Water retention characteristics of coarse porous materials to construct purpose-designed plant growing media

ABSTRACT

Among potential components to construct Technosols for urban greening purposes, the commercially available geogenic coarse porous materials (CPMs) are mainly used in practice because of their high porosity. However, the knowledge of the hydraulic behavior of CPMs as well as of their mixtures with other substrates is limited, provoking their suboptimal usage. Therefore, we determined the water retention characteristics, including the available water capacity (AWC) of six geogenic CPMs: porlith, expanded shale, expanded clay, tuff, pumice, and lava. In order to obtain the water retention characteristics of the CPMs as well as of their mixture with sand (1:4 per volume), the following methods adapted from soil physics were applied over a wide range of pressure heads: Equi-pF apparatus, ceramic tension plates, pressure plate extractors, WP4C apparatus, and water vapor adsorption. The results were used to parametrize the modified Kosugi model (using SHYPFIT 2.0). Porlith and tuff have the highest AWC (0.37 m³ m⁻³ and 0.17 m³ m⁻³, respectively) and are the only ones which can be recommended as effective water-retaining materials. Further materials exhibit an AWC less than 0.10 m³ m⁻³. The CPMs exhibit a bimodal pore size distribution, which can be well described by the applied model, except for pumice and expanded shale. The mixtures present overall low AWCs up to 0.07 m³ m⁻³, with the pure sand having less than 0.03 m³ m⁻³. For practical application a quite high ratio of CPM is needed, and the mixing material must be adapted to the hydraulic properties of the CPMs. The water inside the CPMs may be easily available for plant roots able to penetrate in the CPMs’ coarse pores.

KEY WORDS:

• Technosol
• technogenic substrates
• available water capacity
• urban soil
• plant substrate

1. Introduction

Urban green areas such as parks, gardens, green belts, or allotments help to mitigate the adverse ecological effects derived from the surface sealing in urban areas (Bolund and Hunhammar 1999; Scalenghe and Marsan 2009; Gessner et al. 2014; Morel et al. 2014; Jänicke et al. 2015).

However, the conservation and creation of traditional urban green areas face competitive disadvantages with other urban infrastructure projects. The scarcity of available areas and their often low quality as plant habitats (e.g., highly compacted, polluted, sealed) are among such limitations. Frequently, in order to green such areas, expensive soil remediation techniques are used, or the polluted or sealing layers are removed.

As an alternative, Technosols (IUSS Working Group 2015) specifically designed to function as plant growing media, can be installed in situ as secondary ready-to-use soils or soil layers on these areas (Séré et al. 2008). They can also be used to install secondary urban green on areas not intended to be vegetated before (Nehls et al. 2015), as it is the case for green façades and vegetated roofs.

These purpose-designed Technosols can be composed intentionally of a variety of materials, which in conjunction must fulfill the main functions of soils as plant habitats: provision of a stable rooting zone with adequate aeration, water storage, water drainage, favorable chemical soil conditions (e.g., CEC, pH, EC), as well as nutrient storage and availability.

For this aim, a large variety of materials have been investigated for the construction of Technosols for greening purposes. Some studies focused on natural and processed natural substrates, such as brown coir (coconut fiber; Noguera et al. 1996), compost, peat (Vasenev et al. 2013), biochar (Abel et al. 2013), vermiculite, or clay pellets. Some authors pointed out the relevance of the reuse of recycled or waste materials, such as bricks (Nehls et al. 2013; Farrell et al. 2013), paper mill sludge, rubber, concrete, rubble, or sewage (Séré et al. 2008; Rokia et al. 2014). However, due to strict legal regulations, the contamination of the recycled and waste materials must be previously tested according to the intended use. Additionally, the commercial availability of these materials is limited and their quality can vary largely. For these reasons, the use of uncontaminated and easily available natural porous materials, which are simply mined and crushed (e.g., porlith, tuff, pumice) or additionally processed (e.g., expanded clay, perlite), is a practicable and convenient option for immediate use.

Despite the material origin, its application will strongly depend on the local conditions and the main functions to be fulfilled. For instance, climate zones with high precipitation intensity require a Technosol design which ensures an efficient water drainage, avoiding stagnation, flash flooding, and run off during heavy rain events. Simultaneously, the Technosol has to store enough readily available water for proper plant growth during drought periods. A sandy soil texture can ensure a high saturated hydraulic conductivity for drainage, though it has a very low available water capacity (AWC). Clay or silt would increase the AWC, but their small particles lower significantly the hydraulic conductivity and may also clog the subsequent drainage system of the site. As a solution, the AWC in a sandy Technosol matrix has to be provided by porous materials with rather large grain sizes.

