In-situ time resolved spectrographic measurement using an additively manufactured metallic micro-fluidic analysis platform

Introduction Microfluidic reactionware allows small volumes of reagents to be utilized for highly controlled flow chemistry applications. By integrating these microreactors with onboard analytical systems, the devices change from passive ones to active ones, increasing their functionality and usefulness. A pressing application for these active microreactors is the monitoring of reaction progress and intermediaries with respect to time, shedding light on important information about these real-time synthetic processes. Objective In this multi-disciplinary study the objective was to utilise advanced digital fabrication to research metallic, active microreactors with integrated fibre optics for reaction progress monitoring of solvent based liquids, incompatible with previously researched polymer devices, in combination with on-board Ultraviolet-visible spectroscopy for real-time reaction monitoring. Method A solid-state, metal-based additive manufactured system (Ultrasonic Additive Manufacturing) combined with focussed ion beam milling, that permitted the accurate embedment of delicate sensory elements directly at the point of need within aluminium layers, was researched as a method to create active, metallic, flow reactors with on-board sensing. This outcome was then used to characterise and correctly identify concentrations of UV-active water-soluble B-vitamin nicotinamide and fluorescein. A dilution series was formed from 0.01–1.75 mM; which was pumped through the research device and monitored using UV-vis spectroscopy. Results The results uniquely showed the in-situ ion milling of ultrasonically embedded optical fibres resulted in a metallic microfluidic reaction and monitoring device capable of measuring solvent solutions from 18 μM to 18 mM of nicotinamide and fluorescein, in real time. This level of accuracy highlights that the researched device and methods are capable of real-time spectrographic analysis of a range of chemical reactions outside of those possible with polymer devices.


Preparation of UAM samples
The process chain of sample fabrication is demonstrated in Fig A. The samples to be taken forward into etching were prepared by first depositing five separate Al 3003 H18 foil layers (100μm thick) onto an Al 1050 supporting plate, at 21 + 5 o C, to form a UAM metal matrix.
The parameters used for the deposition of these foils were 1400N normal force, 20μm oscillation amplitude and a 40mm/s welding speed. These parameters were obtained via systematic tests and prior studies focusing on the ultrasonic embedding of metal-coated optical fibres in Al 3003 H18 metal matrices [1]. This temporary fixture was required to stop the fibre moving significantly during the welding process as was noted in preliminary trial runs.

