Preparation and optimization of a lignin-based pressure-sensitive adhesive

Abstract

D-optimal designs were applied to find the best parameters for the preparation of lignin-based pressure-sensitive adhesives (PSA) for sticky notes. Organosolv lignin was directly incorporated into a polycarboxylate polyether (PCE)/water matrix. The independent variables considered in the experimental design were the ratio between PCE, lignin, and water and the curing parameters. The distance traveled by the ball (tack), the peel-off losses and the final water content were the analyzed responses that allowed the optimization of the PSA formulation. The accuracy, the precision and the efficiency of the model were evaluated during the first experimental design for the formulation of the lignin-based adhesive named DES-OL-ADH. This formulation was optimized during the second experimental design abbreviated DES-OL-OPT. The coefficients of determination of the tack, the peel-off losses and the final water content were 0.98, 0.99 and 0.99, respectively. The model was satisfactory which allows the optimization of the PSA formulation. The DES-OL-OPT suggests that lignin-based PSA can be prepared as a sticky note application with 5 wt% of lignin, 84 wt% of PCE and 11 wt% of added water in the oven at 130 °C for 60 min, which shows a higher tackiness and similar peel-off losses than the commercial sticky notes PSAs.

• •Protocol optimization for the preparation of a green pressure sensitive adhesive (PSA) from PCE polymer, lignin, and water.
• •Influence of 5 compositional or processing parameters on adhesive performance through a 2-steps d-optimal experimental design.
• •Development of a new method, based on peel-off losses, to assess the performance of a PSA.

Graphical abstract

Specifications table

Subject area:	Materials Science
More specific subject area:	Lignin-based adhesive
Method name:	Preparation of a lignin-based pressure-sensitive adhesive for a sticky note application
Name and reference of original method:	Sivasankarapillai, G.; Eslami, E.; Laborie, M.-P. Potential of Organosolv Lignin Based Materials in Pressure Sensitive Adhesive Applications. ACS Sustainable Chemistry & Engineering 2019,: 7 (15), 12817-12824
Resource availability:	The resources necessary to reproduce the method are included in this article.

Method details

This method article is related to the research article published in the journal Industrial Crops and Products: Characterization of different types of lignin and their potential use in green adhesives (Gendron et al., 2022). [1] Using three grades of organosolv and one of Kraft lignins, this related article describes how lignin characteristics, such as molecular weight and OH group content, influence their efficiency upon incorporation in an optimized lignin/PCE water-based formulation. The main purpose of the present MethodsX article is to support our previous work by presenting the methodology through which this formulation for a green pressure-sensitive adhesive, made from PCE polymer, lignin, and water, was optimized. The secondary objectives are to analyze the interactions between the adhesive synthesis variables and to develop the correlations between these variables and the properties of the adhesive. The methodological steps used for this study are described below via a formulation protocol and characterization methods of pressure-sensitive adhesives for sticky note applications. The results obtained with the various PSAs prepared in this work are compared to those of a commercial pressure-sensitive adhesive used for sticky notes (Post-it®).

Materials

Mixture components: Polycarboxylate polyether in aqueous solution/sodium salt (ETHACRYL™ HF (PCE)) was provided by Coatex (Arkema Group), and used as received. The solid content of the PCE was 40 wt% as determined by loss on drying (60 °C for 22 h). The organosolv lignin (OL) used in the experiment was purchased from Suzano Canada Inc. (Burnaby, BC, Canada) and was produced from aspen wood (unknown hybrid) according to the Lignol technology in the pilot plant of Ferndale (Washington, USA). [2,3]

Reference- Commercial sticky notes: 3 M commercial Post-it® (code 566) with the dimensions of 50.8 cm by 58.4 cm was used as a point of comparison to assess the characteristics of PSA formulations developed in this work. This commercial Post-it® is made of a paper sheet with a thin layer of PSA deposited alongside one edge of the sheet. PSA layers have a thickness of 14 µm ± 4 µm, as determined by an instant readout digital caliper. This was measured by considering the thickness of a Post-it® pad containing 20 sheets, in order to minimize uncertainties on small values. The thickness of each pad was measured on the side where the adhesive is deposited (thickness of sheets and glue: T_{Pad sheets+glue}) and on the opposite side (to measure the sheets only, T_{Pad sheets}) where there is only paper (Fig. 1). Measurements were taken 20 times to obtain a reliable average. The real glue thickness (T_Glue) is therefore calculated by Eq. (1).

$$T_{\text{Glue}}=\left(\left(\frac{{\overline{T}}_{\text{Pad}\text{sheets}+\text{glue}}}{20}\right)-\left(\frac{{\overline{T}}_{\text{Pad}\text{sheets}}}{20}\right)\right)$$ (1)

Where${\overline{T}}_{\text{Pad}\text{sheets}+\text{glue}}$ and ${\overline{T}}_{\text{Pad}\text{sheets}}$ are the averages of T_{Pad sheets+glue} and T_{Pad sheets}, respectively.

Fig. 1.

Scheme illustrating the execution of the thickness measurement of the commercial adhesive note Post-it® pad (a), the execution of thickness measure (T_{Pad sheets}) with the caliper used for the test (b), and the measurement details (glue and paper sides) (c) are illustrated.

