Using Spectroscopic and Virtual Sensor PAT in Continuous Solid-Dosage Manufacturing

April 1, 2017
Jennifer Markarian

Jennifer Markarian is manufacturing editor of Pharmaceutical Technology.

Pharmaceutical Technology, Pharmaceutical Technology-04-01-2017, Volume 2017 Supplement, Issue 2
Page Number: s26–s29

Soft sensors are powerful tools that can be used along with spectroscopic instruments in on-line measurement.

Process analytical technology (PAT) is helping process engineers to better understand and control their processes by measuring the critical process parameters (CPPs) that affect critical quality attributes (CQAs) of the product. PAT is particularly important for implementing continuous solid-dosage manufacturing, in which material must be measured and tracked as it continuously flows through connected equipment. Spectroscopic PAT is useful, but has some limitations (1). Industry experts say that PAT can be more than spectroscopy; PAT can also include “soft” (i.e., software or virtual) sensors that correlate easily measured process conditions (e.g., temperature, feedrate) to CPPs using first-principle models. Pharmaceutical Technology spoke with Dr. Jim Holman, process development manager, APC Pharma Solids, GEA UK, about the role of PAT in solid-dosage manufacturing and in the GEA ConsiGma continuous manufacturing platform, for which the company won the “Best manufacturing technology and equipment” award at the 2016 CPhI Pharma Awards (2). The ConsiGma platform can produce blends, granules, or coated tablets continuously using either wet granulation or direct compression. 

PAT in continuous solid-dosage manufacturing

PharmTech: What role does PAT play in continuous solid-dosage manufacturing?

Holman (GEA): PAT plays a crucial part in continuous dose manufacture and is already successfully utilized on a number of GEA continuous solid oral dose manufacturing platforms. The mistake that is often made is to automatically assume that PAT refers only to spectroscopy or other added instruments, such as laser diffraction. In theory, a displacement sensor could be called PAT if after data analysis, it allows one to analyze and control the process, and therefore product quality, through the (indirect) measurement of a CQA.

However it’s defined, the real benefit of any PAT system is that it can enable real-time monitoring of the drug product quality attributes that would normally be measured by other methods, usually off-line laboratory testing.  PAT often also ensures that a far higher percentage of the material is measured, thereby increasing the ‘data points per kg’ in comparison to traditional off-line or at-line methods.

Blend uniformity prior to compression, particle size via laser diffraction, moisture content post drying, tablet assay, and tablet coating film thickness are all examples of measurements that have been successfully enabled on GEA continuous lines in the past. The use of PAT enables the rapid development of the process design space without the need for extensive off-line testing, and it allows the development cycle to fully utilize structured scientific methods to assess products in a much shorter timeframe. 

For example, on a GEA ConSigma granulation, drying, blending, and compression line, 105 separate design of experiment points have been conducted in one eight-hour working day. By comparison, the feedback from the project industrial partners at the time was that 3–5 points would be managed, at best, using traditional manufacturing methods. PAT enabled the real-time measurement of granule loss on drying and blend particle size distribution (PSD), while at-line PAT enabled the measurement of tablet assay and other key properties. This testing would have taken weeks to yield the same level of data using off-line systems. 

Spectroscopy, however, is not a golden bullet that will solve this matter for all cases. Not all products can be tracked using near infrared (NIR) spectroscopy.  Secondly, if there is significant change to raw materials/API lots, then models built and verified for one supply may no longer be valid for another. Changes to the process can also cause the model to be invalidated, for example ramping the throughput of a line from 10 kg/hr to 50 kg/hr.  NIR models will be required to cover specific API concentrations, so for multi-dose strength formulations, a number of PAT models will need to be built and maintained.

The key to PAT is consistent sample presentation, and while continuous processing makes it easier to control this, the shifts naturally seen in development programs as the design space is explored can cause issues with some PAT applications.

Online spectroscopy

PharmTech: What are some examples of what is measured by online spectroscopy for continuous tablet production?

Holman (GEA): Currently spectroscopy based-PAT is employed in a number of locations on GEA continuous platforms:

Moisture content (NIR):  NIR spectroscopy focusing on the specific water wavelengths has been integrated into the GEA continuous tableting line for some time now, and is routinely used to give a real-time prediction of the loss on drying of exiting granules. These data can be fed back into the control system of the fluid bed dryer, altering the release criteria and actively controlling the drying process to ensure the correct product quality of the dried granules.

