Динамика атомной и электронной структуры катализаторов на основе палладия при реалистичных технологических условиях тема диссертации и автореферата по ВАК РФ 01.04.15, доктор наук Бугаев Арам Лусегенович

• Бугаев Арам Лусегенович
• доктор наук
• 2021

Оглавление диссертации доктор наук Бугаев Арам Лусегенович

ВВЕДЕНИЕ

РАЗДЕЛ I. Атомная структура гидридных и карбидных фаз наночастиц палладия

1.1. Необратимое образование карбида палладия в ходе реакции гидрирования этилена

1.2. Роль поверхностных и объемных фаз в реакции гидрирования

РАЗДЕЛ II. Структура наночастиц палладия в ходе реакций окисления

2.1. Объемные и поверхностные оксиды в малых нанокластерах палладия в ходе реакции окисления водорода

2.2. Роль оксидных фаз в реакции окисления спиртов на нанокластерах палладия

2.3. Образование карбидной фазы в реакции окисления метана на наночастицах палладия

РАЗДЕЛ 3. Изолированные активные центры палладия, стабилизированные на пористых кристаллических подложках

3.1. Активные центры палладия в металл-органических каркасных структурах для реакций алкенилирования аренов

3.2. Активные центры палладия в TiOP для реакций алкенилирования аренов

3.3. Активные центры палладия в цеолитах для реакции окислительного гомосочетания толуола

ЗАКЛЮЧЕНИЕ

Список цитируемой литературы

Основные публикации автора по теме диссертации

59

Введение диссертации (часть автореферата) на тему «Динамика атомной и электронной структуры катализаторов на основе палладия при реалистичных технологических условиях»

ВВЕДЕНИЕ

Актуальность темы. В работе исследованы нанокатализаторы на основе палладия в ходе промышленно значимых реакций гидрирования, окисления и С-И активации. Данные реакции имеют большое значение для нефтехимической, тонкой химической, автомобильной и фармацевтической промышленностей. Известно, что структура катализатора, формируемая в условиях реакции, может существенно отличаться от структуры исходного материала после синтеза. Важно также понимать, что не все палладиевые центры и не все фазы палладия является активными в тех или иных реакциях. Орвга^в-методология с использованием синхротронного излучения, использованная в данной работе, представляет уникальные возможности по исследованию зависимостей типа «структура-свойства», недоступных к однозначному определению в рамках других методов. Определение таких зависимостей, позволяет выделить активные фазы катализаторов в ходе протекания реакций, установить причины деградации и дезактивации нанокатализаторов, а также развить методы рационального дизайна новых функциональных материалов. Тем не менее, решение этой задачи является нетривиальным, требует использования передовой экспериментальной инфраструктуры и развития эффективных подходов к обработке и анализу данных, в том числе больших данных. В связи с этим, несмотря на огромную практическую значимость, механистическое понимание многих каталитических процессов, в особенности при использовании наноструктурированных материалов, до сих пор мало изучено и остается актуальной задачей не только с практической, но и с фундаментальной научной точки зрения.

Таким образом, тема диссертационного исследования, посвященного определению динамики локальной атомной и электронной структуры активных центров палладия в наноструктурированных катализаторах при реалистичных

технологических условиях непосредственно в ходе протекания каталитических реакций, является актуальной.

Цель работы. Разработка комплексной методики operando- диагностики локальной атомной и электронной структуры активных каталитических центров в наноструктурированных катализаторах на основе палладия в условиях, приближенных к реальным технологическим, с использованием синхротронных методов исследования и суперкомпьютерного моделирования.

Научная новизна. Научная новизна проекта заключается, прежде всего, в том, что в ходе работы впервые в мировой практике предложены методы анализа спектров XANES, на основе которых была получена детальная характеристика локальной атомной и электронной структуры и её динамики для перспективных наноструктурированных материалов на основе палладия непосредственно в ходе протекания каталитических реакций гидрирования углеводородов, окисления спиртов и метана, в режиме operando, т.е. при непосредственном проетакии каталитических реакций. В частности, использование оригинальных подходов для анализа данных позволило однозначно определить формирование гидридных, карбидных и оксидных фаз в ходе протекания каталитических реакций даже в тех случаях, когда эти процессы не приводили к существенному перестроению кристаллической решетки палладия. Определены особенности фазового состава наночастиц в широком спектре реакций гидрирования и окисления. Определена структура изолированных активных центров палладия стабилизированных на пористых кристаллических носителях в серии новых катализаторов для реакций образования С-С связи путем активации С-Н связи (C-H активации).

Была разработана и успешно реализована на практике концепция проведения экспериментальных исследований в режиме operando и с высоким разрешением по времени на источниках синхротронного излучения, которая позволила эффективно исследовать поведение активных каталитических центров наноструктурированных материалов на основе палладия в ходе

протекания каталитических реакций при реалистичных технологических условиях.

Был реализован целый спектр оригинальных методик и подходов для извлечения детальной структурной информации о наноразмерных катализаторах на основе палладия в ходе протекания каталитических реакций при реалистичных технологических условиях. Были успешно использованы алгоритмы анализа больших данных и методы машинного обучения, разработанные при участии соискателя и не имеющие мировых аналогов. Следует отметить, что полученные результаты представляют уникальный пример того, насколько подробную информацию о локальной атомной и электронной структуре наноразмерных объектов, в том числе о наличии в них легких примесей водорода или углерода, можно извлечь с использованием жесткого рентгеновского излучения в режиме operando. Независимым подтверждением высокого научного уровня и научной новизны исследований является тот факт, что заявки на их проведение прошли жесткий конкурсный отбор и были поддержаны ведущим мировым синхротронным центром ESRF (Франция).

Теоретическая и практическая значимость. Наноструктурированные катализаторы на основе палладия являются предпочтительным материалом для широкого спектра промышленно значимых реакций, в том числе в области крупнотоннажного производства, масштабного производства автомобильных катализаторов, тонкой химической и фармацевтической промышленностей. К сожалению, поиск новых, более стабильных и более эффективных катализаторов до сих пор ведется методом проб и ошибок и требует значительных временных затрат. Альтернативой является концепция рационального дизайна, для которого необходимо глубокое понимание структуры и динамики активных центров нанокатализаторов для конкретных каталитических реакций, что является центральной проблемой настоящего исследования. Более того, очевиден тот факт, что структура реально функционирующего катализатора формируется не только на этапе синтеза, но и непосредственно в ходе его работы, что ещё раз

подчеркивает важность развития и применения методик исследования структуры активных центров катализаторов при реалистичных технологических условиях. Для понимания сложности данной задачи можно принять во внимание тот факт, что по тематике исследования нанокатализаторов на основе палладия для ряда каталитических процессов, рассматриваемых в данной диссертационной работе, ежегодно публикуется огромное число научных статей ведущими исследовательскими группами по всему миру. Однако для многих даже относительно простых каталитических реакций механистическое представление их протекания на активных центрах и особенности структуры последних до сих пор остаются предметом активных научных дискуссии. Все это свидетельствует о том, что проблемы, затронутые настоящим исследованием, имеют фундаментальную научную значимость.

Практическая значимость исследования подкрепляется тем фактом, что оно проводилось в тесном сотрудничестве с представителями реального сектора экономики - ведущими мировыми компаниями в области производства и внедрения катализаторов, разработки технологических процессов: Chimet S.p.A. (Италия), Solvay (Бельгия/Китай), Haldor Topsoe (Дания). В рамках сотрудничества, данные предприятия на безвозмездной основе предоставляли реальные промышленные образцы для исследования, участвовали в постановке задач и планировании экспериментальных исследований. Соискатель также является победителем конкурса Фонда содействия развитию малых форм предприятий в научно-технической сфере.

Таким образом, практическая значимость выбранной тематики и перспективы практического использования полученных результатов не вызывают сомнений.

Объекты и методы исследования. В работе исследуются наноструктурированные материалы на основе палладия. Материалы представляют собой преимущественно промышленные образцы, предоставленные крупными мировыми компаниями, занимающимися производством катализаторов (Chimet S.p.A., Solvay), исследование которых

представлено в первом и втором разделах настоящего диссертационного исследования. Выбор индустриальных партнеров и материалов для исследования определялся, во-первых, практической значимостью материалов, используемых в реальных промышленных процессах, во-вторых, возможностью масштабируемого производства и доступности материалов для исследования в количествах, необходимых для проведения многочисленных исследований на источниках синхротронного излучения, и в-третьих, хорошо определенной морфологией и единообразием структуры активных центров, несмотря на промышленные масштабы производства. Во втором разделе исследованы также образцы наночастиц палладия, предоставленные коллегами-химиками из Высшей технологической школы Цюриха (Швейцария). В последнем разделе исследуются новые гетерогенные катализаторы с изолированными центрами палладия, альтернативные своим гомогенным аналогам, на пористых кристаллических подложках, полученные совместно с коллегами-химиками из Католического университета Левена (Бельгия).

Экспериментальные методы исследования основаны на применении синхротронного излучения. В качестве основного метода для определения локальной атомной и электронной структуры активных центров была выбрана спектроскопия рентгеновского поглощения (XAS). Дополнительно использовались рентгеновская дифракция (XRD), малоугловое рассеяние рентгеновских лучей (SAXS), а также широкий спектр лабораторных методов, включая ИК-спектроскопию, просвечивающую электронную микроскопию и другие. Особенностью экспериментальных исследований является использование operando-методологии, т. е. измерение спектральных данных непосредственно в ходе работы катализаторов при реалистичных технологических условиях. Измерения проводились в синхротронных центрах ESRF (Франция), Soleil (Франция), SLS (Швейцария), DESY (Германия), а также в Курчатовском источнике синхротронного излучения (Россия).

Теоретический анализ данных проводился с использованием современного лицензионного программного обеспечения и собственных программных кодов.

Для обработки спектров EXAFS использовался классический Фурье-анализ. XANES область анализировалась путем расчета теоретических спектров из первых принципов методом конечных разностей и последующей подгонки ими экспериментальных спектров с использованием методов машинного обучения. Статистические и хемометрические методы применялись для анализа больших массивов спектральных данных. Теоретические расчеты атомной и электронной структуры катализаторов проводились в рамках теории функционала электронной плотности с использованием метода присоединенных плоских волн, а также подхода молекулярных орбиталей с разложением по базису слетеровского типа.

Положения, выносимые на защиту.

- Разработана комплексная методика орегаи^о-диагностики параметров локальной атомной и электронной структур активных центров наноструктурированных катализаторов на остове одновременного анализа данных рентгеновской спектроскопии и дифракции и результатов теоретического моделирования.

- В условиях протекания каталитической реакции гидрирования этилена происходит формирование фазы карбида палладия в наночастицах палладия. Образование карбидной фазы носит необратимый характер и происходит даже при постоянном избытке молекулярного водорода по отношению к молекулярному этилену.

- Образование карбидной фазы палладия препятствует образованию Р-фазы гидрида палладия. Реакция гидрирования этилена на наночастицах карбида палладия происходит с использованием поверхностно адсорбированного водорода.

- При адсорбции этилена на поверхности наночастиц палладия в диапазоне температур от 20 до 115 °С и атмосферном давлении доминирующим процессом является последовательное дегидрирование этилена в винил, винилиден и этинил с последующим разложением на атомарный углерод. Полученные

производные этилена не участвуют в процессе формирования этана при гидрировании.

- Протекание реакции окисления водорода на поверхности наночастиц палладия приводит к формированию оксида палладия в объеме наночастиц при комнатной температуре. Образование объемной оксидной фазы в аналогичных условиях в чистом кислороде не происходит. Восстановление объемной оксидной фазы в чистом водороде происходит при температурах выше 100 °С.

- Активная фаза катализатора Pd/ШВП в реакции окисления спиртов представлена изолированными Pd2+ центрами и сумнанометровыми кластерами оксида палладия. Агломерация нанокластеров и восстановления палладия до Pd0 приводит к снижению каталитической активности.

- Изолированные четырех-координированные центры палладия, имеющие в ближайшем окружении атомы кислорода, углерода и серы, стабилизированные на пористых кристаллических подложках, являются активными каталитическими центрами в реакциях C-H активации.

Степень достоверности и апробация результатов. Экспериментальные данные были получены на самом современном оборудовании ведущих мировых синхротронных центров: ESRF (Франция), Soleil (Франция), SLS (Швейцария), DESY (Германия), а также в Курчатовском источнике синхротронного излучения (Россия). Теоретический анализ данных проводился либо с использованием современного лицензионного научного программного обеспечения, либо с использованием оригинальных методик и программных кодов, которые также были опубликованы в престижных рецензируемых изданиях и уже получили международное признание. Исследования проводились в сотрудничестве с ведущими международными группами из Университета г. Турина (Италия), Высшей технологической школы Цюриха (Швейцария), Католического университета Левена (Бельгия), Университета Осло (Норвегия). Автор лично представил результаты исследований в качестве устных и приглашенных докладов на многих престижных международных конференциях, преимущественно на английском языке: Европейский конгресс по катализу EuropaCat-2019 (Ахен, Германия), международная конференция по

спектроскопии рентгеновского поглощения XAFS-2018 (Краков, Польша), Международная конференция Operando-VI (Малага, Испания), Международная конференция Faraday Discussions (Лондон, Великобритания), Ежегодные собрания пользователей (Users' Meetings) синхротронных центров ESRF и DESY. Непосредственно по теме представленной диссертации опубликована 41 научная публикация в изданиях, индексируемых в Scopus и Web of Science, из которых 36 - в журналах Q1 и Q2.

Личный вклад. Соискатель внес основной и решающий вклад в планирование и реализацию всех этапов научного исследования за исключением химического синтеза. Автор лично планировал и проводил экспериментальные исследование на источниках синхротронного излучения в России и за рубежом, являлся руководителем заявок на проведение таких исследований. Автор лично проводил обработку и анализ экспериментальных данных, разрабатывал оригинальные программные коды для анализа больших данных и использования алгоритмов машинного обучения в области наук о материалах, проводил сложные квантово-механические расчеты. Является основным автором в подавляющем большинстве статей по тематике исследования. Был руководителем шести научно-исследовательских грантов и ответственным исполнителем в трех грантах, непосредственно связанных с диссертационным исследованием.

РАЗДЕЛ I. Атомная структура гидридных и карбидных фаз наночастиц

палладия

Наночастицы на основе палладия находят широкое применение в промышленно значимых процессах, в качестве катализаторов гидрирования, в том числе гидрирования ненасыщенных углеводородов, таких как алкины, в особенности при необходимости избирательного гидрирования примесей алкинов в потоках алкенов [1-3]. Например, в ходе процесса полимеризации этилена наличие примесей ацетилена в потоке приводит к загрязнению катализатора полимеризации. С другой стороны, необходимо избежать неизбирательного гидрирования этилена в этан - продукта гораздо меньшей экономической стоимости [4]. Благодаря высокой фундаментально научной и практической значимости, пониманию условий и процессов, которые влияют на каталитические свойства палладия посвящено большое число исследований.

Известно, что при взаимодействии с потоками водорода и углеводородов, возможно формирование гидридных и карбидных фаз палладия, что сказывается на его каталитических свойствах [2]. Однако, конкретная роль гидридов и карбидов палладия, как участников каталитического процесса, как и механистическое представление о реакциях гидрирования на поверхности палладия, до сих пор остаются предметом активных обсуждений [1-3, 5-12].

Данный раздел посвящен исследованию динамики наноразмерной структуры промышленных палладиевых катализаторов на углеродной подложке ^/С) и подложке оксида алюминия (Рё/ДЪОз) (Chimet) в ходе реакций гидрирования с использованием передовых экспериментальных возможностей на источниках синхротронного обучения и оригинальных методов анализа рентгеноспектральных данных, отраженных в работах соискателя [А1-А42].

1.1. Необратимое образование карбида палладия в ходе реакции гидрирования этилена

В основу данного подраздела легли два эксперимента, проведенные на станциях BM26 [13] и BM31 [14, 15] синхротронного центра ESRF методами рентгеновской спектроскопии поглощения (XAS) и порошковой рентгеновской дифракции (XRD). Для начала приведем описание типичной экспериментальной установки для проведения каталитического эксперимента в режиме operando, поскольку за исключением незначительных деталей она отражает принципиальную схему проведения большинства экспериментов на источниках синхротронного излучения, представленных в данном диссертационном исследовании.

Ключевым элементом operando-установки является ячейка-микрореактор, позволяющая проводить измерения рентгеновскими методами в контролируемых условиях, в том числе в условиях протекания каталитических реакций. Наиболее часто в данной работе был использован реактор капиллярного типа, позволяющий одновременно проводить измерения рентгеноспектральными и дифракционными методами (см. работы [А3, А5, А7, А9-А12, А14-А17, А20-А24, А26-А27, А30-А32, А36-А39, А41]). Реже использовалась более специализированная каталитическая ячейка, также проточного типа [16] (см. Рисунок 1 и работы [А18, А21, А22]). Также в ряде работ (А28, А40) использовалась ячейка microtomo [17]. Её основным преимуществом являлось использование образца в виде таблетки, что позволяло улучшить соотношение сигнал/шум путем оптимизации толщины образца для рентгеноспектральных измерений. Однако главным недостатком с точки зрения каталитических испытаний являлся большой «мертвый объем». Ячейка подключалась к газовой линии (см. Рисунок 1), оборудованной удаленно управляемыми газовыми расходомерами и электронными кранами. Основными варьируемыми параметрами являлись концентрации и потоки реакционных газов, температура и парциальные и общее давления. Состав газовой смеси после

прохождения через образец контролировался в режиме реального времени средствами масс-спектрометрии. Следует отметить, что отсутствие влияния рентгеновского излучения на ход протекания реакции доказано, во-первых, неизменным составом продуктов в газовой фазе в ходе измерения рентгеновских спектров и при перекрытии рентгеновского пучка, а во-вторых, проведением каталитических тестов с использованием аналогичной установки, но в отсутствии рентгеновского излучения.

