The Use of Molecular Oxygen for Liquid Phase Aerobic Oxidations in Continuous Flow
- 470 Downloads
Molecular oxygen (O2) is the ultimate “green” oxidant for organic synthesis. There has been recent intensive research within the synthetic community to develop new selective liquid phase aerobic oxidation methodologies as a response to the necessity to reduce the environmental impact of chemical synthesis and manufacture. Green and sustainable chemical processes rely not only on effective chemistry but also on the implementation of reactor technologies that enhance reaction performance and overall safety. Continuous flow reactors have facilitated safer and more efficient utilization of O2, whilst enabling protocols to be scalable. In this article, we discuss recent advancements in the utilization of continuous processing for aerobic oxidations. The translation of aerobic oxidation from batch protocols to continuous flow processes, including process intensification (high T/p), is examined. The use of “synthetic air”, typically consisting of less than 10% O2 in N2, is compared to pure O2 (100% O2) as an oxidant source in terms of process efficiency and safety. Examples of homogeneous catalysis and heterogeneous (packed bed) catalysis are provided. The application of flow photoreactors for the in situ formation of singlet oxygen (1O2) for use in organic reactions, as well as the implementation of membrane technologies, green solvents and recent reactor solutions for handling O2 are covered.
KeywordsContinuous flow Flow reactor Continuous processing Aerobic oxidation Molecular oxygen Process intensification Membranes Photochemistry Green solvents
Fluorinated ethylene propylene
High performance liquid chromatography
Light emitting diode
Megatonne per annum
Singlet state oxygen
Triplet state oxygen
Molecular oxygen (O2) is inexpensive, the most readily available oxidant on Earth, and completely harmless to the environment. O2 is therefore perhaps the greenest reagent available to the organic chemist . Furthermore, O2 is a nontoxic gas and is easy to remove after a reaction. Aerobic oxidation reactions are generally very green because they typically display high atom economy and, in most cases, water is the only stoichiometric byproduct. Until very recently, classical oxidation methods using stoichiometric quantities of toxic inorganic oxidants, such as CrO3, KMnO4 and MnO2, were favored in organic synthesis, even though these protocols generally display poor atom economy and use highly energetic oxidants . More recently adopted oxidation approaches use less toxic oxidants, such as dimethylsulfoxide (DMSO) and hypervalent iodine compounds, but are no less green. As social concern regarding the environmental impact of chemical processes gains more interest, there is an increasing demand to design more sustainable chemical methodologies. Anastas introduced the 12 principles of green chemistry, outlining the steps necessary for more sustainable synthesis practices . Over the last 10–15 years, groundbreaking progress has been made in the development of highly selective aerobic oxidation reactions . The replacement of toxic and corrosive stoichiometric oxidants with processes that use O2 combined with catalytic methodologies will ensure atom efficient and selective synthetic oxidation approaches that are sustainable into the future .
Oxidation chemistry utilizing pure O2 or air as the oxidant source is already used extensively within the bulk and commodity chemical manufacturing sector . For example, 6 basic chemicals are produced using pure O2 and 12 chemicals using air at > 2 Mt/a scale. In the bulk and commodity chemicals sector, the use of air and O2 as the oxidant source is driven by the requirement to keep costs as low as possible. However, O2 is underutilized as an oxidant within the fine and pharmaceutical chemical industry. The bulk chemicals sector deals with low value, high volume products and the corresponding production plants are generally designed and engineered as dedicated continuous processes, whereas fine chemicals and the pharmaceutical sector have historically favored the use of multipurpose batch reactors for the manufacture of high value, low volume products . There are unique process challenges associated with handling gas–liquid transformations within multipurpose batch reactors. Efficient mixing between the liquid phase and gas phase is difficult to achieve; therefore, reactions are often mass transfer limited, which leads to problems when scaling up from laboratory to manufacturing scale. The solubility of O2 in water and organic solvents is poor, thus the reactor needs to be pressurized to maximize the amount of gas in solution to reduce mass transfer effects. Typical commercial scale batch reactors can operate between 2 and 6 bar; therefore, higher pressures require more specialized and expensive equipment. In addition, aerobic oxidation reactions are typically highly exothermic, meaning the heat generated needs to be efficiently removed. These challenges, and the fact that the reaction utilizes potentially flammable O2 under certain conditions, unfortunately increase the perceived scale-up risk, which has rendered the use of O2 virtually unacceptable for pharmaceutical and fine chemical synthesis.