Particle Size Distribution (Laser diffraction): Instruments aimed at measuring the PSD of the milled granules has been proven on ConSigma granulation lines.

Blend uniformity (NIR):  By locating a NIR system on the powder inlet to the tablet press, the blend uniformity of a system can be measured in real time just prior to tablet compression. This system is employed on both the continuous direct compression system and the continuous wet granulation line. The measurement of blend uniformity is used for two key purposes:

  • For APIs where an accurate and precise NIR model can be validated, the system can be used to track the actual blend concentration in real time in place of off-line high-performance liquid chromatography (HPLC) testing. 

  • If models can only be used to give indications of the API percentage, NIR can still be a powerful tool in assessing the process design and operating space. Measuring the residence time distribution (RTD) of systems enables the direct modeling of systems using fundamental understand of the process. GEA continuous blending and compression systems have been accurately modelled using RTDs to the point that the blend potency can be predicted using the measured input feed variability and the system RTD, which can be fully aligned with measured blend potency from the online NIR system.

Coating thickness (Raman): On the GEA continuous coater, the thickness of coating layers can be measured as the layer is applied in real time. The measurement is performed by monitoring the drop in the API signal as the tablet is coated and the inherent rise in the coating material signal.

Soft sensors

PharmTech: How can ‘soft’ sensors be used as PAT tools and what information can they provide?

Holman (GEA): Some of the most powerful tools often overlooked as PAT on any continuous solids line start in the basic process data taken from current unit operations, specifically loss-in-weight feeders and the tablet compression systems. The development of virtual (soft) sensors that use robust direct measurements (e.g., temperature, pressure, displacement, force) and turn these data into product quality information is of significant interest at this time and has already been shown to give insight that can potentially replace the function of some spectroscopy-based PAT. 

Data generated from the unique, air-compensated compression stations in a GEA tablet press are shown to track RTD disturbances through the system (to the actual point of compression as opposed to close to point of compression as with NIR). In one case study, a valid NIR model could not be built, so this alternate PAT method was used to quantify the mixing capability of the entire GEA CDC-50 direct compression system. The design space was mapped using the GEA RTD model of continuous blending built on the data from the tablet press only. This method eliminated the need for NIR and allowed for the development to continue without the need for time-intensive HPLC testing.

The advantage of such an analysis is that unlike NIR, the measurement signals from displacement sensors do not need complex multivariate models that required developing, validating, and maintaining. The methods are robust to potential variation in incoming raw materials and scalable over throughput ranges. 

Previous studies have shown that by tracking changes in the filling of individual dies in the tablet press, changes in granule properties out of the granulation and drying system can be directly tracked by comparing the apparent powder bulk density in the press. This method has been shown to have a greater accuracy than continuous PSD measurements via laser diffraction. 

Changes in powder characteristics (e.g., lot-to-lot density or flowability changes) can be directly measured going into a continuous process using the data from the GEA compact feeder, potentially opening the possibility of feed-forward quality control where systems are prepared for product changes.  

PharmTech: What do you foresee as trends in the use of PAT in continuous solid-dosage manufacturing as this type of process begins to be more widely used?

Holman (GEA): The benefits of spectroscopic PAT instruments are well known, as are the limitations. The technology is shown to yield significant benefits into the development system of a product, enabling faster analysis of products. The exponential growth area, however, will be in other virtual sensors that can be considered PAT. Continuous systems generate a significant amount of data, and at this time only basic data are routinely used. Using soft sensors to analysis these data, yielding more insight into the actual product quality, will generate significant interest in the coming years.

We are already seeing a trend of using spectroscopy-based PAT during development to verify fundamental models such as RTD models, but this spectroscopic PAT is excluded for commercial production platforms in favor of soft sensors and mathematical models that are more robust.


  1. J. Markarian, “Overcoming Obstacles in Process Analytical Technology,” Equipment & Processing Report, Feb. 15, 2017.
  2. PharmTech, “Manufacturing Technology and Equipment Prize Goes to GEA Group,” (Oct. 5, 2016),

Article Details

Pharmaceutical Technology
Supplement: Solid Dosage Drug Development and Manufacturing
Vol. 41
April 2017
Pages: s26–s29


When referring to this article, please cite it as J. Markarian, "Solid Dosage Drug Development/Manufacturing 2017," Pharmaceutical Technology Solid Dosage Drug Development and Manufacturing Supplement (April 2017).


download issueDownload Issue : Pharmaceutical Technology-04-01-2017