Рисунок 1 - Общая схема экспериментальной установки для орвгаЫо-измерений на источниках синхротронного излучения

В ходе эксперимента, наиболее полно описанного в работе [А21], ключевой задачей было определить эволюцию гидридных и карбидных фаз в наночастицах палладия в ходе модельной реакции гидрирования этилена, а также установить виляние этих фаз на каталитические свойства. Для этого при постоянном полном потоке в 50 мл/мин через образец подавалась смесь водорода, этилена, и инертного газа-носителя - гелия (Рисунок 2). Количество водорода оставалось постоянным на уровне 15 мл/мин, а поток этилена увеличивался от 0 до 10 мл/мин с шагом 1 мл/мин, таким образом, чтобы образец всегда находился в избытке водорода.

Рисунок 2 - (а) Изменение газовых потоков, протекающих через образец, в ходе одного цикла и (Ь) конверсия этилена (правая ось ординат), оцененная по данным масс-спектрометрии (левая ось ординат)

Примечательно, что конверсия этилена была ниже при низких значениях отношения потоков этилена к водороду и постепенно росла с увеличением потока этилена. Учитывая, что реакция гидрирования этилена имеет нулевой или слабо-отрицательный порядок по этилену [18], такое поведение свидетельствует о влиянии динамически меняющейся структуры катализатора на его каталитические свойства в ходе реакции. Также можно заметить, что для предварительно гидрированного катализатора (первый цикл), наблюдается более медленное увеличение конверсии этана по сравнению со следующими двумя циклами, что также является неожиданным, поскольку Р-гидрид палладия как правило считается наиболее активным [7, 19].

Анализ спектров БХАББ демонстрирует изменение в межатомных расстояниях палладий-палладий, которое постепенно уменьшается от значения 2,81 А, соответствующего гидриду палладия, до 2,75 А, что незначительно выше значения, определенного для металлического состояния исследуемых наночастиц (Рисунок 3). Уже во втором цикле наблюдаются незначительные отклонения в межатомных расстояниях от значений, полученных при аналогичных условиях в первом цикле. Более существенными эти отклонения становятся в третьем цикле (красная кривая на Рис. 3).

2

2.80 -

*

—О— 1-й цикл -■о—2-й цикл

\1

2.76

2.75

2.74 ^Ц—г

0.0 0.1 0.2 0.3 0.4 0.5 0.6 Отношение СгН^Нз

Рисунок 3- (а) Изменение межатомных расстояний Рё-Рё в наночастицах палладия в ходе трех последовательных циклов каталитической реакции при

Для объяснения причины нарушения повторяемости процесса, были проанализированы спектры ХА^ЕБ. Для этого с использованием метода главных компонент было показано, что вся экспериментальная серия спектров (Рисунок 4а-с) может быть представлена как линейная комбинация трех основных компонент (Рисунок 4ё). Затем, испульзуя сингулярное разложоение, реализованное в коде РуБ1Ш [А29, А42], были получены сами спектры этих трех компонент и их концентрационные профили (Рисунок 4). Исходя из формы спектральных особенностей, теоретический анализ которых был проведен в работах [А1, А5, А7, А9, А15, А16, А30-А32], им были поставлены в однозначное соответствие структуры металлического палладия и его гидридной и карбидной фаз. Как видно из Рисунок 4, небольшая доля карбидной фазы образуется уже в конце первого цикла, однако вначале второго её концентрация падает до нуля. В конце второго цикла образуется большая доля карбидной фазы, и этот процесс происходит необратимо: доля карбида сохраняется даже в чистом водороде в начале второго цикла.

увеличивающемся соотношении С2Н4/Н2

Рисунок 4 - (a-c) Спектры XAS, полученные в ходе трех последовательных циклов при увеличивающемся соотношении С2Щ/Н2 (снизу вверх). (d) Результаты анализа главных компонент серии спектров, показанных на частях

(a-c). (e) Спектры чистых компонент, соответствующих металлическому палладию (черный пунктир), гидриду палладий (сплошной синий) и карбиду палладия (точечный красный) и (f) их концентрационные профили в ходе трех последовательных циклов каталитической реакции при увеличивающемся

соотношении C2H4/H2

Таким образом, было показано, что в присутствии этилена карбидная фаза образуется необратимо даже при непрерывном избытке молекулярного водорода [Л21]. Формирование карбидной фазы объясняет нарушение повторяемости в поведении межатомных расстояний от цикла к циклу. Повышенная конверсия этилена в этан при формировании карбидной фазы, свидетельствует о том, что

наличие доли карбидной фазы (возможно, связанной с поверхностными связями Pd-C) необходимо для протекания данной реакции. Исследования были продолжены с использованием одновременно с XAS методики XRD, чувствительной к кристаллическому строению ядра наночастиц. В работе [А39] был подтвержден факт необратимого формирования карбидов в условиях реакции, вплоть до образования полного карбида палладия, а с другой стороны, было обнаружено несколько различных карбидных фаз, отличающихся по параметру решетки. Например, в работе [А40] методика теоретического моделирования спектральных особенностей [A6, A25], связанных с возникновением химической связи Pd-X (X - легкий примесный атом), была успешно расширена для случая образования связей Ru-N на поверхности наночастиц рутения в ходе реакции аминирования спиртов, а в работе A44 успешно исследованы гомогенные катализаторы на основе Ru-X (X=Br,Cl,CO).

1.2. Роль поверхностных и объемных фаз в реакции гидрирования

В рамках данного подраздела была исследована структура потенциальных промежуточных состояний реакции гидрирования, и установлена роль поверхностных и объемных гидридных и карбидных фаз в ходе гидрирования этилена с использованием рентгеновской и ИК-спектроскопий с высоким разрешением по времени. Отличительной особенностью проведенного исследования является концепция последовательного пропускания через образец реакционных газов без непосредственного их смешивания в газовой фазе. Таким образом образование продуктов реакции остается возможным за счет адсорбции реагентов на поверхности катализатора, либо из аккумуляции в его объеме. Данные масс-спектрометрии (Рисунок 5а) свидетельствует о том, что в ходе каждого переключения между потоками водорода и этилена, наблюдается сигнал этана - продукта реакции. При этом интенсивность и форма наблюдаемого сигнала зависит от температуры и направления переключения, что свидетельствует о роли катализатора в данном процессе. На Рис. 5b показана эволюция трех компонент, полученных по методике, описанной в предыдущем

Список литературы диссертационного исследования доктор наук Бугаев Арам Лусегенович, 2021 год

- ■ ■

_ V fl I 1 I 1 1 1 - 1 1 i 1 i ■ 1 1

-60 -40 -20

20 40 60 Time {min)

100 120 140

Fig. 5 Evolution of structural parameters obtained from XRPD, EXAFS and DXANES analysis. Left ordinate axis: variation of the RPd-Pd distance (black squares) obtained from the first-shell EXAFS-analysis and of the average lattice parameter obtained from XRPD refinement (gray circles). Right ordinate axis: stoichiometry of the PdCy (in %) phase, determined by using DXANES modelling and fitting (orange triangles). Exposure of the sample to pure H2, C2H2 and vacuum are highlighted by light-blue, light-red and white areas respectively. In the region below t = 0, the sample was exposed to 1 bar of a hydrogen/acetylene mixture in a 2 : 1 stoichiometric ratio (green area). All the reported experiments have been performed with a sample temperature of 100 °C.

The removal of H2 from the feed mixture (first light-red part in Fig. 5), leads to a complete decomposition of the hydride phase, as all hydride features disappear from both the XANES and DXANES spectra. However, the interatomic distances are higher by 0.5% than in pure Pd NPs. This increase is explained by the formation of the second phase with an increased lattice parameter observed in the XRPD patterns. According to the shape of the DXANES curves this phase was assigned to palladium carbide. There is a slow increase of the lattice parameter of the carbidized sample with progressive C2H2 exposure time at 100 °C (light-red parts in Fig. 5) and that it is not reversed either by successive vacuum or treatment in H2, see Fig. 5 white and light-blue parts, respectively.

To determine the evolution of the y stoichiometry in the PdC^ carbide phase, all experimental DXANES spectra were fitted by the theoretical ones applying a multidimensional interpolation approach. A set of model structures with different y values and lattice parameters (thus different RPd-Pd = a/^2) were initially optimized with the help of the VASP 5.3 code53-55 and used for XANES calculation with the FDMNES code,36-40 as described in Section 2.4. The calculated spectra were then taken as interpolation nodes in the two-dimensional (RPd-Pd, y) space, and used for the construction of a polynomial which describes the shape of the DXANES spectra for any of the RPd-Pd and y values. At the first step, we fitted the experimental XANES spectrum of Pd NPs under vacuum, and used the best fit theoretical spectrum to construct theoretical DXANES curves. All other spectra were fitted by minimizing the root-mean-square deviation F(RPd-Pd, y) between the

theoretical and experimental DXANES spectra, varying RPd-Pd and y values defined

F(Rpd-pd, y) —

A

^[AXANESexp(E,) - AXANEStheo(E;,Rpd-pd, y)]

(3)

where E, are the energy values where the experimental curves have been sampled, E1 = 24 340 eV and EN — 24 440 eV are the first and the last experimental points considered in the fit and N is the total number of experimental points. The 2D distributions of F(RPd-Pd, y) for selected spectra are shown in Fig. 6. For the spectrum taken in H2 (t — -22 min), the minimum of F is achieved for the increased RPd-Pd and indicates zero carbon incorporation (y — -0.01, i.e. y — 0 within the experimental incertitude). This result represents a consistence test of the adopted method, as no Pd-carbide phase is expected to be formed in these conditions.12 An increase of the carbon incorporation is observed after acetylene exposure and the position of the minimum of F shifts towards higher y values from the spectrum taken at t — 0, to that taken at t — 120 min, where the time corresponds to the starting time of each spectrum. The only deviation from the increasing y trend is observed after H2 treatment of the sample (t — 81 min), which leads to a decrease of y by a factor of 2. This difference is observed only in the XANES spectra, while the EXAFS and XRPD values are not affected by H2

1

i= 1

y X 100 y x 100 y X 100

Fig. 6 2D plots of the F(RPd-Pd, y) root-mean-square deviation function between theoretical and experimental DXANES spectra, defined in eqn (3) for different experimental curves collected at a time indicated in the bottom left corner and referred to the feeding conditions defined in Fig. 5. The white dotted lines highlight the position of the minimum for each spectrum corresponding to the best estimation for RPd-Pd and y values of the PdCy phase in the corresponding experimental conditions. The used color scale is quantified in the right panel. The fact that there is one order of magnitude in the intensity differences between the blue and red regions implies that in all cases we are dealing with quite stable minima.

treatment. The reason for such behavior is that the XANES spectra are sensitive not only to carbon atoms, which are inserted in the interstitials of the palladium lattice forming a palladium carbide phase, but also the surface adsorbed acetylene molecules, which do not affect the interatomic distance while they contribute to the number of Pd-C bonds. H2 treatment can be therefore used to remove the surface acetylene through its hydrogenation to ethylene and ethane, while it does not remove the carbon atoms from the palladium lattice.15 This unambiguously shows that the use of XANES spectra, in addition to EXAFS and XRPD, allows the extraction of information on the structure of surface atoms, in addition to the core-shell structure, which is revealed by the combination of the latter two,13 see Section 3.1.

In this case XANES was used as a completely independent technique that provides both structure (RPd-Pd), and stoichiometry (y) of the PdC^ phase. However, given that Fourier-analysis of the EXAFS data has been performed, the RPd-Pd values can be fixed to those obtained by EXAFS analysis. For the current data set, as can be seen from Fig. 6, F(RPd-Pd, y) functions are symmetric with respect to horizontal and vertical dashed lines passing through the minimal point of each spectrum. Thus, for this particular case, having an error in the determination of the RPd-Pd value does not affect the position of the minimum along the y axis, while in the case of an asymmetric distribution of the F function, the use of EXAFS interatomic distances would have been more important.

3.3. Operando XRPD study during the hydrogénation of ethylene: evidence for structural and catalytic oscillating behavior

Operando XRPD patterns were collected every 5 s under steady state gas feeding to the Pd/C catalyst (6, 39 and 5 ml min-1 for He, H2 and C2H4, respectively) allowing monitoring of the structural parameters of the core of the Pd NP during ethylene hydrogenation at 80 °C, see Fig. 7. Part (a) reports the 26 region covering the (220), (311) and (222) Bragg reflections of palladium for a series of 10 subsequent diffraction patterns (from t — 100 to t — 150 s). These data indicate that, during steady state feeding conditions, the core structure of the Pd NPs is not stable, but changes with time. Fig. 7b, left ordinate axis, reports the time evolution of the average lattice parameter, obtained using eqn (1). The structural changes are not random, but follow a clear oscillating behavior with a period of about 150 s. The periodic lattice parameter variation is due to the periodic change in the relative fraction of the b- and a-phases of palladium hydride (see Section 3.1 in general and Fig. 2a in particular) and here reported in Fig. 7b, right ordinate axis. The most interesting aspect of this experiment is the fact that the same periodic oscillations of non-regular shape were observed by using MS monitoring for both the reactants and products, see Fig. 7c. Indeed, the MS signal of the C2H6 product (m/Z — 30) is in perfect phase with the oscillation of the fraction of the b-phase of palladium hydride obtained by using XRPD, while the MS signal of the H2 reac-tant (m/Z — 2) is in perfect antiphase. Although Fig. 7b and c report only 10 min, the oscillations were stable for 30 minutes, until we changed the feed.

This evidence is a direct undisputed proof of a strong structure-reactivity relationship between the structure of the core of the PdHx NPs and the catalyst activity in the ethylene hydrogenation reaction. Also evident is the fact that the catalyst is more active in the H2 + C2H4 / C2H6 reaction when the core of the

Fig. 7 Time resolved, operando XRPD study during the ethylene hydrogenation reaction at 80 °C on a Pd/C catalyst performed under steady state feeding conditions: (6, 39 and 5 ml min-1 for He, H2 and C2H4, respectively). Part (a): selection of XRPD patterns at significant times (from t — 100 s, black curve, to t — 150 s, red curve), evidencing structural changes along the reaction monitored in the 26 range of the most significant Pd Bragg reflections. Part (b): time evolution of the averaged Pd lattice parameter, see eqn (1), and of the fraction of the b-phase of PdHx left and right ordinate axis, respectively. Part (c): time evolution of the catalyst activity monitored by using MS showing a reactant (H2, m/Z — 2) and a product (C2H6, m/Z — 30).

PdHx NPs is in the b-phase, in agreement with previous findings.14 However, the interpretation of the observed oscillating behavior from an atomistic point of view is not straightforward.The set of experimental data summarized in Fig. 7 can be rationalized by the considerations discussed in the following.

In the case of NPs fully in the b-phase (e.g. at t — 200 s in Fig. 7b and c), catalyst activity is maximal. Owing to the exothermicity of the ethylene hydrogenation reaction (DH0 — -136 kJ mol-1),84'85 the NPs undergo a local temperature increase, which results in a partial hydrogen desorption (see Fig. 2), that in turn results in a decrease of the fraction of the b-phase, down to 70% (at t — 270 s). The lower availability of atomic hydrogen in the NP core slows down the reaction, which drives a local temperature decrease, further slowing the reaction and increasing the average H2 partial pressure in the feed. These facts reform the b-phase restoring the higher catalytic activity (t — 355 s). The fact that heating of the NPs plays a role is indicated in that after completing the b-phase formation, there is a continued increase in activity, indicated by the growing ethane signal.

4. Conclusions

In the present study, we have combined synchrotron-based in situ and operando, almost simultaneous, X-ray diffraction and absorption data collection to laboratory volumetric measurements to shed light on the structure and the stoichi-ometry of PdHx and PdC^ phases of Pd NPs during hydrocarbon hydrogenation reactions on a Pd/C catalyst. Six main results have been achieved.

First, the systematic in situ XRPD, EXAFS and volumetric analysis in a wide range of sample temperatures and H2 equilibrium pressures, allowed us to follow

the a-b phase transition diagram of Pd NPs. The structural/stoichiometric ab phase diagram reported in Fig. 2a-c allowed us to determine the PdHx stoi-chiometry from an EXAFS or an XRPD structural datum. Second, the almost simultaneous EXAFS and XRPD set of data allowed us to discriminate the ordered NP core from the disordered shell and to reconstruct the RPd-Pd distance at the amorphous surface of the NP, i.e. the RPd-Pd of the actual active phase, where hydrogenation reactions occur (Fig. 3). Third, while both XRPD and EXAFS are unable to discriminate between palladium hydride and carbide phases, XANES provides unambiguous detection of hydride and carbide phases (Fig. 4a), with the discrimination ability being more evident when the data are reported in difference mode (DXANES, Fig. 4b). Fourth, the combined use of XRPD, EXAFS and XANES, supported by corresponding simulations,86 allowed us to obtain both the core and shell structures and the average y stoichiometry of the PdC^ phase formed upon exposure of Pd NPs to C2H2 for increasing times (Fig. 5 and 6). Fifth, advanced analysis of the DXANES spectra allows the detection of carbon-containing molecules adsorbed at the surface of the NPs. Finally, the collection of operando XRPD patterns allowed us to observe, during the ethylene hydrogenation reaction, periodic oscillations of a non-regular shape of the NPs core lattice parameter, that resulted to be in phase with the MS signal of the C2H6 product and in antiphase with the MS signal of the H2 reactant (Fig. 7), highlighting an interesting direct structure-reactivity relationship.

