用704硅橡膠按圖1所示位置將其封裝在試樣盒中,其工作面積為0.3 cm2。飽和甘汞電極(SCE)和Pt片也按圖1所示位置固定,SCE電極底部、Pt片中心和工作電極暴露面中心在同一水平線上。
2.4.1 加強(qiáng)傳統(tǒng)美德教育。“百善孝為先”。隨著經(jīng)濟(jì)的快速發(fā)展、新思想新觀念的傳入,使得中華民族的傳統(tǒng)美德越來越被人淡忘和不重視,年輕人工作之后對自己的父母不盡孝道,不贍養(yǎng)老人,家庭養(yǎng)老功能弱化,道德制約始終不能像法律制約一樣有效,無法給予不盡贍養(yǎng)義務(wù)的子女一定的處罰,贍養(yǎng)老人得不到重視。針對此,要加強(qiáng)孝文化的宣傳,通過電視、廣播、報紙、網(wǎng)絡(luò)等多種渠道進(jìn)行孝文化建設(shè),讓人充斥在孝文化氛圍濃厚的環(huán)境中,潛移默化地影響人們的觀念。最重要的是,要加大學(xué)校及社會各界對青少年的思想教育,養(yǎng)成孝敬父母、老師、長輩的好習(xí)慣[4]。
測試開始前,使用WDML-30 kN型拉伸機(jī)以10–6s–1的應(yīng)變速率將工作電極加載至不同應(yīng)變狀態(tài),3種材料對應(yīng)的應(yīng)變狀態(tài)如圖2所示。當(dāng)加載至預(yù)定的應(yīng)變狀態(tài)后,向裝置中倒入溶液,進(jìn)行后續(xù)的力學(xué)–電化學(xué)測試。
電化學(xué)工作站采用科斯特CS350H,試驗(yàn)采用三電極體系,輔助電極為鉑片,參比電極為飽和甘汞電極。由于鈦合金在空氣中能自發(fā)鈍化,在表面形成氧化膜,空氣濕度、溫度、放置時間都能對其產(chǎn)生影響。這導(dǎo)致試驗(yàn)開始時工作電極表面狀態(tài)存在偏差,通過陰極極化可以消除這種偏差[7,9,41-43]。測試時,先在–1.2 V(vs. SCE)極化120 s,以除去電極表面的氧化膜,再進(jìn)行1 h的開路電位測試。動電位極化曲線測試的掃描范圍為–0.5 V(vs. OCP)~6 V(vs. SCE),掃描速率為1 mV/s。
圖1 試樣尺寸和裝置
圖2 不同組織對應(yīng)的應(yīng)變狀態(tài)
使用美林?jǐn)?shù)據(jù)技術(shù)股份有限公司的Tempo大數(shù)據(jù)分析平臺進(jìn)行機(jī)器學(xué)習(xí),將1.3小節(jié)中測得的極化曲線數(shù)據(jù)作為數(shù)據(jù)集訓(xùn)練模型,挖掘材料、環(huán)境、應(yīng)變、電位與電流的關(guān)系。由于數(shù)據(jù)量龐大,為節(jié)約運(yùn)算時間,降低學(xué)習(xí)難度,在訓(xùn)練前對數(shù)據(jù)集進(jìn)行預(yù)處理:保留極化曲線測試數(shù)據(jù)電流的正負(fù)號;以每條極化曲線的自腐蝕電位為中心,每隔100 mV提取1個數(shù)據(jù)點(diǎn)。選用決策樹、隨機(jī)森林、梯度提升樹3種模型進(jìn)行機(jī)器學(xué)習(xí),評估預(yù)測值和真實(shí)值的差異,并提取各變量的相對重要性。
對不同應(yīng)變狀態(tài)下的Ti80合金母材在未添加F–的模擬海水(pH=2)中進(jìn)行開路電位和極化曲線測試,結(jié)果如圖3所示。在圖3a中,不同應(yīng)變狀態(tài)下,Ti80合金母材的開路電位隨時間的延長逐漸正移。在前250 s,電位迅速升高;250 s后,電位趨于穩(wěn)定。在相同測試時間條件下,應(yīng)變越大,開路電位越低。這表明Ti80合金母材在未添加F–的模擬海水中可迅速達(dá)到并維持穩(wěn)定狀態(tài),同時應(yīng)變導(dǎo)致開路電位負(fù)移。在圖3b中,各應(yīng)變狀態(tài)下的Ti80合金母材呈現(xiàn)活化–鈍化特性,并且可以保持鈍化狀態(tài)。在測試范圍內(nèi)(6 V,vs. SCE),沒有觀察到破鈍電位,相關(guān)動力學(xué)參數(shù)見表2。由于鈦合金表現(xiàn)出的鈍化特性,自然狀態(tài)下其陽極處于鈍化區(qū),偏離了Tafel斜率描述的活化狀態(tài),因此對于陽極反應(yīng)特征的描述通常使用維鈍電流密度和破鈍電位,而不使用陽極Tafel斜率。根據(jù)表2可知,應(yīng)變對于Ti80合金母材在未添加F–模擬海水中極化曲線的動力學(xué)參數(shù)影響并非單調(diào)的??傮w上,塑性變形條件下的自腐蝕電位低于彈性變形條件下。8%應(yīng)變條件下的維鈍電流密度最大,為14.3 μA/cm2,這一值與其他材料或環(huán)境相比,仍舊很小[40,44-45]。同時,各應(yīng)變狀態(tài)下的維鈍電流密度較為接近,說明應(yīng)變對Ti80在未添加F–模擬海水中維鈍電流密度的影響并不顯著。陰極塔菲爾斜率隨應(yīng)變的增大先增大、后減小,最大值出現(xiàn)在0.16%應(yīng)變條件下。
圖3 不同應(yīng)變狀態(tài)下的Ti80合金母材在未添加F–的模擬海水(pH=2)中的電化學(xué)行為
表2 不同應(yīng)變狀態(tài)下的Ti80合金母材在未添加F–的模擬海水中極化曲線的動力學(xué)參數(shù)
Tab.2 Fitting parameters of potentiodynamic curves for Ti80 alloy base metal under different strain states in simulated seawater without F–addition