Experimental setup and procedure

Sivasankarapillai et al. first synthesized pressure-sensitive adhesives from polycarboxylate polyether (PCE) and organosolv lignin. [4] Their procedure involved to: 1) add the components in a beaker and to stir the system at 100 °C for 15 min on a hot plate, 2) dry the resulting homogeneous blend in a vacuum oven for 1 hour at 150 °C. Our preliminary tests did not show any plus value for heating during step 1. In order to streamline the process and ease handling, this work suggests to use only one heat source: the oven. Mixing is first achieved at room temperature, and the blend is then placed in the oven at the desired temperature for a set time (see the experimental parameters in Table 1). Using only an oven instead of a hotplate favors a steady temperature in the mixture, while facilitating manipulations and scaleup. [5,6] No stirring is then required in the oven, which strengthens the reproducibility of the procedure.

Table 1.

Initial experimental design variables (DES-OL-ADH).

Variables	Role	Coded Levels
		−1	0	+1
Lignin	Mixture	0.05	0.13	0.20
PCE		0.80	0.88	0.95
Added water		0.00	0.08	0.15
Oven temperature ( °C)	Discret digital	90	110	130
Oven time (min)		60	120	180

As a result, four steps are involved in the preparation of PSA samples (Fig. 2). The first is to weigh its components: dried organosolv lignin is weighted directly into the beaker, PCE, and then water are added in the desired ratios. The total weight of the mixture is 40 g. Secondly, the solution is stirred for 30 min at 500 rpm. Thirdly, the mixture is heated in an oven to the desired temperature (90, 110 or 130 °C) for a determined time (60, 120 or 180 min). Fourthly, the adhesive blend is covered and cooled overnight down to room temperature. Adhesive samples were then subjected to a tack and a peel-off losses tests to determine their performance as PSAs. Results were generated and analyzed in the form of an experimental design with statistical analysis.

Fig. 2.

Schematic PSA preparation in experimental design.

Each sample of PSA was analyzed to obtain three responses: distance traveled by the ball (tack distance), peel-off losses, and final water content. Tack, determined with the rolling ball tack method, is the distance traveled by a ball on an adhesive layer according to the ASTM D3121 method (Fig. 3). [7] Parameters of the method are listed in Table 2. Reported values correspond to the distances, as measured from the point where the ball initially contacts the adhesive to its final position when the rolling stops. [7] The adhesive layer thickness was controlled by a rod coating (size 8) corresponding to a thickness of 20 µm ± 2 µm. Although this is slightly higher than the 14 µm ± 4 µm of the commercial adhesive sample, it is worth noting that there is no statistical difference between them.

Fig. 3.

Scheme illustrating the tack test execution according to ASTM D3121 method: a standard steel ball is placed on an inclined plan and activated for the descent by a trigger (a), the ball is rolled on the PSA sample until it stops (b), and the distance traveled by the ball is measured by a rule (c).

Table 2.

Tack test parameters.

Parameter	Unit	Value
Film thickness	µm	20 ± 2
Hard horizontal surface	—	Glass
Coated surface	—	Post-it® 3 M paper
Repetition	—	10 × 2 passing
Specimen's dimensions	mm	127 × 300 (experimental PSA) 38 × 508 (commercial PSA)
T emperature	°C	21 ± 2
Relative humidity	%	50 ± 5

Peel-off losses (POL) were measured by a simple and new method (Fig. 4). Five strips of paper (substrate) coated with a sample of the adhesive were prepared. Their dimensions were 2 cm by 7 cm. The adhesive layer thickness was 20 µm ± 2 µm (controlled with an 8-sized rod coating). Cleaned microscope glasses (Fisherfinest Premium Microscope Slides Superfrost of 25 × 75 × 1 mm) were used as a control surface for the adhesion. Each test was repeated five times for statistical purposes. Both substrate (paper + PSA) and the microscope glass are first weighed separately on a precision scale. The substrate is then deposited gently on the center of the microscope glass, and a standard weight of 1 kg is placed on top of the paper strip for five seconds. The paper strip was then peeled-off using pliers. Both the paper strip and the glass were reweighed after peeling in order to determine whether part of the adhesive had remained bonded to the glass. A calculation was then performed to determine the peel-off losses in mg/cm² (Eq. (2)):

$$POL=\frac{\left({\overline{W}}_a+{\overline{W}}_b\right)}{2A}$$ (2)

Where ${\overline{W}}_a$is the average losses of the adhesive on the paper strip measured by comparing them before (W₁ on the Fig. 4) and after (W₃) the peel-off test. ${\overline{W}}_b$ is the average deposit of adhesive on the glass measured by comparing the glass weight before (W₂) and after (W₄) the peel-off losses test. A is the coated strip surface area (14 cm²).

Fig. 4.

Scheme illustrating the execution of a peel-off losses test: a strip of paper coated with the PSA sample is deposited on a glass surface (a), a standard weight of 1 kg is applied for 5 s on the system (b), paper strip is then peeled off with pliers (c), and the peeled-off paper as well as the glass with PSA deposits are reweighted (d).