A repetition rate of 0.2 Hz (one XRPD pattern every 5 s) was needed to follow the structural oscillation of the core. Palladium hydride phase changes are observed in parallel with the ethylene hydrogenation activity. Local temperature and hydrogen concentrations drive the formation and disappearance of the b-phase and catalytic activity.

Acknowledgements

A. L. B., O. A. U., A. A. G., A. V. S., and C. L. acknowledge the Russian Ministry of Education and Science for financial support (Project RFMEFI58417X0029, Agreement 14.584.21.0029). We are indebted to Vladimir Dmitriev, Herman Emerich, Wouter van Beek and Michela Brunelli for their friendly and competent support during the experiments performed at the BM01B (now BM31) beamline of the ESRF.

Notes and references

1 A. Molnar, A. Sarkany and M. Varga, J. Mol. Catal. A: Chem., 2001,173,185-221.

2 A. Borodzinki, Catal. Rev.: Sci. Eng., 2006, 48, 91-144.

3 A. Borodzinski and G. C. Bond, Catal. Rev.: Sci. Eng., 2008, 50, 379-469.

4 M. Armbruster, M. Behrens, F. Cinquini, K. Fottinger, Y. Grin, A. Haghofer,

B. Klotzer, A. Knop-Gericke, H. Lorenz, A. Ota, S. Penner, J. Prinz,

C. Rameshan, Z. Revay, D. Rosenthal, N. Rupprechter, P. Sautet, R. Schlogl, L. D. Shao, L. Szentmiklosi, D. Teschner, D. Torres, R. Wagner, R. Widmer and G. Wowsnick, ChemCatChem, 2012, 4, 1048-1063.

5 J. A. McCaulley, J. Phys. Chem., 1993, 97, 10372-10379.

6 A. V. Soldatov, S. Dellalonga and A. Bianconi, Solid State Commun., 1993, 85, 863-868.

7 S. Kishore, J. A. Nelson, J. H. Adair and P. C. Eklund, J. Alloys Compd., 2005, 389, 234-242.

8 M. Yamauchi, R. Ikeda, H. Kitagawa and M. Takata, J. Phys. Chem. C, 2008,112, 3294-3299.

9 C. Langhammer, E. M. Larsson, B. Kasemo and I. Zoric, Nano Lett., 2010, 10, 3529-3538.

10 T. Shegai and C. Langhammer, Adv. Mater., 2011, 23, 4409-4414.

11 A. L. Bugaev, A. A. Guda, K. A. Lomachenko, V. V. Srabionyan, L. A. Bugaev, A. V. Soldatov, C. Lamberti, V. P. Dmitriev and J. A. van Bokhoven, J. Phys. Chem. C, 2014, 118, 10416-10423.

12 A. L. Bugaev, A. A. Guda, A. Lazzarini, K. A. Lomachenko, E. Groppo, R. Pellegrini, A. Piovano, H. Emerich, A. V. Soldatov, L. A. Bugaev, V. P. Dmitriev, J. A. van Bokhoven and C. Lamberti, Catal. Today, 2017, 283, 119-126.

13 A. L. Bugaev, A. A. Guda, K. A. Lomachenko, V. V. Shapovalov, A. Lazzarini, J. G. Vitillo, L. A. Bugaev, E. Groppo, R. Pellegrini, A. V. Soldatov, J. A. van Bokhoven and C. Lamberti, J. Phys. Chem. C, 2017, 121, 18202-18213.

14 D. Teschner, J. Borsodi, A. Wootsch, Z. Revay, M. Havecker, A. Knop-Gericke, S. D. Jackson and R. Schlogl, Science, 2008, 320, 86-89.

15 M. W. Tew, M. Nachtegaal, M. Janousch, T. Huthwelker and J. A. van Bokhoven, Phys. Chem. Chem. Phys., 2012, 14, 5761-5768.

16 D. Teschner, Z. Revay, J. Borsodi, M. Havecker, A. Knop-Gericke, R. Schlogl, D. Milroy, S. D. Jackson, D. Torres and P. Sautet, Angew. Chem., Int. Ed., 2008, 47, 9274-9278.

17 D. Teschner, J. Borsodi, Z. Kis, L. Szentmiklosi, Z. Revay, A. Knop-Gericke, R. Schlogl, D. Torres and P. Sautet, J. Phys. Chem. C, 2010, 114, 2293-2299.

18 M. W. Tew, M. Janousch, T. Huthwelker and J. A. van Bokhoven, J. Catal., 2011, 283, 45-54.

19 W. Vogel, W. He, Q. H. Huang, Z. Q. Zou, X. G. Zhang and H. Yang, Int. J. Hydrogen Energy, 2010, 35, 8609-8620.

20 Z. A. Chase, J. L. Fulton, D. M. Camaioni, D. H. Mei, M. Balasubramanian, V. T. Pham, C. Zhao, R. S. Weber, Y. Wang and J. A. Lercher, J. Phys. Chem. C, 2013, 117, 17603-17612.

21 D. G. Narehood, S. Kishore, H. Goto, J. H. Adair, J. A. Nelson, H. R. Gutierrez and P. C. Eklund, Int. J. Hydrogen Energy, 2009, 34, 952-960.

22 A. Filipponi, A. DiCicco and C. R. Natoli, Phys. Rev. B: Condens. Matter Mater. Phys., 1995, 52, 15122-15134.

23 J. J. Rehr and R. C. Albers, Rev. Mod. Phys., 2000, 72, 621-654.

24 J. A. van Bokhoven and C. Lamberti, X-Ray Absorption and X-Ray Emission Spectroscopy: Theory and Applications, John Wiley & Sons, Chichester, UK, 2016.

25 C. Lamberti, Surf. Sci. Rep., 2004, 53, 1-197.

26 S. Bordiga, E. Groppo, G. Agostini, J. A. van Bokhoven and C. Lamberti, Chem. Rev., 2013, 113, 1736-1850.

27 L. Mino, G. Agostini, E. Borfecchia, D. Gianolio, A. Piovano, E. Gallo and C. Lamberti, J. Phys. D: Appl. Phys., 2013, 46, 72.

28 C. Garino, E. Borfecchia, R. Gobetto, J. A. van Bokhoven and C. Lamberti, Coord. Chem. Rev., 2014, 277, 130-186.

29 C. Lamberti and J. A. van Bokhoven, in X-Ray Absorption and X-Ray Emission Spectroscopy: Theory and Applications, ed. J. A. van Bokhoven and C. Lamberti, John Wiley & Sons, Chichester, UK, 2016, vol. II, ch. 13, pp. 353-383.

30 A. Bianconi, J. Garcia and M. Benfatto, Top. Curr. Chem., 1988, 145, 29-67.

31 M. Benfatto and S. Delia Longa, J. Synchrotron Radiat., 2001, 8, 1087-1094.

32 M. Benfatto, S. Delia Longa and C. R. Natoli, J. Synchrotron Radiat., 2003, 10, 51-57.

33 C. R. Natoli, M. Benfatto, S. Della Longa and K. Hatada, J. Synchrotron Radiat., 2003, 10, 26-42.

34 A. L. Ankudinov, B. Ravel, J. J. Rehr and S. D. Conradson, Phys. Rev. B: Condens. Matter Mater. Phys., 1998, 58, 7565-7576.

35 J. J. Rehr and A. L. Ankudinov, Coord. Chem. Rev., 2005, 249, 131-140.

36 Y. Joly, Phys. Rev. B: Condens. Matter Mater. Phys., 2001, 63, 125120.

37 O. Bunau and Y. Joly, J. Phys.: Condens. Matter, 2009, 21, 345501.

38 Y. Joly and S. Grenier, in X-Ray Absorption and X-Ray Emission Spectroscopy: Theory and Applications, ed. J. A. van Bokhoven and C. Lamberti, John Wiley & Sons, Chichester, UK, 2016, vol. I, ch. 4, pp. 73-97.

39 S. A. Guda, A. A. Guda, M. A. Soldatov, K. A. Lomachenko, A. L. Bugaev, C. Lamberti, W. Gawelda, C. Bressler, G. Smolentsev, A. V. Soldatov and Y. Joly, J. Chem. Theory Comput., 2015, 11, 4512-4521.

40 A. A. Guda, S. A. Guda, M. A. Soldatov, K. A. Lomachenko, A. L. Bugaev, C. Lamberti, W. Gawelda, C. Bressler, G. Smolentsev, A. V. Soldatov and Y. Joly, J. Phys.: Conf. Ser., 2016, 712, 012004.

41 A. L. Bugaev, V. V. Srabionyan, A. V. Soldatov, L. A. Bugaev and J. A. van Bokhoven, J. Phys.: Conf. Ser., 2013, 430, 012028.

42 G. Poncelet, P. Jacobs and P. Grange, Preparation of catalysts III, Elsevier, 1983.

43 G. Agostini, C. Lamberti, R. Pellegrini, G. Leofanti, F. Giannici, A. Longo and

E. Groppo, ACS Catal., 2014, 4, 187-194.

44 A. Piovano, A. Lazzarini, R. Pellegrini, G. Leofanti, G. Agostini, S. Rudic, A. L. Bugaev, C. Lamberti and E. Groppo, Adv. Condens. Matter Phys., 2015, 2015, 803267.

45 W. van Beek, O. V. Safonova, G. Wiker and H. Emerich, Phase Transitions, 2011, 84, 726-732.

46 P. M. Abdala, O. V. Safonova, G. Wiker, W. van Beek, H. Emerich, J. A. van Bokhoven, J. Sa, J. Szlachetko and M. Nachtegaal, Chimia, 2012, 66, 699-705.

47 C. W. Andersen, E. Borfecchia, M. Bremholm, M. R. V. Jorgensen, P. N. R. Vennestrom, C. Lamberti, L. F. Lundegaard and B. B. Iversen, Angew. Chem., Int. Ed., 2017, 56, 10367-10372.

48 E. Groppo, G. Agostini, A. Piovano, N. B. Muddada, G. Leofanti, R. Pellegrini, G. Portale, A. Longo and C. Lamberti, J. Catal., 2012, 287, 44-54.

49 C. Lamberti, S. Bordiga, F. Bonino, C. Prestipino, G. Berlier, L. Capello,

F. D'Acapito, F. Xamena and A. Zecchina, Phys. Chem. Chem. Phys., 2003, 5, 4502-4509.

50 B. Ravel and M. Newville, J. Synchrotron Radiat., 2005, 12, 537-541.

51 J. J. Kas, A. P. Sorini, M. P. Prange, L. W. Cambell, J. A. Soininen and J. J. Rehr, Phys. Rev. B: Condens. Matter Mater. Phys., 2007, 76, 195116.

52 S. I. Zabinsky, J. J. Rehr, A. Ankudinov, R. C. Albers and M. J. Eller, Phys. Rev. B: Condens. Matter Mater. Phys., 1995, 52, 2995-3009.

53 G. Kresse and J. Furthmuller, Phys. Rev. B: Condens. Matter Mater. Phys., 1996, 54, 11169-11186.

54 G. Kresse and D. Joubert, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 59, 1758-1775.

55 J. Hafner, J. Comput. Chem., 2008, 29, 2044-2078.

56 J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865-3868.

57 G. Smolentsev and A. Soldatov, J. Synchrotron Radiat., 2005, 13, 19-29.

58 G. Smolentsev and A. V. Soldatov, Comput. Mater. Sci., 2007, 39, 569-574.

59 J. Kieffer and J. P. Wright, Powder Diffr., 2013, 28, S339-S350.

60 V. Petneek, M. Dusek and L. Palatinus, Z. Kristallogr., 2014, 229, 345-352.

61 F. A. Lewis, Platinum Met. Rev., 1982, 26, 70-78.

62 C. Langhammer, V. P. Zhdanov, I. Zorie and B. Kasemo, Chem. Phys. Lett., 2010, 488, 62-66.

63 B. Ingham, M. F. Toney, S. C. Hendy, T. Cox, D. D. Fong, J. A. Eastman, P. H. Fuoss, K. J. Stevens, A. Lassesson and S. Brown, Phys. Rev. B: Condens. Matter Mater. Phys., 2008, 78, 245408.

64 R. Bardhan, L. O. Hedges, C. L. Pint, A. Javey, S. Whitelam and J. J. Urban, Nat. Mater., 2013, 12, 905-912.

65 T. C. Narayan, A. Baldi, A. L. Koh, R. Sinclair and J. A. Dionne, Nat. Mater., 2016, 15, 768-776.

66 G. L. Chiarello and D. Ferri, Phys. Chem. Chem. Phys., 2015, 17, 10579-10591.

67 D. Ferri, M. S. Kumar, R. Wirz, A. Eyssler, O. Korsak, P. Hug, A. Weidenkaffand M. A. Newton, Phys. Chem. Chem. Phys., 2010, 12, 5634-5646.

68 D. Ferri, M. A. Newton, M. Di Miehiel, G. L. Chiarello, S. Yoon, Y. Lu and J. Andrieux, Angew. Chem., Int. Ed., 2014, 53, 8890-8894.

69 A. I. Frenkel, C. W. Hills and R. G. Nuzzo, J. Phys. Chem. B, 2001, 105, 1268912703.

70 A. I. Frenkel, A. Yeviek, C. Cooper and R. Vasie, Annu. Rev. Anal. Chem., 2011, 4, 23-39.

71 A. I. Frenkel, Chem. Soc. Rev., 2012, 41, 8163-8178.

72 G. Agostini, R. Pellegrini, G. Leofanti, L. Bertinetti, S. Bertarione, E. Groppo, A. Zeeehina and C. Lamberti, J. Phys. Chem. C, 2009, 113, 10485-10492.

73 G. Agostini, E. Groppo, A. Piovano, R. Pellegrini, G. Leofanti and C. Lamberti, Langmuir, 2010, 26, 11204-11211.

74 G. Agostini, A. Piovano, L. Bertinetti, R. Pellegrini, G. Leofanti, E. Groppo and C. Lamberti, J. Phys. Chem. C, 2014, 118, 4085-4094.

75 M. Manzoli, F. Vindigni, T. Tabakova, C. Lamberti, D. Dimitrov, K. Ivanov and G. Agostini, J. Mater. Chem. A, 2017, 5, 2083-2094.

76 A. L. Bugaev, A. A. Guda, K. A. Lomaehenko, A. Lazzarini, V. V. Srabionyan, J. G. Vitillo, A. Piovano, E. Groppo, L. A. Bugaev, A. V. Soldatov, V. P. Dmitriev, R. Pellegrini, J. A. van Bokhoven and C. Lamberti, J. Phys.: Conf. Ser., 2016, 712, 012032.

77 R. Campesi, F. Cuevas, R. Gadiou, E. Leroy, M. Hirseher, C. Vix-Guterl and M. Latroehe, Carbon, 2008, 46, 206-214.

78 T. B. Flanagan and W. A. Oates, Annu. Rev. Anal. Chem., 1991, 21, 269-304.

79 R. Griessen, N. Strohfeldt and H. Giessen, Nat. Mater., 2016, 15, 311-317.

80 A. Zuttel, C. Nutzenadel, G. Sehmid, C. Emmenegger, P. Sudan and L. Sehlapbaeh, Appl. Surf. Sci., 2000, 162, 571-575.

81 A. Baldi, T. C. Narayan, A. L. Koh and J. A. Dionne, Nat. Mater., 2014,13,11431148.

82 S. Syrenova, C. Wadell, F. A. A. Nugroho, T. A. Gschneidtner, Y. A. D. Fernandez, G. Nalin, D. Switlik, F. Westerlund, T. J. Antosiewicz, V. P. Zhdanov, K. Moth-Poulsen and C. Langhammer, Nat. Mater., 2015, 14, 1236-1244.

83 C. Zlotea, F. Cuevas, V. Paul-Boncour, E. Leroy, P. Dibandjo, R. Gadiou, C. Vix-Guterl and M. Latroche, J. Am. Chem. Soc., 2010, 132, 7720-7729.

84 G. B. Kistiakowsky and A. G. Nickle, Discuss. Faraday Soc., 1951, 10, 175-187.

85 G. B. Kistiakowsky, H. Romeyn, J. R. Ruhoff, H. A. Smith and W. E. Vaughan, J. Am. Chem. Soc., 1935, 57, 65-75.

86 A. L. Bugaev, O. A. Usoltsev, A. A. Guda, K. A. Lomachenko, I. A. Pankin, Y. V. Rusalev, H. Emerich, E. Groppo, R. Pellegrini, A. V. Soldatov, J. A. van Bokhoven and C. Lamberti, J. Phys. Chem. C, 2018, 122, DOI: 10.1021/ acs.jpcc.7b11473.