對不同應(yīng)變狀態(tài)下的Ti80合金母材在添加0.001 mol/L F–的模擬海水中進(jìn)行開路電位和極化曲線測試,結(jié)果如圖4所示。在圖4a中,不同應(yīng)變狀態(tài)下,Ti80合金母材的開路電位隨時間的延長負(fù)移至某一值后趨于穩(wěn)定,應(yīng)變越大,開路電位越低。8%應(yīng)變條件下,開路電位最負(fù),其值為–810.0 mV(vs. SCE),遠(yuǎn)低于不添加F–的海水條件下的開路電位。其余應(yīng)變狀態(tài)下的開路電位較為接近。在圖4b中,各應(yīng)變狀態(tài)下的Ti80合金母材呈現(xiàn)活化–鈍化特性,并且可以保持鈍化狀態(tài),在測試范圍內(nèi)(6 V,vs. SCE)沒有觀察到破鈍電位,相關(guān)動力學(xué)參數(shù)見表3。根據(jù)表3可知,應(yīng)變促進(jìn)自腐蝕電位負(fù)移,對于維鈍電流密度的影響不大,8%應(yīng)變條件下的維鈍電流密度最大,為46.8 μA/cm2。陰極塔菲爾斜率隨應(yīng)變的增大先增大、后減小,最大值出現(xiàn)在2%應(yīng)變條件下。
圖4 不同應(yīng)變狀態(tài)下的Ti80合金母材在添加0.001 mol/L F–的模擬海水中的電化學(xué)行為
對不同應(yīng)變狀態(tài)下的Ti80合金母材在添加0.01 mol/L F–的模擬海水中進(jìn)行開路電位和極化曲線測試,結(jié)果如圖5所示。在圖5a中,不同應(yīng)變狀態(tài)下的Ti80合金母材開路電位在極短時間內(nèi)負(fù)移至–900 mV附近,隨后逐漸正移至–800 mV附近,并且在前1 000 s范圍內(nèi)伴隨有劇烈的電位波動,這可能對應(yīng)著點(diǎn)蝕的萌生或鈍化膜的破壞。在圖5b中,各應(yīng)變狀態(tài)下的Ti80合金母材呈現(xiàn)活化–鈍化特性,并且可以保持鈍化狀態(tài),在測試范圍內(nèi)沒有觀察到破鈍電位,相關(guān)動力學(xué)參數(shù)見表4。根據(jù)表4可知,應(yīng)變對Ti80合金母材在添加0.01 mol/L F–模擬海水中的自腐蝕電位的影響不大。塑性變形條件下的維鈍電流密度高于彈性變形條件,6%應(yīng)變條件下的維鈍電流密度最大,為197.2 μA/cm2,遠(yuǎn)高于未添加和添加0.001 mol/L F–的模擬海水條件。陰極塔菲爾斜率隨應(yīng)變的增大先增大、后減小,最大值出現(xiàn)在0.2%應(yīng)變條件下。
表3 不同應(yīng)變狀態(tài)下的Ti80合金母材在添加0.001 mol/L F–的模擬海水中極化曲線的動力學(xué)參數(shù)
Tab.3 Fitting parameters of potentiodynamic curves for Ti80 alloy base metal under different strain states in simulated seawater with the addition of 0.001 mol/L F–
圖5 不同應(yīng)變狀態(tài)下的Ti80合金母材在添加0.01 mol/L F–模擬海水中的電化學(xué)行為
對不同應(yīng)變狀態(tài)下的900 ℃熱模擬組織在未添加F–的模擬海水中進(jìn)行開路電位和極化曲線測試,結(jié)果如圖6所示。在圖6a中,不同應(yīng)變狀態(tài)下,900 ℃熱模擬組織的開路電位隨時間的延長逐漸正移。在前250 s時,電位迅速升高;250 s后,電位升高速率減緩;測試1 h后,開路電位隨應(yīng)變的增加而降低。這表明900 ℃熱模擬組織在未添加F–的模擬海水中可維持穩(wěn)定狀態(tài),同時應(yīng)變導(dǎo)致開路電位負(fù)移。在圖6b中,各應(yīng)變狀態(tài)下的900 ℃熱模擬組織呈現(xiàn)活化–鈍化特性,并且可以保持鈍化狀態(tài),在測試范圍內(nèi)沒有觀察到破鈍電位,相關(guān)動力學(xué)參數(shù)見表5。根據(jù)表5可知,應(yīng)變對自腐蝕電位的影響較為復(fù)雜,這是由陰陽極反應(yīng)共同作用的結(jié)果。維鈍電流密度隨應(yīng)變的增大而增大,8%應(yīng)變條件下的維鈍電流密度最大,為17.7 μA/cm2,大約是無應(yīng)變條件下維鈍電流密度的2倍。陰極塔菲爾斜率隨應(yīng)變的增大先增大、后減小,最大值出現(xiàn)在0.2%應(yīng)變條件下。
表4 不同應(yīng)變狀態(tài)下的Ti80合金母材在添加0.01 mol/L F–的模擬海水中極化曲線的動力學(xué)參數(shù)
Tab.4 Fitting parameters of potentiodynamic curves for Ti80 alloy base metal under different strain states in simulated seawater with the addition of 0.01 mol/L F–
圖6 不同應(yīng)變狀態(tài)下的Ti80合金900 ℃熱模擬組織在不添加F–模擬海水中的電化學(xué)行為
表5 不同應(yīng)變狀態(tài)下的Ti80合金900 ℃熱模擬組織在未添加F–的模擬海水中極化曲線的動力學(xué)參數(shù)
Tab.5 Fitting parameters of potentiodynamic curves for 900 ℃ simulated Ti80 microstructure under different strain states in simulated seawater without F– addition