There are several mineral coarse porous materials (CPMs) commercialized as plant-growing media that could be used. However, the description and quantification of their hydraulic properties are often insufficient up to now, are poorly described, constrained only to few parameters (Bilderback et al. 2005; Raviv and Lieth 2008), or to narrow ranges of horticultural interest (up to pF 2.0; Wallach et al. 1992). In contrast, for organic materials used in horticulture, there are more studies available, describing their water retention characteristics more in detail (Fields et al. 2016; Schindler et al. 2016a, 2016b).

In order to increase the knowledge-based design of Technosols that sustain vegetation, the objective of this work is to accurately characterize the hydraulic properties of six CPMs of natural origin in their commercial forms that can potentially increase the AWC of a sandy Technosol. Porlith, expanded shale, expanded clay, tuff, pumice, and lava were chosen due to their availability and use in the German market, their porosity, predictable qualities, lightweight, structural stability, not having extreme pH, and their large grain sizes. An accurate water retention capacity of these porous materials will allow to optimize their use within an appropriate ratio in composition with sand (or other materials), according to local climatic conditions (dryness, excess of precipitation).

2. Materials and methods

2.1. Materials

Six commonly used CPMs plus two mixing materials of natural origin, sand and silt, in their commercial presentation have been analyzed:

• Porlith is a consolidated smectite clay-schist scoria. It is a mineral by-product of the oil exploitation from an organic-containing Cenozoic sediment, deposited in a crater lake of volcanic origin (see Roth-Kleyer 2005). During exploitation process, the material was highly heated, resulting in a mixture of partial melted red claystone residue and porous lava-type material. The production process of porlith, as well as the resulting material, is comparable with the ones of bricks. Nowadays it is used as a growing medium. The pebbles used in this study had a diameter of 2–7 mm.

• Pumice is a solidified lava with a low density due to the fast cooling process and the immediate depressurization and expansion after eruption (pebbles of 2–12 mm).

• Tuff is formed naturally of consolidated volcanic ashes and lapilli, and was used in a grain size of 3–8 mm.

• Lava is solidified magma, crushed to 5–15 mm grain size.

• Expanded clay is produced by heating natural clay to temperatures of 1100–1200°C in a rotating kiln. The gases produced by the heating expand the clay, creating small bubbles in its inner core, providing it with a foamy structure. Due to the rotation, the pebbles are roundish and have a lightweight ceramic shell (Hammer et al. 2000). Expanded clay is available in different sizes and densities. In this study it was used in a grain size of 4–8 mm.

• Expanded shale (used in 2–7 mm grain size) is a shale light aggregate expanded by a thermal process with the formation of open pores (Bender 1986).

• The quartz sand (SiO₂ > 96%), used for mixing with the porous materials has a particle size distribution of 0.06–0.6 mm, with 94% of particles between 0.1 and 0.4 mm (Sand Schulz®).

• The silt used in order to embed the CPM at certain hydraulic tests was obtained by grinding pure quartz sand. It has a maximum particle size of 0.2 mm.

Expanded clay and shale are of natural origin but further processed, while the rest of the materials are of natural origin, only mined and crushed, but not further processed.

2.2. Methods

2.2.1. General characterization

The pH (Knick 761 Calimatic pH-meter) and the electric conductivity (InoLab) were measured according to EN ISO 10523:2012 with the materials crushed to pass 2 mm. The densities ($\rho$) and porosities ($\varphi$) of the CPMs were characterized as follows. The bulk density (${\rho }_{\mathrm{b}\mathrm{u}\mathrm{l}\mathrm{k}}$) was measured with a pycnometer flask filled with the dry materials and weighed. The apparent density (${\rho }_{app}$) of the CPM was measured dividing their mass by their apparent volume. The apparent volume excludes the large voids among CPMs. It was determined as the difference between the bulk volume (pycnometer or volumetric flask) and the volume that haloysite clay spheres (Dragonite™; 50–100 μm; ${\rho }_{bulk}$ 1.85 g cm⁻³) occupy of the CPMs inter-particle voids. Therefore, CPMs were mixed with Dragonite™ and compacted. The compaction of the mixture was reached by vibration, refilling with Dragonite™ until reaching a constant filling of the volumetric flask. The measurement of the CPM particle density (${\rho }_p$) was performed with a Multi-Pycnometer (QuantaChrome MVP-1), and the materials were crushed in order to avoid encapsulated pores. Total porosity (${\varphi }_t$) was calculated by $1-\frac{{\rho }_{bulk}}{{\rho }_p}$. It accounts for the voids among the CPM plus their intrinsic porosity (inter and intra-particle porosity, respectively; Fig. 1). The intra-particle porosity (${\varphi }_{intra}$) was calculated by $1-\frac{{\rho }_{app}}{{\rho }_p}$.