ICatalvsis

© Cite This: ACS Catal. 2018, 8, 6870-6881

Research Article

pubs.acs.org/acscatalysis

Dynamic Behavior of Pd/P4VP Catalyst during the Aerobic Oxidation of 2-Propanol: A Simultaneous SAXS/XAS/MS Operando Study

Elena Groppo,*^ Andrea Lazzarini,1^® Michele Carosso,1 Aram Bugaev,§® Maela Manzoli,"® Riccardo Pellegrini,^ Carlo Lamberti,§,#® Dipanjan Banerjee,V and Alessandro Longo*,•

^Department of Chemistry, INSTM and NIS Centre, University of Turin, via Quarello 15, Turin I-10135, Italy

^Centre for Materials Science and Nanotechnology, Department of Chemistry, University of Oslo, Sem Saelands vei 26, Oslo N-0315, Norway

§The Smart Materials Research Center, Southern Federal University, Zorge Street 5, Rostov-on-Don 344090, Russia "Department of Drug Science and Technology, NIS Centre and INSTM, University of Turin, Via Pietro Giuria 9, Turin I-10125, Italy

^Chimet SpA - Catalyst Division, Via di Pescaiola 74, Viciomaggio Arezzo I-52041, Italy

#Department of Physics and CrisDi Interdepartmental Centre, University of Turin, via Pietro Giuria 1, Turin 10125, Italy VDepartment of Chemistry, KU Leuven, Celestijnenlaan 200F box 2404, Leuven 3001, Belgium •Netherlands Organization for Scientific Research at ESRF, BP 220, Grenoble F-38043 Cedex 9, France

^ Supporting Information

ABSTRACT: The behavior of a Pd(OAc)2/P4VP catalyst submitted to different pretreatments (prereduced, preoxidized, and untreated) during the aerobic oxidation of 2-propanol to acetone in the gas phase has been investigated. Synchronous, time-resolved, SAXS/XAS/MS techniques coupled with operando DRIFT spectroscopy (which gave information on the destiny of the acetate ligands) and ex situ HR-TEM (to detect the formation of Pd nanoparticles and to obtain their size distribution) were employed to accomplish a dynamical picture of the changes occurring to the Pd phase under transient reaction conditions. In addition, the catalytic performances were qualitatively explored by means of a CATLAB microreactor, with the final aim to establish structure—activity relationships. Our approach clearly demonstrates that highly isolated Pd2+ cationic species, either atomically dispersed or in the form of ultrasmall Pd2+—oxo clusters, are efficient and very stable active sites for the gas-phase aerobic oxidation of 2-propanol to acetone. Noticeably, the behavior of the Pd(OAc)2/P4VP catalyst in reaction conditions is influenced by the nature of the support. On one hand, the presence of the pyridyl functional groups is fundamental to stabilize the cationic Pd2+ species; on the other hand, the porous structure of the P4VP polymer efficiently confines the active Pd2+ species in the presence of the reagents. As such, our catalyst is situated at the confluence between its homogeneous and heterogeneous analogues.

KEYWORDS: alcohol oxidation, palladium, SAXS, XAS, operando

1. INTRODUCTION

Selective aerobic oxidation of alcohols to their corresponding

aldehydes over noble-metal-based catalysts is an environmentally benign process in fine chemistry, but also a reaction particularly demanding, because it requires the activation of molecular oxygen and C—O bonds in close proximity, at temperatures typically below 160 °C.1—5 The economic and

environmental advantages of molecular oxygen as a chemical

oxidant are readily apparent: oxygen is abundant, inexpensive, and thermodynamically potent. However, effective solutions to this problem must overcome the intrinsic reactivity and selectivity challenges posed by the chemistry of O2: O2 is a four-electron oxidant when it is reduced to water but most desired reactions are 2-electron oxidations, and partially reduced oxygen species are typically more reactive and potent

oxidants than O2 itself. — Both heterogeneous and homogeneous Pd-based catalysts are largely employed in selective alcohol oxidations, on account of the Pd ability to perform selective oxidations at temperatures typically between 60 and 160 °C and atmospheric oxygen pressure.1-6 Albeit significant progress has been achieved in understanding the role of Pd-based catalysts by using in situ or operando spectroscopic10-12 and microscopic13'14 tools, the mechanism of the reaction is still a matter of discussion.15-17

It has long been accepted that the reaction proceeds following an oxidase-style mechanism consisting of two steps:

Received: April 11, 2018 Revised: June 8, 2018 Published: June 13, 2018

ACS Publications © 2018 American Chemical Society

6870

(l) a Pd-mediated oxidation of the alcohol by dehydrogen-ation, with the formation of the corresponding aldehyde and of a Pd-hydride intermediate; and (2) the aerobic oxidation of the reduced catalyst. The two stages occur independently and in sequence.1- ^ ,19 According to this mechanism, molecular oxygen is not directly involved in the substrate oxidation, but has the double role of reoxidizing the Pd-hydrides (supported by the observation that O2 can be replaced by a hydrogen acceptor),2^21 and suppressing the decarbonylation of the oxidation products over metallic Pd.1-5 Competing reactions may occur at specific surface sites19 during the catalytic cycles (such as the decarbonylation or the overoxidation of the products, the overoxidation of the catalyst, and the aggregation of Pd species to inactive bulk metal), with a consequent decrease in selectivity and deactivation of the catalyst.

In both homogeneous and heterogeneous cases, the unambiguous identification of the oxidation state (Pd(ll) or Pd(0)) and of the aggregation of the Pd species during the catalysis has a pivotal importance. The aerobic oxidation of alcohols was investigated both in the liquid phase (aqueous

media or organic solvents)

and in the gas phase, ' by

X-ray Absorption Spectroscopy (XAS) coupled with vibrational spectroscopies and/or electrochemical measurements. For Pd-based heterogeneous catalysts, it was discovered earlier that the rate of alcohol oxidation is higher on a reduced metal surface than on an oxidized one. ' ' Several works indicate that metallic Pd is the catalytically active phase. Interestingly, in most of these studies, it was also reported that the introduction of oxygen in the reactant feed causes a sudden increase of the alcohol conversion, an observation that was attributed to a cleaning of the Pd surface by oxygen, rather than to a change in the Pd oxidation state. Other studies30-35 indicate that surface PdO is the active phase in alcohol oxidation reactions, and that the oxide-to-metal structural transition is accompanied by catalyst deactivation through secondary decarbonylation of the products.

The nature of the active Pd sites during alcohol oxidation is similarly uncertain also for homogeneous Pd-based cata-

lysts.

6,36-41

For the Pd(OAc)2/pyridine system, Steinhoff et

al. detected (by means of in situ spectroscopy) a Pd(ll) resting state.17'42'43 However, Uemura et al.44 proposed an alternative hypothesis, wherein Pd remains in the +2 oxidation state throughout the catalytic cycle. The above-mentioned Pd-(OAc)2/pyridine homogeneous system is one of the most efficient and selective catalysts for oxidation of alcohols to

aldehydes.

17,36,42-46

All the major classes of alcohols (primary,

secondary, benzylic, and allylic) are oxidized in toluene solution at 80 °C, generally in good-to-excellent yields (80-

100%).36 In this system, the labile pyridine ligands greatly facilitate the reductive elimination of the aldehyde from the Pd-H intermediate and therefore significantly enhance the rate of Pd(0) oxidation by molecular oxygen. When the efficiency of this second step diminishes (i.e., in the absence or in defect of oxygen), the catalyst is deactivated through the formation of Pd nanoparticles.

Recently, we have been involved in the study of the chemistry of Pd(OAc)2 inside a porous, divinylbenzene cross-linked, pyridine-containing polymer (P4VP).47-50 In the freshly prepared form, this system might be seen as the heterogeneous counterpart of Pd(OAc)2/pyridine, and this stimulated us to explore its behavior during the aerobic oxidation of an alcohol in the gas phase, as a function of the reaction conditions. In particular, we applied synchronous,

time-resolved, SAXS/XAS/MS techniques to obtain a dynamical picture of the changes occurring to the Pd species (in terms of oxidation state, aggregation, and local structure) under transient reaction conditions. The SAXS/XAS/MS measurements were complemented with two additional characterization techniques: (1) operando DRIFT spectroscopy, to interrogate the destiny of the acetate ligands; and (2) ex situ HR-TEM, to directly visualize the eventual presence of Pd nanoparticles and to evaluate their particle size distribution. Finally, the catalytic performances were qualitatively explored by means of a CATLAB microreactor, with the final aim to correlate the catalyst properties to the catalytic performances. We selected 2-propanol as reactant, because of its high vapor pressure that allows performing the experiments in the gas phase, thus avoiding competitive solvent effects. In addition, it is easy to handle, nontoxic, with a low boiling point and a flash point at higher temperatures with respect to other alcohols. Finally, 2-propanol is the simplest secondary alcohol, and it can be oxidized solely to acetone, thus simplifying the catalytic study. Although oxidation of 2-propanol to acetone is not relevant from a technological point of view (acetone is produced directly or indirectly from propylene, mainly via the cumene process), our spectroscopic results indicate the occurrence of a single site mechanism for the alcohol oxidation reaction that might be of potential help in developing more performant catalysts for selective alcohol oxidation.

2. EXPERIMENTAL SECTION

2.1. Catalyst Preparation. The Pd(OAc)2/P4VP catalyst was prepared in the Chimet laboratories starting from Pd(II) acetate (hereafter Pd(OAc)2) and a poly-4-vinylpyridine 25% cross-linked with divinylbenzene (Sigma-Aldrich, hereafter P4VP), showing a specific surface area of about 50 m2 g-1. P4VP (in the form of microspheres) was added to an orange solution of Pd(OAc)2 in acetonitrile containing 4 wt % of Pd51 with respect to the support and left under stirring at room temperature overnight.47,48 The solution appeared completely decolored, the sample was filtered, successively dried at room temperature, and mildly ground in an agate mortar when necessary. We demonstrated previously48 that the Pd(OAc)2 precursor is stabilized inside the P4VP scaffold through the coordination of the pyridyl groups to the Pd2+ cations, with the consequent rupture of the trimeric structure characteristic for solid Pd(OAc)2, and the restructuring of the acetate ligands in a monodentate coordination.

2.2. Treatment Protocols. Before investigating the catalyst changes during the aerobic oxidation of 2-propanol, the effect of the two reagents alone was explored. Reduction in 2-propanol was accomplished by feeding in the reactor (either the capillary in SAXS/XAS/MS measurements or the DRIFTS cell for the FT-IR measurements) the vapors of the alcohol stripped by an inert flow (He, 20 mL/min) at 50 °C, and increasing the temperature to 200 °C (ramp 2 °C/min). Oxidation of the catalyst was achieved by flowing in the reactor a 15% O2 in He flow (20 mL/min of total flow) at 50 °C and successively increasing the temperature to 200 °C (ramp 2 °C/ min). The aerobic oxidation of 2-propanol was conducted on the catalysts (1) prereduced in 2-propanol, (2) preoxidized in O2, and (3) fresh. The reaction was performed by feeding in the reactor the vapors of the alcohol stripped by a 15% O2/He flow at 50 °C, and successively increasing the temperature to 180 °C. The temperature was limited to 180 °C due to the

6871

observation that at higher temperature the complete combustion of 2-propanol was favored.

2.3. Characterization Techniques and Data Analysis.

2.3.1. Synchronous SAXS/XAS/MS Measurements. Synchronous SAXS/XAS/MS measurements were performed on the BM26A beamline at the ESRF facility (Grenoble, France), by using the experimental setup reported previously.48'52'53 The catalyst powder was placed in a 2 mm glass capillary, having upstream and downstream two small pieces of quartz wool. The capillary was connected to the BM26A gas rig with mass-flow controllers for gas delivery,54 and heated with a heat gun. The evolution of the gaseous products of reaction was monitored with an online mass spectrometer (MS) at the end of the capillary by sampling a fraction of the out-streamflow.

Transmission XAS spectra at the Pd K edge (24350 eV) were collected using an ionization chamber before the sample and an X-ray sensitive photodiode placed in the center of the SAXS detector. The white beam was monochromatized using a Si(l1l) double crystal, and harmonic rejection was performed by using Pt coated mirrors (horizontal acceptance 2 mrad). The beam was focused to achieve 1.5 X 1 mm dimension on the sample. The energy was calibrated measuring the XANES spectrum of a palladium foil. The XAS spectra were acquired in the 24 200-24 600 eV range with an energy step of 3 eV and an integration time of 1 s/point in the pre-edge region, 1.5 eV step and 3 s/point in the XANES region, while the step in the EXAFS region was chosen to obtain a 0.05 A-1 step in the k-space with the acquisition time increasing quadratically from 3 to 9 s/point. Each spectrum required an acquisition time of about 10 min as a compromise between fast acquisition and quality of the spectra. It is worth noticing that, according to the literature,18'27 the reaction-induced reconstruction of Pd nanoparticles equilibrates within ca. 10 s during the vapor phase selective oxidation of crotyl alcohol. Hence, the time scale for spectra acquisition is long enough to allow our system to equilibrate at each temperature. The spectra were normalized and analyzed in the frame of multiple scattering theory with the GNXAS package software.55,56 The details of the data analysis are reported in section S2.

Simultaneously with XAS, SAXS patterns were collected by using a 2D Mar CCD detector. The sample—detector distance was calibrated according to the peak position of a standard Ag behenate powder sample. The energy change between the start and the end of the XAS spectrum (about 400 eV) is irrelevant to SAXS so that the incident beam wavelength can be treated as constant, A = 0.509(1) A. At this energy, we cover a q range (q = 4n sin 8/A; 0.01—0.3 A-1) big enough to get information on the size and the shape of eventually formed nanoparticles. A SAXS pattern was collected for each XAS spectrum. The patterns were integrated with Fit2D57 and modeled with a

homemade code.53,58

The SAXS data have been analyzed by fitting the experimental patterns with the function described by eq 1:

I(q) = A + 4 + cfD(r)j(qR )

2 6, r dr

(1)

where the term A + B/q4 describes the Porod function,59 simulating the polymer contribution; D(r) corresponds to the Weibull function, accounting for the particle size distribution, which is in turn defined as D(r) = (r/R)-1 exp(-r/R)b, in which R is the average radius of the particles and j(qR) is the

spherical first-order Bessel function, accounting for the spherical shape of the metal clusters.

2.3.2. DRIFT Spectroscopy. FT-IR spectra were collected in diffuse reflectance mode (DRIFT) on a Nicolet 6700 instrument, equipped with a MCT detector. A Thermo-Fisher environmental chamber was used to record the FT-IR spectra under reaction conditions. The cell was connected to a gas-flow system (under atmospheric pressure), equipped with electronic mass flow controllers (MFC). Each FT-IR spectrum required an acquisition time of about 2 min. FT-IR spectra were recorded at regular time intervals, during the whole process at a spectral resolution of 4 cm-1. Just at the end of each experiment, the catalyst powder was recovered and immediately measured by HR-TEM, trying to minimize the exposure to air.

2.3.3. High-Resolution Transmission Electron Microscopy. High-Resolution Transmission Electron Micrographs (HR-TEM) were obtained using a JEOL 3010-UHR instrument operating at 300 kV, equipped with a LaB6 filament and fitted with X-ray EDS analysis by a Link ISIS 200 detector. Digital micrographs were acquired by a 2k X 2k pixel Gatan US1000 CCD camera. Samples were quickly deposited (in the dry form, i.e., without using any solvent) on a copper grid covered with a lacey carbon film. Histograms of the particle size distribution were obtained by considering a statistical representative number of particles on the HR-TEM images, and the mean particle diameter (dTEM) was calculated as

(dTEM)

£ dini

(2)

where ni was the number of particles of diameter di.

2.4. Catalytic Tests. For the catalytic tests, an integrated quartz microreactor and mass spectrometer system (CATLAB from Hiden) was adopted. The system features a fast-response, low thermal mass furnace with integrated air-cooling, a precision quadrupole mass spectrometer, and a quartz inert capillary with "hot zone" inlet for continuous close-coupled catalyst sampling with minimal dead volume and memory effects. The catalyst was loaded into the quartz reactor with an inner diameter of 10 mm. The reaction temperature was monitored by using an in-bed thermocouple that ensures optimal measurement of catalyst temperature. The reactant gases were fed through electronic mass flow controllers. Feed and product analysis were performed by using a Pfeiffer OmniStar quadrupole mass spectrometer, monitoring the following ionic masses: (m/z) = 4 (He), 18 (H2O), 28 (CO2), 32 (O2), 43 (acetone), 45 (2-propanol), 60 (acetic acid). It is worth noticing that 2-propanol contributes also to the intensity of mass 43, which however is the most intense fragment for acetone.

3. RESULTS AND DISCUSSION

3.1. Pd(OAc)2/P4VP Reduced in 2-Propanol and Its Catalytic Performances. 3.1.1. Reactivity of Pd(OAc)2/P4VP with 2-Propanol. The ability of simple alcohols in reducing palladium oxide in mild conditions has been known for a long time.1 Newton et al.60 have recently proved that also ethanol-water, a prototypical "green" solvent mixture, cannot be considered as innocent toward supported Pd nanoparticles. Even dehydrogenation of the simplest secondary alcohol, 2-propanol, leads to PdO reduction and may poison metallic Pd already at room temperature.1 Reduction of Pd(OAc)2 is less

6872

Figure 1. (a) 2D map showing the evolution of the DRIFT spectra as a function of the reaction temperature, for the Pd(OAc)2/P4VP reacting with 2-propanol. The intensity increases from blue to white. Dotted lines highlight the evolution of the absorption bands assigned to Pd(OAc)2 and to the pyridyl groups of P4VP interacting with the hosted Pd2+ cations. (b) Representative HR-TEM micrograph of Pd(OAc)2/P4VP after reaction with 2-propanol at 200 °C and the corresponding particle size distribution.