The final water content (wt%) of the sample was measured immediately after step 4 of the protocol (Fig. 2). Given the composition of the mixture and the fact that the protocol does not involve any chemical reaction, this work assumes that all volatile matters to be water molecules. The losses on drying (LOD) may then directly correlated to water evaporation. LOD are defined as the variation in mixture weight between the end of steps 1 (W_I initial total sample weight) and 4 (W_F final total sample weight, see Fig. 2) and are calculated according to Eq. (3).

$$LOD=W_I-W_F$$ (3)

The final water content (FW) may then be calculated according to Eq. (4):

$$FW=\left(\frac{((W_{PCE}*C_{H_{2}O})+W_{H_{2}O})-LOD}{W_F}\right)*100\%$$ (4)

Where W_PCE is the weight of PCE added in the preparation, $C_{H_{2}O}$ is the water content in the PCE and the $W_{H_{2}O}$ is the weight of the water added in the preparation.

Pressure-sensitive adhesives must be able to form strong physical bonds (high tackiness) with almost any surface through a simple contact, be removable with minimal effort without leaving any residue behind, and have minimum creep. [4,8] This is especially true for sticky note PSAs, where peel-off must be easy (without fracturing the bonded surface) and for peel-off losses have to be minimal. [9] The composition of the mixture is also important to take in account: we are looking for a PSA that, while optimizing the aforementioned characteristic, has the highest water and lignin contents. This is because increasing the water content reduces the cost and the environmental impact of the resin. Lignin, a biobased heteropolymer, also contributes to decrease the share of synthetic polymer matrices.

PSA experimental sample appearance was visually evaluated in terms of apparent viscosity, lignin dispersion into PCE/water matrix and spreadability. Apparent viscosity is classified in three categories which are pasty (PSA sample was impossible to stir with a glass rod), more viscous (we felt resistance when we stirred the PSA sample with a glass rod) and less viscous (PSA sample was easy to stir with a glass rod). When no lignin particle was visible in the sample blend, a good dispersion of the lignin was considered. A good spreadability was obtained when the PSA was easy to spread with the coating rod and the resulting PSA layer was homogeneous on paper.

Experimental designs

Experimental designs allow for optimization of scientific research trials: they maximize the amount of information that can be derived from a limited number of experiments. Experimental design seeks to establish a relationship between the quantities of interest Y (response) and some experimental variables, X. [10] This work uses a d-optimal experimental design. . The selection of experimental points was then assisted by the software JMP Pro 15  [11] which calculates, using an exchange algorithm, the best plan for the study. [10] Two experimental d-optimal designs were generated for this work. The first (DES-OL-ADH) was used to assess the impact of each variable on the responses in order to determine the optimal zones. The second experimental design (DES-OL-OPT) was built from the analyses DES-OL-ADH to refine the model. From the three mixture components (lignin, PCE and water), two are continuous and independent, while the last is dependent of the two others (sum of the weight fractions is equal to “1”). Other variables, oven temperature and oven time, are continuous and independent of each other. Three coded levels (−1, 0, +1), were initially examined for each variable (Table 1). Continuous variables, X_i are hence analyzed over the experimental range $X_{\mathrm{i}}\in [-1;+1]$, and the level 0 is taken as the midpoint of this interval. The various $X_{\mathrm{i}}\in [-1;+1]$ intervals were determined from preliminary experiments and literature⁴: $[0.05;0.20]$, $[0.80;0.95]$, and $[0.00;0.15]$ for lignin, PCE, and water contents respectively; $[90;130]$ C for the oven temperature; and of $[60;180]$ min for the oven time (Table 1). For the second experimental design DES-OL-OPT, these intervals and the number of coded levels were refined to $[0.05;0.10]$ (lignin, 3 levels), $[0.80;0.91]$ (PCE, 5 levels), $[0.03;0.15]$ (water, 5 levels), $[90;130]$ C (oven temperature, 3 levels), and $\left\{60\right\}$ min (oven time, 1 level). The lower and upper limits for the corresponding responses, listed in Table 3, were determined experimentally, and implemented in the software. Experimental characteristics measured for the commercial Post-it® adhesive, listed in Table 4, were considered as objectives to construct the model. The DES-OL-ADH and DES-OL-OPT were composed respectively of 24 (Table 6) and 12 (Table 7) experiments. One experiment (#1) is common to both experimental designs. The DES-OL-ADH design had a central point and four repeated points. In the second design DES-OL-OPT, each sample was repeated twice to refine the model.

Table 3.

Experimental design responses to optimization.

Responses	Units	Lower Limit	Upper Limit	Goal
Distance traveled by the ball	cm	15	30a	Match Target
Peel-off losses	mg/cm2	0	10	Minimize

^aLimited by 8-sized rod coating used for the experimental sample tack test.

Table 4.

Characterization of a commercial sticky note adhesive.

Characterization	Distance Traveled by the Ball (cm)a	Peel-off Losses (mg/cm2)a
Commercial sticky note PSA	31 ± 9	0.1 ± 0.1

^aObtained from the method detailed in this article and used for the tests. The commercial PSA was used, as received.

Table 6.

First experimental design d-optimal DES-OL-ADH.