The challenges associated with handling O2 are better addressed by using continuous processing than multipurpose batch reactors [7, 8]. There is a current paradigm shift in the pharmaceutical industry from traditional batch manufacturing to continuous processing for the preparation of active pharmaceutical ingredients (APIs) [9, 10, 11, 12]. This paradigm shift is reflected by a new focus in the pharmaceutical industry on process intensification, sustainability, product quality, safety, energy usage and cost . The United States Food and Drug Administration (FDA) is taking proactive steps to facilitate the implementation of continuous manufacturing within the pharmaceutical industry as an attempt to improve product quality and reduce the environmental impact of pharmaceutical manufacture . The University of Wisconsin-Madison Oxidation Consortium (MadOx) involving Eli Lilly and Co., Merck and Pfizer was established in 2012 as a precompetitive collaboration aimed at solving the challenges associated with aerobic oxidations in pharmaceutical manufacturing . In particular, the consortium focused on the development of safe and scalable continuous flow technologies for aerobic oxidation reactions. Recent reviews have provided overviews of the significant progress made in the last decade towards the utilization of O2 within continuous flow environments [16, 17, 18].
A significant obstacle to the uptake of aerobic oxidation reactions is that undergraduate organic chemistry textbooks still teach classical oxidation methods, which use toxic inorganic oxidants in stoichiometric quantities rather than more recently developed greener aerobic oxidation strategies. Therefore, organic chemists lack the necessary knowledge to implement these new greener methods. There are hurdles to the implementation of large-scale aerobic oxidations owing to the lack of experience and equipment within pharmaceutical manufacturing. In this article, we highlight selected synthetic examples of liquid phase aerobic oxidation reactions under continuous flow conditions. The first section deals with the process aspects associated with utilizing aerobic oxidation reactions, and also gives an overview of a typical continuous flow setup for performing aerobic oxidations. Subsequently, homogeneous catalysis and heterogeneous catalysis examples are discussed. The utilization of photochemistry for the in situ formation of singlet oxygen (1O2) from ground state triplet oxygen (3O2) is treated only briefly, owing to the large number of examples published. The use of supercritical fluids and liquid carbon dioxide (CO2) as green solvents for aerobic oxidations is examined. Membrane technologies, new reactor developments and scale-up strategies are discussed. The advantages and challenges associated with the utilization of continuous processing for liquid phase aerobic oxidations are highlighted throughout.
2 Process Aspects
2.1 Mass and Heat Transfer
Interfacial area to volume ratio for different reactor types
(data from )
Type of reactor
Interfacial area to volume ratio (m2 m−3)
Impinging jet absorbers
Packed columns, concurrent
Packed columns, counter current
Laboratory scale stirred tank (Fig. 1)
Tube reactors, horizontal and coil (Fig. 2)
Tube reactors, vertical
Gas–liquid microchannel contactor
2.3 Using Diluted O2
Limiting oxygen concentration (LOC) data for organic solvents. NMP N-Methyl-2-pyrrolidone, DMSO dimethylsulfoxide, 2-MeTHF 2-methyltetrahydrofuran
(data from )
LOC (vol %)
2.4 Ability to Use Pure O2
The minimum ignition energy (MIE) is the lowest energy required for an oxygen/organic vapor mixture to spontaneously ignite . The MIE of flammable mixtures are over ten orders of magnitude lower for pure O2 than for air. Most safety studies carried out in microreactors examine the use of O2 for reactions occurring in the gas phase. Veser demonstrated for a Pt-catalyzed H2/O2 reaction to H2O2 that explosion propagation can be completely suppressed at channel sizes below the millimeter range, and thus the process is inherently safe . However, at larger channel dimensions (> 0.4 cm) Poliakoff and co-workers, when investigating the catalytic dehydrogenation of 4-vinylcyclohexane, observed periodic temperature spikes near the surface of the Pd/Al2O3 catalyst bed that indicated the occurrence of cycles of propagating flames . In the case of liquid phase aerobic oxidation reactions, O2 is substoichiometric to solvent, which significantly reduces the likelihood of an explosion. Small oxygen segments alleviate the likelihood that autoignition will occur, because the small channel dimensions do not exceed typical quenching distances for explosion propagation. Furthermore, the solvent plays a role as a heat sink. Unlike batch reactors, tubular flow reactors possess no headspace; therefore there is no headspace containing a large volume of potentially combustible oxygen/organic vapor. Nonetheless, the safety associated with a process should be assessed carefully on a case by case basis. Safe operation can be ensured by employing a properly designed continuous flow reactor that can withstand an explosion event in a worst case scenario . A key benefit of continuous processing is that, generally, a far smaller inventory of the overall material to be processed is present within the system at any one time. Miniaturization reduces the risks and allows for secondary containment of the reactor in the case of an explosion event.