對不同應(yīng)變狀態(tài)下的900 ℃熱模擬組織在添加0.001 mol/L F–的模擬海水中進(jìn)行開路電位和極化曲線測試,結(jié)果如圖7所示。在圖7a中,不同應(yīng)變狀態(tài)下,900 ℃熱模擬組織的開路電位隨時間的延長逐漸正移。其中,前250 s電位迅速升高,250 s后電位升高速率減緩,測試1 h后,開路電位隨應(yīng)變的增加而降低。在圖7b中,各應(yīng)變狀態(tài)下的900 ℃熱模擬組織呈現(xiàn)活化–鈍化特性,并且可以保持鈍化狀態(tài),在測試范圍內(nèi)沒有觀察到破鈍電位,相關(guān)動力學(xué)參數(shù)見表6。根據(jù)表6可知,應(yīng)變對自腐蝕電位的影響較為復(fù)雜,這是由陰陽極反應(yīng)共同作用的結(jié)果。應(yīng)變增加導(dǎo)致維鈍電流密度增大,8%應(yīng)變條件下的維鈍電流密度最大,為26.1 μA/cm2,大約是無應(yīng)變條件下維鈍電流密度的3倍。陰極塔菲爾斜率隨應(yīng)變的增大先增大、后減小,最大值出現(xiàn)在0.2%應(yīng)變條件下。
圖7 不同應(yīng)變狀態(tài)下的Ti80合金900 ℃熱模擬組織在添加0.001 mol/L F–的模擬海水中的電化學(xué)行為
表6 不同應(yīng)變狀態(tài)下的Ti80合金900 ℃熱模擬組織在添加0.001 mol/L F–的模擬海水中極化曲線的動力學(xué)參數(shù)
Tab.6 Fitting parameters of potentiodynamic curves for 900 ℃ simulated Ti80 microstructure under different strain states in simulated seawater with the addition of 0.001 mol/L F–
對不同應(yīng)變狀態(tài)下的900 ℃熱模擬組織在添加0.01 mol/L F–的模擬海水中進(jìn)行開路電位和極化曲線測試,結(jié)果如圖8所示。在圖8a中,有應(yīng)變的900 ℃熱模擬組織,開路電位在極短時間內(nèi)負(fù)移至–900 mV附近,隨后逐漸正移至–760 mV附近,這對應(yīng)著表面鈍化膜的破壞–再生過程;無應(yīng)變的900 ℃熱模擬組織,開路電位先正移至–509 mV,之后迅速負(fù)移至–780 mV,最后逐漸正移至–740 mV。兩者的差異表明,應(yīng)變可促進(jìn)表面鈍化膜的破壞。在圖8b中,各應(yīng)變狀態(tài)下的900 ℃熱模擬組織呈現(xiàn)活化–鈍化特性,并且可以保持鈍化狀態(tài),在測試范圍內(nèi)沒有觀察到破鈍電位,相關(guān)動力學(xué)參數(shù)見表7。根據(jù)表7可知,應(yīng)變整體上促進(jìn)自腐蝕電位負(fù)移,維鈍電流密度增加。8%應(yīng)變條件下的維鈍電流密度最大,為127.4 μA/cm2,遠(yuǎn)高于未添加和添加0.001 mol/L F–的模擬海水條件,卻明顯低于Ti80合金在添加0.01 mol/L F–的模擬海水條件下的維鈍電流密度。不同應(yīng)變狀態(tài)下的陰極塔菲爾斜率十分接近,整體上呈現(xiàn)隨應(yīng)變的增大先增大、后減小的趨勢,最大值出現(xiàn)在0.2%應(yīng)變條件下。
圖8 不同應(yīng)變狀態(tài)下的Ti80合金900 ℃熱模擬組織在添加0.01 mol/L F–模擬海水中的電化學(xué)行為
表7 不同應(yīng)變狀態(tài)下的Ti80合金900 ℃熱模擬組織在添加0.01 mol/L F–的模擬海水中極化曲線的動力學(xué)參數(shù)
Tab.7 Fitting parameters of potentiodynamic curves for 900 ℃ simulated Ti80 microstructure under different strain states in simulated seawater with the addition of 0.01 mol/L F–
對不同應(yīng)變狀態(tài)下的1 500 ℃熱模擬組織在未添加F–的模擬海水中進(jìn)行開路電位和極化曲線測試,結(jié)果如圖9所示。在圖9a中,不同應(yīng)變狀態(tài)下,900 ℃熱模擬組織的開路電位隨時間的延長逐漸正移。前250 s電位迅速升高,250 s后電位升高速率減緩,測試1 h后,開路電位隨應(yīng)變的增加而降低。這表明 1 500 ℃熱模擬組織在未添加F–的模擬海水中可維持穩(wěn)定狀態(tài),同時應(yīng)變導(dǎo)致開路電位負(fù)移。在圖9b中,各應(yīng)變狀態(tài)下的1 500 ℃熱模擬組織呈現(xiàn)活化–鈍化特性,并且可以保持鈍化狀態(tài),在測試范圍內(nèi)沒有觀察到破鈍電位,相關(guān)動力學(xué)參數(shù)見表8。根據(jù)表8可知,應(yīng)變促進(jìn)自腐蝕電位負(fù)移,維鈍電流密度增加,陰極塔菲爾斜率增加。8%應(yīng)變條件下的維鈍電流密度最大,為9.8 μA/cm2,大約是無應(yīng)變條件下維鈍電流密度的2倍。陰極塔菲爾斜率增加表明,同等過電位條件下,陰極電流密度減小,即應(yīng)變抑制陰極反應(yīng)的進(jìn)行。