Tests	Mixture Variables			Continuous Variables		Responses
	Lignin (wt%)	PCE (wt%)	Added water (wt%)	Oven Ta ( °C)	Oven time (min)	Distanceb (cm)	Peel-off Losses (mg/cm2)	Final Water (wt%)
1	0.05	0.80	0.15	90	60	> 30	0.12 ± 0.05	60
2	0.20	0.80	0.00	130	60	2.0 ± 0.3	5 ± 2	36
3	0.05	0.88	0.08	90	120	19 ± 4	0.5 ± 0.2	54
4	0.20	0.80	0.00	90	180	—c	— c	39
5	0.05	0.95	0.00	90	180	14 ± 2	0.6 ± 0.3	49
6	0.05	0.95	0.00	130	180	2.2 ± 0.3	3.2 ± 0.7	20
7	0.20	0.80	0.00	130	180	6 ± 2	10.6 ± 0.1	6
8	0.13	0.80	0.08	110	60	8 ± 3	0.6 ± 0.2	52
9	0.10	0.85	0.05	130	60	7 ± 2	0.9 ± 0.2	46
10	0.20	0.80	0.00	90	60	— c	— c	45
11	0.05	0.95	0.00	130	60	15 ± 2	1.1 ± 0.1	48
12	0.13	0.80	0.08	130	120	1.5 ± 0.5	5.0 ± 0.7	30
13	0.13	0.88	0.00	110	120	7 ± 2	2.4 ± 0.2	42
14	0.13	0.88	0.00	90	60	8 ± 3	0.5 ± 0.1	50
15	0.05	0.95	0.00	90	60	17 ± 3	0.1 ± 0.1	54
16	0.05	0.80	0.15	90	180	18 ± 5	0.4 ± 0.1	54
17	0.05	0.80	0.15	130	180	2.7 ± 0.6	2.7 ± 0.3	24
18	0.05	0.80	0.15	130	180	2.7 ± 0.7	3.1 ± 0.5	24
19	0.20	0.80	0.00	130	180	5 ± 3	10.3 ± 0.8	6
20	0.05	0.88	0.08	110	180	14 ± 2	1.3 ± 0.2	45
21	0.05	0.80	0.15	130	60	22 ± 2	0.5 ± 0.2	54
22	0.05	0.95	0.00	90	180	16 ± 2	0.8 ± 0.2	48
23	0.13	0.80	0.08	90	180	5.3 ± 0.7	1.0 ± 0.2	48
24	0.05	0.95	0.00	130	180	4.0 ± 0.6	3.3 ± 0.6	26

^aOven temperature.

^bDistance traveled by the ball (tack test).

^cSample was too pasty to test.

Table 7.

Second experimental Design d-optimal DES-OL-OPT.

Tests	Mixture Variables			Continuous Variables		Responses
	Lignin (wt%)	PCE (wt%)	Added water (wt%)	Oven Ta ( °C)	Oven Time (min)	Distanceb (cm)	Peel-off Losses (mg/cm2)	Final Water (wt%)
1c	0.05	0.80	0.15	90	60	> 30	0.12 ± 0.05	60
25	0.05	0.80	0.15	90	60	>30	0.07 ± 0.04	59
26	0.06	0.91	0.03	90	60	26 ± 3	0.05 ± 0.06	55
27	0.06	0.91	0.03	90	60	>30	0.04 ± 0.03	55
28	0.05	0.85	0.11	110	60	>30	0.09 ± 0.09	57
29	0.05	0.85	0.11	110	60	>30	0.11 ± 0.09	57
30	0.05	0.84	0.11	130	60	19 ± 5	0.4 ± 0.3	53
31	0.05	0.84	0.11	130	60	18 ± 3	0.4 ± 0.2	52
32	0.10	0.85	0.06	110	60	16 ± 4	0.5 ± 0.2	54
33	0.10	0.85	0.06	110	60	14 ± 3	0.3 ± 0.1	53
34	0.06	0.82	0.12	90	60	>30	0.04 ± 0.04	58
35	0.06	0.82	0.12	90	60	>30	0.04 ± 0.03	58

^aOven temperature.

^bDistance traveled by the ball (tack test).

^cTest #1 is common to both experimental designs.

The quality of the experimental design fitting was controlled by the coefficient of determination, R², and the adjusted coefficient of determination R². [12] The precision and reproducibility of the design results were evaluated by repeating some experiments and by the lack of fit, the pure and total errors.

Method validation and optimization

Experimental results of experiments for the DES-OL-ADH design are listed in Table 6. Distance traveled by the ball (tack distance), peel-off losses and final water content varied from 2.0 to >30 cm, 0.1 to 10.6 mg/cm² and 6 to 60 wt% respectively. Statistical analysis made it possible to model these results and to determine the optimal adhesive formulation. Model parameters were fitted through a standard least squares regression that gave emphasis to parsimony, which means that the software minimized the number of parameters required to describe a phenomenon by analyzing how much extra-variance can be explained by adding a further parameter. A coefficient of determination (R²), which expresses the quality of the model fit can be determined for each response. [13] Closer R² is from 1 (0 ≤ R² ≤ 1), more is accurate the fitted model (R² = 1 corresponding to an exact fit). [14] The adjusted R² (Adj. R²) corrects R² values to account for the number of parameters used in the model: it is a measure of the model's parsimony. [12] A small difference between R² and Adj. R² points to a significant model whose parameters are highly orthogonal (independent from each other). These coefficients are listed in Table 8 and their respective correlations are represented in Fig. 5 (a to c) (removal efficiency plots). The results of experiments in the form of removal rate of tack distances (Y₁ as Eq. (5)), POL (Y₂ as Eq. (6)) and FW (Y₃ as Eq. (7)) are in accordance with Fig. 5(a), Fig. 5(b) and (c) respectively. The equation terms are simplified by symbols listed in the Table 5 to lighten each equation. The square exponent is the square terms (T² and t²), and the interaction terms are the multiplication of two model terms (LP, LH, LT, Lt, PH, PT, Pt, HT, Ht and Tt).