Figure 1. Graphic representation of the calculated porosities. The pores are represented in white. Two main pore domains are observed: the pores inside the CPMs, or intra-particle porosity (${\varphi }_{intra}$), as well as the voids among particles, or inter-particle porosity (${\varphi }_{inter}$). The (A) total porosity (${\varphi }_t$) accounts for both and it was calculated by $1-\frac{{\rho }_{bulk}}{{\rho }_p}$. The (B) intra-particle porosity (${\varphi }_{intra}$) excludes the ${\varphi }_{inter}$ and was calculated by $1-\frac{{\rho }_{app}}{{\rho }_p}$.

2.2.2. Sample preparation

2.2.2.1. CPM as pure materials

The pure CPMs were packed into steel cylinders of 100 and 385 cm³ at their ${\rho }_{bulk}$. Besides, an additional set of samples were prepared mixing the CPM with silt in steel cylinders of 100 cm³. This procedure was done in order to ensure capillary continuum and proper drainage within the limited time span of the experiment.

2.2.2.2. Sand

The sand was packed into steel cylinders of 100 cm³ and compacted to a ${\rho }_{bulk}$ of 1.48 g cm⁻³ (94% of proctor density in accordance with ASTM D698).

2.2.2.3. CPM mixed with sand

The CPMs were mixed with sand (coarse texture matrix) with a ratio of 1:4 per volume. The samples were prepared by mixing the corresponding volume of 0.75 m³m⁻³ of sand at ${\rho }_{bulk}$ of 1.48 g cm⁻³ with 0.25 m³m⁻³ the CPMs, homogeneously mixed and compacted afterwards.

2.2.3. Capillarity testing

The feasibility to determine the water retention characteristics of the CPMs with standard laboratory methods was tested in advance. In this context, the capillarity among CPMs, meaning the conductivity of water from one particle to the other, is a crucial prerequisite to ensure proper saturation and drainage of the material within the range of applied pressure heads. The capillary rise within the CPM was tested by packing each material into steel cylinders of 5 cm height (385 cm³) placed in a 2 mm water table, and monitored for more than 24 h photographing the surface in a 2 min interval. Capillary rise was assumed as sufficient when 90% of the surface was visibly wet. The fastest capillary rise was measured in expanded shale (70 min) and the slowest for expanded clay (1295 min), while for the other materials it was in between 150 and 365 min. As the capillarity was verified, we proceeded with further tests.

2.2.4. Water retention characteristics

We applied four different methods with specific features and measuring ranges in order to obtain detailed data needed to properly describe the water retention curve from saturation to wilting point of (a) the pure sand, (b) the pure CPMs, and (c) the CPMs mixed with sand.

1) Equi-pF, Streat Instruments lt. This automated soil moisture release curve apparatus works with hanging water columns. It quantifies water retention characteristics at low pressure heads, including water saturation, air entry point, and field capacity. The resolution of pressure heads was 10 hPa, with two drying and one wetting loops. However, its measuring range is limited to a pressure head of pF 2.0.

2) Porous ceramic plate extractors, Eijkelkamp, Giesbeek. With this standardized method (ISO 11274) the water content of the materials was quantified at pF 1.8, and intermediate pressure heads pF 2.5 and 3.0. All measurements were done at least in triplicate.

3) Dewpoint Hygrometer WP4C apparatus. With this method, the water content at permanent wilting point (pF 4.2) was determined using the chilled mirror dewpoint technique. ¹ All measurements were done in triplicate.

4) Water vapor adsorption. Pressure heads higher than the permanent wilting point water (>pF 4.2) were determined by measuring the water vapor adsorbed by the materials, controlling the relative humidity in a glass desiccator with a saturated solution of K₂SO₄ at 20°C (Wexler and Hasegawa 1954). All measurements were done in triplicate.