2.5 Scale-up and Manufacture
When there is sufficient understanding of a reaction system and adequate process design to address safety concerns and mitigate risks, aerobic oxidations, even using pure O2, can be adopted at large scales through the utilization of appropriate continuous-flow processing systems. Experiments including microcalorimetry, differential scanning calorimetry (DSC) and autoclave explosion pressure measurements should focus on minimizing the perceived scale-up risk through contingency planning for worst case scenarios . There are a number of scale-up strategies that can be applied, including: (1) running the process for a longer time in the same equipment (scale-out); (2) a larger reactor volume with the same channel diameter but faster flow rates; (3) unit parallelization (numbering up); and (4) channel dimension enlarging to provide a larger volume through smart dimensioning . Examples of all of these strategies are shown below.
3 Homogeneous Catalysis
3.1 Pd-Catalyzed Reactions
Oxygen in its ground state, triplet oxygen (3O2), displays relatively low reactivity and poor selectivity; therefore, a catalyst system and/or elevated temperatures and pressures are required to increase reaction rates and improve selectivity. Palladium is perhaps the most studied metal for homogeneous catalyzed aerobic oxidations. A broad range of homogeneous Pd-catalyzed aerobic oxidations reactions have been developed over the last 10–15 years . Palladium catalysts are very sensitive to the oxygen concentration. Pd(II) is reduced to Pd(0) species, which aggregate to form inactive Pd black . This phenomenon causes a significant challenge when attempting to scale-up this chemistry under batch conditions due to poor mixing and temperature control. The direct oxidation of Pd(0) by O2 is kinetically unfavored. With this in mind, the utilization of continuous flow reactors that provide good heat and mass transfer can be beneficial for this type of chemistry by preventing catalyst decomposition through the rapid reoxidation of Pd(0) to Pd(II).
The same group also reported the connection of two different C–H bonds via a cross-dehydrogenative Heck reaction of indoles and alkenes to prepare vinylindoles . A small library of 3-vinylindole derivatives was prepared in residence times between 10 and 20 min under continuous flow conditions (Scheme 4b).
3.2 Cu-Catalyzed Reactions
As stated above in the section Using Diluted O2, a limitation of using O2 diluted with N2 is that the O2 is competing with N2 for dissolution in the liquid phase. Favre-Réguillon and co-workers studied the same Cu(I)/TEMPO alcohol oxidation but used pure O2 as the oxygen source (Scheme 6b) . They argued that by utilizing pure O2 it would be possible to operate the system at a lower pressure and decrease the reaction temperature, because the reaction would be less likely to be mass transfer limited. A substantially enhanced reaction rate and similar yields were achieved by using higher concentrations of O2. The same residence time could be used to obtain similar yields at 5 bar pressure and room temperature. Superior space time yields (i.e., the product yield per unit of time and per reactor volume) were achievable by using pure O2 when compared to using diluted O2.
By using a similar continuous flow configuration, Pieber and Kappe also developed a flow protocol for the Fe-catalyzed aerobic oxidation of 2-benzylpyrdines to their corresponding ketones (Scheme 7b) . Propylene carbonate could be used as solvent instead of using environmentally less desirable DMSO. Propylene carbonate is a nontoxic and biodegradable cyclic carbonate.