Electrochemical cell manufacture and initial exposure testing
Due to the fragile nature of optical fibres when subjected to methods such as CNC milling, a means of forming fluidic channels around these embedded fibres needed to be established in order for them to be cross-sectioned at a later stage. As a result, electrochemical etching was selected as a test method for producing fluidic pathways around embedded features.
This process begins by first etching the narrowest channel possible into the surface of the polymer etch resist applied to the surface of the UAM substrates. This value was determined to be ca. 550 μm and was achieved using a combination of 20% laser power and 1100 mm/s scanning speed. Using these parameters, multiple substrates were produced and taken forward into testing to establish the effects of a various parameters in an ECE cell on the width and depth of channels produced.
A DOE approach was taken to the formation of fluidic channels onto the surface of the UAM substrates utilizing the electrochemical etching cell in order to establish the most suitable range of parameters to be used in later optimisations. The responses examined from these different combinations of parameters centred on both the width and depth of the produced channel. This was achieved through non-contact focus variation microscopy using an Alicona surface profiling system. The desired result of these trials was to produce a channel as close to the width of the overlying aperture in the tape as possible through reducing the effects of under-etching and producing a depth whereby the core of a silica fibre could be exposed (ca. 200 μm) for later work to fibre cross-sectioning. The results of these initial DOE trials are displayed graphically as response contour plots within the appendix.
The sample producing the cleanest channel with the narrowest diameter and appropriate depth (> 200 μm) was the sample with parameter combinations of 750mA and an exposure time of 5 minutes. This combination produced a channel depth of 375 μm and a width of 841 μm. As a result, an optimisation DOE was performed, with this parameter combination acting as the design space central point. Both time and current increments were equally spaced either side of this centre point in order to try to optimise the etching result. As with previous trials, experimental runs were both randomised and repeated. The parameters used for this are located below in Table 1. Samples identical to those used in previous trials were formed and used in this optimisation and again measured in order to determine the response in width and depth of the channels formed because of changes in current and exposure time. The resulting response contour plots for the responses within the range of samples are located the appendix.
In contrast to earlier trials, the channels formed with the lowest exposure time (2.5 min) at each of the different current settings (625, 750 and 875 mA) were seen to exhibit consistent channels with clean edges and consistent depths. Images of these optimised samples are located in Fig B whilst the resulting measurements are displayed in Table 2.
taken forward for etching trials.  Initially CNC machining had been trialled as a means of creating an aperture between embedded optical fibres in order to spectroscopically sample the passing solution, but was demonstrated Once the cell was manufactured, a Synrad CO 2 laser-marking system was employed (Synrad Inc., Mukilteo, United States) to form channel layouts onto the UAM in channel sizes more commonly found in flow reactor systems (< 1mm). These CO 2 marking systems are often used in order to achieve high-resolution markings on the surface of various polymer substrates. Due to the highly reflective nature of the aluminium substrate along with its high thermal conductivity, the CO 2 laser has no effect on the metallic substrate at the powers used in this work (10-20 W). The laser system was used to thermally remove channel structures into the adhesive polymer layer placed onto the surface of the substrates so that the exposed section was open to etching by the ECE cell.
A design of experiments (DOE) approach was undertaken in order to establish the relationship between channel formation and the various parameters associated with these electrolytic cells.  Table 3. The channels producing the most consistent etching of appropriate dimensions (100-500μm width and a depth > 200μm so that the entire fibre is exposed for later works in which the fibre will be cut) were then further optimised in order to produce a channel suitable for UV-vis This cell design ensured a constant anode-cathode separation, as well as negating the requirement to protect metal connections at both the anode and cathode, as these are also susceptible to etching when exposed to the cell. The anode was formed from Al 1050 due to the high degree of conductivity and corrosion resistance.