The measurement of the pure CPMs was challenging. Although the capillarity among CPMs was verified at saturation, at higher pressure heads the capillarity is most probably insufficient to assure proper dewatering from the intrinsic pores. Further, the CPMs have a heterogeneous contact with the ceramic plate due to their large sizes. In order to ensure capillarity, the CPMs were mixed with silt at the pressure heads pF 1.8, 2.5, and 3.0 measurements. For the measurements of the permanent wilting point and higher pressure heads, we applied methods which determine water content via dew point and thus are independent from capillary water transport. In contrast to the pure CPMs, the pure sand and CPM–sand mixtures have an adequate capillarity for proper dewatering. Altogether, the measurements obtained with the four methods mentioned above provided sufficient data to adequately describe the water retention characteristics of the materials.

For the composite samples with sand, as well as for the measurements in which CPMs were mixed with silt, the water content of the sand and silt was also measured and later subtracted in order to calculate the water content corresponding just to the water held by the CPMs, using the following equation:

(1) ${\theta }_{CPM}=[{\theta }_t-((1-X)\mathrm{x}{\theta }_{s{a}_i})/X]\mathrm{x}(1-{\varphi }_{inter})$(1)

where ${\theta }_{CPM}$ is the volumetric water content of the CPM [m³m⁻³], ${\theta }_t$ is the total volumetric water content, X is the ratio of CPM mixed with sand or silt [m³m⁻³], and ${\theta }_{s{a}_i}$ is the volumetric water content of sand or silt at a corresponding pressure head. The term 1 − ${\varphi }_{inter}$ accounts for the correction of the inter-particle porosity among the CPMs packed in the steel cylinders. In order to emphasize the water retained exclusively inside the CPMs ${\varphi }_{intra}$ when mixed with sand and silt at pF 1.8, ${\varphi }_{inter}$ was not considered, and therefore set equal to 0. This approach was applied to the results obtained via the ceramic tension plate.

The water retention curve of the pure materials was derived using water content values obtained with the Equi-pF up to pF 2, and with the ceramic tension plates for pF 2.5 and 3.0. The AWC was set equal to water held in between pF 1.8 (field capacity, FC), and pF 4.2 (permanent wilting point, see, e.g., Blume et al. 2009).

2.2.4.1. Parameter estimation

We used the experimental data to fit the parameters of the water retention function according to Peters–Durner–Iden (Peters 2013; Peters 2014; Iden and Durner 2014) using the software SHYPFIT 2.0 (Peters and Durner 2006). The applied model describes the water retained in capillaries, the film water, and the adsorbed water. The model for capillary water retention is based on the model of Kosugi (1996) for a unimodal pore size distribution and on the modified Kosugi model (Durner 1994; Romano et al. 2011) for a bimodal pore size distribution, with the two distinct pore domains corresponding to the inter-particle pores and the intra-particle pores of the CPMs (Fig. 1).

The basic functions used are the unimodal function of Kosugi (1996) for sand, as well as the bimodal form of it for the CPM and the composite samples with sand. The unimodal Kosugi retention function is given by:

(2) $\mathrm{\Gamma }(h)=\frac{1}{2}erfc\left[\frac{ln\left(\frac{h}{h_m}\right)}{\sqrt{2}\sigma }\right]$(2)

where $h_m$ is the pressure head corresponding to the median pore radius, $\sigma$[–] is the standard deviation (SD) of the log-transformed pore-size distribution density function, and $erfc$[–] denotes the complementary error function. The bimodal functions are weighted sums of the unimodal functions:

(3) $\mathrm{\Gamma }(h)=\sum _{i=1}^2{w}_i{\mathrm{\Gamma }}_i$(3)

where ${\mathrm{\Gamma }}_i$ are the weighted subfunctions, expressed by unimodal functions, and $w_i$[–] are the weighting factors for the subfunctions (0 $<w_i<$ 1).

3. Results and discussion

3.1. General characterization

The general characteristics of the materials are summarized in Table 1. The CPMs have appropriate pH and EC values to be used as fertile plant substrates. The pH of the materials ranges from 5.2 to 7.7, favoring the availability of most macro-nutrients, promoting biological activity, and preventing mobilization of toxic metals (Bradham et al. 2006; Schaetzl and Thompson 2015). The values are in line with the ones reported by Roth-Kleyer (2007) for similar materials. The low values of EC (<1 dS m⁻¹) for all materials indicate their low salinity and sodicity.

Table 1. General characteristics of the CPMs and the sand. EC = electric conductivity, ${\rho }_{bulk}$ = bulk density, ${\rho }_{app}$ = apparent density, ${\varphi }_t$ = total porosity and ${\varphi }_{intra}$ = intra-particle porosity (only the porous of the materials, without the surroundings).