$$\begin{array}{cc}\hfill Y_1& =5.58L+13.99P+19.48H-2.57T^2+1.14t^2-8.40LP-29.86LH-0.14LT+3.45Lt+2.12PH-3.96PT-3.32Pt\hfill \\ & -5.57HT-7.68Ht-1.76Tt\hfill \end{array}$$ (5)

$$\begin{array}{cc}\hfill Y_2& =5.56L+1.58P+1.25H+0.36T^2-0.65t^2-3.41LP-3.57LH+2.60LT+1.94Lt-1.56PH+0.77PT+0.68Pt\hfill \\ & +0.75HT+0.69Ht+0.58Tt\hfill \end{array}$$ (6)

$$\begin{array}{cc}\hfill Y_3& =34.72L+46.78P+51.42H-4.48T^2+1.12t^2+4.96LP+1.01LH-10.75LT-8.58Lt+7.74PH-7.60PT-7.52Pt\hfill \\ & -9.09HT-8.63Ht-5.54Tt\hfill \end{array}$$ (7)

Table 8.

Statistical parameters of the DES-OL-ADH design (see Table 6).

Parameters	Distance Traveled by the Ball	Peel-off Losses	Final Water Content
R squared	0.98	0.99	0.99
R squared adjusted	0.95	0.98	0.98
Lack of fit	16.31	1.04	13.80
Pure error	3.64	0.14	20.36
Total error	19.95	1.18	34.16
Mean of response	10 cm	2.5 mg/cm2	40 wt%

Fig. 5.

Removal efficiency plots of DES-OL-ADH experimental design (where blue lines indicated the mean response): Distance traveled by the ball (tack test) (a), peel-off losses (b), and final water content (c).

Table 5.

Symbolic representation of the variables found in the prediction expressions (5) to (7).

Symbolic representation	Model Terms
L	$\left(\frac{Lignin\left(wt\%\right)-0.05}{0.15}\right)$
P	$\left(\frac{PCE\left(wt\%\right)-0.8}{0.15}\right)$
H	$\left(\frac{Addedwater\left(wt\%\right)}{0.15}\right)$
T	$\left(\frac{Oventemperature\left({}^{\circ }C\right)-110}{20}\right)$
t	$\left(\frac{Oventime\left(min\right)-120}{60}\right)$

All R² are greater than 0.98, which demonstrates a high correlation between observed (experimental) and predicted values for the experimental design DES-OL-ADH. Moreover, the Adj. R² values were greater than 0.95, demonstrating a high significance for the model. The quality of the fit is also illustrated by the total error of the fit, which is the sum of the pure error and of the lack of fit. [12] The pure error is an estimate of the error variance for exact replicates (variability in the responses at replicated conditions). [12,15] It is a measure of the uncertainty contained within the experimental data set used to fit the model. Lack of fit represents the error variance induced by restricting the total number of parameters. These statistical parameters were determined directly by the JMP Pro 15 software. [11] In Table 8, we can see that the pure errors (3.64 and 0.14) are smaller than lacks of fit (16.31 and 1.04) for the responses of distance traveled by the ball and peel-off losses. This means that the model, generated by our analysis of the DES-OL-ADH experimental design, simultaneously has high R² and pure errors considerably lower than lacks of fit for its responses. The model is then validated and can be employed to optimize the formulation of the adhesive in terms of tack distance and peel-off losses. In the case of the final water content, however, the pure error (20.36) accounts for more than half of the total error (34.16). The model is therefore less suited to predict this response. Since it is of less interest for the optimization of the formulation, we did not perform any additional tests to further adjust the model. The final water content is, indeed, primarily relevant as a measure of the bio-based content of the adhesive formulation and of the overall viscosity of the mixture (water decreases the mixture viscosity, facilitating handling and spreading). With an Adj.R² of 0.98, the model was estimated sufficient to meet our needs in terms of final water content prediction despite its high pure error.

Prediction profilers (Fig. 6) make it possible to visualize the behavior of our adhesive as a function of the five entry parameters considered by the model. This prediction tool facilitates the optimization of the adhesive formulation. As mentioned earlier, the prediction is expected to be highly precise for tack distances and peel-off losses responses and slightly less so for the final water content. These profilers will then be used to establish the second experimental design DES-OL-OPT by minimizing the difference between predicted responses and target values, which correspond to those of commercial sticky notes PSAs. This is an additional precaution taken to optimize the model and to increase its accuracy. Fig. 6(a) indicates that PSA tackiness increases monotonically with the content of organosolv lignin. Lignin hence increases the adhesiveness of PSA. It is also the case for oven time and temperature, although temperature appears to have little effect on tackiness between 90 and 110 °C. On the order hand, the tackiness decreased (so, distance traveled by the ball increased) with water content, while there was an optimum for PCE content around 85 wt%. The evolution of tackiness with the initial water content, oven temperature and oven time is consistent with the fact that diluting the final formulation reduces the adhesiveness of the resin. This is confirmed by matching these parameters with the appearance (more or less viscous) of the various resins. A qualitative evaluation of the resin appearance in terms of viscosity, lignin dispersion, and spreadability are provided in Tables 9 and 10, which summarized the impact of each parameter in the various responses.