4 Heterogeneous Catalysis
A limitation of homogeneous catalysis is that the product needs to be separated from the catalyst after the reaction. The use of a heterogeneous catalyst is one method to prevent the product from becoming contaminated by the catalyst because it is in a different phase, provided that leaching into the liquid phase does not occur. The active metal is dispersed on a support, such as carbon, metal oxide or other inorganic material as a packed bed within a flow system . A number of techniques are used for the preparation of catalysts, including impregnation, adsorption, precipitation or ion exchange . The stabilization of a catalyst on an inert solid support can also improve the thermal stability of catalysts. The improved thermal stability is particularly beneficial given the high temperatures often employed within continuous flow reactors. The incorporation of one or more promoters, derived from the early transition metals, lanthanides and/or main group elements, can further modulate activity and selectivity. However, additional challenges exist for heterogeneous catalyst systems compared to their homogeneous counterparts. Isothermal temperature control can be difficult to obtain and the efficient mixing between the gas, liquid and solid phases can be difficult to achieve . A high steady-state conversion is sometimes not possible to achieve due to catalyst deactivation and/or leaching . An additional difficulty regarding their widespread uptake is that the preparation of heterogeneous catalysts is often outside the skill set of a standard organic chemist.
5 Uncatalyzed Reactions
6 Organomagnesium and Organolithium Reagents
7 Membrane Technologies
A membrane acts as a gas permeable contact interface between the liquid phase and gas phase to enable the liquid phase to be saturated with dissolved gas . Membranes have been used to pre-load the liquid phase with O2 before a reaction or to continuously introduce O2 to the liquid phase throughout the duration of a reaction. The nature of the contacting method ensures the process is inherently safe because the liquid phase and gaseous O2 are in different channels, thus flammable organic solvents are never in the presence of gaseous O2. In addition, this method also enables better control over residence time at different gas flow rates compared to flow regimes containing undissolved O2, such as a segmented gas–liquid flow regime, within a single microchannel. Membranes are designed to have a large interfacial area to minimize mass transfer effects.
7.1 Tube-in-Tube Reactor
7.2 Tube-in-Shell Reactor
7.3 Dual- and Triple- Channel Microreactor
8 Photochemistry and Singlet Oxygen
Photochemical reactions pose an additional challenge for scale-up, because the light penetration depth has a critical influence on the performance of photooxygenations . Noël and co-workers presented an interesting numbering-up approach to facilitate the scale-up of a photochemical aerobic oxidation of thiols to disulfides by placing eight capillaries in parallel that are irradiated with white LEDs (Scheme 26b) . A further challenge in parallelization of gas liquid reactions is that efficient and uniform gas liquid distribution can be difficult to achieve within parallelized reactor configurations. The lack of uniform mixing in flow can cause stoichiometry imbalance and poor control of residence in the channels. In the study by Noël and co-workers, the calculated standard deviation for yield was less than 10% across the different channels.
Noël and co-workers reported a mild and selective direct oxidation of activated and unactivated C(sp3)–H bonds enabled by decatungstate photocatalysis (Scheme 26c) . Hydrogen atom transfer (HAT) can be utilized for the production of highly reactive radical species, which can be trapped to give synthetically useful products. Decatungstate is a versatile and inexpensive HAT catalyst that readily performs hydrogen abstraction on C(sp3)–H upon photochemical activation. Initial optimization studies using tetrabutylammonium decatungstate (TBADT) demonstrated that full conversion could not be achieved in batch probably caused by the slow diffusion of oxygen into the liquid reaction mixture and the limited light penetration. Nonetheless, significantly improved results were observed within a continuous flow environment. In particular, the flow approach was successful for the oxidation of natural scaffolds such as (−)-ambroxide, pregnenolone acetate, (+)-sclareolide and artemisinin.