	Size		EC	${\rho }_{bulk}$	${\rho }_{app}$	${\rho }_p$	${\varphi }_t$	${\varphi }_{intra}$
Material	(mm)	pH	($\mu$S cm^–1)	(g$\cdot$cm^−3)	(g$\cdot$cm^−3)	(g$\cdot$cm^−3)	(cm${ }^3\cdot$cm^−3)	(cm${ }^3\cdot$cm^−3)
Porlith	2–7	5.7	2.5	0.65	1.24	2.77	0.76	0.55
Expanded shale	2–7	7.4	109.9	0.78	1.42	2.41	0.68	0.42
Expanded clay	4–8	7.7	807.0	0.41	1.24	2.15	0.81	0.42
Tuff	3–8	6.9	81.5	0.88	1.47	2.55	0.66	0.42
Pumice	2–12	7.3	173.9	0.57	0.81	2.12	0.73	0.61
Lava	5–15	7.5	49.5	0.86	1.65	3.05	0.72	0.46
Sand	0.1–0.5	5.2	2.0	1.48	–	2.60	0.43	–

The ${\rho }_{bulk}$ of sand adjusted to 1.48 g cm⁻³ at the composite sample preparation has a corresponding ${\varphi }_t$ of 0.43. The ${\rho }_{bulk}$ of the pure CPMs presents lower values ranging from 0.41 to 0.88 g cm⁻³. Their ${\rho }_p$ are high (2.15–3.05 g cm⁻³), and therefore, the corresponding ${\varphi }_t$ of the CPMs reaches 0.76 and 0.81 for porlith and expanded clay, respectively, as the highest values, and 0.65 for tuff as the lowest (Table 1).

The values for ${\varphi }_t$ measured in this study are in line with reported results. For example, Asdrubali and Horoshenkov (2002) report a ${\varphi }_t$ of 0.74–0.84 for expanded clay, and Beardsell et al. (1979) report a ${\varphi }_t$ of 0.58 for volcanic scoria (similar to tuff). Comparing ${\varphi }_t$ of the CPMs with non-mineral materials their values are lower (even of expanded clay). Wood fibers, for example, exhibit a quite high ${\varphi }_t$ of 0.93 (Gruda and Schnitzler 2004), and Beardsell et al. (1979) report a ${\varphi }_t$ of 0.77 for pinebark and 0.95 for peat.

The high ${\varphi }_t$ of the CPMs of our study specifically reflects the large particle size and the shape of the CPMs, which results in large inter-particle pore sizes. Consequently, the high ${\varphi }_t$ is not effective if the CPMs are included in mixtures with fine textured fractions, where smaller particles would easily fill the large voids among the CPMs, lowering ${\varphi }_t$. This is the case for the composite samples of CPMs and sand, where the addition of sand decreases considerably the ${\varphi }_t$. Such effects have to be considered when CPMs are assessed as plant substrate components which are planned to be mixed with smaller particles. Instead, the ${\varphi }_{intra}$ is a more adequate reference to describe the effective porosity of CPM pebbles in mixtures. Porlith and pumice present the highest ${\varphi }_{intra}$ of the CPMs, with 0.55 and 0.61, respectively, while further materials range between 0.42 and 0.46 (Table 1).

3.2. Water retention characteristics

3.2.1. Water retention characteristics of the pure materials

Figure 2 shows the water retention curves of the CPMs (pure sand included) and the corresponding fitted unimodal or bimodal retention functions.

Figure 2. Observed water retention data and fitted retention curves for the sand, composite samples and pure CPMs. Sand: ($\times$) measured, ($\cdot \cdot \cdot \cdot$) fitted, sand mixed with CPM: ($\bullet$) measured, (- -) fitted, CPMs: ($\circ$) measured, (—) fitted (uncertainty is symbolized by - -). The graph of porlith (upper left) shows the measured points for sand. In the graphs can be observed the inflections of the air entry point(s), and the water held in the range of available water content (AWC) from pF 1.8 to pF 4.2.

3.2.1.1. Sand

As pore sizes are directly related to the water potential (Kutlek and Nielsen 1994), the relatively low air entry point and a steep slope of the water retention curve of sand is related to a narrow inter-particle pore size distribution.

