Physicochemical Design Rules for the Formulation of Novel Salt

Graphical abstract


102
Eight diverse salt samples produced using a range of processing methods were evaluated. 103 Their physicochemical properties, the efficiency of transfer to product and adhesion to 104 product, the release of sodium ions during dissolution and subsequent saltiness perception 105 was assessed.   The bulk density (ρb) of salt samples were obtained gravimetrically using a dry glass 10 152 mL graduated cylinder at 20°C (room temperature) and was calculated using the weight 153 and corresponding volume according to Equation 2. The tapped density (ρt) was 154 measured in the same way, but the measuring cylinder was tapped strenuously until no 155 further change in volume took place (Basu & Athmaselvi, 2018 Germany) and external cross-calibration between pulse-counting and analogue detector 204 modes were used when required. Internal standards, used to correct for instrumental drift,

Extracted TI parameters 286
A number of parameters were extracted from TI curves relating to saltiness intensity, rate 287 and duration, including perceived maximum intensity of saltiness (Imax), area under the TI 288 curve (AUCsensory) and the maximum perceived saltiness over time to Imax (rate Imax).

289
Extracted parameters are further detailed in Fig. 3 and Fig. 4.    The size of RS directly impacted bulk density and tapped density, with larger particle 384 fractions having higher densities (Table 1). This is expected as larger particles flow and 385 pack more readily. This is also observed for CI %, where the larger particles have a CI % 386 of < 11%, indicating "excellent" or "good" flow, the smallest RS (RS <106 µm) has a CI % 387 of 30.8%, which indicates "poor" flow properties.

388
Dendritic salt had a bulk density, tapped density and CI% that in all cases is similar or sits 389 between RS<106 µm and RS 106-425 µm. Given that dendritic salt has a mean particle 390 size (239 µm) that also sits between these two fractions (44 µm and 299 µm), it can be 391 assumed that it behaves similarly to RS particles.     (Table 1). Similarly to transfer efficiency results, smaller particle sizes remain 437 adhered to the peanuts with less particle loss. Whilst the global correlation between mean 438 particle diameter and adhesion after packaging (r=0.60) was weaker than transfer 439 efficiency and was not significant (p=0.11, Table 1), there was a significant impact of 440 particle size on adhesion losses for RS (P<0.05) and FMS (P<0.05).

441
As mentioned previously in section 3.6.1, adhesion forces between particles and food surface. In the first instance, larger salt crystals are more exposed to mutual contact than 448 smaller particles, so the larger particles in this study are lost first. Larger particles are more 449 likely to pack more closely. When coated in fat, they cling to each other, thereby further 450 overcoming adhesion forces.

451
In the second instance, as the peanuts fall and impact the bottom, they 'shake' off some of 452 the salt particles. This is due to the transfer of kinetic energy from peanuts to salt crystals.

453
The kinetic energy of a salt particle is proportional to its mass and hence its volume. 454 Therefore, larger particle sizes achieve greater kinetic energy than smaller particles with  Oral processing is a rapid event. In most cases, salt crystals cannot fully dissolve before a 467 bolus is formed and swallowed (Tian and Fisk, 2012). This incomplete dissolution limits 468 potential saltiness perception. To evaluate this, the salt particles were applied to a real 469 food matrix, oiled peanuts, and dissolution of salt was observed by the change in 470 conductivity of the dissolution media (RO water). Raw conductivity data was converted to a 471 percentage of total conductivity to observe comparative dissolution kinetics between 472 samples over time. The dissolution graph in Fig. 2 shows a slow increase in conductivity 473 (%) in all samples until 5 seconds and then followed by a rapid increase in conductivity (%) 474 between 5 s -20 s. After 20 s, the increase in conductivity slows again.

475
As shown in Fig. 2 R to elucidate relationships between variables (Fig. 3). In general, samples can be seen to 486 be separated on the biplot by particle size along the axis t1 and by NaCl content and 487 processing type on axis t2.

