Quality Control | Build a More Effective Tablet Press Process
How one company solved a problem with tablets sticking
a company that develops, manufactures, markets, distributes, and sells vitamins, nutritional supplements, and sports nutrition products, needed to solve an intermittent problem: A multimineral tablet was sticking inside the dies of a high-speed rotary tablet press. The company also hoped to improve the tablet’s hardness to better withstand packaging and shipping.
Joe Weider founded the sports nutrition industry in 1940. Schiff Nutrition began as the nutritional products division of Weider Health and Fitness (formerly called Weider Nutrition). The company manufactures its products in a highly controlled, state-of-the-art manufacturing facility constructed in 1997 at its headquarters. Schiff follows a standardized set of good manufacturing practices (GMPs) for nutritional supplements developed by the National Nutritional Foods Association (NNFA), the largest dietary supplement trade association in the United States.
Schiff’s multimineral tablet is made using the following process. First, a mixture of various powders is weighed and blended and discharged into bulk sacks. Next, these sacks are moved to another processing area, where the blended material is loaded into hoppers that feed the tablet presses.
A graphic showing the response surface for critical hardness response.The production staff varied press parameters without success, then developed a statistically designed experiment using a mixture design that allowed specific constraints on each of the constituents. The experimental results indicated that the standard formulation provided close to optimal properties for the multimineral tablet. "It was clear that the formulation itself was not causing the problem," said Jarom Webster, research and development scientist for Schiff Nutrition. "So we searched for other factors that could be playing a role and discovered that when we moved the operation from a single-sided to a double-sided press the problem went away."
The primary active ingredients of the tablet are magnesium and chelated zinc. It requires three different binders that make up approximately half of its composition—49.3% by weight. Additional ingredients include a lubricant, a glidant, and a disintegrant. Webster experimented with four components:
• binder 1 (from 0 to 49.3 weight percent of total tablet weight);
• binder 2 (0 to 49.3%); and
• binder 3 (0 to 49.3%).
The responses for the experiment included the tablet hardness at three different levels of compression in the tablet press as measured on a Dr. Schleuniger 6D hardness tester. Webster also evaluated the time required for the tablets to disintegrate in a 37°C water bath. He tested punch tightness, which measures the force exerted by the lower punch to eject the tablet from the die; this test was important because, under certain circumstances, powder can stick to the die wall and potentially jam the press.
It was clear that the formulation itself was not causing the problem. So we searched for other factors that could be playing a role and discovered that when we moved the operation from a single-sided to a double-sided press, the problem went away.
Jarom Webster, Schiff NutritionHere is the list of responses:
• tablet hardness at 15 kiloNewtons (kN) compression force, measured in kilopons;
• hardness at 30 kN;
• hardness at 45 kN;
• disintegration time at 45 kN;
• punch tightness (Newtons); and
• tablet weight relative standard deviation (RSD, %).
After identifying the factors and responses, Webster used Design-Expert software from Stat-Ease Inc. (Minneapolis) to design the experiment. The program laid out 20 formulations, including a number of replicates for estimating experimental error while keeping the three ingredients within their specified total. The test matrix evaluated all factors simultaneously and probed higher order and multiple component interaction effects.
Webster entered the experimental results into Design-Expert, which analyzed them, created statistically validated predictive models to describe each of the responses, and provided graphs to visualize the outputs. For example, Figure 1 (see p. 46) shows the response surface for the tablet hardness at 45 kN. Webster specified what he desired and the relative weight to be placed on each of the tableting attributes. The software then identified the composition that would maximize overall desirability and predicted the results. The flag in Figure 2 (see p. 46) pinpoints the ideal composition.
This color-coded graph displays relative desirability of alternative formulations."Based on this experiment, the optimized mixture turned out to be very close to what we were already running," Webster said. "The results also showed that there was a considerable amount of variability that was not accounted for by the experiment. So we began looking outside the box for other factors that we did not consider in our initial experimental design."
Webster asked production to run an informal experiment by tracking whether the tablets were pressed on single- or double-sided presses and recording the results. The problem went away when double-sided presses were used. "While design of experiments did not formally solve this problem, it led to the solution by conclusively determining that factors outside the ones we had been considering were the cause of problem," Webster said. n