**1. Introduction**

The two conventional processes for manufacturing ceramic cellular foam structures are the replica, as well as the space holder method [1,2]. These methods use organic scaffolds, which have to be burnt out. A novel manufacturing route for ceramic foam structures, called Freeze Foaming, that avoids the use of organic additives, has been developed by the Fraunhofer Institute for Ceramic Technologies and Systems (IKTS) [3,4]. The cell structure of a sample manufactured using Freeze Foaming is defined by a pressure-induced and pressure-controlled foaming process, followed by subsequent freeze drying, of a ceramic suspension in a vacuum. There are two different foaming agents—as ambient pressure drops, a reduced boiling point leads to the evaporation of water out of the aqueous suspension. The other one is air that is introduced during the manufacturing of the suspension. While the pressure is reduced (and the foam expands), the suspension's temperature follows the line of equilibrium in the phase diagram of water to the triple point. Since the pressure is reduced further, the temperature falls beneath the equilibrium temperature in the triple point of our suspension, which causes our created structure to be instantly frozen, and dries via sublimation. This freezing step can result in cryogenic structures similar to typical freeze cast structures [5,6] and accounts for the microporosity of foamed structures. Possible applications of foams and porous parts made using Freeze Foaming encompass a

wide spectrum, including biomedical uses, like artificial bones [7–9] as well as carrier materials for catalytic converters [10], biosensors and drugs, or thermal [11] and acoustic insulators [12]. Freeze Foaming enables the processing of biocompatible materials while offering the unique possibility of creating foams exhibiting a multimodal pore-size distribution and interconnectivity. These factors offer good conditions for the cultivation of organic cells. Previous work [3,4,7,13] has shown a particular suitability for an application in artificial bones due to their special structural properties.

As the foaming is influenced by a complex interaction of several process and material parameters, further research into the foam formation during Freeze Foaming is needed. A reproducible manufacturing of tailored foams with a specified structure is not possible as of now. To adjust the properties according to applications, and develop process and quality guidelines, it is important to further examine the influence of relevant process parameters on the foam's structure. With the presented work, the authors aim at finding a solution to a tunable shaping method, which enables the manufacturing of highly controllable pore structures to be used, e.g., as bone-mimicking scaffolds for an ever-older population [14].

To that end, an in situ μCT extension for an existing scanner was developed, which allows the analysis of different steps of the very foaming process during the manufacturing. Conventionally, examining changes in a sample using X-ray computed tomography is done by scanning before and after a change in state or structure [15,16]. The conventional method does not enable observation of the foam development of Freeze Foams. In material research focusing on damage and degradation analysis, significant improvements were made in the last years, due to progressive in situ techniques [15,17,18].

In the investigations from [19], the sintering process of ceramics could be analyzed with the use of CT. It was sufficient to analyze the process every 30 min to ge<sup>t</sup> a statement about the sintering theories. Further investigations in the field of in situ analyses are described in [20]. In so-called in situ X-ray nanotomography systems, pixel sizes of 100 nm are achieved in 20 s, with the use of a focus size of 50 nm. The sintering stages of metals and ceramics were also analyzed. In the investigations presented here for the analysis of the formation process of Freeze Foams, such a resolution is not necessary. In the investigations from [21], the hydration of gypsum plaster setting was investigated with in situ X-ray tomography. The scan duration was 200 s. However, the entire structure formation of the Freeze Foaming process (i.e., foaming and freezing) takes only about 60 s. Therefore, a novel CT setup had to be developed, firstly, in order to visualize the foaming process per se and, secondly, to introduce measures making tomographic image acquisition possible. The process had to be designed in such a way that it could be stopped at certain process steps and fixed for the CT imaging. This entire experimental setup—a new controllable laboratory freeze dryer in a computer tomography scanner—is one of the fundamental novelties of this work. In the first phase, a process-optimized testing device was developed [22]. It is suitable for 2-dimensional examinations using X-ray radiography (for real-time observation of the foaming progress), as well as three-dimensional scans to evaluate structural phenomena. Using the now-reproducible manufacturing of a model suspension [23], detailed results of in situ foam structure analysis are presented.

#### **2. Materials and Methods**

The ceramic suspensions used in this work are composed of water and dispersion agen<sup>t</sup> (Dolapix CE 64, Co. Zschimmer & Schwarz Mohsdorf GmbH & Co. KG, Burgstädt, Germany), added hydroxyapatite powder (Sigma-Aldrich now Merck KGaA, Darmstadt, Germany; BET = 70.01 m2/g, d50 = 2.64 μm), binder (polyvinyl alcohol), and rheological modifier (Tafigel PUR40, Co. Münzing Chemie GmbH, Heilbronn, Germany) [23]. The choice of suspension and composition was derived from preliminary tests on the basis of different suspensions which, after Freeze Foaming, resulted in reproducible foam structures [23]. The detailed manufacturing process of the suspension is described in [24]. For the investigations of this contribution and with regard to its possible influence on the foaming process and structure formation, three ceramic suspensions with different temperatures were used (5, 23, and 40 ◦C).

An in situ device, to be used inside a v|tome|x L450 (General Electric, Cincinnati, OH, USA), was developed during the first phase of research [22]. It allows the material to be subjected to phenomenological analysis, and for detection and characterization of pores during the foaming process. To examine developmental steps during the formation of the foam's structure, the device has to be leak-proof under vacuum (Figure 1). Using different foaming molds, the device can be used for either X-ray radiography (2D) or gaining spatial information (μCT, 3D) about the foam's structure. The resolution was set to 22 μm/vx using an acceleration voltage of 100 kV, and a beam current of 300 μA. To fix the foaming suspension for the time of the CT scan (720 projections with 250 ms exposure time each, 3 min total measurement time), the foaming is stopped using a pressure control system with dedicated software, developed in-house, and an adjustable bypass. The pressure is kept at a constant level for the duration of the CT, in order to stabilize the structure. The vacuum chamber itself is rotationally symmetrical, and made of low-absorbing polymer to ensure optimal image quality. The choice of polymethylmethacrylate (PMMA) for the chamber prevented stabilizing the foam by means of externally freezing, as the material's thermal conductivity is very low.

**Figure 1.** In situ μCT device: CAD model (**a**) and mounting situation (**b**).

In previous examinations [22], the pressure reduction rate's influence on the foam structure—and especially the orientation of pores—has been shown. Due to the concave nature of the bottom of the mold, foaming led to a high number of samples exhibiting large pores near the bottom, which distorted the results of the foam analysis. Therefore, a new mold with a flat bottom was designed (Figure 2). Furthermore, water vapor emitted from the suspension decreased the pressure reduction rate in the lower pressure range of 25 mbar and below (near equilibrium of water vapor at 20 ◦C). The reduced pressure drop led to a higher amount of coalescence effects in the finished foam structure. To accelerate the pressure reduction, a cold trap has been used. For this purpose, a cylinder made from aluminum, with channels, was manufactured. It was cooled down with liquid nitrogen and placed on top of the mold (Figure 2, right picture). The water vapor condenses on its surface, which significantly reduces the time to fall below the triple point.

**Figure 2.** Improvements to the experiment.

Besides the process analysis compression tests, in situ μCT scans under compressive load were also conducted on the conducted foams. In situ compressive scans were performed using a Finetec FCTS 160 IS (Garbsen, Germany) with an acceleration voltage of 50 kV and a beam current of 250 μA. The resolution was around 11 μm/vx, and the exposure time 625 ms. Mechanical testing of prepared cylindrical samples took place using an universal testing machine Zwick 1475 (Ulm, Germany). A preload of 2 N and a traverse speed of 2 mm/min were chosen.