圖9 不同應(yīng)變狀態(tài)下的Ti80合金1 500 ℃熱模擬組織在未添加F–模擬海水中的電化學(xué)行為
對不同應(yīng)變狀態(tài)下的1 500 ℃熱模擬組織在含0.001 mol/L F–的模擬海水中進(jìn)行開路電位和極化曲線測試,結(jié)果如圖10所示。在圖10a中,不同應(yīng)變狀態(tài)下,1 500 ℃熱模擬組織的開路電位隨時間的延長逐漸正移。前250 s電位迅速升高,250 s后電位升高速率減緩,測試1 h后,開路電位隨應(yīng)變增加而降低。在圖10b中,各應(yīng)變狀態(tài)下的1 500 ℃熱模擬組織呈現(xiàn)活化–鈍化特性,并且可以保持鈍化狀態(tài),在測試范圍內(nèi)沒有觀察到破鈍電位,相關(guān)動力學(xué)參數(shù)見表9。根據(jù)表9可知,應(yīng)變促進(jìn)自腐蝕電位負(fù)移、維鈍電流密度增加。2%應(yīng)變條件下的維鈍電流密度最大,為36.4 μA/cm2,明顯大于不含F(xiàn)–的模擬海水條件,同時明顯大于相同應(yīng)變、環(huán)境條件下的Ti80合金和900 ℃熱模擬組織。陰極塔菲爾斜率隨應(yīng)變的增大先增大、后減小,最大值出現(xiàn)在0.2%應(yīng)變條件下。
表8 不同應(yīng)變狀態(tài)下的Ti80合金1 500 ℃熱模擬組織在未添加F–的模擬海水中極化曲線的動力學(xué)參數(shù)
Tab.8 Fitting parameters of potentiodynamic curves for 1 500 ℃ simulated Ti80 microstructure under different strain states in simulated seawater without F–addition
圖10 不同應(yīng)變狀態(tài)下的Ti80合金1 500 ℃熱模擬組織在添加0.001 mol/L F–模擬海水中的電化學(xué)行為
表9 不同應(yīng)變狀態(tài)下的Ti80合金1 500 ℃熱模擬組織在添加0.001 mol/L F–的模擬海水中極化曲線的動力學(xué)參數(shù)
Tab.9 Fitting parameters of potentiodynamic curves for 1 500 ℃ simulated Ti80 microstructure under different strain states in simulated seawater with the addition of 0.001 mol/L F–
對不同應(yīng)變狀態(tài)下的1 500 ℃熱模擬組織在添加0.01 mol/L F–的模擬海水中進(jìn)行開路電位和極化曲線測試,結(jié)果如圖11所示。在圖11a中,不同應(yīng)變狀態(tài)下,1 500 ℃熱模擬組織的開路電位迅速負(fù)移至–800 mV附近,1 h后逐漸正移至–735 mV附近。在這一過程中,除無應(yīng)變試樣的開路電位曲線較為平滑外,其余應(yīng)變狀態(tài)的開路電位曲線均伴隨劇烈波動,這表明應(yīng)變可促進(jìn)表面鈍化膜的破壞。在圖11b中,各應(yīng)變狀態(tài)下的1 500 ℃熱模擬組織呈現(xiàn)活化–鈍化特性,并且可以保持鈍化狀態(tài),在測試范圍內(nèi)沒有觀察到破鈍電位,相關(guān)動力學(xué)參數(shù)見表10。根據(jù)表10可知,應(yīng)變促進(jìn)自腐蝕電位負(fù)移,維鈍電流密度增加,陰極塔菲爾斜率增加。2%應(yīng)變條件下的維鈍電流密度最大,為170.2 μA/cm2,遠(yuǎn)高于未添加和添加0.001 mol/L F–的模擬海水條件,同時高于相同應(yīng)變、環(huán)境條件下的Ti80合金和900 ℃熱模擬組織。
圖11 不同應(yīng)變狀態(tài)下的Ti80合金1 500 ℃熱模擬組織在添加0.01 mol/L F–模擬海水中的電化學(xué)行為
表10 不同應(yīng)變狀態(tài)下的Ti80合金1 500 ℃熱模擬組織在添加0.01 mol/L F–的模擬海水中極化曲線的動力學(xué)參數(shù)
Tab.10 Fitting parameters of potentiodynamic curve for 1 500 ℃ simulated Ti80 microstructure under different strain states in simulated seawater with the addition of 0.01 mol/L F–
結(jié)合表2—10的數(shù)據(jù)可知,塑性變形顯著增加了電極的表面活性,提高了維鈍電流密度,而彈性變形對電化學(xué)行為的影響不顯著。這是由于鈦合金表面有一層以其氧化物為主的鈍化膜,當(dāng)基體發(fā)生塑性變形時,鈍化膜中的缺陷增加,形成更多活性位點(diǎn),促進(jìn)陰陽極反應(yīng)。當(dāng)發(fā)生彈性變形時,晶格中原子間距增加,并不產(chǎn)生大量缺陷,O2–的擴(kuò)散通道也沒有顯著增加,因此對于電化學(xué)活性的影響并不顯著。
本文中涉及的反應(yīng)有:
Ti+O2→4TiO2(1)
TiO2+4H++4F–→TiF4+H2O (2)
TiF4+2F–aq→TiF62–(3)