Fig. 6.

Prediction profiler for each response of the DES-OL-ADH design (the red dotted lines indicate an arbitrary point): Distance traveled by the ball (tack test) (a), peel-off losses (b), and final water content (c).

Table 9.

Visual evaluation of adhesive's appearance for experiments of DES-OL-ADH (Table 6) and DES-OL-OPT (Table 7). Gray cases indicate whether the corresponding experiment were: pasty (column 2), more or less viscous average (columns 3 and 4), poorly dispersed (with visible lignin agglomerates, column 5), easily spreadable (column 6).

Table 10.

Impact of each variable on experiment responses, whether quantitative (tack, peel-off losses, and final water content) or qualitative (lignin dispersion, spreadability). The impact is either absent (blank case), positively low (+), positively moderate (++), positively strong (+++), negatively low (-), or negatively moderate (–). Where negatively means decrease and positively means increase.

Responses		Variables
		Oven Temperature	Oven Time	Lignin	PCE	Water
Appearance	Lignin incorporation	++	++	–	+	–
	Spreadability	+	+	–	+	–
Properties	Tack	++	++	+++	+	–
	Peel-off losses	–	–	–
	Final water content	–	–	–	+	++

Adhesive blends behaved quite differently when it came to peel-off losses test according to the prediction profiler (Fig. 6(b)) since their composition had very little influence on this response. Minimal losses were observed for low lignin contents (˂ 11 wt%), while PCE and water contents had a negligible influence. However, peel-off losses were, negatively correlated with the oven heating temperature and time (Fig. 6(b), Tables 9 and 10), which means that the peel-off losses increase with the oven temperature and time. We suggest that it is possible for lignin to act as a physical cross-linker within the resin. Hydroxyl groups of lignin might be able to form hydrogen bonds with the negatively charged oxygens of PCE's carboxylate groups. Being a polyol, lignin may hence create bridges between otherwise electrostatically stabilized PCE chains. This bridging effect is expected to increase with lignin content and with the overall solid content of the mixture (as water evaporates, the intermolecular distance decreases, which is favorable to the formation of weak interactions between the solutes).

For findings regarding the final water content (Fig. 6(c)), it is positively correlated with the initial water content and negatively correlated to the heating temperature and time (i.e., to the intensity of the evaporation). It is worth noting that the final water content is positively correlated with PCE content, but negatively with lignin. This is because the lignin employed in this work is a near-dry material, while the PCE salt solution initially contains 60 wt% of water, which means that PCE actually contributes to the total water content of the initial mixture. The final water content in most of the PSA formulations stood at around 50 wt%, which is too high for condensation to be significant. Handling of these adhesives revealed a liquid, but viscous, behavior, which tends to support that no significant cross-linking occurred. However, in some instances where the final water content was lower, such as in experiments 2 and 7, a pasty behavior was observed below 40 wt%_water. While the increase in solid content might explain the change in viscosity, the transition from a liquid to a pasty state displaying a yield stress could be consistent with an increase in the prevalence of physical (hydrogen bonds) and/or chemical (condensation) cross-linking. From Tables 9 and 10, it can be concluded that the heating temperature, the heating time, and the lignin content are the variables that most influence the characteristics (responses) of the prepared PSAs. This can be confirmed by analyzing the effects of each variable with the Pareto charts. Pareto's law is a tool that makes possible to identify the actions whose impacts will be the most significant. In this work's statistical analysis, Pareto's 80/20 rule is used, which states that roughly 80% of all effects come from 20% of the potential causes. The 80/20 rule is often interpreted as an instance of the Pareto distribution in mathematical analyzes. [12,16] Our analysis (Fig. 7(a)) shows that for tack distances, the water initially added (mixture variable) to the blend has the greatest effect (22%), closely followed by the PCE (21%). They are followed by the cross-effect of the water added to the oven time and temperature which have a significant effect on the tack distances. While lignin content is in the fifth position for the tack test, it has by far the most influence on the peel-off losses (Fig. 7(b)) since it accounts for more than 30% of all effect on the result (compared with less than 10% for all other variables). Note that the importance of the lignin content was also observed in the prediction profiler. Mixture variables (formulation components: PCE, water added, and lignin) represent more than 60% of the total effect on the final water content (Fig. 7(c)). Based on the statistical analysis and the prediction profilers of the DES-OL-ADH experimental design, six promising experimental conditions were identified, which narrowed the range over which parameters must be varied to achieve a commercial level of acceptability. They correspond to a high tack distance (31 ± 9 cm, see Table 4) and very weak peel-off losses (0.1 ± 0.1 mg/cm²). Note that final contents in water and lignin must be as high as possible to increase the biobased nature of the adhesive. Profilers (Fig. 6) predict that those conditions are most likely to be met over the following parameter ranges: 5 to 10 wt% for lignin content, 80–91 wt% for PCE, and 3–15 wt% for water, with an oven temperature of 90 to 130 °C and a fixed time of 60 min. According to these values ​​identified as more promising during the first experimental plan, six corresponding adhesive formulations, listed in Table 7, were prepared, and tested in duplicate to increase the precision of the model. Duplicating these experiments is an additional precaution taken to refine the model and increase its accuracy. The comparison of the predicted results from the DES-OL-ADH versus the observed results (actual) of the DES-OL-OPT was listed on the Tables 11 and 12. Experiment #1 from DES-OL-ADH was chosen as a reference because of its low amount of synthetic polymer (PCE) and its high-water content, which resulted in good properties: high-biosourced content, high tack distance, and low peel-off losses. These conditions were repeated once (experiment #25) to increase the biobased character, the distance traveled by the ball was higher and the peel-off losses were low. The actual results are relatively similar to the predicted results with 9% and 0% of relative errors for the peel-off losses and final water content, respectively. The experiments #26 and #27 had the lowest peel-off losses on the prediction profiler. The experimental result (actual result) was higher than the predicted one. The experiments #28 and #29 had the higher tack distance for the oven temperature of 110 °C and low peel-off losses. The #30 and #31 profiles were chosen because the heating temperature of 130 °C allow better lignin incorporation and a higher spreadability on paper than the other temperatures (90 °C and 110 °C) (Table 9). The tack distance decreases (higher adhesiveness) compared to other profiles, but the peel-off losses are low. The experiments #32 and #33 contain more lignin (10 wt%) but had a high predicted tack. For the last profile, which corresponds to experiments #34 and #35, it was a compromise between a low predicted tack and peel-off losses, but a low heating temperature (90 °C). The lignin content of 6 wt% of these tests was not studied in other experiments, so it was interesting to consider it. The relative errors between the predicted result versus the actual results are higher in case of the peel-off losses, but this can be explained by the very low values obtained for this test. The relative errors between the predicted and the actual results are higher for the peel-off losses than for the two other indicators (Table 11). We believe that it could be explained by the fact that the predicted POL values for the optimized formulations have the same order of magnitude than the method's uncertainty, measured to be in the range of ± 0.03 to ± 0.3 mg/cm². Hence, while the method was efficient at discriminating between unoptimized formulations, whose POL ranged from 0.04 to 10.6 mg/cm², or on nearly 2 orders of magnitude (Table 7), its precision proved insufficient to differentiate the various optimized formulations (predicted POL ranging from 0.01 to 0.32 mg/cm²). The uncertainty of the method could be improved by performing an automated, instead of a manual, peel-off measurements, but such equipment was not available in the context of this study. Here, the measurements were made using pliers. While a single operator was selected to perform all the tests to reduce uncertainty and ensure a certain repeatability in terms of rate and angle peeling. The fact that the process was manual will have added to the uncertainty of the method. Therefore, while our POL methodology proved and useful indicator to screen the unoptimized PSA formulation (Table 7), we recommended to rely on the other indicators (distance traveled by the ball and final water content) to discriminate between the various optimized formulations (Table 11).We conclude that the optimization of the low values of peel-off losses was necessary to specify the model. The final quantities of water obtained are similar to these predicted by the DES-OL-ADH experimental design (Table 12), even though this first experimental design showed that the model was less suited to predict final water content.