Sand has no intra-particle porosity, and its inter-particle pores are predominantly in the range >50 μm, so that only very little water is held at pF > 1.8, namely <0.03 m³m⁻³. In comparison with naturally textured sands (0.16–0.19 m³m⁻³; Ad-Hoc-Arbeitsgruppe Boden 2005) such value is comparably low and it is attributed to the well-sorted particle size distribution of the used sand (mainly between 0.1 and 0.4 mm).

3.2.1.2. CPMs

All CPMs show a first air entry point that indicates an abrupt drainage of a pore domain in the range of >0.3 mm (draining at pF < 1; Kutlek and Nielsen 1994), corresponding mainly to the large inter-particle voids and may also include visible large pores of lava and expanded shale (pictures included in Fig. 2). Only porlith and tuff show a clearly pronounced second air entry point implying a higher amount of smaller pores in the range between 3 and 30 μm related to their ${\varphi }_{intra}$. For expanded clay and lava, this is less pronounced. Expanded shale shows no distinctive inflection of a secondary air entry point, but a constant water release rate at pressure heads >1.0, which indicates a wide pore size distribution of its ${\varphi }_{intra}$. Pumice up to pF 2.0 shows no secondary air entry point.

Due to the coarse grained texture of the materials, it can be assumed that the water in the large inter-particle pores held at FC is negligible, whereas the water at this point is mainly retained by the intra-particulate pores.

Despite all materials are commercialized either as amendments to increase the water retention capacity (WRC) of soils, or as soilless medium substrates (Gül and Sevgican 1993; Beattie and Berghage 2004; Sailor and Hagos 2011; Sailor et al. 2008; Roth-Kleyer 2007), only porlith has a high water content at FC (0.44 m³m⁻³), followed by tuff (0.23 m³m⁻³). Further CPMs have values ≤0.11 m³m⁻³ (Table 2). Porlith and tuff also present the highest values of AWC (0.37 and 0.17 m³m⁻³, respectively), while the rest of the CPMs do not exceed 0.10 m³m⁻³.

Table 2. Available water capacity (AWC), field capacity (FC), and parameters of unimodal (applied just for sand) and bimodal Peters–Durner–Iden model (2015) fitted to the data of pure materials and the materials mixed with sand.

	Parameter
Material	$h_{m1}$	${\sigma }_1$	$h_{m2}$	${\sigma }_2$	${\theta }_r$	${\theta }_s$	$w_1$	RMSE	AWC	FC
Porlith	6.99	0.286	284.20	0.205	0.17	0.76	0.50	2.37$\cdot$10^−3	0.37	0.44
Exp. shale	6.25	0.200	3.36	0.830	0.11	0.68	0.66	4.88$\cdot$10^−2	0.08	0.09
Exp. clay	2.69	0.200	120.70	1.091	0.00	0.81	0.18	3.34$\cdot$10^−2	0.09	0.10
Tuff	908.42	0.208	3.54	0.567	0.11	0.66	0.73	1.43$\cdot$10^−2	0.17	0.23
Pumice	2.68	0.355	2.68	0.338	0.15	0.73	0.60	4.12$\cdot$10^−2	0.09	0.11
Lava	3.49	0.499	56.97	0.207	0.07	0.72	0.04	0.84$\cdot$10^−2	0.06	0.07
Sand	20.45	0.691	–	–	0.02	0.43	–	7.03$\cdot$10^−3	0.03	0.03
Porlith + sand	29.38	2.791	29.38	0.200	0.11	0.46	0.80	0.69$\cdot$10^−2	0.08	0.13
Exp. shale + sand	176.51	1.248	27.68	0.307	0.04	0.42	0.86	0.44$\cdot$10^−2	0.06	0.08
Exp. clay + sand	27.68	0.200	27.68	0.201	0.09	0.47	0.75	0.15$\cdot$10^−2	0.06	0.09
Tuff + sand	26.07	0.297	186.78	1.856	0.07	0.43	0.25	0.58$\cdot$10^−2	0.10	0.13
Pumice + sand	26.92	1.678	22.19	0.401	0.02	0.48	0.85	0.23$\cdot$10^−2	0.07	0.09
Lava + sand	121.15	1.170	24.17	0.305	0.01	0.44	0.86	0.33$\cdot$10^−2	0.05	0.05

For simplification, ${\theta }_s$ was set equal to ${\varphi }_t$ and thus not fitted. The parameter $h_0$ was set to 10^6.8 cm, which is the suction at oven dryness for 105°C (Schneider and Goss 2012). $h_i$ and ${\sigma }_i$ are shape parameters. The first can be related to the air entry point and the latter can be related to the inclination. $w$ is a weighing factor.