10 Green Solvents
There has been a recent drive within the pharmaceutical and fine chemicals industry towards the utilization of more environmentally benign solvents . GlaxoSmithKline, among others, have published green solvent guides to support the implementation of greener solvents at the discovery and manufacturing stages in API development [111, 112, 113, 114]. As discussed in the Introduction, the main safety challenge associated with liquid phase aerobic oxidations is the utilization of O2 in the presence of flammable organic solvent. A benefit of using continuous flow reactors is the ability to safely operate above the boiling point of a solvent through the pressurization of the system. This feature enables access to elevated reaction rates that would be less accessible, or even forbidden, within batch reactors. The solubility of O2 decreases with an increase in temperature; therefore, the ability to pressurize the system is very important. Another limitation is that the mass transfer of O2 from the gas to the liquid phase is often the rate-limiting step in these processes. Significant efforts have focused on identifying solvents that, under certain conditions, can dissolve all the reaction components (substrate, catalyst and O2) within a single phase. One such solution is to operate within the supercritical regime for a solvent to generate a single phase. CO2 and H2O are both nonflammable and have both been employed in this manner. The application of continuous flow technologies facilitates the effective and safe use of these green supercritical solvents at elevated temperatures and pressures, which would otherwise require more cumbersome, expensive and specialized batch reactors. Another way to reduce the impact of chemical manufacture is to utilize no solvent whatsoever, and to conduct reactions with neat reagents. However neat systems can add complications in terms of thermal management, impurity formation and on the physical properties of the reagents for a particular reaction. Ionic liquids and fluorous solvents are also potentially green solvents for conducting aerobic oxidations [115, 116]; however, as of yet there are no reported aerobic oxidation examples that use these solvents within continuous-flow reactors.
10.1 Carbon Dioxide
Poliakoff and co-workers reported the optimization of the oxidation of 2-octanol over a packed bed reactor (5% Pt + 1% Bi on Al2O3) using scCO2 (Scheme 28b) . The system afforded 2-octanone in a consistent 75% yield for 5 h with no evidence for catalyst deactivation and no formation of the octene shown. Subsequently, the system was applied for the oxidation of a number of secondary alcohols (11 examples, 10–75% yield).
Water is a green and non-flammable solvent, thus, on these terms, water is the ideal solvent for aerobic oxidations. However, a significant limitation associated with using water within flow reactors is that the inherent carbon richness of organic substrates mean that most do not dissolve in water, causing slow reaction rates. Uozumi and co-workers studied the aerobic oxidation of alcohols in H2O within a catalyzed by platinum nanoparticles dispersed in an amphiphilic polymer within a continuous flow reactor, but very low substrate concentrations were used (10–100 µM) . The likelihood of multiple phases complicates the development of a flow process due to the multiple phases present. One approach to avoid multiple phases is through the dissolution of organic compounds and O2 within a single phase by operating in the supercritical regime for water.
11 Novel Reactor Developments
The recent developments in reactor technologies for handling O2 have aimed to address the challenge associated with the poor solubility of O2 in water and organic solvents. Thus, efforts have focused on reducing residence times through the enhancement in the mass transfer from the gas to the liquid phase without the need to employ high pressures, which are often undesired. Recent reactor designs have focused on maximizing the interfacial area between the gas and liquid phase to enhance mass transfer.
A nebulizer-based continuous-flow reactor has been developed for the photochemical 1O2 chemistry (NebPhotOX) by Vassilikogiannakis and co-workers at the University of Crete . A solution containing the substrate and the photosensitizer is nebulized by using pure O2 or air into a chamber that is enclosed by LED light strips to form 1O2 as the reactive intermediate. The NebPhotOx system was used for the photooxidation of β-citronellol (Scheme 31b). The pneumatic nebulizer generates aerosols consisting of fine droplets with an approximate 60 μm average diameter corresponding to a droplet-specific surface areas of 100,000 m2 m−3. The same group also reported the synthesis of cyclopent-2-enones from furans using the NebPhotOx reactor in a similar manner .
Raston and co-workers have developed a vortex fluidic device (VFD) for accelerating and increasing the efficiencies of organic reactions . The dynamic thin film is generated by continuously adding a fluid from jet feeds to a rapidly rotating surface (Scheme 31c). The reaction mixture is rotated at very high speeds (up to 9000 rpm) to produce the liquid phase as a thin film, thus providing a large surface area between the liquid and gas phases. The reactor system was demonstrated on the aerobic oxidation of thiols to disulfides . In particular, the aerobic oxidation of N-acetyl-l-cysteine in water was investigated. Full conversion was achieved in less than 2.5 min residence time within a VFD. The aerobic oxidation within a VFD configuration performed significantly better compared to in batch where only 5% conversion was observed after 1 h reaction time. However, the system was not compared to a segmented flow reactor setup.