Fig. 7.

Standardized Pareto charts for distance traveled by the ball (tack test) (a), peel-off losses (b), and final water content (c) of the DES-OL-ADH experimental design. Terms of the model (X-axis, listed inTable 5) are ranked by orthogonal estimate (blue bars, whose intensity is read on the left Y-axis). The cumulative explained variance (orange line) is read on the right Y-axis, the blue dot represents 80% of the variance 5, 8 and 7 model terms are respectively necessary to represents 80% of the variance in the tack test, peel-off losses, and final water content.

Table 11.

Predicted results from DES-OL-ADH vs. Actual results obtained from DES-OL-OPT^a.

Profileb	Sample	Distancec (cm)		Peel-off Losses (mg/cm2)		Final Water Content (wt%)
		Predicted	Actual	Predicted	Actual	Predicted	Actual
1	#1 / #25	30	>30	0.11	0.10	60	60
2	#26 / #27	20	26 and >30	0.01	0.05	56	55
3	#28 / #29	26	>30	0.32	0.10	61	57
4	#30 / #31	20	18.5	0.22	0.40	53	53
5	#32 / #33	14	15	0.01	0.40	56	54
6	#34 / #35	25	>30	0.10	0.04	59	58

^aThe actual results were the average of the repetitions for the same profile.

^bProfiles from the design DES-OL-ADH (JMP Pro 15).

^cDistance traveled by the ball (tack test).

Table 12.

Relative error of the samples obtained from DES-OL-OPT compared to the predicted results from DES-OL-ADH.

Profile	Sample	Relative Error (%)
		Distancea	Peel-off Losses	Final Water Content
1	#1 / #25	—b	9	0
2	#26 / #27	—b	400	2
3	#28 / #29	—b	69	7
4	#30 / #31	8	82	0
5	#32 / #33	7	3900	4
6	#34 / #35	—b	60	2

^aDistance traveled by the ball (tack test).

^bOne or both results are greater than 30 cm which is the method limit.