488
Extracted parameters from the dissolution study are clearly separated on axis t1. T25%,

489
T50% and T75% are closely correlated and are negatively presented on axis t1. T90% is 490 less well resolved. This is partly due to the dissolution kinetics of RS <106 µm sample, 491 which follows a different dissolution profile, as shown in Fig. 2. The dissolution for this 492 sample slows more quickly than the other samples. This is proposed to be due to the 493 strong adhesion forces between surface oil and the highly compact salt particles of small 494 particle size.

495
Particle size is positively correlated with the time to reach 25% conductivity (r=0.65), and 496 this was almost significant at p=0.06. The linear regression is weaker than expected due to 497 the outlying trend of RS <106 µm mentioned previously. ANOVA results confirmed 498 significant differences between different particle sizes for curve parameters (further 499 information can be found in supplementary Table 1). Samples on the negative side of axis 500 1, RS and FMS 425-600µm, can be described as dissolving more slowly due to their larger 501 particle size and lower surface area.

502
FMS samples are presented closely to T25%, T50% and T75%, indicating that foam-mat   Table 2; however, due to their importance in  Table 2.

537
Stop and max. duration). In general, group 1 is related to the intensity of saltiness and total 538 saltiness. Group 2 is related to the temporal aspects of saltiness perception (correlation 539 highest with slower time to maximum saltiness). These two groups are highly negatively 540 correlated.

541
SODA-LO® followed by <106 µm samples of both RS and FMS are positively correlated 542 with group 1 TI parameters (Fig. 3) with the highest peak intensity of mean TI curves (Fig.   543 4). ANOVA on extracted parameters from the TI curves confirmed that these salts have 544 the highest values for Group 1 parameters (Table 2). In comparison, 425-600 µm samples 545 of both RS and FMS resulted in the lowest mean TI curve peaks (Fig. 4), saltiness 546 intensity (Table 2) and are positioned further away from Group 1 parameters in the PLS-R 547 biplot (Fig. 3). This supports previous studies showing a reduction in particle size results in the Na + and changing the hydrophobicity of the particle which ultimately restricts Na + 564 release and dissolution (Supplementary Table 1).  Table   567 2). The initial slope gradient extracted from between 0 and 20 s of the dissolution curves is 568 positively correlated with rate to Imax (r=0.85, p=0.01), which is a key marker of Group 1.

572
Dissolution slope is significantly inversely correlated with the group 2 sensory parameters, properties were used to explore the impact of particle design on adhesion to product, loss in-pack, rate of dissolution and saltiness perception, ultimately to generate a series of 605 design rules that address each of the initial three phases proposed as potential routes to 606 optimise saltiness perception.

607
Phase I: Adhesion during application and before packaging: 608 Key Finding: Transfer efficiency is driven by particle size (r=-0.85, p=0.008), bulk 609 density (r=-0.801, p<0.05) and flow properties (r=0.77, p=0.015) 610 • Decreasing regular table salt particle size increased transfer efficiency during 611 coating, likely due to increased interaction with surface fat on the product.

612
• Foam mat processing increased transfer efficiency indicating this is due to reduced 613 bulk density.

614
• Flow properties were correlated with transfer efficiency suggesting particle-particle 615 interactions also play a role.

616
Phase II: Adhesion during packaging and transport:

617
Key Finding: Loss from the product in packaging is driven by particle size (p<0.05) 618 • Smaller particle sizes exhibited less loss due to enhanced adhesion energy 619 between surface oil on the product and the smaller salt crystals. • Smaller particles sizes were associated with faster sodium dissolution rates; 624 however, this was compromised for highly dense small particles due to high levels 625 of interaction with surface fats.

626
• Smaller particle sizes had a greater saltiness intensity (Imax) due to faster 627 dissolution in saliva.

628
• Greater particle hydrophobicity resulted in slower sodium release.

629
In summary, to maximise potential perceived saltiness, salt particles should be designed 630 with small particle size, low density and hydrophobicity and have a particle shape 631 associated with optimal flow properties. Also, the in vitro sodium dissolution method used 632 in this study was able to predict key parameters associated with in vivo saltiness time-

633
intensity. Future studies should investigate these design rules within a commercial product 634 context and seek to validate the potential for sodium reduction whilst retaining consumer 635 acceptability.

636
In addition to salt, these physicochemical design rules may apply to new product