Figure 2. 2D maps showing the evolution of the XANES spectra (a), of the particle size distribution as determined by the analysis of the SAXS data (b), and amplitude of the phase-uncorrected FT of the EXAFS spectra (c) as a function of the reaction temperature, for the Pd(OAc)2/P4VP reacting with 2-propanol. The intensity increases from black to yellow. The labels refer to the main assignments. Parts d and d' show the evolution of gaseous acetic acid (m/z = 60) as detected by online MS and by the independent catalytic test performed with the CATLAB microreactor.

easy than reduction of PdO: it does not occur at room temperature, but requires reaching ca. 110 °C, and it is completed around 180 °C. In our previous work,50 we followed the reduction of Pd(OAc)2 in 2-propanol by means of DRIFT spectroscopy, by monitoring the disappearance of the IR bands characteristic of the acetate ligands and by characterizing the obtained Pd nanoparticles by using CO as a probe molecule. The sequence of DRIFT spectra collected during the reaction is shown in Figure S1 and here represented in a 2D map as a function of the temperature in Figure 1a, in the 1800-1250 cm-1 region. The two absorption bands at ca. 1365 and 1300 cm-1, assigned to the vsym(COO) mode of two slightly

different terminal acetate ligands (dotted lines in Figure 1a labeled as Pd(OAc)2),48 gradually decrease in intensity and are no more observed around 170 °C. Concomitantly, a shrinkage of the very intense band centered at 1596 cm-1 (assigned to the 8a vibrational mode of the pyridyl functional group) is observed (dotted line in Figure 1a labeled as py--Pd2+), due to the disappearance of the shoulder at ca. 1640 cm-1. This shoulder was previously considered as the fingerprint of a chemical interaction between the pyridyl groups of P4VP and the Pd2+ cations of the hosted Pd(OAc)2,48 similar to what was reported for several P4VP/metal complexes.61-66

6873

Figure 3. (a) 2D map showing the evolution of the DRIFT spectra as a function of the reaction temperature, for the Pd(OAc)2/P4VP catalyst prereduced in 2-propanol during the aerobic oxidation of 2-propanol to acetone. The intensity increases from blue to yellow. (b) Two representative HR-TEM images of prereduced Pd(OAc)2/P4VP after the aerobic oxidation of 2-propanol at 180 °C and the corresponding particle size distribution.

At the end of the DRIFTS experiment, the sample was analyzed by HR-TEM. A representative micrograph is shown in Figure 1b, along with the particle size distribution determined by counting about 800 particles. Very small Pd nanoparticles, homogeneously distributed in the polymer, and with a spherical shape and a regular size are observed. Most of them have a diameter smaller than 2 nm and are hardly detectable by our TEM instrument. The average particle size ((dTEM) = 1.4 ± 0.3 nm) is very similar to that of Pd particles obtained upon reducing the same system in

H2,47,48 and

reflects the stabilization effect of the pyridyl ligands in P4VP.

To get insights into the mechanism of Pd(OAc)2/P4VP reduction, the reaction of the catalyst with 2-propanol was followed by means of simultaneous, time-resolved, SAXS/ XAS/MS measurements. Figure S2 shows the sequence of normalized XANES spectra, SAXS patterns, and EXAFS spectra at the Pd K-edge collected simultaneously during the reaction of the Pd(OAc)2/P4VP catalyst with 2-propanol upon increasing the temperature from 50 to 200 °C. A gradual change is observed by all three techniques starting from ca. 110 °C, while they did not evolve anymore once the temperature reached ca. 180 °C. This evolution indicates that the Pd2+ precursor is progressively reduced with the consequent formation of Pd0 nanoparticles. The temperature interval during which Pd(OAc)2 is reduced to Pd0 nanoparticles is in very good agreement with that previously determined by DRIFT spectroscopy (Figure 1a). To better visualize the spectral changes as a function of the reaction temperature, the same data have been reported in 2D maps in Figure 2a-c, together with the evolution of the gaseous products as simultaneously monitored by online MS (part d), and in the independent experiment with the CATLAB microreactor (part d'). Starting from the XANES spectra (Figure 2a), the following changes are observed during the reaction: (1) the edge (border between red and black regions) progressively shifts to lower energy; (2) the peak at ca. 24 372 eV (labeled as Pd2+), characteristic of Pd(OAc)2, vanishes; and (3) that peak is replaced by a peak at ca. 24 386 eV (labeled as Pd0), which is assigned to the first EXAFS oscillation of palladium atoms arranged in a face centered cubic (fcc) local structure, typical

of Pd0 nanoparticles.47,48,67 Figure 2b shows the contribution

of the spherical particles (either Pd(OAc)2 or Pd0, or both) with respect to the modeled background as determined by the analysis of the SAXS data. A gradual increase of the average

particle size ((dSAXS)) is observed throughout the reaction, from ca. (dSAXS) = 1.27 ± 0.05 nm to ca. 1.79 ± 0.05 nm, in good agreement with that determined by means of HR-TEM at the end of the reaction.

As far as the EXAFS spectra are concerned (Figure 2c), the peak initially at ca. 1.45 A (not phase corrected, labeled as Pd-OAc), attributed to the first-shell Pd-O contribution of Pd(OAc)2, decreases in intensity and shifts at slightly higher distances (ca. 1.52 A), while at the same time a new peak appears at ca. 2.55 A (labeled as Pd-Pd), which is due to the first-shell Pd-Pd contribution of the Pd nanoparticles. The persistence of a signal around ca. 1.52 A reveals that a substantial fraction of the Pd atoms interacts with low-Z elements. Because all of the Pd(OAc)2 has been reduced, this peak is attributed to the interaction of Pd either with the nitrogen of the pyridine ligands in P4VP47,48 or with the carbon of carbonaceous species derived from the dehydrogen-ation of 2-propanol. Hereafter, we will refer to this peak as Pd-X contribution, where X states either for O, N, or C (which are not distinguishable by EXAFS). The results of the EXAFS fits for each spectrum of the series are reported in Table S1. Starting from ca. 130 °C, the fraction of the Pd atoms in interaction with other Pd atoms (%Pd-Pd in Table S1) gradually increases up to ca. 84%. In these conditions, all of the Pd(OAc) 2 has been reduced to Pd nanoparticles. The average coordination number NPd-Pd = 2.6 ± 0.5 is extremely small as compared to NPd-Pd = 12 in bulk Pd metal, as expected for very small nanoparticles.68-73 Notably the results agree with literature data on similar systems74-76 (more details are reported in the Supporting Information).

In summary, Pd(OAc)2 in P4VP is reduced by 2-propanol at elevated temperature to Pd0 nanoparticles, which are stabilized by the pyridyl moieties in P4VP. As far as the reduction mechanism is concerned, synchronous MS measurements detect small traces of acetic acid (m/z = 60) during the reduction of Pd(OAc)2 by 2-propanol (Figure 2d). This is confirmed by an independent catalytic test performed with the CATLAB microreactor (Figure 2d'). Acetic acid originates from the hydrogenation of the acetate ligands and indicates the occurrence of alcohol dehydrogenation.

3.1.2. Catalytic Performances of the Pd(OAc)2/P4VP Prereduced in 2-Propanol. The Pd(OAc)2/P4VP catalyst prereduced in 2-propanol efficiently catalyzes the oxidation of 2-propanol to acetone. The reaction was initially followed by

6874

Figure 4. 2D maps showing the evolution of the XANES spectra (a), of the particle size distribution as determined by the analysis of the SAXS data (b), and of the amplitude of phase-uncorrected FT of the EXAFS spectra (c) as a function of the reaction temperature, for the Pd(OAc)2/P4VP prereduced in 2-propanol during the aerobic oxidation of 2-propanol. The intensity increases from black to yellow. The labels refer to the main assignments. Parts d and d' show the conversion of 2-propanol (m/z = 45) into acetone (m/z = 43) as detected by online MS and by an independent catalytic test performed with the CATLAB microreactor, respectively.

means of DRIFT spectroscopy. Figure S3 reports the evolution of the DRIFT spectra as a function of temperature, whereas Figure 3a shows the corresponding 2D map in the 1800-1680 cm-1 region, which is the region where acetone can be easily detected, because the characteristic v(C=O) absorption band at ca. 1730 cm-1 does not overlap neither with those of the catalyst nor with those of 2-propanol. Acetone starts to be spectroscopically detected already around 80-90 °C, but the maximum conversion is achieved only at ca. 170-180 °C, in well agreement with the MS data (vide infra, Figure 4d and d'). After the reaction, the catalyst was also analyzed by HR-TEM. A representative image and the particle size distribution are reported in Figure 3b. Despite the fact that the majority of the Pd NPs have preserved the initial small size, some very big Pd agglomerates, whose composition was checked by EDX analysis, have been observed (inset). With respect to the same catalyst before the aerobic oxidation of 2-propanol, a slight increase of the average particle size is detected ((dTEM) = 1.6 ± 0.4 nm).

The main question, however, is whether the catalyst works in the oxidized or in the reduced state. This is readily apparent looking to Figure 4a-c, which shows the evolution of the XANES spectra (part a), of the particle size distribution as determined by the analysis of the SAXS data (part b), and of the amplitude of Fourier transforms (FT) of the EXAFS spectra (part c) during the oxidation of 2-propanol over the prereduced Pd(OAc)2/P4VP catalyst as a function of the reaction temperature. The raw spectra are reported in Figure S4. As soon as the reaction starts (around 80-90 °C), the contributions due to metal Pd (i.e., the Pd0 peak at ca. 24 386

eV in the XANES spectra and the Pd-Pd contribution at ca. 2.55 A in the EXAFS spectra) rapidly decrease in intensity and reach the minimum values around 110 °C. At the same time, the contributions ascribed to Pd2+ (the Pd2+ peak at ca. 24 372 eV in the XANES spectra and the Pd-X contribution around 1.52 A in the EXAFS spectra) increase in intensity, reaching the maximum values at ca. 160 °C. The fraction of the Pd atoms in interaction with other Pd atoms (%Pd-Pd in Table S2), as determined by fitting the EXAFS data, drastically decreases to less than 15%. The whole set of data clearly indicates that, in the presence of the reaction mixture, the Pd0 nanoparticles are rapidly oxidized to PdO nanoparticles. The oxidation is almost complete because of their very small size. In the whole 110-170 °C interval, which corresponds to the best performance of the catalyst (Figure 4d and d'), the average oxidation state of palladium remains 2+, while the particle size changes are negligible ((dSAXS) increases from 1.89 ± 0.05 to 1.96 ± 0.05 nm, Figure 4b).

Around 170-180 °C, a very sudden change is observed in all of the spectra. The PdO nanoparticles are reduced back to Pd0, and at the same time the particle size abruptly increases ((dSAXS) goes from 1.96 ± 0.05 to 2.45 ± 0.05 nm, Figure 4b)77 and the %Pd-Pd (Table S1) reaches again a value of about 70%. The phenomenon is associated with a slight loss of the catalytic activity (Figure 4d and d'). The whole sequence of data presented so far converges in indicating that (1) the prereduced Pd(OAc)2/P4VP catalyst is rapidly oxidized in the presence of the reaction mixture and remains oxidized during the conversion of 2-propanol to acetone in the whole 110-170 °C temperature range, in agreement with recent studies that

6875

Figure 5. (a) 2D map showing the evolution of the DRIFT spectra as a function of the reaction temperature, for the Pd(OAc)2/P4VP catalyst during the reaction with molecular oxygen. The intensity increases from blue to yellow. Dotted lines highlight the evolution of the absorption bands assigned to Pd(OAc)2 and to the pyridyl groups of P4VP in interaction with Pd(OAc)2. (b) A representative HR-TEM micrograph of Pd(OAc)2/ P4VP after the reaction with molecular oxygen at 200 °C.

Figure 6. 2D maps showing the evolution of the XANES spectra (a), of the particle size distribution as determined by the analysis of the SAXS data (b), and of the amplitude of the phase-uncorrected FT of the EXAFS spectra (c) as a function of the reaction temperature, for the Pd(OAc)2/P4VP reacting with molecular oxygen. The intensity increases from black to yellow. The labels refer to the main assignments.

have implicated surface PdO as the active phase in the selective oxidation of alcohols;30-35 and (2) the prereduced Pd(OAc)2/ P4VP catalyst is not stable during the aerobic oxidation of 2-propanol at 170-180 °C. The reason for the instability of the Pd phase in these reaction conditions is not completely clear, but it might be associated with a sudden increase of the temperature at the catalyst surface. It is interesting to notice that the catalyst deactivation is not irreversible. Once the catalyst is reoxidized at 180 °C, the activity is almost completely restored (data not shown), in a way similar to what was reported in the literature for supported PdOx

nanoparticles.26

3.2. Pd(OAc)2/P4VP Oxidized in Molecular O2 and Its Catalytic Performances. 3.2.1. Reactivity of Pd(OAc)2/P4VP with Oxygen. Successively, we explored the reactivity of Pd(OAc) 2 toward molecular oxygen, which is the second reagent in the investigated reaction. Figure 5a shows the evolution of the DRIFT spectra for Pd(OAc)2/P4VP in the presence of oxygen at increasing temperature from 50 to 200 °C (the corresponding spectra are reported in Figure S5). The two absorption bands at ca. 1365 and 1300 cm-1 characteristic of the acetate ligands48 start to decrease in intensity when the temperature approaches 180 °C, although they do not completely disappear even at 200 °C. This indicates that a fraction of the acetate ligands is removed. At the same time, the shoulder at ca. 1640 cm-1, indicative of the chemical interaction between the pyridyl groups of P4VP and the Pd2+ cations of Pd(OAc)2, does not disappear (contrary to what is observed for the reaction of Pd(OAc)2/P4VP with 2-

propanol), but further shifts to ca. 1665 cm-1 (dotted line in Figure 5a). A larger upward shift of the 8a vibrational mode of the pyridyl moieties indicates that these functional groups in P4VP interact with stronger acid sites, as it might be the case for Pd2+ cations that have lost one or both the acetate ligands. Interestingly, almost no nanoparticles were detected by means of HR-TEM at the end of the oxidation reaction (Figure 5b), signifying either that the acetate ligands are burnt off leaving isolated Pd2+ cations stabilized by the P4VP environment, or that extremely dispersed PdO clusters, with size below the detection limit of our TEM instrument, are formed.

The results of the synchronous XAS/SAXS/MS experiment in the presence of molecular oxygen are reported in Figure 6, while the raw experimental data are shown in Figure S6. Essentially, no changes are observed in the edge position of the XANES spectra along the whole reaction (Figure 6a), but only a slight increase of the signal around 24 395 eV is detected. The particle size distribution determined by analyzing the SAXS data (Figure 6b) remains unchanged. In the EXAFS spectra, a slight shift of the Pd-X contribution toward longer distances is observed (from ca. 1.40 to 1.50 A, not phase corrected), as well as a tiny decrease in the peak intensity. The fit of the EXAFS data (Table S3) confirms that RPd - X moves from ca. 2.04 ± 0.05 A, typical for RPd-O in Pd(OAc)2, to ca. 2.08 ± 0.05 A. This change is compatible with a reconstruction of the local environment around the Pd2+ cations. Only traces of acetic acid were detected by MS, at a temperature of 180 °C.

3.2.2. Catalytic Performances of the Pd(OAc)2/P4VP Preoxidized in Molecular Oxygen. Also, the Pd(OAc)2/

6876

C)

Figure 7. 2D maps showing the evolution of the XANES spectra (a), of the particle size distribution as determined by the analysis of the SAXS data (b), and of the amplitude of the phase-uncorrected FT of the EXAFS spectra (c) as a function of the reaction temperature, for the Pd(OAc)2/P4VP preoxidized in molecular oxygen during the aerobic oxidation of 2-propanol. The intensity increases from black to yellow. Parts d and d' show the conversion of 2-propanol (m/z = 45) into acetone (m/z = 43) as detected by online MS and by an independent catalytic test performed with the CATLAB microreactor, respectively.

P4VP catalyst preoxidized in molecular oxygen efficiently catalyzes the aerobic oxidation of 2-propanol to acetone. The reaction starts around 80-90 °C, as determined by DRIFT spectroscopy (appearance of the absorption band around 1730 cm-1 due to acetone, Figure S7) and by MS measurements (Figure 7d and d'). The maximum conversion is achieved around 150 °C, that is, at a slightly lower temperature than for the prereduced catalyst. Figure 7 displays the results of the synchronous XAS/SAXS/MS experiment during the aerobic oxidation of 2-propanol on the preoxidized Pd(OAc)2/P4VP catalyst, while the raw experimental spectra are reported in Figure S8. During the whole reaction, no changes are detected by any technique, indicating that highly dispersed Pd2+ cations stabilized by the pyridyl groups in P4VP are the active sites in the reaction. Noticeably, the catalyst is highly stable also at 180 °C, in contrast to what was observed after prereduction.