As displayed in Table 7 for the DES-OL-OPT experimental design, experiments #25 to #29 and #34 and #35 yielded high tack distance and low peel-off losses, which was consistent with our predictions. Those results are similar to commercial sticky notes PSAs (reference). Several of those experiments had, however, a tack distance above 30 cm, which corresponds to the limit of the method (for practical reasons: the rod used to control the thickness of the glue samples had a limit length, which did not allow to extend the track beyond 30 cm). The commercial sticky notes PSAs were not affected by this rod limit because the PSA was already on the paper substrate like a Post-it®. The commercial sticky notes were used, as received. To compare these experiments between them and to commercial PSA, the exact results must be known, so the experiments with a distance $\le$ 30 cm were selected like #30 to #33. The samples #30 and #31 have tack distances close to the commercial one, comparatively to the distance of samples #32 and #33. Even though their peel-off losses are higher than the other tests, they meet those of the commercial PSAs with their standard deviations. The appearance (listed in Table 9) has mostly favored the samples #30 and #31 of the final PSA formulation choice. At 130 °C, the incorporation of lignin into PCE/H₂O matrix is more complete (good lignin dispersion) and the texture of the glue allows a good spreadability on paper substrate (Tables 9 and 10).

The DES-OL-OPT model adequacy was evaluated by the relative standard deviation (RSD) of the repetitions (Table 13). The experiments #30 and #31 of 5 wt% of organosolv lignin, 84 wt% PCE and 11 wt% added water have a low RSD for the three quantified responses. For the peel-off losses test, a heating temperature $\le$ 110 °C appears to increase the RSD (#1 & #25 and #32 & #33 comparatively to #30 & #31). The RSD are higher for the peel-off losses test because the responses are very small values. In addition to their good reproducibility, samples #30 and #31 have low peel-off losses, a considerably high tack distance (high adhesiveness), compared to commercial sticky notes PSAs and a good spreadability on paper substrate. Their preparation conditions (ratio of component mixture, oven time and oven temperature) also promote good incorporation of lignin into the PCE/water mixture. A comparison between the results obtained with the optimized formulation (Experiments #30 and #31) and those of the literature is provided in the related research article. [1] Sivasankarapillai et al. (2019) [4] have investigated different lignin contents in PCE matrix. They reported a bell-like curve for the tack-distance with an initial decrease from 0 to 10 wt% of lignin in PCE, followed by an overall increase from 10 to 30 wt%, which means that adhesiveness reaches an optimum for 10 wt% of lignin. At 40 wt% of lignin (PCE:lignin weight ratio of 3:2), they observed that the miscibility was too poor for the mixture to be tested. In this study, we found 20 wt% to be the upper limit above which a lack of miscibility of the lignin in the mixture is observed. The main difference of the both formulations lies within the final water content: Sivasankarapillai et al. [4] has a PSA moisture level below 3 wt%, while we obtained an optimized formulation with a final water content of ∼ 50 wt%. In other words, this means that the maximum content of 20 wt% in lignin reported in our work corresponds to a PCE:lignin weigth ratio of roughly 2:1, which is closer to each other comparatively to the 3:2 reported in the research work of Sivasankarapillai et al. [4] By increasing the final water content of our PSA to ∼ 50 wt%, we reduce the share of fossil-based PCE in the final adhesive. In our optimization work, we observed that a lignin content of 5 wt% is most suitable for sticky notes application. [1] Please note that shear resistance results for the PCE/lignin/water optimized adhesive are reported in the related research article Gendron et al. (2022). [1] All these results confirm that the optimal formulation turns out to be that of samples #30 and #31 (5 wt% OL, 84 wt% PCE and 11 wt% water added at 130 °C for 60 min). Optimization of the first model made it possible to obtain the best formulation conditions for lignin-based PSA for sticky notes application.

Table 13.

Relative standard deviation for DES-OL-ADH and DES-OL-OPT responses.

Repeat Tests	Relative Standard Deviation (%)
	Distancea (cm)	Peel-off Losses (mg/cm2)	Final Water Content (wt%)
#1 & #25	—b	37.2	1.2
#26 & #27	—b	15.7	0.0c
#28 & #29	—b	14.1	0.0c
#30 & #31	3.8	0.0c	1.3
#32 & #33	9.4	35.4	1.3
#34 & #35	—b	0.0c	0.0c

^aDistance traveled by the ball (tack test).

^bMethod limit (> 30 cm).

^cSame result for the repetition.

Conclusions

This study demonstrated that the d-optimal experimental design is a good tool to optimize formulation parameters for lignin-based sticky notes pressure-sensitive adhesive. The standard least squares were appropriate for the prediction of the PSA characteristics: tack, peel-off losses and final water content. The study showed a good correlation between the experimental data and the predicted values. The optimum formulation parameters were the preparation made with 5 wt% of organosolv lignin, 84 wt% of PCE and 11 wt% of added water and heated at 130 °C for 60 min. Its peel-off losses were similar to those of a commercial sticky notes PSAs. Its tack was higher than the commercial one and can be explained by good interactions created between PCE and lignin. A new peel-off losses method was developed to rapidly obtain an appropriate and representative result.

CRediT authorship contribution statement

Jeanne Gendron: Conceptualization, Methodology, Formal analysis, Investigation, Validation, Writing – original draft, Writing – review & editing. Charles Bruel: Validation, Writing – original draft, Writing – review & editing, Supervision. Yacine Boumghar: Resources, Writing – review & editing, Supervision, Project administration, Funding acquisition. Daniel Montplaisir: Conceptualization, Resources, Writing – review & editing, Supervision, Project administration, Funding acquisition.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

• Data will be made available on request.