在pH=2的海水中,發(fā)生反應(yīng)(1),形成鈍化膜,覆蓋在電極表面,隔絕金屬基體和溶液。由于TiO2性質(zhì)穩(wěn)定,難以溶解,陽極電流密度非常小。當(dāng)環(huán)境中存在更多F–時,TiO2發(fā)生反應(yīng)(2)、(3)溶解,促進(jìn)Ti氧化生成TiO2,增大維鈍電流密度。同時,更多陽極反應(yīng)生成的電子穿過鈍化膜,到達(dá)膜/溶液界面,促進(jìn)陰極反應(yīng)。因此,增加F–濃度,可顯著提高鈦合金的電化學(xué)活性。
通過機(jī)器學(xué)習(xí),挖掘應(yīng)變、環(huán)境、組織與電化學(xué)行為之間的關(guān)系。使用決策樹、隨機(jī)森林、梯度提升樹模型分別對3種組織在含氟模擬海水中的極化曲線進(jìn)行擬合,結(jié)果如圖12所示。由圖12可知,決策樹和梯度提升樹模型的擬合效果較好,散點(diǎn)大都落在斜率為1的直線上,表明預(yù)測值與真實(shí)值接近;而隨機(jī)森林模型的擬合效果較差,預(yù)測值與真實(shí)值偏離較大。這與其他學(xué)者[32,46-47]的研究結(jié)果不同,可能是由于輸入變量不同導(dǎo)致的。
使用決策樹、隨機(jī)森林、梯度提升樹3種模型訓(xùn)練擬合過程中各變量的相對重要性因子如圖13所示。3種模型的變量相對重要性排序都是電位>F–濃度>應(yīng)變>組織,其中決策樹和梯度提升樹模型變量的相對重要性一致,隨機(jī)森林模型中組織的相對重要性比另外2種模型更高,這可能是導(dǎo)致其預(yù)測值偏離真實(shí)值的部分原因。根據(jù)數(shù)據(jù)挖掘的結(jié)果,在應(yīng)變–F–環(huán)境耦合的條件下,F(xiàn)–濃度對電化學(xué)行為影響最大,應(yīng)變狀態(tài)次之,材料的組織影響最小。
圖12 模型訓(xùn)練集真實(shí)值與預(yù)測值
圖13 變量的相對重要性
本文通過對不同應(yīng)變狀態(tài)的Ti80合金在含氟模擬海水中進(jìn)行開路電位和極化曲線測試,研究了應(yīng)變、F–濃度、組織對Ti80合金電化學(xué)行為的影響。結(jié)果表明,塑性變形顯著提高Ti80在模擬海水中的電化學(xué)活性,而彈性變形的影響并不明顯;F–濃度增加顯著提高電化學(xué)活性;根據(jù)數(shù)據(jù)挖掘的結(jié)果,在應(yīng)變–F–環(huán)境耦合的條件下,F(xiàn)–濃度對電化學(xué)行為的影響最大,應(yīng)變狀態(tài)次之,材料的組織影響最小。
本文的機(jī)器學(xué)習(xí)部分內(nèi)容使用了美林?jǐn)?shù)據(jù)技術(shù)股份有限公司的Tempo大數(shù)據(jù)分析平臺,在此表示感謝!
[1] XIONG Jian-yu, TAN M Y, FORSYTH M. The Corrosion Behaviors of Stainless Steel Weldments in Sodium Chloride Solution Observed Using a Novel Electrochemical Meas-urement Approach[J]. Desalination, 2013, 327: 39-45.
[2] ZHANG Ying-xiao, FAN Lin, LIU Zhi-yong, et al. Effect of Alternating Magnetic Field on Electrochemical Beha-vior of 316L and TA2 in Simulated Seawater[J]. Journal of Materials Engineering and Performance, 2021, 30(12): 9377-9389.
[3] HWANG M J, PARK E J, MOON W J, et al. Character-ization of Passive Layers Formed on Ti-10wt% (Ag, Au, Pd, or Pt) Binary Alloys and Their Effects on Galvanic Corrosion[J]. Corrosion Science, 2015, 96: 152-159.
[4] TSAI W T, LIN C L, PAN S J. Susceptibility of Ti-6Al-4V Alloy to Stress Corrosion Cracking in a Lewis-Neutral Aluminium Chloride-1-Ethyl-3-Methylimidazolium ChlorideIonic Liquid[J]. Corrosion Science, 2013, 76: 494-497.
[5] SU Bao-xian, WANG Bin-bin, LUO Liang-shun, et al. The Corrosion Behavior of Ti-6Al-3Nb-2Zr-1Mo Alloy: Effects of HCl Concentration and Temperature[J]. Journal of Materials Science & Technology, 2021, 74: 143-154.
[6] HE Sheng-tong, ZENG Wei-dong, ZHAO Zi-bo, et al. Analysis of Anisotropy Mechanism in Relation with Slip Activity in near α Titanium Alloy Pipe after Pilger Cold Rolling[J]. Journal of Alloys and Compounds, 2022, 909: 164785.