It is difficult to determine whether the Pd2+ cations are isolated or aggregated in small Pd-oxo clusters. In this respect, it is worth noticing that atomically dispersed Pd2+ species in a ultradiluted mesoporous 0.03 wt % Pd/Al2O3 catalyst showed exceptional activity in the selective oxidation of alcohols.34 Pd1O4 single-sites anchored on the internal surface of micropores in a microporous silicate exhibit high selectivity and activity in the partial oxidation of CH4 to CH3OH with H2O2.78 On the other hand, a trinuclear [(LPd")3(^3-O)2]2+ intermediate compound has been recently identified during O2 activation by Pd complexes and shown to be a chemically and kinetically competent intermediate in catalytic alcohol oxidation reactions.79 Our experimental data are compatible with the presence of both atomically dispersed Pd2+ species and ultrasmall Pd-oxo clusters.

3.3. Aerobic Oxidation of 2-Propanol over Untreated

Pd(OAc)2/P4VP. As a final step, we investigated the behavior of the untreated Pd(OAc)2/P4VP catalyst in the oxidation of 2-propanol to acetone in the same reaction conditions adopted for the prereduced and preoxidized catalysts. Three successive reaction cycles were performed to test the catalyst stability. Figure 8 summarizes the main results. During the first cycle, 2-propanol starts to be oxidized to acetone only at ca. 160-170 °C, as determined by DRIFT spectroscopy (Figure 8a1) and synchronous MS (Figure 8b1), as well as by the independent catalytic test performed with the CATLAB microreactor (Figure 8c1). The reason is that the Pd2+ cations need to lose (at least partially) the acetate ligands. Indeed, the XANES spectra (Figure 8d1) slightly change upon increasing the temperature, in the same way as was observed during the treatment in only O2 (Figure 6a). This indicates that the active phase is not Pd(OAc)2, but the cationic Pd2+ species in a different environment. In the successive cycles (second, Figure 8a2-d2, and third, Figure 8a3-d3), the oxidation of 2-propanol to acetone starts around 100-110 °C, reaching the maximum conversion at ca. 150 °C. The XANES spectra (as well as the EXAFS spectra and the SAXS patterns) do not change anymore (Figure 8d2,d3). The catalyst is highly stable even at 180 °C.

4. CONCLUSIONS

We have investigated the behavior of a heterogeneous Pd(OAc)2/P4VP catalyst that mimics the most famous and widely employed homogeneous Pd(OAc)2/pyridine system, during the gas-phase aerobic oxidation of 2-propanol to acetone. The combined use of synchronous SAXS/XAS/MS,

6877

Figure 8. (a1) 2D map showing the evolution of the DRIFT spectra as a function of the reaction temperature, for the untreated Pd(OAc)2/P4VP catalyst during the aerobic oxidation of 2-propanol. The intensity increases from blue to yellow. Parts b1 and c1 show conversion of 2-propanol (m/ z = 45) into acetone (m/z = 43) as detected by online MS and by an independent catalytic test performed with the CATLAB microreactor, respectively. (d1) 2D map showing the evolution of the XANES spectra as a function of the reaction temperature, for the untreated Pd(OAc)2/ P4VP during the aerobic oxidation of 2-propanol. The intensity increases from black to yellow. Parts a2-d2 and a3-d3 are the same for the second and third reaction cycles, respectively.

coupled with operando DRIFT spectroscopy and HR-TEM, reveals to be a strategic approach to unravel simultaneously the oxidation state and the aggregation of the Pd phase under reaction conditions, at the same time monitoring the catalytic performances. In particular, we have demonstrated the following:

(1) Pd(OAc) 2 in P4VP is reduced by 2-propanol in the 110-170 °C temperature range, leading to highly dispersed Pd0 nanoparticles with a very homogeneous particle size, which are stabilized by the pyridyl ligands in P4VP, similarly to what

was previously found in the presence of H2.47,48

(2) These Pd0 nanoparticles are rapidly and almost completely oxidized in the presence of the 2-propanol-O2 reaction mixture (at least when the O2 concentration is 15 vol %), already at low temperature (50 °C). The oxidized PdO nanoparticles efficiently oxidize 2-propanol into acetone starting from ca. 110 °C. No variation in both particle size and Pd oxidation state is registered until 170 °C. This demonstrates the fundamental role of surface PdO in the adopted reaction conditions, in good agreement with the recent literature on the Pd-catalyzed selective oxidation of alcohols.30-35 At temperature higher than 170 °C, the PdO nanoparticles are suddenly reduced and aggregate to form

larger particles, with a consequent slight decrease of the catalytic activity.

(3) In the presence of only O2 at ca. 180-200 °C, Pd(OAc) 2 in P4VP loses a fraction of the acetate ligands, with the consequent formation of isolated Pd2+ cations or ultrasmall Pd-oxo clusters or both, stabilized by the pyridyl ligands in P4VP.

(4) These Pd2+ cationic species catalyze the oxidation of 2-propanol to acetone starting from ca. 100 °C (i.e., at a temperature slightly lower than the prereduced catalyst). The catalyst remains in the reaction mixture stable also when reaching 180 °C.

(5) The untreated catalyst does not work in the reaction until a fraction of the acetate ligands are removed, which occurs around 170-180 °C. At that point, the catalyst behaves as the preoxidized one: it converts 2-propanol to acetone starting from ca. 100 °C, without any change in the Pd oxidation state and aggregation, even at 180 °C and for several reaction cycles, proving that the reaction occurs at a single site even though a heterogeneous system is used.

Observations (3)-(5) clearly demonstrate that highly isolated Pd2+ cationic species, either atomically dispersed or in the form of ultrasmall Pd2+-oxo clusters, are efficient and

6878

very stable active sites for the gas-phase aerobic oxidation of 2-propanol. These results are in very good agreement with recent findings that both atomically dispersed Pd2+ species in a heterogeneous ultradiluted catalyst and multinuclear Pd-oxo homogeneous complexes39,79 are active species implicated in Pd-catalyzed aerobic oxidation reactions. It is important to remark, however, that our experiments have been conducted in excess of oxygen, and that a different outcome could have been obtained in other experimental conditions. Noticeably, the behavior of the Pd(OAc)2/P4VP catalyst under reaction conditions is influenced by the nature of the support. On the one hand, the presence of the pyridyl functional groups is fundamental to stabilize the cationic Pd2+ species, exactly as for the ligand-modulated homogeneous Pd2+ complexes. On the other hand, the porous structure of the P4VP polymer efficiently confines the active Pd2+ species in the presence of the reagents. We can state that our Pd(OAc)2/P4VP catalyst is situated at the confluence between its homogeneous and heterogeneous analogues.

■ ASSOCIATED CONTENT ^ Supporting Information

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acscatal.8b01421.

Raw experimental DRIFTS, SAXS, and XAS data; details on the EXAFS data analysis; and results of the EXAFS fit (PDF)

■ AUTHOR INFORMATION Corresponding Authors

*E-mail: [e-mail]. *E-mail: [e-mail]. ORCID®

Elena Groppo: 0000-0003-4153-5709 Andrea Lazzarini: 0000-0002-0404-6597 Aram Bugaev: 0000-0001-8273-2560 Maela Manzoli: 0000-0002-4427-7939 Carlo Lamberti: 0000-0001-8004-2312 Notes

The authors declare no competing financial interest.

■ ACKNOWLEDGMENTS

A.B. and C.L. acknowledge the Russian Ministry of Education and Science for financial support (project RFMEFI58417X 0029, agreement 14.584.21.0029) and for funding the research in Rostov-on-Don.

■ REFERENCES

(1) Mallat, T.; Baiker, A. Oxidation Of Alcohols With Molecular Oxygen On Solid Catalysts. Chem. Rev. 2004, 104, 3037-3058.

(2) Lee, A. F. Mechanistic Studies Of Alcohol Selective Oxidation. In Heterogeneous Catalysts for Clean Technology. Spectroscopy, Design, and Monitoring; Wilson, K., Lee, A. F., Eds.; Wiley-VCH Verlag & Co. Kgaa: Weinheim, Germany, 2014; pp 11-38.

(3) Durndell, L. J.; Lee, A. F.; Bailie, D. S.; Muldoon, M. J. Selective Palladium-Catalysed Aerobic Oxidation Of Alcohols. In Transition Metal Catalysis in Aerobic Alcohol Oxidation; Cardona, F., Parmeggiani, C., Eds.; The Royal Society Of Chemistry: Cambridge, UK, 2015; pp 92-132.

(4) Muhler, M. Oxidative Dehydrogenation Of Alcohols To Aldehydes, Ketones And Carboxylic Acids. In Handbook of

Heterogeneous Catalysis; Ertl, G., Knozinger, H., Weitkamp, J., Eds.; VCH: Weinheim, 1997; Vol. 5, p 2274.

(5) Mallat, T.; Baiker, A. Oxidation Of Alcohols With Molecular Oxygen. In Handbook of Heterogeneous Catalysis; Ertl, G., Knozinger, H., Weitkamp, J., Eds.; Wiley-VCH Verlag Gmbh & Co. Kgaa: New York, 2008; pp 3521-3538.

(6) Stahl, S. S. Palladium-Catalyzed Oxidation Of Organic Chemicals With O2. Science 2005, 309, 1824.

(7) Cavani, F.; Teles, J. H. Sustainability In Catalytic Oxidation: An Alternative Approach Or A Structural Evolution? ChemSusChem 2009, 2, 508-534.

(8) Montemore, M. M.; Van Spronsen, M. A.; Madix, R. J.; Friend, C. M. O2 Activation By Metal Surfaces: Implications For Bonding And Reactivity On Heterogeneous Catalysts. Chem. Rev. 2018, 118, 2816-2862.

(9) Centi, G.; Cavani, F.; Trifiro, F. In Selective Oxidation by Heterogeneous Catalysis; Twigg, M. V., Spencer, M. S., Eds.; Springer Science+Business Media, LLC: New York, 2001.

(10) Beale, A. M.; Jacques, S. D. M.; Weckhuysen, B. M. Chemical Imaging Of Catalytic Solids With Synchrotron Radiation. Chem. Soc. Rev. 2010, 39, 4656-4672.

(11) Grunwaldt, J.-D.; Schroer, C. G. Hard And Soft X-Ray Microscopy And Tomography In Catalysis: Bridging The Different Time And Length Scales. Chem. Soc. Rev. 2010, 39, 4741-4753.

(12) Jung, U.; Elsen, A.; Li, Y.; Smith, J. G.; Small, M. W.; Stach, E. A.; Frenkel, A. I.; Nuzzo, R. G. Comparative In Operando Studies In Heterogeneous Catalysis: Atomic And Electronic Structural Features In The Hydrogenation Of Ethylene Over Supported Pd And Pt Catalysts,. ACS Catal. 2015, 5, 1539-1551.

(13) Gai, P. L.; Sharma, R.; Ross, F. M. Environmental (S)TEM Studies Of Gas-Liquid-Solid Interactions Under Reaction Conditions. MRS Bull. 2008, 33, 107-114.

(14) Li, Y.; Zakharov, D.; Zhao, S.; Tappero, R.; Jung, U.; Elsen, A.; Baumann, P.; Nuzzo, R. G.; Stach, E. A.; Frenkel, A. I. Complex Structural Dynamics Of Nanocatalysts Revealed In Operando Conditions By Correlated Imaging And Spectroscopy Probes. Nat. Commun. 2015, 6, 1.

(15) Ferri, D.; Baiker, A. Advances In Infrared Spectroscopy Of Catalytic Solid-Liquid Interfaces: The Case Of Selective Alcohol Oxidation,. Top. Catal. 2009, 52, 1323-1333.

(16) Grunwaldt, J. D.; Caravati, M.; Baiker, A. Oxidic Or Metallic Palladium: Which Is The Active Phase In Pd-Catalyzed Aerobic Alcohol Oxidation? J. Phys. Chem. B 2006, 110, 25586-25589.

(17) Steinhoff, B. A.; Guzei, I. A.; Stahl, S. S. Mechanistic Characterization Of Aerobic Alcohol Oxidation Catalyzed By Pd(Oac)<Inf > 2</Inf>/Pyridine Including Identification Of The Catalyst Resting State And The Origin Of Nonlinear [Catalyst] Dependence. J. Am. Chem. Soc. 2004, 126, 11268-11278.

(18) Williams, R. M.; Medlin, J. W. Benzyl Alcohol Oxidation On Pd(111): Aromatic Binding Effects On Alcohol Reactivity. Langmuir 2014, 30, 4642-4653.

(19) Ferri, D.; Mondelli, C.; Krumeich, F.; Baiker, A. Discrimination Of Active Palladium Sites In Catalytic Liquid-Phase Oxidation Of Benzyl Alcohol. J. Phys. Chem. B 2006, 110, 22982-22986.

(20) Hayashi, M.; Yamada, K.; Nakayama, S.-Z. Catalytic Hydrogen-Transfer Reactions Of Benzylic And Allylic Alcohols With Palladium Compounds In The Presence Of Vinyl Acetate Or Under An Ethylene Atmosphere,. J. Chem. Soc. Perkin. Trans. 2000, 1, 15011503.

(21) Keresszegi, C.; Mallat, T.; Baiker, A. Selective Transfer Dehydrogenation Of Aromatic Alcohols On Supported Palladium. New J. Chem. 2001, 25, 1163-1167.

(22) Grunwaldt, J.-D.; Keresszegi, C.; Mallat, T.; Baiker, A. In Situ EXAFS Study Of Pd/Al2O3 During Aerobic Oxidation Of Cinnamyl Alcohol In An Organic Solvent. J. Catal. 2003, 213, 291-295.

(23) Keresszegi, C.; Grunwaldt, J. D.; Mallat, T.; Baiker, A. In Situ EXAFS Study On The Oxidation State Of Pd/Al2O3 And Bi-Pd/ Al2O3 During The Liquid-Phase Oxidation Of 1-Phenylethanol. J. Catal. 2004, 222, 268-280.

6879

(24) Keresszegi, C.; Ferri, D.; Mallat, T.; Baiker, A. On The Role Of CO Formation During The Aerobic Oxidation Of Alcohols On Pd/ Al2O3: An In Situ Attenuated Total Reflection Infrared Study. J. Catal. 2005, 234, 64-75.

(25) Mondelli, C.; Ferri, D.; Grunwaldt, J.-D.; Krumeich, F.; Mangold, S.; Psaro, R.; Baiker, A. Combined Liquid-Phase ATR-IR And XAS Study Of The Bi-Promotion In The Aerobic Oxidation Of Benzyl Alcohol Over Pd/Al2O3,. J. Catal. 2007, 252, 77-87.

(26) Lee, A. F.; Ellis, C. V.; Naughton, J. N.; Newton, M. A.; Parlett, C. M. A.; Wilson, K. Reaction-Driven Surface Restructuring And Selectivity Control In Allylic Alcohol Catalytic Aerobic Oxidation Over Pd. J. Am. Chem. Soc. 2011, 133, 5724-5727.

(27) Gaskell, C. V.; Parlett, C. M. A.; Newton, M. A.; Wilson, K.; Lee, A. F. Redox-Controlled Crotyl Alcohol Selective Oxidation: In Situ Oxidation And Reduction Dynamics Of Catalytic Pd Nano-particles Via Synchronous XANES/MS. ACS Catal. 2012, 2, 22422246.

(28) Dirkx, J. M. H.; Van Der Baan, H. S. The Oxidation Of Glucose With Platinum On Carbon As Catalyst,. J. Catal. 1981, 67, 1-13.

(29) Gangwal, V. R.; Van Der Schaaf, J.; Kuster, B. F. M.; Schouten, J. C. Noble-Metal-Catalysed Aqueous Alcohol Oxidation: Reaction Start-Up And Catalyst Deactivation And Reactivation. J. Catal. 2005, 232, 432-443.

(30) Lee, A. F.; Wilson, K. Structure-Reactivity Correlations In The Selective Aerobic Oxidation Of Cinnamyl Alcohol: In Situ XAFS. Green Chem. 2004, 6, 37-42.

(31) Lee, A. F.; Hackett, S. F. J.; Hargreaves, J. S. J.; Wilson, K. On The Active Site In Heterogeneous Palladium Selox Catalysts. Green Chem. 2006, 8, 549-555.

(32) Parlett, C. M. A.; Bruce, D. W.; Hondow, N. S.; Lee, A. F.; Wilson, K. Support-Enhanced Selective Aerobic Alcohol Oxidation Over Pd/Mesoporous Silicas. ACS Catal. 2011, 1, 636-640.

(33) Parlett, C. M. A.; Keshwalla, P.; Wainwright, S. G.; Bruce, D. W.; Hondow, N. S.; Wilson, K.; Lee, A. F. Hierarchically Ordered Nanoporous Pd/SBA-15 Catalyst For The Aerobic Selective Oxidation Of Sterically Challenging Allylic Alcohols. ACS Catal.

2013, 3, 2122-2129.

(34) Hackett, S. F. J.; Brydson, R. M.; Gass, M. H.; Harvey, I.; Newman, A. D.; Wilson, K.; Lee, A. F. High-Activity, Single-Site Mesoporous Pd/Al2O3 Catalysts For Selective Aerobic Oxidation Of Allylic Alcohols. Angew. Chem. 2007, 119, 8747-8750.