Porlith has a high AWC of 0.38 m³m⁻³, which is in a comparable magnitude with the one for bricks (up to 0.4 m³m⁻³; Blume and Runge 1978; Nehls et al. 2013). These comparable water retention characteristics can be attributed to similarities of the structure of these two materials.

For tuff, the measured FC (0.23 m³m⁻³) exceeds the reported values of 0.10 m³m⁻³ (Da Silva et al. 1993) and 0.12 m³m⁻³ (Wallach et al. 1992). The tuff studied in this work has, accordingly, more pores able to retain water in the range of the AWC.

Expanded clay, expanded shale, and pumice have similarly low FC and AWC compared to tuff and porlith. The expanded clay has a ceramic cover, which is likely to disrupt the capillary continuum and thus hinders the uptake of water (see the structure of the material in Fig. 2). Expanded shale shows a low and narrow amount of pores retaining water in between pF 1.8 and pF 3.0, draining almost completely at pF > 3.0.

Regarding pumice, previous research on its porosity shows that it has a wide pore size spectrum and that shape and connectivity of the pores depends on the origin of the pumice (Whitham and Sparks 1986; Lura et al. 2004). Gunnlaugsson and Adalsteinsson (1994) report that the majority of pumice ${\varphi }_{intra}$ is composed by closed, occluded, or dead-end pores not available for water storage. This particular porosity may also be present in our studied material, and it can be reflected in the high variance of the WRC (Fig. 2) at pF ≤ 2.5 (SD = 0.007 m³m⁻³), and pF 3.0, with a SD = 0.053 m³m⁻³. For higher suctions (pF > 4.0), the values are more consistent as the materials were crushed prior measurement.

Lava has a very low AWC. Its water retention characteristics are quite similar to the ones of the sand of our study. The low AWC is mainly attributed to its large pore sizes >50 μm, some of them clearly visible, as depicted in Fig. 2.

In order to test the equivalence of the two experimental setups (pure CPMs, measured with Equi pF apparatus vs. CPMs mixed with silt, measured with ceramic pressure plates) as well as the accuracy of the applied model fitted to the experimental data, the water content of the CPMs at pF 1.8 of both setups was compared. The differences in the water content values calculated between the two methods are smaller than 0.055 m³m⁻³ for all CPMs, except for pumice and expanded shale. This high difference points to an inconsistency between the two applied setups. Since for the parametrization of the water retention model we used the outcomes of both setups, this inconsistency results in a high uncertainty of the model. Thus, for pumice and expanded shale, the water retention curve can be given only for rather low suctions. In Fig. 2, the mentioned uncertainty is symbolized by the dash dotted line. The difference in the experimental setups points to the need for a specially adapted method which accounts for a higher representative volume of CPMs.

3.2.2. Water retention characteristics of the mixtures

The water retention characteristics of the composite samples show just slight differences when compared to the sand. The data indicate two pore size domains for the composite samples: a highly pronounced pore size domain related to the sand matrix, and a secondary pore size domain, introduced by the CPMs. The latter is discernible by a moderate, but constant water release at pressure heads within the range of AWC (pF 1.8–4.2), with corresponding pore sizes between 0.2 and 50 μm.

The overall increase of AWC is quite moderate (maximum 0.07 m³m⁻³; Table 2); however, the increase can be clearly related to the added CPMs, as AWC for the pure sand is 0.03 m³m⁻³. In order to emphasize the effective contribution of CPMs to the increase of AWC, it is necessary to quantify the water retained in their ${\varphi }_{intra}$.

3.2.2.1. Water content of the CPM ${\varphi }_{intra}$

This retained water calculated with Equation (1)(1) ${\theta }_{CPM}=[{\theta }_t-((1-X)\mathrm{x}{\theta }_{s{a}_i})/X]\mathrm{x}(1-{\varphi }_{inter})$(1) shows a moderate to high water retention of CPMs at FC, with porlith and tuff being the most effective ones (Fig. 3). We found that the FC and the AWC for pebbles mixed with silt are higher than for those mixed with sand, except for expanded clay. This may be due to an additional porosity created in the contact zone between the surface of the CPMs and the mixing material, which occurs in a higher extent for silt than for sand. The calculation of this additional porosity was not included in Equation (1)(1) ${\theta }_{CPM}=[{\theta }_t-((1-X)\mathrm{x}{\theta }_{s{a}_i})/X]\mathrm{x}(1-{\varphi }_{inter})$(1) , and cannot be reliably determined by the methods applied.