[7] YANG Xiao-jia, DU Cui-wei, WAN Hong-xia, et al. Infl-u-ence of Sulfides on the Passivation Behavior of Titanium Alloy TA2 in Simulated Seawater Environments[J]. App-lied Surface Science, 2018, 458: 198-209.
[8] ZHANG Ying-xiao, YAN Ting-ting, FAN Lin, et al. Effect of pH on the Corrosion and Repassivation Behavior of TA2 in Simulated Seawater[J]. Materials, 2021, 14(22): 6764.
[9] CUI Zhong-yu, WANG Li-wei, ZHONG Ming-yuan, et al. Electrochemical Behavior and Surface Characteristics of Pure Titanium during Corrosion in Simulated Desulfuri-zed Flue Gas Condensates[J]. Journal of the Electroche-mical Society, 2018, 165(9): C542-C561.
[10] CUI Zhong-yu, LIU Zhi-yong, WANG Li-wei, et al. Effect of Plastic Deformation on the Electrochemical and Stress Corrosion Cracking Behavior of X70 Steel in Near- Neutral pH Environment[J]. Materials Science and Engin-eering: A, 2016, 677: 259-273.
[11] KRAWIEC H, VIGNAL V, SCHWARZENBOECK E, et al. Role of Plastic Deformation and Microstructure in the Micro-Electrochemical Behaviour of Ti-6Al-4V in Sodium Chloride Solution[J]. Electrochimica Acta, 2013, 104: 400-406.
[12] YAZDI R, GHASEMI H M, ABEDINI M, et al. Interplay between Mechanical Wear and Electrochemical Corrosion during Tribocorrosion of Oxygen Diffusion Layer on Ti-6Al-4V in PBS Solution[J]. Applied Surface Science, 2020, 518: 146048.
[13] VENKATESH S, JOY N, MAGESHWARAN G, et al. Investigation on the Electrochemical Characteristics of Ti-6Al-4?V Weldment[J]. Materials Today: Proceedings, 2021, 44: 3727-3731.
[14] SU Bao-xian, WANG Bin-bin, LUO Liang-shun, et al. Corrosion Behaviour of a Wrought Ti-6Al-3Nb-2Zr-1Mo Alloy in Artificial Seawater with Various Fluoride Conc-entrations and pH Values[J]. Materials & Design, 2022, 214: 110416.
[15] REN Shuai, DU Cui-wei, LIU Zhi-yong, et al. Effect of Fluoride Ions on Corrosion Behaviour of Commercial Pure Titanium in Artificial Seawater Environment[J]. Applied Surface Science, 2020, 506: 144759.
[16] IMANI A, ASSELIN E. Fluoride Induced Corrosion of Ti-45Nb in Sulfuric Acid Solutions[J]. Corrosion Science, 2021, 181: 109232.
[17] ZHANG Hong-wei, MAN Cheng, DONG Chao-fang, et al. The Corrosion Behavior of Ti6Al4V Fabricated by Selective Laser Melting in the Artificial Saliva with Diffe-rent Fluoride Concentrations and pH Values[J]. Corrosion Science, 2021, 179: 109097.
[18] SHEN Zhi-xin, MA Ai-bin, JIANG Jing-hua, et al. Electr-ochemical Corrosion Behavior of Ultrafine-Grained Mg Alloy ZE41A through Severe Plastic Deformation[J]. Procedia Engineering, 2012, 27: 1817-1822.
[19] WANG Ying, JIN Jun-song, ZHANG Mao, et al. Influe-nce of Plastic Deformation on the Corrosion Behavior of CrCoFeMnNi High Entropy Alloy[J]. Journal of Alloys and Compounds, 2022, 891: 161822.
[20] SACCO E A, ALVAREZ N B, CULCASI J D, et al. Effect of the Plastic Deformation on the Electrochemical Beha-vior of Metal Coated Steel Sheets[J]. Surface and Coat-ings Technology, 2003, 168(2-3): 115-122.
[21] JANDAGHI M R, POURALIAKBAR H. Elucidating the Microscopic Origin of Electrochemical Corrosion and Electrical Conductivity by Lattice Response to Severe Plastic Deformation in Al-Mn-Si Alloy[J]. Materials Research Bulletin, 2018, 108: 195-206.
[22] LI Yong, PEI Zi-bo, ZAMAN B, et al. Effects of Plastic Deformations on the Electrochemical and Stress Corro-sion Cracking Behaviors of TC2 Titanium Alloy in Simu-lated Seawater[J]. Materials Research Express, 2018, 5(11): 116516.
[23] HU Jun, DU Lin-xiu, XIE Hui, et al. Effect of Weld Peak Temperature on the Microstructure, Hardness, and Transf-ormation Kinetics of Simulated Heat Affected Zone of Hot Rolled Ultra-Low Carbon High Strength Ti-Mo Ferritic Steel[J]. Materials & Design, 2014, 60: 302-309.
[24] MOON J, KIM S J, LEE C. Effect of Thermo-Mechanical Cycling on the Microstructure and Strength of Lath Mar-tensite in the Weld CGHAZ of HSLA Steel[J]. Materials Science and Engineering: A, 2011, 528(25/26): 7658-7662.
[25] OR?OWSKA M, PIXNER F, HüTTER A, et al. Local Changes in the Microstructure, Mechanical and Electro-chemical Properties of Friction Stir Welded Joints from Aluminium of Varying Grain Size[J]. Journal of Materials Research and Technology, 2021, 15: 5968-5987.
[26] MA H C, LIU Z Y, DU C W, et al. Stress Corrosion Crac-king of E690 Steel as a Welded Joint in a Simulated Mari-ne Atmosphere Containing Sulphur Dioxide[J]. Corrosion Science, 2015, 100: 627-641.