(35) Parlett, C. M. A.; Durndell, L. J.; Wilson, K.; Bruce, D. W.; Hondow, N. S.; Lee, A. F. Selective Oxidation Of Allylic Alcohols Over Highly Ordered Pd/Meso-Al2O3 Catalysts. Catal. Commun.

2014, 44, 40-45.

(36) Stahl, S. S. Palladium Oxidase Catalysis: Selective Oxidation Of Organic Chemicals By Direct Dioxygen-Coupled Turnover. Angew. Chem., Int. Ed. 2004, 43, 3400-3420.

(37) Conley, N. R.; Labios, L. A.; Pearson, D. M.; Mccrory, C. C. L.; Waymouth, R M. Aerobic Alcohol Oxidation With Cationic Palladium Complexes: Insights Into Catalyst Design And Decomposition. Organometallics 2007, 26, 5447-5453.

(38) Nielsen, R. J.; Goddard, W. A. Mechanism Of The Aerobic Oxidation Of Alcohols By Palladium Complexes Of N-Heterocyclic Carbenes. J. Am. Chem. Soc. 2006, 128, 9651-9660.

(39) Ingram, A. J.; Walker, K. L.; Zare, R N.; Waymouth, R M. Catalytic Role Of Multinuclear Palladium-Oxygen Intermediates In Aerobic Oxidation Followed By Hydrogen Peroxide Disproportiona-tion. J. Am. Chem. Soc. 2015, 137, 13632-13646.

(40) Scheuermann, M. L.; Goldberg, K. I. Reactions Of Pd And Pt Complexes With Molecular Oxygen. Chem. - Eur. J. 2014, 20, 1455614568.

(41) Jaworski, J. N.; Mccann, S. D.; Guzei, I. A.; Stahl, S. S. Detection Of Palladium(I) In Aerobic Oxidation Catalysis. Angew. Chem., Int. Ed. 2017, 56, 3605-3610.

(42) Steinhoff, B. A.; Fix, S. R.; Stahl, S. S. Mechanistic Study Of Alcohol Oxidation By The Pd(Oac)<Inf > 2</Inf>/O < Inf > 2</ Inf>/DMSO Catalyst System And Implications For The Develop-

ment Of Improved Aerobic Oxidation Catalysts. J. Am. Chem. Soc. 2002, 124, 766-767.

(43) Steinhoff, B. A.; Stahl, S. S. Mechanism Of Pd(Oac)<Inf > 2</ Inf>/DMSO-Catalyzed Aerobic Alcohol Oxidation: Mass-Transfer-Limitation Effects And Catalyst Decomposition Pathways. J. Am. Chem. Soc. 2006, 128, 4348-4355.

(44) Nishimura, T.; Onoue, T.; Ohe, K.; Uemura, S. Palladium(ll)-Catalyzed Oxidation Of Alcohols To Aldehydes And Ketones By Molecular Oxygen. J. Org. Chem. 1999, 64, 6750-6755.

(45) Steinhoff, B. A.; Stahl, S. S. Ligand-Modulated Palladium Oxidation Catalysis: Mechanistic Insights Into Aerobic Alcohol Oxidation With The Pd(Oac)2/Pyridine Catalyst System. Org. Lett. 2002, 4, 4179-4181.

(46) Steinhoff, B. A.; King, A. E.; Stahl, S. S. Unexpected Roles Of Molecular Sieves In Palladium-Catalyzed Aerobic Alcohol Oxidation. J. Org. Chem. 2006, 71, 1861-1868.

(47) Groppo, E.; Liu, W.; Zavorotynska, O.; Agostini, G.; Spoto, G.; Bordiga, S.; Lamberti, C.; Zecchina, A. Sub-Nanometric Pd Particles Stabilized Inside Highly Cross-Linked Polymeric Supports. Chem. Mater. 2010, 22, 2297-2308.

(48) Groppo, E .; Agostini, G.; Borfecchia, E.; Lazzarini, A.; Liu, W.; Lamberti, C.; Giannici, F.; Portale, G.; Longo, A. The Pyridyl Functional Groups Guide The Formation Of Pd Nanoparticles Inside A Porous Poly(4-Vinyl-Pyridine),. ChemCatChem 2015, 7, 21882195.

(49) Longo, A.; Lamberti, C.; Agostini, G.; Borfecchia, E.; Lazzarini, A.; Liu, W.; Giannici, F.; Portale, G.; Groppo, E. Pd Nanoparticles Formation Inside Porous Polymeric Scaffolds Followed By In Situ XANES/SAXS. J. Phys.: Conf. Ser. 2016, 712, 012039.

(50) Lazzarini, A.; Groppo, E.; Agostini, G.; Borfecchia, E.; Giannici, F.; Portale, G.; Longo, A.; Pellegrini, R.; Lamberti, C. Formation And Growth Of Palladium Nanoparticles Inside Porous Poly(4-Vinyl-Pyridine) Monitored By Operando Techniques: The Role Of Different Reducing Agents,. Catal. Today 2017, 283, 144-150.

(51) Note that the Pd loading was selected to maximize the amount of Pd, at the same time keeping the Pd dispersion as high as possible. 4 wt % corresponds to one-half of the adsorption capacity of the P4VP support.

(52) Nikitenko, S.; Beale, A. M.; Van Der Eerden, A. M. J.; Jacques, S. D. M.; Leynaud, O.; O'Brien, M. G.; Detollenaere, D.; Kaptein, R.; Weckhuysen, B. M.; W. B. Implementation Of A Combined SAXS/ WAXS/QEXAFS Set-Up For Time-Resolved In Situ Experiments. J. Synchrotron Radiat. 2008, 15, 632-640.

(53) Longo, A.; Portale, G.; Bras, W.; Giannici, F.; Ruggirello, A. M. Turco Liveri, V., Structural Characterization Of Frozen N-Heptane Solutions Of Metal-Containing Reverse Micelles. Langmuir 2007, 23, 11482-11487.

(54) Martis, V.; Beale, A. M.; Detollenaere, D.; Banerjee, D.; Moroni, M.; Gosselin, F.; Bras, W. A High-Pressure And Controlled-Flow Gas System For Catalysis Research. J. Synchrotron Radiat. 2014, 21, 462-463.

(55) Rehr, J. J.; Albers, R. C. Theoretical Approaches To X-Ray Absorption Fine Structure. Rev. Mod. Phys. 2000, 72, 621-654.

(56) Filipponi, A.; Di Cicco, A.; Natoli, C. R. X-Ray-Absorption Spectroscopy And N-Body Distribution Functions In Condensed Matter. I. Theory. Phys. Rev. B: Condens. Matter Mater. Phys. 1995, 52, 15122-15134.

(57) Hammersley, A. FIT2D: A Multi-Purpose Data Reduction, Analysis And Visualization Program,. J. Appl. Crystallogr. 2016, 49, 646-652.

(58) ESRF Internal Report. ESRF98HA01T, FIT2D V9.129 Reference Manual V3.1; 1998.

(59) Tessier, D.; Rakai, A.; Bozonverduraz, F. Spectroscopic Study Of The Interaction Of Carbon-Monoxide With Cationic And Metallic Palladium In Palladium Alumina Catalysts. J. Chem. Soc., Faraday Trans. 1992, 88, 741-749.

(60) Newton, M. A.; Brazier, J. B.; Barreiro, E. M.; Parry, S.; Emmerich, H.; Adrio, L. A.; Mulligan, C. J.; Hellgardt, K.; Hii, K. K. Operando XAFS Of Supported Pd Nanoparticles In Flowing Ethanol/

6880

Water Mixtures: Implications For Catalysis. Green Chem. 2016, 18, 406-411.

(61) Mccurdie, M. P.; Belfiore, L. A. Spectroscopic Analysis Of Transition-Metal Coordination Complexes Based On Poly(4-Vinyl-pyridine) And Dichlorotricarbonylruthenium(ll). Polymer 1999, 40, 2889-2902.

(62) Wu, K. H.; Wang, Y. R.; Hwu, W. H. FTIR And TGA Studies Of Poly(4-Vinylpyridine-Co-Divinylbenzene)-Cu(ll) Complex. Polym. Degrad. Stab. 2003, 79, 195-200.

(63) Pardey, A. J.; Rojas, A. D.; Yanez, J. E.; Betancourt, P.; Scott, C.; Chinea, C.; Urbina, C.; Moronta, D.; Longo, C. Spectroscopic Characterization Of Coordination Complexes Based On Dichlorocopper(II) And Poly(4-Vinylpyridine): Application In Catalysis. Polyhedron 2005, 24, 511-519.

(64) Belfiore, L. A.; Mccurdie, M. P.; Ueda, E. Polymeric Coordination-Complexes Based On Cobalt, Nickel, And Ruthenium That Exhibit Synergistic Thermal-Properties. Macromolecules 1993, 26, 6908-6917.

(65) Belfiore, L. A.; Pires, A. T. N.; Wang, Y. H.; Graham, H.; Ueda, E. Transition-Metal Coordination In Polymer Blends And Model Systems. Macromolecules 1992, 25, 1411-1419.

(66) Groppo, E.; Uddin, M. J.; Bordiga, S.; Zecchina, A.; Lamberti, C. Structure And Redox Activity Of Copper Sites Isolated In A Nanoporous P4VP Polymeric Matrix,. Angew. Chem., Int. Ed. 2008, 47, 9269-9273.

(67) Groppo, E.; Agostini, G.; Borfecchia, E.; Wei, L.; Giannici, F.; Portale, G.; Longo, A.; Lamberti, C. Formation And Growth Of Pd Nanoparticles Inside A Highly Cross-Linked Polystyrene Support: Role Of The Reducing Agent. J. Phys. Chem. C 2014, 118, 84068415.

(68) Frenkel, A. I. Solving The Structure Of Nanoparticles By Multiple-Scattering EXAFS Analysis. J. Synchrotron Radiat. 1999, 6, 293-295.

(69) Frenkel, A. I.; Hills, C. W.; Nuzzo, R. G. A View From The Inside: Complexity In The Atomic Scale Ordering Of Supported Metal Nanoparticles,. J. Phys. Chem. B 2001, 105, 12689-12703.

(70) Frenkel, A. Solving The 3D Structure Of Metal Nanoparticles. Z. Kristallogr. - Cryst. Mater. 2007, 222, 605-611.

(71) Witkowska, A.; Di Cicco, A.; Principi, E. Local Ordering Of Nanostructured Pt Probed By Multiple-Scattering XAFS. Phys. Rev. B: Condens. Matter Mater. Phys. 2007, 76, 1 DOI: 10.1103/Phys-RevB.76.104110.

(72) Agostini, G.; Pellegrini, R.; Leofanti, G.; Bertinetti, L.; Bertarione, S.; Groppo, E.; Zecchina, A.; Lamberti, C. Determination Of The Particle Size, Available Surface Area, And Nature Of Exposed Sites For Silica-Alumina-Supported Pd Nanoparticles: A Multitechnical Approach,. J. Phys. Chem. C 2009, 113, 10485-10492.

(73) Agostini, G.; Piovano, A.; Bertinetti, L.; Pellegrini, R.; Leofanti, G.; Groppo, E.; Lamberti, C. Effect Of Different Face Centered Cubic Nanoparticle Distributions On Particle Size And Surface Area Determination: A Theoretical Study,. J. Phys. Chem. C 2014, 118, 4085-4094.

(74) Sun, Y.; Frenkel, A. I.; Isseroff, R.; Shonbrun, C.; Forman, M.; Shin, K. W.; Koga, T.; White, H.; Zhang, L. H.; Zhu, Y. M.; Rafailovich, M. H.; Sokolov, J. C. Characterization Of Palladium Nanoparticles By Using X-Ray Reflectivity, EXAFS, And Electron Microscopy. Langmuir 2006, 22, 807-816.

(75) Lee, A. F.; Ellis, P. J.; Fairlamb, I. J. S.; Wilson, K. Surface Catalysed Suzuki-Miyaura Cross-Coupling By Pd Nanoparticles: An Operando XAS Study. Dalton Trans. 2010, 39, 10473-10482.

(76) Mouat, A. R.; Whitford, C. L.; Chen, B.-R.; Liu, S.; Perras, F. A.; Pruski, M.; Bedzyk, M. J.; Delferro, M.; Stair, P. C.; Marks, T. J. Synthesis Of Supported Pd0 Nanoparticles From A Single-Site Pd2+ Surface Complex By Alkene Reduction. Chem. Mater. 2018, 30, 1032-1044.

(77) It is worth noticing that the apparent discrepancy of the (dSAXS) and (dTEM) values can be explained by considering that the SAXS signal is weighted for r6 (see eq 1); hence bigger particles contribute to the SAXS signal much more than the smaller ones. However, the

particle size distribution determined by HR-TEM relies only on the small particles.

(78) Huang, W.; Zhang, S.; Tang, Y.; Li, Y.; Nguyen, L.; Li, Y.; Shan, J.; Xiao, D.; Gagne, R.; Frenkel, A. I.; Tao, F. F. Low-Temperature Transformation Of Methane To Methanol On Pd < Inf > 1</Inf > O < Inf > 4</Inf > Single Sites Anchored On The Internal Surface Of Microporous Silicate. Angew. Chem., Int. Ed. 2016, 55, 13441-13445.

(79) Ingram, A. J.; Solis-Ibarra, D.; Zare, R. N.; Waymouth, R. M. Trinuclear Pd3O2 Intermediate In Aerobic Oxidation Catalysis. Angew. Chem., Int. Ed. 2014, 53, 5648-5652.

6881

ISSN 1600-5775

Fluorescence-detected XAS with sub-second time resolution reveals new details about the redox activity of Pt/CeO2 catalyst

Alexander A. Guda,a Aram L. Bugaev,a,b Rene Kopelent,c Luca Braglia,

a,b

Received 23 January 2018 Accepted 4 April 2018

Edited by S. Diaz-Moreno, Diamond Light Source, UK

Keywords: time-resolved XAS; fluorescence detection; heterogeneous catalysis; transient kinetics; CO oxidation; automotive catalysts; ceria; platinum..

Supporting information: this article has supporting information at journals.iucr.org/s

Alexander V. Soldatov/ Grigory Smolentsevc*

Maarten Nachtegaal,c Olga V. Safonovac* and

aThe Smart Materials Research Center, Southern Federal University, Sladkova 174/28, Rostov-on-Don 344090, Russian Federation, bDepartment of Chemistry, NIS and CrisDi Interdepartmental Centres, asn INST Reference Center, University of Turin, Via P. Giuria 7, Turin 10125, Italy, and cPaul Scherer Institute, Villigen 5232, Switzerland. ^Correspondence e-mail: [e-mail], [e-mail]

A setup for fluorescence-detected X-ray absorption spectroscopy (XAS) with sub-second time resolution has been developed. This technique allows chemical speciation of low-concentrated materials embedded in highly absorbing matrices, which cannot be studied using transmission XAS. Using this setup, the reactivity of 1.5 wt% Pt/CeO2 catalyst was studied with 100 ms resolution during periodic cycling in CO- and oxygen-containing atmospheres in a plug-flow reactor. Measurements were performed at the Pt L3- and Ce L3-edges. The reactivity of platinum and cerium demonstrated a strong correlation. The oxidation of the catalyst starts on the ceria support helping the oxidation of platinum nanoparticles. The new time-resolved XAS setup can be applied to various systems, capable of reproducible cycling between different states triggered by gas atmosphere, light, temperature, etc. It opens up new perspectives for mechanistic studies on automotive catalysts, selective oxidation catalysts and photocatalysts.

1. Introduction

OPEN Q ACCESS

When performing a catalytic reaction, the atoms in the catalyst's active site periodically change their local coordination and often the oxidation state. Detection of intermediate species and distinguishing them from inactive spectators allows the reaction mechanisms to be unravelled and helps the rational design of better catalysts. This type of research requires the development of highly sensitive spectroscopic methods (Weckhuysen, 2003; Meunier, 2010; Urakawa, 2016; Beale et al., 2010) combined with advanced theory (Campbell, 2017). Spectroscopic methods with sub-second time resolution are needed to detect relevant intermediates (Gott & Oyama, 2009; Burch et al., 2011; Kopelent et al., 2015). Some important catalytic materials also operate under transient conditions. These are, for example, automotive exhaust catalysts (Trovarelli, 2002) and catalysts used in chemical looping processes (Guo et al., 2014). Time-resolved spectroscopy is crucial to understand the functioning of these materials on the atomic scale (Yamamoto et al., 2007; Nagai et al., 2009; Newton et al., 2010, 2012; Ferri et al., 2014).

The use of time-resolved X-ray absorption spectroscopy (XAS) for in situ and operando studies of catalysts has increased significantly during the last few decades (Frenkel et al., 2013; Newton & Dent, 2013). Several beamlines pioneering in this field (Frahm, 1988; Kaminaga et al., 1981) are now

complemented by new facilities equipped with oscillating channel-cut monochromators (Stotzel et al., 2010; Fonda et al., 2012; Nonaka et al., 2012; MUller et al., 2016), energy-dispersive (ED) polychromators (Pascarelli et al., 2016; Kong et al., 2012) and double-crystal monochromators performing rapid continuous energy scans (van Beek et al., 2011; Nikitenko et al., 2008; Dent et al., 2009). For scanning monochromators, the acquisition time per 1000 eV range XAS spectrum can be varied depending on the setup from minutes to milliseconds. A time of 10 ms for high-quality XAS spectra was recently achieved by combining a novel oscillating channel-cut mono-chromator with fast ionization chambers (Muller et al., 2016). ED polychromators operating under static conditions nowadays allow XAS spectra to be measured even faster, with sub-millisecond resolution (Pascarelli et al., 2016; Pascarelli & Mathon, 2010).