Focused ion beam cross sectioning of embedded fibres
A novel method based upon ion beam milling and lift out procedures was then developed as a result of previous publications utilising ion beam for various experimentation regarding both UAM [3] and fibre machining [4,5].
Due to the embedded optical fibre being difficult to access with traditional tools such as scribes, an appropriate means of cross sectioning the fibre needed to be established in order for it to be useful as an optical waveguide for future spectroscopic applications. Focused ion beam (FIB) milling was previously shown by Abdi et. al. to be a highly suitable method in the cleaving of polymer optical fibres (POF) in the development of strain sensors [6,7]. FIB machining has also been used in a number of other fibre-based applications, including; modification of waveguide properties [8], fabrication of long period gratings [9], micromachining of the fibre end tip for Fabry-Perot cavities in sensing applications [10], precision cutting of photonic crystal fibres [11], formation of optical lenses for efficient fibre-to-waveguide coupling [5], and the writing of gratings in silicon optical waveguides [12].
FIB offers several important advantages such as; precise site selection, milling control, and high quality surface finish. These are all properties that are highly desirable when designing a system for spectroscopic interrogation inside fluidic channels. In light of this, this section presents a novel technique for the cross sectioning of optical fibres embedded within UAM metal matrices via DBFIB technology.
A Nova 600 NanoLab Dual Beam system equipped with an SEM and uses a liquid gallium ion source was then employed for both cross sectioning and imaging of embedded fibres. This apparatus delivers a focused Ga + ion beam with energy 5-30 keV and a probe current of 1 pA -20 nA. The FIB technique started with coating the exposed fibre sample embedded with approximately 100 nm of palladium-gold using a Quorum Q150T ES sputter coater (Quorum Technologies, Lewes, UK) (Fig D -A). Cross-sectioning of the exposed fibre was comprised of two major stages: firstly a mass removal stage in which 240 μm long cross sections were milled in order to remove a large central fibre section, followed by cross-section polishing of the newly exposed surfaces in order to produce flat clean edges and ensure the aperture distance of 250 μm. The ion beam for both the ion milling and imaging was set at an accelerating voltage of 10 kV and a 10 × 20 μm 2 area removed, in stages, in the mass removal step using a 20 nA aperture and a rectangular milling box.
In the mass removal stage, sections were removed from edge to edge of the fibre as opposed to performing one large single cut (Fig E -B). This was due to the edges of the fibre tending to mill at a faster rate than the central regions due to re-deposition of milled material back onto the bulk, whereas material at the edges typically does not redeposit in the local area as there is less surface area for it to redeposit on. A small amount of material was left un-milled in the second cut in order to keep this cut section connected to the bulk of the fibre (Fig E -C). The coaxial OmniProbe nano-manipulator integrated into this DBFIB system was then used to manipulate this fibre section and detach it from the bulk. These probes are ideal for nanoscale positioning and manipulations and are commonly called upon to execute FIB lift-out procedures.
A 5 nA aperture was used for cross-section polishing of fibre faces, with an assisted cleaning cross-section milling box providing a line-by-line scan. The initial distance between the two sets of cuts was set at 240 μm, with an additional 10 μm being removed in cross-section polishing. This distance was chosen in order to form a balance between the concentrations of analytes/reagents it is possible to monitor in the device. This is based upon standard practice within UV-vis spectroscopy whereby smaller distances typically excel for low concentration monitoring whilst larger distances are more suitable for higher concentrations. Imaging was then performed using an ion beam with a 100 pA aperture and SEM. An ion beam was chosen to perform the imaging due to the lack of secondary electron signal generated by the sample because of the insulating nature of the cross-section surface.  The fibre was initially cross-section using the ion beam to mill away small rectangular segments, from one side of the fibre to the other (Fig F). Due to small degrees of charging of the substrate because of the non-conductive nature of the newly exposed fibre, drifting of the image was noted during milling. This resulted in a slightly more distorted cut that had initially been planned. Due to issues with the highly charged surfaces occasionally causing milled away sections to stick to the remaining bulk material, a small section of the fibre was left intact in order to allow manual removal of this cross-section fragment. Employing the DBFIB coaxial OmniProbe needle to manipulate this cross-section segment, the segment was removed to create an aperture of 240 μm between the two new faces of the fibre ( Fig G). The faces of these fibres were coarse and required additional polishing in order to produce both clean flat faces and adjust the separation distance to the desired 250 μm (Fig G).

Fig H. (A) Course unpolished face of milled fibre (B) Rectangular milling box used in cross-section polishing
Using a reduced aperture, these faces were highly polished to yield clean, flat surfaces that are highly desirable for the transmission and receiving of light across a band gap (Fig I). In spectroscopy, this reduces coupling losses and could potentially yield increases in sensitivity.

Fig I. (A)-(B) Clean faces produced through cross-section polishing highly suitable for spectroscopic applications
Concluding this final polishing stage, the etching channel and fibre were encapsulated with a covering Al 3003 foil layer at the same parameters and previously used for substrate manufacture. The total time to cut through the fibre and polish both faces of the fibre was approximately eight hours. Once embedded, the system was taken forward for spectroscopic testing in regards to the transmission performance of the cross-sectioned fibres.

Manufacture and testing of UAM UV-Vis flow system
Combining all of the methods described in this work, the first reported case of a UAM UV-Vis flow cell was manufactured.
UAM processed Al 3003-H18 substrates were first manufactured at process parameters proven conducive to achieving high degrees of mechanical performance whilst still allowing functionality of embedded optical fibres. These substrates were then exposed to an electrochemical etching cell in order to form channels with dimensions similar to that commonly employed in flow chemistry. This allowed for an evaluation of how the main variables of the cell affected the dimensions of the newly formed channel. This was achieved through a combination of DOE, non-contact focus variation microscopy and spectroscopic measurements.
This research has outlined how UAM can be used to develop the flow reactors featuring discretely embedded sensors within specifically selected materials whilst maintaining the design freedom associated with additive manufacturing. These advantages could enable a much wider range of reactions and experimentation to be investigated and is more likely to be applicable to modern research and industrial chemistry due to their robust nature lower cost.
This can potentially move away from the concept of a reactor chip in a lab and move toward the desired goal of a lab-on-a-chip.