[27] VARVARA S, BERGHIAN-GROSAN C, BOSTAN R, et al. Experimental Characterization, Machine Learning Anal-ysis and Computational Modelling of the High Effective Inhibition of Copper Corrosion by 5-(4-Pyridyl)-1,3,4- Oxadiazole-2-Thiol in Saline Environment[J]. Electro-chimica Acta, 2021, 398: 139282.
[28] BONGIORNO V, GIBBON S, MICHAILIDOU E, et al. Exploring the Use of Machine Learning for Interpreting Electrochemical Impedance Spectroscopy Data: Evalu-ation of the Training Dataset Size[J]. Corrosion Science, 2022, 198: 110119.
[29] ZHU Shan, SUN Xin-yang, GAO Xiao-yang, et al. Equi-valent Circuit Model Recognition of Electrochemical Impedance Spectroscopy via Machine Learning[J]. Journal of Electroanalytical Chemistry, 2019, 855: 113627.
[30] LI Jian-kuan, SUN Chong, SHUANG Shuo, et al. Invest-igation on the Flow-Induced Corrosion and Degradation Behavior of Underground J55 Pipe in a Water Production Well in the Athabasca Oil Sands Reservoir[J]. Journal of Petroleum Science and Engineering, 2019, 182: 106325.
[31] ZHANG Jin-rui, ZHANG Meng-xi, DONG Bi-qin, et al. Quantitative Evaluation of Steel Corrosion Induced Deter-ioration in Rubber Concrete by Integrating Ultrasonic Testing, Machine Learning and Mesoscale Simulation[J]. Cement and Concrete Composites, 2022, 128: 104426.
[32] AGHAAMINIHA M, MEHRANI R, COLAHAN M, et al. Machine Learning Modeling of Time-Dependent Corro-sion Rates of Carbon Steel in Presence of Corrosion Inh-ibitors[J]. Corrosion Science, 2021, 193: 109904.
[33] GONG Xiao-yu, DONG Chao-fang, XU Jia-jin, et al. Ma-c-hine Learning Assistance for Electrochemical Curve Simulation of Corrosion and Its Application[J]. Materials and Corrosion, 2020, 71(3): 474-484.
[34] PEI Zi-bo, ZHANG Da-wei, ZHI Yuan-jie, et al. Towards Understanding and Prediction of Atmospheric Corrosion of an Fe/Cu Corrosion Sensor via Machine Learning[J]. Corrosion Science, 2020, 170: 108697.
[35] YANG Xiao-jia, YANG Ying, SUN Mei-hui, et al. A New Understanding of the Effect of Cr on the Corrosion Resi-stance Evolution of Weathering Steel Based on Big Data Technology[J]. Journal of Materials Science & Techn-ology, 2022, 104: 67-80.
[36] SU Bao-xian, LUO Liang-shun, WANG Bin-bin, et al. Annealed Microstructure Dependent Corrosion Behavior of Ti-6Al-3Nb-2Zr-1Mo Alloy[J]. Journal of Materials Science & Technology, 2021, 62: 234-248.
[37] DAVIS J R. ASM Specialty Handbook: Cast Irons[M]. [s. l.]: ASM International, 1996.
[38] CUI Z Y, LIU Z Y, WANG X Z, et al. Crack Growth Behaviour and Crack Tip Chemistry of X70 Pipeline Steel in Near-Neutral pH Environment[J]. Corrosion Engine-ering, Science and Technology, 2016, 51(5): 352-357.
[39] CAI Bao-ping, LIU Yong-hong, TIAN Xiao-jie, et al. An Experimental Study of Crevice Corrosion Behaviour of 316L Stainless Steel in Artificial Seawater[J]. Corrosion Science, 2010, 52(10): 3235-3242.
[40] CUI L Y, LIU Z Y, XU D K, et al. The Study of Micro-biologically Influenced Corrosion of 2205 Duplex Stain-less Steel Based on High-Resolution Characterization[J]. Corrosion Science, 2020, 174: 108842.
[41] CUI Zhong-yu, CHEN Shuang-shuai, DOU Yun-peng, et al. Passivation Behavior and Surface Chemistry of 2507 Super Duplex Stainless Steel in Artificial Seawater: Influ-ence of Dissolved Oxygen and pH[J]. Corrosion Science, 2019, 150: 218-234.
[42] CUI Zhong-yu, CHEN Shuang-shuai, WANG Li-wei, et al. Passivation Behavior and Surface Chemistry of 2507 Super Duplex Stainless Steel in Acidified Artificial Seaw-ater Containing Thiosulfate[J]. Journal of the Electro-chemical Society, 2017, 164(13): C856-C868.
[43] DOU Yun-peng, HAN Si-ke, WANG Li-wei, et al. Chara-c-terization of the Passive Properties of 254SMO Stainless Steel in Simulated Desulfurized Flue Gas Condensates by Electrochemical Analysis, XPS and ToF-SIMS[J]. Corrosion Science, 2020, 165: 108405.
[44] LIU Z Y, HAO W K, WU W, et al. Fundamental Invest-igation of Stress Corrosion Cracking of E690 Steel in Simulated Marine Thin Electrolyte Layer[J]. Corrosion Science, 2019, 148: 388-396.
[45] YANG Xiao-jia, SHAO Jia-min, LIU Zhi-yong, et al. Stress-Assisted Microbiologically Influenced Corrosion Mechanism of 2205 Duplex Stainless Steel Caused by Sulfate-Reducing Bacteria[J]. Corrosion Science, 2020, 173: 108746.