However, XAS data acquisition with sub-second resolution is typically performed in transmission mode. This limits the application of time-resolved XAS for many types of industrially relevant catalytic materials which either contain low concentration of the element of interest or are supported on a highly X-ray absorbing matrix. Typical examples of such materials are metal-loaded ceria-zirconia catalysts used in the automotive industry and a large variety of oxide-supported catalysts containing less than 0.5% of the element of interest. Limitations can also arise from specific experimental conditions and reactor design not allowing for transmission detection. Therefore, the development of setups for measuring time-resolved fluorescence-detected XAS is of great importance for the catalysis community. Significant efforts have already been made in the development of such setups using PIPS silicon photodiodes combined with ED polychromators (Nagai et al., 2009) and oscillating monochromators (Haumann et al., 2005; Yao et al., 2014; Zhang et al., 2004; Konig et al., 2014). However, XAS spectra obtained with silicon diodes contain strong background due to the lack of energy resolution. Therefore, these setups are not ideal for low-concentrated samples and detection of small spectral changes. Techniques for sub-second XAS acquisition using single-photon-counting fluorescence detectors are currently available for the 30 ns to 1 ms time range (Smolentsev et al, 2014) and used for pump-and-probe studies of homogeneous photocatalysts excited by laser light. Using an X-ray emission spectrometer in the von Hamos geometry one can also measure X-ray absorption and emission spectra in fluorescence mode with sub-second time-resolution (Szlachetko et al., 2012, 2013; Kopelent et al., 2016). However, the efficiency of this detection system in terms of statistics is typically worse than that of standard single-photon-counting fluorescence detectors. This is due to the high-energy-resolution optics which selects only a small part of fluorescence photons and due to the low solid angle of currently available emission spectrometers. Therefore this detection system is also not ideal for low-concentrated samples.

In this work we developed a setup for time-resolved XAS acquisition with 20 ms to 1 s resolution. Using this method we revealed new details about the initial redox kinetics of a

1.5 wt% Pt/CeO2 catalyst under transient conditions. Noble metals on ceria-based supports are widely used by the automotive industry to reduce the content of CO, nitrous oxides and unreacted hydrocarbons in the car exhaust. These catalysts operate under fast oscillations of the gas composition (Trovarelli, 2002); therefore, it is important to unravel the role of each component in such catalysts with sub-second timeresolution. The low concentration of metal nanoparticles, as well as the highly absorbing nature of ceria, makes it difficult to gain a detailed understanding of the structure-function relationships. Until now, the dynamic structure of these catalysts was typically analysed by time-resolved high-energy X-ray diffraction and infrared spectroscopy (Newton et al., 2010, 2012; Ferri et al., 2014) which could not provide quantitative information on the oxidation state and local coordination of all elements of interest. Studies by time-resolved X-ray absorption methods on these catalysts are scarce (Yamamoto et al., 2007; Nagai et al., 2009; Marchionni et al., 2016; Gibson et al., 2017) due to the compromises between the optimal catalyst composition, the reaction conditions and the concentration of elements of interest in the X-ray beam. The setup developed in this contribution opens new opportunities for various catalyst compositions and allows for quantitative correlations between the reactivity of the noble metal and the support in plug-flow reactors down to 100 ms time resolution.

2. Materials and method

2.1. X-ray source

Ce L3 and Pt L3 XAS spectra were measured at the SuperXAS beamline (Abdala et al., 2012) of the Swiss Light Source (SLS) at Paul Scherrer Institute, Switzerland. The incident beam was provided by a 2.9 T super-bend magnet. The SLS was running in the standard top-up mode at 2.4 GeV and with an average current of 400 mA. The Si surface of a collimating mirror at 2.5 mrad was used for harmonic rejection at the Ce L3-edge. For measurements at the Pt L3-edge we used the Rh-coated surface of this mirror. The energy was selected by a Si(111) channel-cut monochromator, which provides an energy resolution of AE/E = 2.0 x 10~4. Ce L3-edge static data were collected from 5620 eV to 5920 eV with a step of 1 eV near the edge and 3 eV in the EXAFS region. Pt L3-edge static data were collected from 11455 eV to 11755 eV with a step of 1 eV near the edge and 3 eV in the EXAFS region. The beam was focused down to 0.1 mm in the vertical direction and 0.5 mm in the horizontal direction by a Rh-coated toroidal mirror. A smaller vertical beam size was used to scan the catalyst bed along the flow direction. The photon flux obtained at the sample was about 3-5 x 1011 photons s_1 at the Ce L3-edge and 5-7 x 1011 photons s_1 at the Pt L3-edge.

2.2. Data acquisition system

Fig. 1 shows a detailed scheme of the data acquisition (DAQ) system. In this DAQ, the fluorescence signal from

Fiber optic Figure 1

Scheme of the data acquisition system. SDD: five-element silicon drift detector; APD: avalanche photodiode; DXP: digital X-ray processor XIA XMAP operating in the multi-channel analyser mapping mode; ADC: analog-to-digital converter; AIC: analog input card. Blue boxes refer to the detectors; green boxes indicate the signal processing devices; red boxes correspond to the computers. The thick arrows show the communication lines between the devices with possible time delays, and the thin arrows refer to the delay-free connections. The control and analysis computer initiates an acquisition cycle, which is afterwards controlled only by a signal generator.

plug-flow reactor cell (Chiarello et al., 2014) (Fig. 2) between two quartz wool plugs. High-surface-area ceria (85 m2 g_1) in the shape of truncated octahedral particles was prepared by a hydrothermal method. Platinum nano-particles of 1.2 (± 0.2) nm diameter were supported on ceria by incipient wetness impregnation by tetraammine platinum(II) nitrate (Aldrich, 99.995%) followed by calcination in air (400°C, 4 h) and reduction in 5% H2 flow (300°C, 4 h). Further details of sample preparation and characterization are described by Kopelent et al. (2015). The reactor was connected to 5% CO in argon (CO 4.7 purity, Ar 5.0 purity), 21% O2 in argon (O2 4.5 purity, Ar 5.0 purity) and argon (4.8 purity). The vertical orientation of the reactor

the sample was detected by a five-element SGX silicon drift detector (SDD) and processed by XIA electronics (Digital X-ray Processor DXP-XMAP) operating in the multi-channel analyser mapping mode. In comparison, the pump-sequential-probe method with time-tagged photon counting uses a different, so-called 'list mapping', mode, allowing for 30 ns to 1 ms time resolution (Smolentsev et al, 2014). The energy resolution of the detector was used to eliminate the elastic scattering and fluorescence signal from other elements.

During the time-resolved experiments the gas composition in the reactor was periodically switched by two three-way switching valves (Parker, Series 9) which were triggered by a TTL signal generated by the beamline control system using an analog input card (AIC) (Hytec Electronics, Model VTB8204) (Fig. 2). The same TTL signal was sent simultaneously to a signal generator (Berkeley Nucleonics, Model 645, 50 MHz function/arbitrary waveform generator). Using the arrival of the TTL signal as the starting time, the signal generator produced a train of square pulses of equal duration and sent it to the gate of the XIA electronics. Each pulse triggered the collection of photons by the SDD detector. The width of each pulse determines the length of acquisition in each time point and, thus, the overall time-resolution of the experiment. After measuring the last time point the monochromator was moved to the next energy point and the time-resolved measurements, as described above, were repeated. To produce reliable data the reproducibility of the chemical processes in the catalytic cell is essential. To control it, we were always repeating transient experiments several times at one fixed energy (where the largest spectral changes were expected) before performing an experiment over the full energy range.

2.3. Sample and catalytic setup

34 mg of 1.5 wt% Pt/CeO2 catalyst powder (Kopelent et al., 2015) sieved to 100-150 mm grain size was placed into an in situ

29 30 31 sample position, mm Figure 2

(a) Scheme of the experimental setup: a TTL signal initiates periodic switches between two gas flows in the cell and triggers the signal generator (SG) to produce a train of a fixed number of pulses that control the data acquisition system for the SDD detector (Det). (b) Vertical scan along the reactor cell using the Ce La fluorescence signal showing the beginning and the end of the catalyst bed with respect to the direction of the gas flow. The inset shows a TEM image of 1.5 wt% Pt/CeO2 catalyst (Pt nanoparticles are indicated by arrows).

enabled scanning of the reactor cell with a focused X-ray beam along the gas flow direction with 0.1 mm resolution. Mass flow controllers were producing two gas mixtures with the same total flow (50 ml min-1). One flow was directed to the cell while the other was going to the exhaust using two three-way switching valves (Fig. 2). A Pfeiffer Omnistar mass spectrometer was connected to the reactor outlet to analyse the reaction products. The reaction gases have not been preheated. However, we controlled the temperatures of the cell body and monitored the sample temperature at the beginning of the catalyst bed using a separate thermocouple. Working at relatively low temperatures, the difference between the cell and the sample temperatures did not exceed 1-2° C and changes in the catalyst temperature upon gas switching were also below 2°C. Thus, we do not expect a strong temperature gradient along the catalyst bed.

3. Results and discussion 3.1. Performance of the setup

The setup presented in this work uses a step-by-step energy scanning mode of the monochromator. It is designed under the assumption that the system under study can be repro-ducibly cycled between different states. In contrast, fast-scanning quick-XAS and static ED XAS setups can also be applied for studies of fast irreversible and non-reproducible processes, especially when the experiment can be performed in transmission mode on concentrated samples with optimized thickness. For low-concentrated samples this does not work due to insufficient statistics of the weaker fluorescence signal. Therefore, reproducibility and periodic cycling strategy (pump-and-probe strategy) are needed for low-concentrated systems to obtain high-quality time-resolved XAS spectra. As a step-by-step energy scanning does not bring additional limitations to the type of catalytic systems that can be studied, we have chosen this mode as it does not require any synchronization between the monochromator and the DAQ system.

The normalization to the incoming beam intensity (10) is important for the detection of small spectral differences as in the top-up mode of the SLS storage ring the electron injections occur every three minutes changing the intensity by approximately 0.5%. We implemented two options for 10 normalization: (i) an avalanche photodiode (APD) detecting the signal from the elastic scattering of the incoming beam by air and registering it by an additional channel of the XIA electronics, and (ii) an ionization chamber placed before the sample and registering the transmission signal by a separate DAQ based on a fast ADC (National Instruments, Model PXIe-6366) card synchronized with the XIA DAQ by the TTL signal. The first option is simpler but the maximal count rate for one channel of the XIA is limited to 500000 counts s_1. If the concentration of the element of interest in the sample is low and the summed count rate for a selected fluorescence line is significantly lower than 500000 counts s_1, this method of

normalization is sufficient. For concentrated samples (e.g. CeO2 support), when the count rate for the selected fluorescence line is much higher, 10 measured with this method will contribute to the noise level of the normalized spectra significantly. In this case, the use of the ionization chamber that measures 10 with good signal-to-noise ratio 10~4) is preferable. For example, when measuring time-resolved Ce L3 XAS spectra of a 0.5 wt% Pt/CeO2 catalyst with option (i) at 5726 eV we accumulated 200000 counts per time point with an APD detector and 120000 counts per time point from the Ce La fluorescence line. This gives the noise level of 10 of 0.3% and therefore this 10 signal could not be used to correct for 0.5% intensity jumps related to top-up mode of the synchrotron (see Fig. S1 of the supporting information). These intensity jumps were apparent in the data when the changes in the oxidation state of cerium during redox cycling of catalyst were less than 5%. With option (ii) for the 10 normalization, the noise level arising from the ionization chamber is 50 times smaller (0.006%).

The proposed DAQ scheme requires minimal communication between the control software and the detector, which allows any millisecond time delays or jitters to be avoided. The user defines the time that the sample spends in the different gas environments that are being periodically cycled and the overall time-resolution, which is then used to program the signal generator. In the multi-channel analyser mapping mode the full energy spectrum of each channel of the fluorescence detector is buffered at the XIA electronics for each time point, then streamed to the detector computer via an optical fiber connection, and finally saved to the file server in HDF5 format. A Python script reads the HDF5 files and processes the data in the energy and time dimensions for each detector channel to obtain kinetic and spectroscopic information. The script works with two threads: one controls the energy scan and acquisition, while the other is responsible for processing of the output files and visualization. There are several factors that can potentially limit the time resolution of the setup:

(i) time resolution of the detection and the DAQ system,

(ii) speed of the triggering, and (iii) speed of the data transfer and processing. The DXP card is rather fast: it has a 20 ns sampling time and thus provides a time resolution of the same order. SDD detectors are a bit slower: they have a resolution of about 1 ms, which is still sufficient for the present setup. Triggering has a nanosecond precision. The speed of data transfer and on-the-fly processing can be a limiting issue under some conditions. During preliminary tests with a slow and unstable 1 Gb network connection, the time resolution of the setup was limited to 100 ms. After a successful upgrade to a faster 10 Gb connection, the file transfer speed and the processing are not the limiting factors even for 20 ms resolution. As a possibility for further optimization, the DXP can be switched from the multi-channel analyser to the singlechannel analyser mapping mode. In this case, the detector will only record the sum of the counts within a certain region of interest instead of the whole energy spectrum. This will significantly reduce the amount of data but will limit the possibilities for data reprocessing after the experiment.

In the current setup, the exchange of the gas atmosphere in the plug-flow reactor limits the time resolution of the whole system. We could detect changes in the catalyst state by XAS after 100 ms after the switch. The full replacement of the gas composition in the cell takes even longer; 90% gas exchange at 50 ml min-1 flow rate takes about 2 s. We demonstrated this by switching between 1% O2 in argon and pure argon and monitoring the decay of the m/z signal of 32 with mass spectrometry. These values can still be optimized by a further reduction of the dead volume in the cell and by increasing the flow rate. For some catalytic systems, alternative faster methods to trigger the changes in the reaction conditions can be used. For example, for photocatalytic systems, switching the light on and off can be used as a reaction trigger. With a LED-based source, the switching time can be reduced down to 1 ms, so that the speed of the reaction triggering will no longer be the limiting factor.

o.o-r—_,_,_r—J

11560 11580 11600 11620 Energy, eV

Figure 4

Pt L3 XAS spectra for 1.5 wt% Pt/CeO2 catalyst under reducing conditions (1% CO in argon at 150° C) and oxidizing conditions (4% O2 in argon at 150° C) are compared with the spectra of Pt foil and PtO2 references.

3.2. Quantitative analysis of time-resolved XAS spectra

Fig. 3 shows the static Ce L3 and Pt L3 XAS spectra of the 1.5 wt% Pt/CeO2 catalyst in oxidizing (4% O2 in argon) and reducing (1% CO in argon) conditions at 150°C. Due to the high concentration of cerium in the catalyst, the Ce L3 XAS spectra demonstrate significant self-absorption. According to our previous studies (Safonova et al., 2014), in 4% O2 at 150°C, the catalyst is fully oxidized to Ce4+ and when exposed to 1% CO it contains 11% Ce3+. This information allows for quan-

11580 11600 Energy, eV

Figure 3

Static XAS spectra of 1.5 wt% Pt/CeO2 catalyst above the Ce L3 (a) and Pt L3 (b) edges measured under reducing (1% CO in argon) and oxidizing (4% O2 in argon) conditions at 150°C.

tification of the Ce + concentration in transient experiments without self-absorption correction. Static XAS spectra at the Pt L3-edge also demonstrate significant changes as a function of the gas atmosphere composition. A comparison of the spectrum of the catalyst measured at 150°C in 4% O2 with the spectra of the platinum foil and PtO2 references (Fig. 4) suggests that the platinum nanoparticles 1.2 nm in diameter) on ceria are partially oxidized.

In 1% CO atmosphere, the Pt nanoparticles are reduced to the metallic state. The broad shape of the Pt L3-edge white line suggests CO adsorption at the platinum surface (Niki-tenko et al, 2008; Safonova et al., 2006; Small et al., 2012). The static Pt L3 XAS spectra of the catalyst in oxidizing (4% O2) and reducing (1% CO) atmospheres are almost independent of the temperature indicating that the static CO and oxygen coverage does not change significantly in the temperature range 25-150° C. Therefore, we preferred to quantify the changes in the coverage of platinum by oxygen rather than the oxidation state of platinum.

By periodically changing the gas composition the catalyst was periodically reduced and oxidized. At each energy point of the full XAS spectrum these switches were performed and the signal was recorded with a 100 ms time resolution. The resulting reconstructed time-resolved Ce L3 XAS data were analysed quantitatively using the static XAS spectra plotted in Fig. 3. The 900 time-resolved XAS spectra (corresponding to a 180 s cycle with 100 ms time resolution) at 32 energy points (marked as stars in Fig. 5b) were normalized to the edge jump of one and processed by using principal component analysis (PCA) and linear combination fit within the Fitlt software package (Smolentsev & Soldatov, 2007; Smolentsev et al, 2009). The spectra for the pure components were determined using all time-resolved XAS spectra plus the static XAS spectra of the same catalyst containing 0% and 11% of Ce3+. In addition, we applied the following constraints: the concentrations cannot be negative and the sum of all components in each spectrum should be equal to 100%.