Figure 3. Volumetric water content of the CPMs intra-particle porosity (${\varphi }_{intra}$) held at matric force of pF 1.8 (considered as field capacity) and AWC (cross hatched columns) when embedded in sand compared when embedded in silt. The pebbles of the CPMs were embedded in silt to ensure capillary contact when measured with ceramic tension plates at matric forces ranging from 1.8 to 3. Therefore, the water content of the pure sand and silt was obtained and subsequently subtracted with Equation (1)(1) ${\theta }_{CPM}=[{\theta }_t-((1-X)\mathrm{x}{\theta }_{s{a}_i})/X]\mathrm{x}(1-{\varphi }_{inter})$(1) . The result is the water stored specifically inside the pebbles.

For expanded clay, the sandy matrix is not able to drain the water from the shell fine pores, while in the silt matrix this is possible. Therefore, the water in the inner pore system can drain when embedded in silt, but is insulated when embedded in sand. The water insulated in the CPM of the composite samples can be favorable. It could solve the problem of the low field capacity and could store water, which otherwise would drain fast and would be lost for plants, notwithstanding the prevention of flash flooding due to heavy rain events. The roots of the plants can colonize the CPMs to obtain the water and the nutrients stored there, as it has been proved in the case of bricks (Nehls et al. 2013). Therefore, the capacity of the substrates to be rooted should be tested in further works.

3.2.3. Resume for practical use

Despite the moderate retaining water performance of four out of six of the analyzed CPMs, most of them are widely used (Roth-Kleyer 2007) as the main components of green roof substrates where they are readily available (Sailor et al. 2008; Ampim et al. 2010; Sailor and Hagos 2011). However, they are mostly used in mixtures with organic material and other mineral components, improving their performance.

In the design of a Technosol profile with the purpose of sustaining vegetation, a mixture of sand and porlith or tuff, which considerably increases the AWC, can accomplish the function of a good quality subsoil layer. In order to construct a proper topsoil, a further addition of organic material (e.g., compost) is needed. The addition of organic material increases the overall AWC, along with the nutrient retention capacity, and improves the texture. In contrast, mono-substrates of CPMs commonly used for certain horticultural purposes (e.g., for germination, in pots, or in green roofs) hinders soil aggregation and provokes excessive aeration due to the large size of the materials.

With the knowledge of potential AWC of each CPM, an optimization of the composite substrates in terms of the overall AWC can be conducted. The results obtained in this work indicate this can be achieved by a higher mixing ratio CPM : sand. For instance, a soil with a FC of about 0.30 m³m⁻³ (recommended by the German Landscape Research, Development and Construction Society, FLL 2002 as a minimum for green roof substrates) can be reached mixing porlith with naturally textured sand (FC 0.16–0.19 m³m⁻³; Ad-Hoc-Arbeitsgruppe Boden 2005) at a ratio of 1:1. Additionally, the soil can be complemented with organic material to reach even higher water contents.

4. Conclusions

The CPMs are commonly used and commercialized for greening purposes due to their high porosity. The AWC is high for porlith, moderate for tuff, but low for pumice, lava, expanded clay and expanded shale. Hence, the water retention characteristics of sandy soils can just be weakly improved by the addition of the last four mentioned. Tuff and porlith positively affect the hydraulic characteristics of sandy soils; however, the applied ratio between sand and CPM has to be adapted in order to gain the optimum in AWC.

Nevertheless, a high water content in the intra-particle pores of the CPM when mixed with sand was calculated for most CPMs, despite the low overall AWC in the mixtures. Such retained water could be available for plant roots entering the CPMs.

The pore sizes of the CPMs show a bimodal distribution for pure materials as well as the composite samples with sand. There are two distinctive pore sizes, one for the inter-particle pores among the CPMs or within the embedding sand matrix, and the other for the intra-particle pores of the CPMs.

We conclude that the behavior of the components must be known in detail and that the commercialized mixtures have to be investigated in a high resolution, probably using the here introduced methods in order to adjust the mixtures according to the purpose.

Acknowledgments

This work was supported by the Deutscher Akademische Austauschdienst (DAAD, German Academic Exchange Service) under Grant A/10/80575. We thank Dr. Peter Dominik for his contribution of ideas and for the constructive discussion.

Additional information

Funding

This work was supported by the Deutscher Akademischer Austauschdienst (DAAD, German Academic Exchange Service) under Grant A/10/80575.

Notes

1. See www.ums-muc.de/assets-ums/00A4R.pdf.