[46] HOU Y, ALDRICH C, LEPKOVA K, et al. Analysis of Electrochemical Noise Data by Use of Recurrence Qua-ntif-ication Analysis and Machine Learning Metho-ds[J]. Electrochimica Acta, 2017, 256: 337-347.
[47] HUO Wei-wei, LI Wei-er, ZHANG Ze-hui, et al. Perfor-mance Prediction of Proton-Exchange Membrane Fuel Cell Based on Convolutional Neural Network and Random Forest Feature Selection[J]. Energy Conversion and Ma-nagement, 2021, 243: 114367.
Mechanical-electrochemical Study of Ti80 and Heat Treatment Simulated Microstructure in Fluoride-contained Simulated Seawater Environment
1a,1a,1a,2,1
(1. a. Institute for Advanced Materials and Technology, b. Key Laboratory for Corrosion and Protection (MOE), University of Science and Technology Beijing, Beijing 100083, China; 2. School of Chemistry and Chemical Engineering, Guangzhou University, Guangzhou 510006, China)
In this work, effects of strain, environment, and microstructure on the electrochemical behavior of Ti80 alloy and its simulated heat treatment microstructures in fluoride-contained simulated seawater were studied to provide data support for the safe service of marine engineering equipment. Machine learning method was used to study the influence and compare the relative importance of the affected factors. Results depict that potentiodynamic curves of Ti80 alloy could be accurately predicted under different strain states without additional measurement. To simulate the microstructures of heat affected zone, base metals were kept at 900 ℃ and 1 500 ℃ for 5 min, and then cooled by air to room temperature, as called 900 ℃ and 1 500 ℃ simulated Ti80 microstructure. Tensile test specimens with three different microstructures (base metal, 900 ℃ and 1 500 ℃ simulated Ti80 microstructure) were sectioned, grounded with 2000 grits silicon paper, ultrasonically cleaned by acetone and ethanol, and embedded in a sealant (KAFUTER 704 RTV) to provide 0.3 cm2as working area. Specimens were loaded to different strain states on a WDML-30 kN with a strain rate of 10–6s–1before electrochemical measurement. The simulated seawater in ASTM D1141-98(2013) was used to deploy the solution. The pH value of seawater was adjusted to 2 by HCl. NaF was added to increase F–concentration with two levels: 0.001 mol/L and 0.01 mol/L. After polarized at 1.2 V for 120 s, open circuit potential and potentiodynamic curve were tested under different strain states by a CS350H. The machine learning method (Tempodata from Meritdata) was used to mine the relationship between electrochemical behavior and strain, environment, and microstructure. To speed up the model construction, data of current density from potentiodynamic curves were preprocessed in this way: generate a data point every 100 mV form the corrosion potential. Decision tree, random forest, and gradient boosting tree were trained by current density of potentiodynamic curve. Accuracy and relative importance of models were compared. The results showed that open circuit potential shifted negatively as strain increased in seawater without F–addition and with the addition of 0.001 mol/L F–. But strain had little effect on open circuit potential in seawater with the addition of 0.01 mol/L F–. On the whole, strain promoted the increase of passive current density. The condition of 1 500 ℃ simulated Ti80 microstructure in seawater with the addition of 0.01 mol/L F–was the most severely affected by strain, whose passive current density in the maximum strain was about 3 times that without strain. The maximum value of the cathode Tafel slope appeared near the yield point. The increase of F–concentration significantly increased the passive current density. Decision tree and gradient boosting tree algorithms were more accurate in predicting the current value of the polarization curve, while the random forest algorithm was less accurate. In the relative importance comparison, F–concentration had the greatest effect on electrochemical behavior, followed by strain state, and the microstructure had the least effect. In summary, plastic deformation significantly improves the electrochemical activity of Ti80 in simulated seawater, while the effect of elastic deformation is not obvious. The increase in F–concentration significantly promotes the electrochemical activity. The decision tree and gradient boosting tree algorithm could be used to accurately predict potentiodynamic curves with different strains, fluoride ion concentrations, and microstructures of Ti80. For Ti80 in simulated fluoride-contained seawater, the order of importance that affects the electrochemical behavior is: F–concentration> strain> microstructure.
Ti80 alloy; mechanical-electrochemical; machine learning
TG146.2+3
A
1001-3660(2022)05-0049-12
10.16490/j.cnki.issn.1001-3660.2022.05.006
2022–04–04;
2022–04–19
2022-04-04;
2022-04-19
張穎驍(1990—),男,博士研究生,主要研究方向?yàn)殁伜辖鸬母g與防護(hù)。
ZHANG Ying-xiao (1990-), Male, Doctoral candidate, Research focus: corrosion and protection of titanium alloy.
李曉剛(1963—),男,博士,教授,主要研究方向?yàn)榻饘俨牧献匀画h(huán)境腐蝕及耐蝕鋼的研發(fā)。
LI Xiao-gang (1963-), Male, Doctor, Professor, Research focus: corrosion of metal materials in natural environment and development of low alloy steel.
張穎驍, 張梓楊, 宋龍飛, 等. Ti80合金及其熱模擬組織在含氟模擬海水中的力學(xué)電化學(xué)行為研究[J]. 表面技術(shù), 2022, 51(5): 49-60.
ZHANG Ying-xiao, ZHANG Zi-yang, SONG Long-fei, et al. Mechanical-electrochemical Study of Ti80 and Heat Treatment Simulated Microstructure in Fluoride-contained Simulated Seawater Environment[J]. Surface Technology, 2022, 51(5): 49-60.
責(zé)任編輯:劉世忠