George, Poliakoff and co-workers at the University of Nottingham recently reported the construction of a thermal and photochemical “vortex reactor” that uses a rapidly rotating cylinder to generate Taylor vortices (Scheme 31d) . The vortices result in a high interfacial area between the gas and liquid phases, thus enabling rapid dissolution of oxygen into the liquid phase. An interesting feature of the reactor system is that it draws air in from the laboratory so does not specifically need pressurized oxygen from a cylinder, with the optimal uptake of air observed at 4000 rpm. The reactor was demonstrated for a number of reaction systems that utilize 1O2 as a reagent, including the photooxygenations of α-terpinene and furfuryl alcohol and the photodeborylation of phenyl boronic acid. The system was also applied successfully to develop a single process for a three-step synthesis of artemisinin from artemisinic acid.
Liquid phase aerobic oxidation reactions offer a valuable alternative to classical oxidation methods using stoichiometric quantities of toxic inorganic oxidants. The challenges (efficient mixing, safety, catalyst decomposition) associated with the use of O2 for organic synthesis can be better addressed through the implementation of continuous flow technology, which can improve reaction reproducibility and provide robust scale-up options. The selection of examples summarized in this review is clear evidence that many aerobic oxidation transformations can be performed effectively and safely under continuous flow conditions. Even pure O2, as opposed to synthetic air, can be safely harnessed in particular instances to provide highly convincing synthetic and manufacturing benefits. Nevertheless, the utilization of continuous flow reactors still poses significant challenges in terms of cost and lack of available infrastructure and expertise available within the synthetic chemistry community. We are convinced that, for environmental, economic, regulatory and synthetic reasons, continuous flow aerobic oxidations will be embraced by scientists and engineers within academic laboratories, and the pharmaceutical and fine chemical manufacturing industries, where further exciting developments can be anticipated in the coming years.
Open access funding provided by University of Graz. The CCFLOW Project (Austrian Research Promotion Agency FFG No. 862766) is funded through the Austrian COMET Program by the Austrian Federal Ministry of Transport, Innovation and Technology (BMVIT), the Austrian Federal Ministry of Science, Research and Economy (BMWFW), and by the State of Styria (Styrian Funding Agency SFG). We are grateful to Dr. Doris Dallinger for carefully proofreading this chapter.
- 1.Jones AB, Wang J, Hamme AT, Han W (2013) Oxygen. In: Encyclopedia of reagents for organic synthesis. https://doi.org/10.1002/047084289x.ro028.pub3
- 25.Osterberg PM, Niemeier JK, Welch CJ, Hawkins JM, Martinelli JR, Johnson TE, Root TW, Stahl SS (2015) Experimental limiting oxygen concentrations for nine organic solvents at temperatures and pressures relevant to aerobic oxidations in the pharmaceutical industry. Org Process Res Dev 19:1537–1543PubMedCrossRefPubMedCentralGoogle Scholar
- 27.Veser K (2001) Experimental and theoretical investigation of H oxidation in a high-temperature catalytic microreactor. AIChE J 56:1265–1273Google Scholar
- 36.Roberge DM, Gottsponer M, Eyholzer M, Kockmann N (2009) Industrial design, scale-up, and use of microreactors. Chim Oggi 27:8–11Google Scholar
- 73.De Angelis S, Hone CA, Degennaro L, Celestini P, Luisi R, Kappe CO (2018) Sequential α-lithiation and aerobic oxidation of an arylacetic acid—continuous-flow synthesis of cyclopentyl mandelic acid. J Flow Chem 8:109–116Google Scholar
- 79.Chaudhuri SR, Hartwig J, Kupracz L, Kodanek T, Wegner J, Kirschning A (2014) Oxidations of allylic and benzylic alcohols under inductively-heated flow conditions with gold-doped superparamagnetic nanostructured particles as catalyst and oxygen as oxidant. Adv Synth Catal 356:3530–3538CrossRefGoogle Scholar
- 87.Wu G, Cao E, Ellis P, Constantinou A, Kuhn S, Gavriilidis A (2018) Development of a flat membrane microchannel packed-bed reactor for scalable aerobic oxidation of benzyl alcohol in flow. Chem Eng J 1–9Google Scholar
- 114.Byrne FP, Jin S, Paggiola G, Petchey THM, Clark JH, Farmer TJ, Hunt AJ, Mcelroy CR, Sherwood J (2016) Tools and techniques for solvent selection: green solvent selection guides. Sustainable chemical processes. Springer, Berlin, pp 1–24Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.